diff options
Diffstat (limited to 'openssl/crypto/bn')
56 files changed, 26466 insertions, 16213 deletions
diff --git a/openssl/crypto/bn/Makefile b/openssl/crypto/bn/Makefile index 6dd136be5..0cdbd2016 100644 --- a/openssl/crypto/bn/Makefile +++ b/openssl/crypto/bn/Makefile @@ -77,6 +77,12 @@ sparcv9a-mont.s: asm/sparcv9a-mont.pl $(PERL) asm/sparcv9a-mont.pl $(CFLAGS) > $@ sparcv9-mont.s: asm/sparcv9-mont.pl $(PERL) asm/sparcv9-mont.pl $(CFLAGS) > $@ +vis3-mont.s: asm/vis3-mont.pl + $(PERL) asm/vis3-mont.pl $(CFLAGS) > $@ +sparct4-mont.S: asm/sparct4-mont.pl + $(PERL) asm/sparct4-mont.pl $(CFLAGS) > $@ +sparcv9-gf2m.S: asm/sparcv9-gf2m.pl + $(PERL) asm/sparcv9-gf2m.pl $(CFLAGS) > $@ bn-mips3.o: asm/mips3.s @if [ "$(CC)" = "gcc" ]; then \ @@ -104,6 +110,10 @@ x86_64-gf2m.s: asm/x86_64-gf2m.pl $(PERL) asm/x86_64-gf2m.pl $(PERLASM_SCHEME) > $@ modexp512-x86_64.s: asm/modexp512-x86_64.pl $(PERL) asm/modexp512-x86_64.pl $(PERLASM_SCHEME) > $@ +rsaz-x86_64.s: asm/rsaz-x86_64.pl + $(PERL) asm/rsaz-x86_64.pl $(PERLASM_SCHEME) > $@ +rsaz-avx2.s: asm/rsaz-avx2.pl + $(PERL) asm/rsaz-avx2.pl $(PERLASM_SCHEME) > $@ bn-ia64.s: asm/ia64.S $(CC) $(CFLAGS) -E asm/ia64.S > $@ @@ -125,14 +135,15 @@ ppc-mont.s: asm/ppc-mont.pl;$(PERL) asm/ppc-mont.pl $(PERLASM_SCHEME) $@ ppc64-mont.s: asm/ppc64-mont.pl;$(PERL) asm/ppc64-mont.pl $(PERLASM_SCHEME) $@ alpha-mont.s: asm/alpha-mont.pl - (preproc=/tmp/$$$$.$@; trap "rm $$preproc" INT; \ + (preproc=$$$$.$@.S; trap "rm $$preproc" INT; \ $(PERL) asm/alpha-mont.pl > $$preproc && \ - $(CC) -E $$preproc > $@ && rm $$preproc) + $(CC) -E -P $$preproc > $@ && rm $$preproc) # GNU make "catch all" -%-mont.s: asm/%-mont.pl; $(PERL) $< $(PERLASM_SCHEME) $@ +%-mont.S: asm/%-mont.pl; $(PERL) $< $(PERLASM_SCHEME) $@ %-gf2m.S: asm/%-gf2m.pl; $(PERL) $< $(PERLASM_SCHEME) $@ +armv4-mont.o: armv4-mont.S armv4-gf2m.o: armv4-gf2m.S files: diff --git a/openssl/crypto/bn/asm/armv4-gf2m.pl b/openssl/crypto/bn/asm/armv4-gf2m.pl index c52e0b75b..8f529c95c 100644 --- a/openssl/crypto/bn/asm/armv4-gf2m.pl +++ b/openssl/crypto/bn/asm/armv4-gf2m.pl @@ -20,48 +20,26 @@ # length, more for longer keys. Even though NEON 1x1 multiplication # runs in even less cycles, ~30, improvement is measurable only on # longer keys. One has to optimize code elsewhere to get NEON glow... +# +# April 2014 +# +# Double bn_GF2m_mul_2x2 performance by using algorithm from paper +# referred below, which improves ECDH and ECDSA verify benchmarks +# by 18-40%. +# +# Câmara, D.; Gouvêa, C. P. L.; López, J. & Dahab, R.: Fast Software +# Polynomial Multiplication on ARM Processors using the NEON Engine. +# +# http://conradoplg.cryptoland.net/files/2010/12/mocrysen13.pdf while (($output=shift) && ($output!~/^\w[\w\-]*\.\w+$/)) {} open STDOUT,">$output"; -sub Dlo() { shift=~m|q([1]?[0-9])|?"d".($1*2):""; } -sub Dhi() { shift=~m|q([1]?[0-9])|?"d".($1*2+1):""; } -sub Q() { shift=~m|d([1-3]?[02468])|?"q".($1/2):""; } - $code=<<___; #include "arm_arch.h" .text .code 32 - -#if __ARM_ARCH__>=7 -.fpu neon - -.type mul_1x1_neon,%function -.align 5 -mul_1x1_neon: - vshl.u64 `&Dlo("q1")`,d16,#8 @ q1-q3 are slided $a - vmull.p8 `&Q("d0")`,d16,d17 @ a·bb - vshl.u64 `&Dlo("q2")`,d16,#16 - vmull.p8 q1,`&Dlo("q1")`,d17 @ a<<8·bb - vshl.u64 `&Dlo("q3")`,d16,#24 - vmull.p8 q2,`&Dlo("q2")`,d17 @ a<<16·bb - vshr.u64 `&Dlo("q1")`,#8 - vmull.p8 q3,`&Dlo("q3")`,d17 @ a<<24·bb - vshl.u64 `&Dhi("q1")`,#24 - veor d0,`&Dlo("q1")` - vshr.u64 `&Dlo("q2")`,#16 - veor d0,`&Dhi("q1")` - vshl.u64 `&Dhi("q2")`,#16 - veor d0,`&Dlo("q2")` - vshr.u64 `&Dlo("q3")`,#24 - veor d0,`&Dhi("q2")` - vshl.u64 `&Dhi("q3")`,#8 - veor d0,`&Dlo("q3")` - veor d0,`&Dhi("q3")` - bx lr -.size mul_1x1_neon,.-mul_1x1_neon -#endif ___ ################ # private interface to mul_1x1_ialu @@ -159,56 +137,17 @@ ___ # void bn_GF2m_mul_2x2(BN_ULONG *r, # BN_ULONG a1,BN_ULONG a0, # BN_ULONG b1,BN_ULONG b0); # r[3..0]=a1a0·b1b0 - -($A1,$B1,$A0,$B0,$A1B1,$A0B0)=map("d$_",(18..23)); - +{ $code.=<<___; .global bn_GF2m_mul_2x2 .type bn_GF2m_mul_2x2,%function .align 5 bn_GF2m_mul_2x2: -#if __ARM_ARCH__>=7 +#if __ARM_MAX_ARCH__>=7 ldr r12,.LOPENSSL_armcap .Lpic: ldr r12,[pc,r12] tst r12,#1 - beq .Lialu - - veor $A1,$A1 - vmov.32 $B1,r3,r3 @ two copies of b1 - vmov.32 ${A1}[0],r1 @ a1 - - veor $A0,$A0 - vld1.32 ${B0}[],[sp,:32] @ two copies of b0 - vmov.32 ${A0}[0],r2 @ a0 - mov r12,lr - - vmov d16,$A1 - vmov d17,$B1 - bl mul_1x1_neon @ a1·b1 - vmov $A1B1,d0 - - vmov d16,$A0 - vmov d17,$B0 - bl mul_1x1_neon @ a0·b0 - vmov $A0B0,d0 - - veor d16,$A0,$A1 - veor d17,$B0,$B1 - veor $A0,$A0B0,$A1B1 - bl mul_1x1_neon @ (a0+a1)·(b0+b1) - - veor d0,$A0 @ (a0+a1)·(b0+b1)-a0·b0-a1·b1 - vshl.u64 d1,d0,#32 - vshr.u64 d0,d0,#32 - veor $A0B0,d1 - veor $A1B1,d0 - vst1.32 {${A0B0}[0]},[r0,:32]! - vst1.32 {${A0B0}[1]},[r0,:32]! - vst1.32 {${A1B1}[0]},[r0,:32]! - vst1.32 {${A1B1}[1]},[r0,:32] - bx r12 -.align 4 -.Lialu: + bne .LNEON #endif ___ $ret="r10"; # reassigned 1st argument @@ -260,8 +199,72 @@ $code.=<<___; moveq pc,lr @ be binary compatible with V4, yet bx lr @ interoperable with Thumb ISA:-) #endif +___ +} +{ +my ($r,$t0,$t1,$t2,$t3)=map("q$_",(0..3,8..12)); +my ($a,$b,$k48,$k32,$k16)=map("d$_",(26..31)); + +$code.=<<___; +#if __ARM_MAX_ARCH__>=7 +.arch armv7-a +.fpu neon + +.align 5 +.LNEON: + ldr r12, [sp] @ 5th argument + vmov.32 $a, r2, r1 + vmov.32 $b, r12, r3 + vmov.i64 $k48, #0x0000ffffffffffff + vmov.i64 $k32, #0x00000000ffffffff + vmov.i64 $k16, #0x000000000000ffff + + vext.8 $t0#lo, $a, $a, #1 @ A1 + vmull.p8 $t0, $t0#lo, $b @ F = A1*B + vext.8 $r#lo, $b, $b, #1 @ B1 + vmull.p8 $r, $a, $r#lo @ E = A*B1 + vext.8 $t1#lo, $a, $a, #2 @ A2 + vmull.p8 $t1, $t1#lo, $b @ H = A2*B + vext.8 $t3#lo, $b, $b, #2 @ B2 + vmull.p8 $t3, $a, $t3#lo @ G = A*B2 + vext.8 $t2#lo, $a, $a, #3 @ A3 + veor $t0, $t0, $r @ L = E + F + vmull.p8 $t2, $t2#lo, $b @ J = A3*B + vext.8 $r#lo, $b, $b, #3 @ B3 + veor $t1, $t1, $t3 @ M = G + H + vmull.p8 $r, $a, $r#lo @ I = A*B3 + veor $t0#lo, $t0#lo, $t0#hi @ t0 = (L) (P0 + P1) << 8 + vand $t0#hi, $t0#hi, $k48 + vext.8 $t3#lo, $b, $b, #4 @ B4 + veor $t1#lo, $t1#lo, $t1#hi @ t1 = (M) (P2 + P3) << 16 + vand $t1#hi, $t1#hi, $k32 + vmull.p8 $t3, $a, $t3#lo @ K = A*B4 + veor $t2, $t2, $r @ N = I + J + veor $t0#lo, $t0#lo, $t0#hi + veor $t1#lo, $t1#lo, $t1#hi + veor $t2#lo, $t2#lo, $t2#hi @ t2 = (N) (P4 + P5) << 24 + vand $t2#hi, $t2#hi, $k16 + vext.8 $t0, $t0, $t0, #15 + veor $t3#lo, $t3#lo, $t3#hi @ t3 = (K) (P6 + P7) << 32 + vmov.i64 $t3#hi, #0 + vext.8 $t1, $t1, $t1, #14 + veor $t2#lo, $t2#lo, $t2#hi + vmull.p8 $r, $a, $b @ D = A*B + vext.8 $t3, $t3, $t3, #12 + vext.8 $t2, $t2, $t2, #13 + veor $t0, $t0, $t1 + veor $t2, $t2, $t3 + veor $r, $r, $t0 + veor $r, $r, $t2 + + vst1.32 {$r}, [r0] + ret @ bx lr +#endif +___ +} +$code.=<<___; .size bn_GF2m_mul_2x2,.-bn_GF2m_mul_2x2 -#if __ARM_ARCH__>=7 +#if __ARM_MAX_ARCH__>=7 .align 5 .LOPENSSL_armcap: .word OPENSSL_armcap_P-(.Lpic+8) @@ -269,10 +272,18 @@ $code.=<<___; .asciz "GF(2^m) Multiplication for ARMv4/NEON, CRYPTOGAMS by <appro\@openssl.org>" .align 5 +#if __ARM_MAX_ARCH__>=7 .comm OPENSSL_armcap_P,4,4 +#endif ___ -$code =~ s/\`([^\`]*)\`/eval $1/gem; -$code =~ s/\bbx\s+lr\b/.word\t0xe12fff1e/gm; # make it possible to compile with -march=armv4 -print $code; +foreach (split("\n",$code)) { + s/\`([^\`]*)\`/eval $1/geo; + + s/\bq([0-9]+)#(lo|hi)/sprintf "d%d",2*$1+($2 eq "hi")/geo or + s/\bret\b/bx lr/go or + s/\bbx\s+lr\b/.word\t0xe12fff1e/go; # make it possible to compile with -march=armv4 + + print $_,"\n"; +} close STDOUT; # enforce flush diff --git a/openssl/crypto/bn/asm/armv4-mont.pl b/openssl/crypto/bn/asm/armv4-mont.pl index f78a8b5f0..1d330e9f8 100644 --- a/openssl/crypto/bn/asm/armv4-mont.pl +++ b/openssl/crypto/bn/asm/armv4-mont.pl @@ -1,7 +1,7 @@ #!/usr/bin/env perl # ==================================================================== -# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL # project. The module is, however, dual licensed under OpenSSL and # CRYPTOGAMS licenses depending on where you obtain it. For further # details see http://www.openssl.org/~appro/cryptogams/. @@ -23,6 +23,21 @@ # than 1/2KB. Windows CE port would be trivial, as it's exclusively # about decorations, ABI and instruction syntax are identical. +# November 2013 +# +# Add NEON code path, which handles lengths divisible by 8. RSA/DSA +# performance improvement on Cortex-A8 is ~45-100% depending on key +# length, more for longer keys. On Cortex-A15 the span is ~10-105%. +# On Snapdragon S4 improvement was measured to vary from ~70% to +# incredible ~380%, yes, 4.8x faster, for RSA4096 sign. But this is +# rather because original integer-only code seems to perform +# suboptimally on S4. Situation on Cortex-A9 is unfortunately +# different. It's being looked into, but the trouble is that +# performance for vectors longer than 256 bits is actually couple +# of percent worse than for integer-only code. The code is chosen +# for execution on all NEON-capable processors, because gain on +# others outweighs the marginal loss on Cortex-A9. + while (($output=shift) && ($output!~/^\w[\w\-]*\.\w+$/)) {} open STDOUT,">$output"; @@ -52,16 +67,40 @@ $_n0="$num,#14*4"; $_num="$num,#15*4"; $_bpend=$_num; $code=<<___; +#include "arm_arch.h" + .text +.code 32 + +#if __ARM_MAX_ARCH__>=7 +.align 5 +.LOPENSSL_armcap: +.word OPENSSL_armcap_P-bn_mul_mont +#endif .global bn_mul_mont .type bn_mul_mont,%function -.align 2 +.align 5 bn_mul_mont: + ldr ip,[sp,#4] @ load num stmdb sp!,{r0,r2} @ sp points at argument block - ldr $num,[sp,#3*4] @ load num - cmp $num,#2 +#if __ARM_MAX_ARCH__>=7 + tst ip,#7 + bne .Lialu + adr r0,bn_mul_mont + ldr r2,.LOPENSSL_armcap + ldr r0,[r0,r2] + tst r0,#1 @ NEON available? + ldmia sp, {r0,r2} + beq .Lialu + add sp,sp,#8 + b bn_mul8x_mont_neon +.align 4 +.Lialu: +#endif + cmp ip,#2 + mov $num,ip @ load num movlt r0,#0 addlt sp,sp,#2*4 blt .Labrt @@ -191,14 +230,447 @@ bn_mul_mont: ldmia sp!,{r4-r12,lr} @ restore registers add sp,sp,#2*4 @ skip over {r0,r2} mov r0,#1 -.Labrt: tst lr,#1 +.Labrt: +#if __ARM_ARCH__>=5 + ret @ bx lr +#else + tst lr,#1 moveq pc,lr @ be binary compatible with V4, yet bx lr @ interoperable with Thumb ISA:-) +#endif .size bn_mul_mont,.-bn_mul_mont -.asciz "Montgomery multiplication for ARMv4, CRYPTOGAMS by <appro\@openssl.org>" +___ +{ +sub Dlo() { shift=~m|q([1]?[0-9])|?"d".($1*2):""; } +sub Dhi() { shift=~m|q([1]?[0-9])|?"d".($1*2+1):""; } + +my ($A0,$A1,$A2,$A3)=map("d$_",(0..3)); +my ($N0,$N1,$N2,$N3)=map("d$_",(4..7)); +my ($Z,$Temp)=("q4","q5"); +my ($A0xB,$A1xB,$A2xB,$A3xB,$A4xB,$A5xB,$A6xB,$A7xB)=map("q$_",(6..13)); +my ($Bi,$Ni,$M0)=map("d$_",(28..31)); +my $zero=&Dlo($Z); +my $temp=&Dlo($Temp); + +my ($rptr,$aptr,$bptr,$nptr,$n0,$num)=map("r$_",(0..5)); +my ($tinptr,$toutptr,$inner,$outer)=map("r$_",(6..9)); + +$code.=<<___; +#if __ARM_MAX_ARCH__>=7 +.arch armv7-a +.fpu neon + +.type bn_mul8x_mont_neon,%function +.align 5 +bn_mul8x_mont_neon: + mov ip,sp + stmdb sp!,{r4-r11} + vstmdb sp!,{d8-d15} @ ABI specification says so + ldmia ip,{r4-r5} @ load rest of parameter block + + sub $toutptr,sp,#16 + vld1.32 {${Bi}[0]}, [$bptr,:32]! + sub $toutptr,$toutptr,$num,lsl#4 + vld1.32 {$A0-$A3}, [$aptr]! @ can't specify :32 :-( + and $toutptr,$toutptr,#-64 + vld1.32 {${M0}[0]}, [$n0,:32] + mov sp,$toutptr @ alloca + veor $zero,$zero,$zero + subs $inner,$num,#8 + vzip.16 $Bi,$zero + + vmull.u32 $A0xB,$Bi,${A0}[0] + vmull.u32 $A1xB,$Bi,${A0}[1] + vmull.u32 $A2xB,$Bi,${A1}[0] + vshl.i64 $temp,`&Dhi("$A0xB")`,#16 + vmull.u32 $A3xB,$Bi,${A1}[1] + + vadd.u64 $temp,$temp,`&Dlo("$A0xB")` + veor $zero,$zero,$zero + vmul.u32 $Ni,$temp,$M0 + + vmull.u32 $A4xB,$Bi,${A2}[0] + vld1.32 {$N0-$N3}, [$nptr]! + vmull.u32 $A5xB,$Bi,${A2}[1] + vmull.u32 $A6xB,$Bi,${A3}[0] + vzip.16 $Ni,$zero + vmull.u32 $A7xB,$Bi,${A3}[1] + + bne .LNEON_1st + + @ special case for num=8, everything is in register bank... + + vmlal.u32 $A0xB,$Ni,${N0}[0] + sub $outer,$num,#1 + vmlal.u32 $A1xB,$Ni,${N0}[1] + vmlal.u32 $A2xB,$Ni,${N1}[0] + vmlal.u32 $A3xB,$Ni,${N1}[1] + + vmlal.u32 $A4xB,$Ni,${N2}[0] + vmov $Temp,$A0xB + vmlal.u32 $A5xB,$Ni,${N2}[1] + vmov $A0xB,$A1xB + vmlal.u32 $A6xB,$Ni,${N3}[0] + vmov $A1xB,$A2xB + vmlal.u32 $A7xB,$Ni,${N3}[1] + vmov $A2xB,$A3xB + vmov $A3xB,$A4xB + vshr.u64 $temp,$temp,#16 + vmov $A4xB,$A5xB + vmov $A5xB,$A6xB + vadd.u64 $temp,$temp,`&Dhi("$Temp")` + vmov $A6xB,$A7xB + veor $A7xB,$A7xB + vshr.u64 $temp,$temp,#16 + + b .LNEON_outer8 + +.align 4 +.LNEON_outer8: + vld1.32 {${Bi}[0]}, [$bptr,:32]! + veor $zero,$zero,$zero + vzip.16 $Bi,$zero + vadd.u64 `&Dlo("$A0xB")`,`&Dlo("$A0xB")`,$temp + + vmlal.u32 $A0xB,$Bi,${A0}[0] + vmlal.u32 $A1xB,$Bi,${A0}[1] + vmlal.u32 $A2xB,$Bi,${A1}[0] + vshl.i64 $temp,`&Dhi("$A0xB")`,#16 + vmlal.u32 $A3xB,$Bi,${A1}[1] + + vadd.u64 $temp,$temp,`&Dlo("$A0xB")` + veor $zero,$zero,$zero + subs $outer,$outer,#1 + vmul.u32 $Ni,$temp,$M0 + + vmlal.u32 $A4xB,$Bi,${A2}[0] + vmlal.u32 $A5xB,$Bi,${A2}[1] + vmlal.u32 $A6xB,$Bi,${A3}[0] + vzip.16 $Ni,$zero + vmlal.u32 $A7xB,$Bi,${A3}[1] + + vmlal.u32 $A0xB,$Ni,${N0}[0] + vmlal.u32 $A1xB,$Ni,${N0}[1] + vmlal.u32 $A2xB,$Ni,${N1}[0] + vmlal.u32 $A3xB,$Ni,${N1}[1] + + vmlal.u32 $A4xB,$Ni,${N2}[0] + vmov $Temp,$A0xB + vmlal.u32 $A5xB,$Ni,${N2}[1] + vmov $A0xB,$A1xB + vmlal.u32 $A6xB,$Ni,${N3}[0] + vmov $A1xB,$A2xB + vmlal.u32 $A7xB,$Ni,${N3}[1] + vmov $A2xB,$A3xB + vmov $A3xB,$A4xB + vshr.u64 $temp,$temp,#16 + vmov $A4xB,$A5xB + vmov $A5xB,$A6xB + vadd.u64 $temp,$temp,`&Dhi("$Temp")` + vmov $A6xB,$A7xB + veor $A7xB,$A7xB + vshr.u64 $temp,$temp,#16 + + bne .LNEON_outer8 + + vadd.u64 `&Dlo("$A0xB")`,`&Dlo("$A0xB")`,$temp + mov $toutptr,sp + vshr.u64 $temp,`&Dlo("$A0xB")`,#16 + mov $inner,$num + vadd.u64 `&Dhi("$A0xB")`,`&Dhi("$A0xB")`,$temp + add $tinptr,sp,#16 + vshr.u64 $temp,`&Dhi("$A0xB")`,#16 + vzip.16 `&Dlo("$A0xB")`,`&Dhi("$A0xB")` + + b .LNEON_tail2 + +.align 4 +.LNEON_1st: + vmlal.u32 $A0xB,$Ni,${N0}[0] + vld1.32 {$A0-$A3}, [$aptr]! + vmlal.u32 $A1xB,$Ni,${N0}[1] + subs $inner,$inner,#8 + vmlal.u32 $A2xB,$Ni,${N1}[0] + vmlal.u32 $A3xB,$Ni,${N1}[1] + + vmlal.u32 $A4xB,$Ni,${N2}[0] + vld1.32 {$N0-$N1}, [$nptr]! + vmlal.u32 $A5xB,$Ni,${N2}[1] + vst1.64 {$A0xB-$A1xB}, [$toutptr,:256]! + vmlal.u32 $A6xB,$Ni,${N3}[0] + vmlal.u32 $A7xB,$Ni,${N3}[1] + vst1.64 {$A2xB-$A3xB}, [$toutptr,:256]! + + vmull.u32 $A0xB,$Bi,${A0}[0] + vld1.32 {$N2-$N3}, [$nptr]! + vmull.u32 $A1xB,$Bi,${A0}[1] + vst1.64 {$A4xB-$A5xB}, [$toutptr,:256]! + vmull.u32 $A2xB,$Bi,${A1}[0] + vmull.u32 $A3xB,$Bi,${A1}[1] + vst1.64 {$A6xB-$A7xB}, [$toutptr,:256]! + + vmull.u32 $A4xB,$Bi,${A2}[0] + vmull.u32 $A5xB,$Bi,${A2}[1] + vmull.u32 $A6xB,$Bi,${A3}[0] + vmull.u32 $A7xB,$Bi,${A3}[1] + + bne .LNEON_1st + + vmlal.u32 $A0xB,$Ni,${N0}[0] + add $tinptr,sp,#16 + vmlal.u32 $A1xB,$Ni,${N0}[1] + sub $aptr,$aptr,$num,lsl#2 @ rewind $aptr + vmlal.u32 $A2xB,$Ni,${N1}[0] + vld1.64 {$Temp}, [sp,:128] + vmlal.u32 $A3xB,$Ni,${N1}[1] + sub $outer,$num,#1 + + vmlal.u32 $A4xB,$Ni,${N2}[0] + vst1.64 {$A0xB-$A1xB}, [$toutptr,:256]! + vmlal.u32 $A5xB,$Ni,${N2}[1] + vshr.u64 $temp,$temp,#16 + vld1.64 {$A0xB}, [$tinptr, :128]! + vmlal.u32 $A6xB,$Ni,${N3}[0] + vst1.64 {$A2xB-$A3xB}, [$toutptr,:256]! + vmlal.u32 $A7xB,$Ni,${N3}[1] + + vst1.64 {$A4xB-$A5xB}, [$toutptr,:256]! + vadd.u64 $temp,$temp,`&Dhi("$Temp")` + veor $Z,$Z,$Z + vst1.64 {$A6xB-$A7xB}, [$toutptr,:256]! + vld1.64 {$A1xB-$A2xB}, [$tinptr, :256]! + vst1.64 {$Z}, [$toutptr,:128] + vshr.u64 $temp,$temp,#16 + + b .LNEON_outer + +.align 4 +.LNEON_outer: + vld1.32 {${Bi}[0]}, [$bptr,:32]! + sub $nptr,$nptr,$num,lsl#2 @ rewind $nptr + vld1.32 {$A0-$A3}, [$aptr]! + veor $zero,$zero,$zero + mov $toutptr,sp + vzip.16 $Bi,$zero + sub $inner,$num,#8 + vadd.u64 `&Dlo("$A0xB")`,`&Dlo("$A0xB")`,$temp + + vmlal.u32 $A0xB,$Bi,${A0}[0] + vld1.64 {$A3xB-$A4xB},[$tinptr,:256]! + vmlal.u32 $A1xB,$Bi,${A0}[1] + vmlal.u32 $A2xB,$Bi,${A1}[0] + vld1.64 {$A5xB-$A6xB},[$tinptr,:256]! + vmlal.u32 $A3xB,$Bi,${A1}[1] + + vshl.i64 $temp,`&Dhi("$A0xB")`,#16 + veor $zero,$zero,$zero + vadd.u64 $temp,$temp,`&Dlo("$A0xB")` + vld1.64 {$A7xB},[$tinptr,:128]! + vmul.u32 $Ni,$temp,$M0 + + vmlal.u32 $A4xB,$Bi,${A2}[0] + vld1.32 {$N0-$N3}, [$nptr]! + vmlal.u32 $A5xB,$Bi,${A2}[1] + vmlal.u32 $A6xB,$Bi,${A3}[0] + vzip.16 $Ni,$zero + vmlal.u32 $A7xB,$Bi,${A3}[1] + +.LNEON_inner: + vmlal.u32 $A0xB,$Ni,${N0}[0] + vld1.32 {$A0-$A3}, [$aptr]! + vmlal.u32 $A1xB,$Ni,${N0}[1] + subs $inner,$inner,#8 + vmlal.u32 $A2xB,$Ni,${N1}[0] + vmlal.u32 $A3xB,$Ni,${N1}[1] + vst1.64 {$A0xB-$A1xB}, [$toutptr,:256]! + + vmlal.u32 $A4xB,$Ni,${N2}[0] + vld1.64 {$A0xB}, [$tinptr, :128]! + vmlal.u32 $A5xB,$Ni,${N2}[1] + vst1.64 {$A2xB-$A3xB}, [$toutptr,:256]! + vmlal.u32 $A6xB,$Ni,${N3}[0] + vld1.64 {$A1xB-$A2xB}, [$tinptr, :256]! + vmlal.u32 $A7xB,$Ni,${N3}[1] + vst1.64 {$A4xB-$A5xB}, [$toutptr,:256]! + + vmlal.u32 $A0xB,$Bi,${A0}[0] + vld1.64 {$A3xB-$A4xB}, [$tinptr, :256]! + vmlal.u32 $A1xB,$Bi,${A0}[1] + vst1.64 {$A6xB-$A7xB}, [$toutptr,:256]! + vmlal.u32 $A2xB,$Bi,${A1}[0] + vld1.64 {$A5xB-$A6xB}, [$tinptr, :256]! + vmlal.u32 $A3xB,$Bi,${A1}[1] + vld1.32 {$N0-$N3}, [$nptr]! + + vmlal.u32 $A4xB,$Bi,${A2}[0] + vld1.64 {$A7xB}, [$tinptr, :128]! + vmlal.u32 $A5xB,$Bi,${A2}[1] + vmlal.u32 $A6xB,$Bi,${A3}[0] + vmlal.u32 $A7xB,$Bi,${A3}[1] + + bne .LNEON_inner + + vmlal.u32 $A0xB,$Ni,${N0}[0] + add $tinptr,sp,#16 + vmlal.u32 $A1xB,$Ni,${N0}[1] + sub $aptr,$aptr,$num,lsl#2 @ rewind $aptr + vmlal.u32 $A2xB,$Ni,${N1}[0] + vld1.64 {$Temp}, [sp,:128] + vmlal.u32 $A3xB,$Ni,${N1}[1] + subs $outer,$outer,#1 + + vmlal.u32 $A4xB,$Ni,${N2}[0] + vst1.64 {$A0xB-$A1xB}, [$toutptr,:256]! + vmlal.u32 $A5xB,$Ni,${N2}[1] + vld1.64 {$A0xB}, [$tinptr, :128]! + vshr.u64 $temp,$temp,#16 + vst1.64 {$A2xB-$A3xB}, [$toutptr,:256]! + vmlal.u32 $A6xB,$Ni,${N3}[0] + vld1.64 {$A1xB-$A2xB}, [$tinptr, :256]! + vmlal.u32 $A7xB,$Ni,${N3}[1] + + vst1.64 {$A4xB-$A5xB}, [$toutptr,:256]! + vadd.u64 $temp,$temp,`&Dhi("$Temp")` + vst1.64 {$A6xB-$A7xB}, [$toutptr,:256]! + vshr.u64 $temp,$temp,#16 + + bne .LNEON_outer + + mov $toutptr,sp + mov $inner,$num + +.LNEON_tail: + vadd.u64 `&Dlo("$A0xB")`,`&Dlo("$A0xB")`,$temp + vld1.64 {$A3xB-$A4xB}, [$tinptr, :256]! + vshr.u64 $temp,`&Dlo("$A0xB")`,#16 + vadd.u64 `&Dhi("$A0xB")`,`&Dhi("$A0xB")`,$temp + vld1.64 {$A5xB-$A6xB}, [$tinptr, :256]! + vshr.u64 $temp,`&Dhi("$A0xB")`,#16 + vld1.64 {$A7xB}, [$tinptr, :128]! + vzip.16 `&Dlo("$A0xB")`,`&Dhi("$A0xB")` + +.LNEON_tail2: + vadd.u64 `&Dlo("$A1xB")`,`&Dlo("$A1xB")`,$temp + vst1.32 {`&Dlo("$A0xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A1xB")`,#16 + vadd.u64 `&Dhi("$A1xB")`,`&Dhi("$A1xB")`,$temp + vshr.u64 $temp,`&Dhi("$A1xB")`,#16 + vzip.16 `&Dlo("$A1xB")`,`&Dhi("$A1xB")` + + vadd.u64 `&Dlo("$A2xB")`,`&Dlo("$A2xB")`,$temp + vst1.32 {`&Dlo("$A1xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A2xB")`,#16 + vadd.u64 `&Dhi("$A2xB")`,`&Dhi("$A2xB")`,$temp + vshr.u64 $temp,`&Dhi("$A2xB")`,#16 + vzip.16 `&Dlo("$A2xB")`,`&Dhi("$A2xB")` + + vadd.u64 `&Dlo("$A3xB")`,`&Dlo("$A3xB")`,$temp + vst1.32 {`&Dlo("$A2xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A3xB")`,#16 + vadd.u64 `&Dhi("$A3xB")`,`&Dhi("$A3xB")`,$temp + vshr.u64 $temp,`&Dhi("$A3xB")`,#16 + vzip.16 `&Dlo("$A3xB")`,`&Dhi("$A3xB")` + + vadd.u64 `&Dlo("$A4xB")`,`&Dlo("$A4xB")`,$temp + vst1.32 {`&Dlo("$A3xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A4xB")`,#16 + vadd.u64 `&Dhi("$A4xB")`,`&Dhi("$A4xB")`,$temp + vshr.u64 $temp,`&Dhi("$A4xB")`,#16 + vzip.16 `&Dlo("$A4xB")`,`&Dhi("$A4xB")` + + vadd.u64 `&Dlo("$A5xB")`,`&Dlo("$A5xB")`,$temp + vst1.32 {`&Dlo("$A4xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A5xB")`,#16 + vadd.u64 `&Dhi("$A5xB")`,`&Dhi("$A5xB")`,$temp + vshr.u64 $temp,`&Dhi("$A5xB")`,#16 + vzip.16 `&Dlo("$A5xB")`,`&Dhi("$A5xB")` + + vadd.u64 `&Dlo("$A6xB")`,`&Dlo("$A6xB")`,$temp + vst1.32 {`&Dlo("$A5xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A6xB")`,#16 + vadd.u64 `&Dhi("$A6xB")`,`&Dhi("$A6xB")`,$temp + vld1.64 {$A0xB}, [$tinptr, :128]! + vshr.u64 $temp,`&Dhi("$A6xB")`,#16 + vzip.16 `&Dlo("$A6xB")`,`&Dhi("$A6xB")` + + vadd.u64 `&Dlo("$A7xB")`,`&Dlo("$A7xB")`,$temp + vst1.32 {`&Dlo("$A6xB")`[0]}, [$toutptr, :32]! + vshr.u64 $temp,`&Dlo("$A7xB")`,#16 + vadd.u64 `&Dhi("$A7xB")`,`&Dhi("$A7xB")`,$temp + vld1.64 {$A1xB-$A2xB}, [$tinptr, :256]! + vshr.u64 $temp,`&Dhi("$A7xB")`,#16 + vzip.16 `&Dlo("$A7xB")`,`&Dhi("$A7xB")` + subs $inner,$inner,#8 + vst1.32 {`&Dlo("$A7xB")`[0]}, [$toutptr, :32]! + + bne .LNEON_tail + + vst1.32 {${temp}[0]}, [$toutptr, :32] @ top-most bit + sub $nptr,$nptr,$num,lsl#2 @ rewind $nptr + subs $aptr,sp,#0 @ clear carry flag + add $bptr,sp,$num,lsl#2 + +.LNEON_sub: + ldmia $aptr!, {r4-r7} + ldmia $nptr!, {r8-r11} + sbcs r8, r4,r8 + sbcs r9, r5,r9 + sbcs r10,r6,r10 + sbcs r11,r7,r11 + teq $aptr,$bptr @ preserves carry + stmia $rptr!, {r8-r11} + bne .LNEON_sub + + ldr r10, [$aptr] @ load top-most bit + veor q0,q0,q0 + sub r11,$bptr,sp @ this is num*4 + veor q1,q1,q1 + mov $aptr,sp + sub $rptr,$rptr,r11 @ rewind $rptr + mov $nptr,$bptr @ second 3/4th of frame + sbcs r10,r10,#0 @ result is carry flag + +.LNEON_copy_n_zap: + ldmia $aptr!, {r4-r7} + ldmia $rptr, {r8-r11} + movcc r8, r4 + vst1.64 {q0-q1}, [$nptr,:256]! @ wipe + movcc r9, r5 + movcc r10,r6 + vst1.64 {q0-q1}, [$nptr,:256]! @ wipe + movcc r11,r7 + ldmia $aptr, {r4-r7} + stmia $rptr!, {r8-r11} + sub $aptr,$aptr,#16 + ldmia $rptr, {r8-r11} + movcc r8, r4 + vst1.64 {q0-q1}, [$aptr,:256]! @ wipe + movcc r9, r5 + movcc r10,r6 + vst1.64 {q0-q1}, [$nptr,:256]! @ wipe + movcc r11,r7 + teq $aptr,$bptr @ preserves carry + stmia $rptr!, {r8-r11} + bne .LNEON_copy_n_zap + + sub sp,ip,#96 + vldmia sp!,{d8-d15} + ldmia sp!,{r4-r11} + ret @ bx lr +.size bn_mul8x_mont_neon,.-bn_mul8x_mont_neon +#endif +___ +} +$code.=<<___; +.asciz "Montgomery multiplication for ARMv4/NEON, CRYPTOGAMS by <appro\@openssl.org>" .align 2 +#if __ARM_MAX_ARCH__>=7 +.comm OPENSSL_armcap_P,4,4 +#endif ___ +$code =~ s/\`([^\`]*)\`/eval $1/gem; $code =~ s/\bbx\s+lr\b/.word\t0xe12fff1e/gm; # make it possible to compile with -march=armv4 +$code =~ s/\bret\b/bx lr/gm; print $code; close STDOUT; diff --git a/openssl/crypto/bn/asm/mips-mont.pl b/openssl/crypto/bn/asm/mips-mont.pl index caae04ed3..a33cdf411 100644 --- a/openssl/crypto/bn/asm/mips-mont.pl +++ b/openssl/crypto/bn/asm/mips-mont.pl @@ -46,7 +46,7 @@ # ($s0,$s1,$s2,$s3,$s4,$s5,$s6,$s7)=map("\$$_",(16..23)); # ($gp,$sp,$fp,$ra)=map("\$$_",(28..31)); # -$flavour = shift; # supported flavours are o32,n32,64,nubi32,nubi64 +$flavour = shift || "o32"; # supported flavours are o32,n32,64,nubi32,nubi64 if ($flavour =~ /64|n32/i) { $PTR_ADD="dadd"; # incidentally works even on n32 diff --git a/openssl/crypto/bn/asm/mips.pl b/openssl/crypto/bn/asm/mips.pl index d2f3ef7bb..acafde5e5 100644 --- a/openssl/crypto/bn/asm/mips.pl +++ b/openssl/crypto/bn/asm/mips.pl @@ -48,7 +48,7 @@ # has to content with 40-85% improvement depending on benchmark and # key length, more for longer keys. -$flavour = shift; +$flavour = shift || "o32"; while (($output=shift) && ($output!~/^\w[\w\-]*\.\w+$/)) {} open STDOUT,">$output"; @@ -1872,6 +1872,41 @@ ___ ($a_4,$a_5,$a_6,$a_7)=($b_0,$b_1,$b_2,$b_3); +sub add_c2 () { +my ($hi,$lo,$c0,$c1,$c2, + $warm, # !$warm denotes first call with specific sequence of + # $c_[XYZ] when there is no Z-carry to accumulate yet; + $an,$bn # these two are arguments for multiplication which + # result is used in *next* step [which is why it's + # commented as "forward multiplication" below]; + )=@_; +$code.=<<___; + mflo $lo + mfhi $hi + $ADDU $c0,$lo + sltu $at,$c0,$lo + $MULTU $an,$bn # forward multiplication + $ADDU $c0,$lo + $ADDU $at,$hi + sltu $lo,$c0,$lo + $ADDU $c1,$at + $ADDU $hi,$lo +___ +$code.=<<___ if (!$warm); + sltu $c2,$c1,$at + $ADDU $c1,$hi + sltu $hi,$c1,$hi + $ADDU $c2,$hi +___ +$code.=<<___ if ($warm); + sltu $at,$c1,$at + $ADDU $c1,$hi + $ADDU $c2,$at + sltu $hi,$c1,$hi + $ADDU $c2,$hi +___ +} + $code.=<<___; .align 5 @@ -1920,21 +1955,10 @@ $code.=<<___; sltu $at,$c_2,$t_1 $ADDU $c_3,$t_2,$at $ST $c_2,$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_2,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_1,$a_1 # mul_add_c(a[1],b[1],c3,c1,c2); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at +___ + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0, + $a_1,$a_1); # mul_add_c(a[1],b[1],c3,c1,c2); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_3,$t_1 @@ -1945,67 +1969,19 @@ $code.=<<___; sltu $at,$c_1,$t_2 $ADDU $c_2,$at $ST $c_3,2*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_3,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_1,$a_2 # mul_add_c2(a[1],b[2],c1,c2,c3); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_3,$at - $MULTU $a_4,$a_0 # mul_add_c2(a[4],b[0],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at +___ + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0, + $a_1,$a_2); # mul_add_c2(a[1],b[2],c1,c2,c3); + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1, + $a_4,$a_0); # mul_add_c2(a[4],b[0],c2,c3,c1); +$code.=<<___; $ST $c_1,3*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_1,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_3,$a_1 # mul_add_c2(a[3],b[1],c2,c3,c1); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_1,$at - $MULTU $a_2,$a_2 # mul_add_c(a[2],b[2],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at +___ + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0, + $a_3,$a_1); # mul_add_c2(a[3],b[1],c2,c3,c1); + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1, + $a_2,$a_2); # mul_add_c(a[2],b[2],c2,c3,c1); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_2,$t_1 @@ -2016,97 +1992,23 @@ $code.=<<___; sltu $at,$c_3,$t_2 $ADDU $c_1,$at $ST $c_2,4*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_2,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_1,$a_4 # mul_add_c2(a[1],b[4],c3,c1,c2); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_2,$at - $MULTU $a_2,$a_3 # mul_add_c2(a[2],b[3],c3,c1,c2); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $MULTU $a_6,$a_0 # mul_add_c2(a[6],b[0],c1,c2,c3); - $ADDU $c_2,$at - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at +___ + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0, + $a_1,$a_4); # mul_add_c2(a[1],b[4],c3,c1,c2); + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1, + $a_2,$a_3); # mul_add_c2(a[2],b[3],c3,c1,c2); + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1, + $a_6,$a_0); # mul_add_c2(a[6],b[0],c1,c2,c3); +$code.=<<___; $ST $c_3,5*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_3,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_5,$a_1 # mul_add_c2(a[5],b[1],c1,c2,c3); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_3,$at - $MULTU $a_4,$a_2 # mul_add_c2(a[4],b[2],c1,c2,c3); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_3,$at - $MULTU $a_3,$a_3 # mul_add_c(a[3],b[3],c1,c2,c3); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at +___ + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0, + $a_5,$a_1); # mul_add_c2(a[5],b[1],c1,c2,c3); + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1, + $a_4,$a_2); # mul_add_c2(a[4],b[2],c1,c2,c3); + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1, + $a_3,$a_3); # mul_add_c(a[3],b[3],c1,c2,c3); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_1,$t_1 @@ -2117,112 +2019,25 @@ $code.=<<___; sltu $at,$c_2,$t_2 $ADDU $c_3,$at $ST $c_1,6*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_1,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_1,$a_6 # mul_add_c2(a[1],b[6],c2,c3,c1); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_1,$at - $MULTU $a_2,$a_5 # mul_add_c2(a[2],b[5],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_1,$at - $MULTU $a_3,$a_4 # mul_add_c2(a[3],b[4],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_1,$at - $MULTU $a_7,$a_1 # mul_add_c2(a[7],b[1],c3,c1,c2); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at +___ + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0, + $a_1,$a_6); # mul_add_c2(a[1],b[6],c2,c3,c1); + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1, + $a_2,$a_5); # mul_add_c2(a[2],b[5],c2,c3,c1); + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1, + $a_3,$a_4); # mul_add_c2(a[3],b[4],c2,c3,c1); + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1, + $a_7,$a_1); # mul_add_c2(a[7],b[1],c3,c1,c2); +$code.=<<___; $ST $c_2,7*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_2,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_6,$a_2 # mul_add_c2(a[6],b[2],c3,c1,c2); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_2,$at - $MULTU $a_5,$a_3 # mul_add_c2(a[5],b[3],c3,c1,c2); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_2,$at - $MULTU $a_4,$a_4 # mul_add_c(a[4],b[4],c3,c1,c2); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at +___ + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0, + $a_6,$a_2); # mul_add_c2(a[6],b[2],c3,c1,c2); + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1, + $a_5,$a_3); # mul_add_c2(a[5],b[3],c3,c1,c2); + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1, + $a_4,$a_4); # mul_add_c(a[4],b[4],c3,c1,c2); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_3,$t_1 @@ -2233,82 +2048,21 @@ $code.=<<___; sltu $at,$c_1,$t_2 $ADDU $c_2,$at $ST $c_3,8*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_3,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_3,$a_6 # mul_add_c2(a[3],b[6],c1,c2,c3); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_3,$at - $MULTU $a_4,$a_5 # mul_add_c2(a[4],b[5],c1,c2,c3); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_3,$at - $MULTU $a_7,$a_3 # mul_add_c2(a[7],b[3],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at +___ + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0, + $a_3,$a_6); # mul_add_c2(a[3],b[6],c1,c2,c3); + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1, + $a_4,$a_5); # mul_add_c2(a[4],b[5],c1,c2,c3); + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1, + $a_7,$a_3); # mul_add_c2(a[7],b[3],c2,c3,c1); +$code.=<<___; $ST $c_1,9*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_1,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_6,$a_4 # mul_add_c2(a[6],b[4],c2,c3,c1); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_1,$at - $MULTU $a_5,$a_5 # mul_add_c(a[5],b[5],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at +___ + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0, + $a_6,$a_4); # mul_add_c2(a[6],b[4],c2,c3,c1); + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,1, + $a_5,$a_5); # mul_add_c(a[5],b[5],c2,c3,c1); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_2,$t_1 @@ -2319,52 +2073,17 @@ $code.=<<___; sltu $at,$c_3,$t_2 $ADDU $c_1,$at $ST $c_2,10*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_2,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_5,$a_6 # mul_add_c2(a[5],b[6],c3,c1,c2); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_2,$at - $MULTU $a_7,$a_5 # mul_add_c2(a[7],b[5],c1,c2,c3); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at +___ + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0, + $a_5,$a_6); # mul_add_c2(a[5],b[6],c3,c1,c2); + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,1, + $a_7,$a_5); # mul_add_c2(a[7],b[5],c1,c2,c3); +$code.=<<___; $ST $c_3,11*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_3,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_6,$a_6 # mul_add_c(a[6],b[6],c1,c2,c3); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at +___ + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0, + $a_6,$a_6); # mul_add_c(a[6],b[6],c1,c2,c3); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_1,$t_1 @@ -2375,21 +2094,10 @@ $code.=<<___; sltu $at,$c_2,$t_2 $ADDU $c_3,$at $ST $c_1,12*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_1,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_7,$a_7 # mul_add_c(a[7],b[7],c3,c1,c2); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at +___ + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0, + $a_7,$a_7); # mul_add_c(a[7],b[7],c3,c1,c2); +$code.=<<___; $ST $c_2,13*$BNSZ($a0) mflo $t_1 @@ -2457,21 +2165,10 @@ $code.=<<___; sltu $at,$c_2,$t_1 $ADDU $c_3,$t_2,$at $ST $c_2,$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_2,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_1,$a_1 # mul_add_c(a[1],b[1],c3,c1,c2); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at +___ + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0, + $a_1,$a_1); # mul_add_c(a[1],b[1],c3,c1,c2); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_3,$t_1 @@ -2482,52 +2179,17 @@ $code.=<<___; sltu $at,$c_1,$t_2 $ADDU $c_2,$at $ST $c_3,2*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_3,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_1,$a_2 # mul_add_c(a2[1],b[2],c1,c2,c3); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at - mflo $t_1 - mfhi $t_2 - slt $at,$t_2,$zero - $ADDU $c_3,$at - $MULTU $a_3,$a_1 # mul_add_c2(a[3],b[1],c2,c3,c1); - $SLL $t_2,1 - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_1,$t_1 - sltu $at,$c_1,$t_1 - $ADDU $t_2,$at - $ADDU $c_2,$t_2 - sltu $at,$c_2,$t_2 - $ADDU $c_3,$at +___ + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,0, + $a_1,$a_2); # mul_add_c2(a2[1],b[2],c1,c2,c3); + &add_c2($t_2,$t_1,$c_1,$c_2,$c_3,1, + $a_3,$a_1); # mul_add_c2(a[3],b[1],c2,c3,c1); +$code.=<<___; $ST $c_1,3*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_1,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_2,$a_2 # mul_add_c(a[2],b[2],c2,c3,c1); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_2,$t_1 - sltu $at,$c_2,$t_1 - $ADDU $t_2,$at - $ADDU $c_3,$t_2 - sltu $at,$c_3,$t_2 - $ADDU $c_1,$at +___ + &add_c2($t_2,$t_1,$c_2,$c_3,$c_1,0, + $a_2,$a_2); # mul_add_c(a[2],b[2],c2,c3,c1); +$code.=<<___; mflo $t_1 mfhi $t_2 $ADDU $c_2,$t_1 @@ -2538,21 +2200,10 @@ $code.=<<___; sltu $at,$c_3,$t_2 $ADDU $c_1,$at $ST $c_2,4*$BNSZ($a0) - - mflo $t_1 - mfhi $t_2 - slt $c_2,$t_2,$zero - $SLL $t_2,1 - $MULTU $a_3,$a_3 # mul_add_c(a[3],b[3],c1,c2,c3); - slt $a2,$t_1,$zero - $ADDU $t_2,$a2 - $SLL $t_1,1 - $ADDU $c_3,$t_1 - sltu $at,$c_3,$t_1 - $ADDU $t_2,$at - $ADDU $c_1,$t_2 - sltu $at,$c_1,$t_2 - $ADDU $c_2,$at +___ + &add_c2($t_2,$t_1,$c_3,$c_1,$c_2,0, + $a_3,$a_3); # mul_add_c(a[3],b[3],c1,c2,c3); +$code.=<<___; $ST $c_3,5*$BNSZ($a0) mflo $t_1 diff --git a/openssl/crypto/bn/asm/ppc-mont.pl b/openssl/crypto/bn/asm/ppc-mont.pl index f9b6992cc..da69c6aaa 100644 --- a/openssl/crypto/bn/asm/ppc-mont.pl +++ b/openssl/crypto/bn/asm/ppc-mont.pl @@ -325,6 +325,7 @@ Lcopy: ; copy or in-place refresh .long 0 .byte 0,12,4,0,0x80,12,6,0 .long 0 +.size .bn_mul_mont_int,.-.bn_mul_mont_int .asciz "Montgomery Multiplication for PPC, CRYPTOGAMS by <appro\@openssl.org>" ___ diff --git a/openssl/crypto/bn/asm/ppc.pl b/openssl/crypto/bn/asm/ppc.pl index 1249ce229..04df1fe5c 100644 --- a/openssl/crypto/bn/asm/ppc.pl +++ b/openssl/crypto/bn/asm/ppc.pl @@ -392,6 +392,7 @@ $data=<<EOF; .long 0 .byte 0,12,0x14,0,0,0,2,0 .long 0 +.size .bn_sqr_comba4,.-.bn_sqr_comba4 # # NOTE: The following label name should be changed to @@ -819,6 +820,7 @@ $data=<<EOF; .long 0 .byte 0,12,0x14,0,0,0,2,0 .long 0 +.size .bn_sqr_comba8,.-.bn_sqr_comba8 # # NOTE: The following label name should be changed to @@ -972,6 +974,7 @@ $data=<<EOF; .long 0 .byte 0,12,0x14,0,0,0,3,0 .long 0 +.size .bn_mul_comba4,.-.bn_mul_comba4 # # NOTE: The following label name should be changed to @@ -1510,6 +1513,7 @@ $data=<<EOF; .long 0 .byte 0,12,0x14,0,0,0,3,0 .long 0 +.size .bn_mul_comba8,.-.bn_mul_comba8 # # NOTE: The following label name should be changed to @@ -1560,6 +1564,7 @@ Lppcasm_sub_adios: .long 0 .byte 0,12,0x14,0,0,0,4,0 .long 0 +.size .bn_sub_words,.-.bn_sub_words # # NOTE: The following label name should be changed to @@ -1605,6 +1610,7 @@ Lppcasm_add_adios: .long 0 .byte 0,12,0x14,0,0,0,4,0 .long 0 +.size .bn_add_words,.-.bn_add_words # # NOTE: The following label name should be changed to @@ -1720,6 +1726,7 @@ Lppcasm_div9: .long 0 .byte 0,12,0x14,0,0,0,3,0 .long 0 +.size .bn_div_words,.-.bn_div_words # # NOTE: The following label name should be changed to @@ -1761,6 +1768,7 @@ Lppcasm_sqr_adios: .long 0 .byte 0,12,0x14,0,0,0,3,0 .long 0 +.size .bn_sqr_words,.-.bn_sqr_words # # NOTE: The following label name should be changed to @@ -1866,6 +1874,7 @@ Lppcasm_mw_OVER: .long 0 .byte 0,12,0x14,0,0,0,4,0 .long 0 +.size bn_mul_words,.-bn_mul_words # # NOTE: The following label name should be changed to @@ -1991,6 +2000,7 @@ Lppcasm_maw_adios: .long 0 .byte 0,12,0x14,0,0,0,4,0 .long 0 +.size .bn_mul_add_words,.-.bn_mul_add_words .align 4 EOF $data =~ s/\`([^\`]*)\`/eval $1/gem; diff --git a/openssl/crypto/bn/asm/ppc64-mont.pl b/openssl/crypto/bn/asm/ppc64-mont.pl index a14e769ad..68e3733e3 100644 --- a/openssl/crypto/bn/asm/ppc64-mont.pl +++ b/openssl/crypto/bn/asm/ppc64-mont.pl @@ -1,7 +1,7 @@ #!/usr/bin/env perl # ==================================================================== -# Written by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL +# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL # project. The module is, however, dual licensed under OpenSSL and # CRYPTOGAMS licenses depending on where you obtain it. For further # details see http://www.openssl.org/~appro/cryptogams/. @@ -65,6 +65,14 @@ # others alternative would be to break dependence on upper halves of # GPRs by sticking to 32-bit integer operations... +# December 2012 + +# Remove above mentioned dependence on GPRs' upper halves in 32-bit +# build. No signal masking overhead, but integer instructions are +# *more* numerous... It's still "universally" faster than 32-bit +# ppc-mont.pl, but improvement coefficient is not as impressive +# for longer keys... + $flavour = shift; if ($flavour =~ /32/) { @@ -110,6 +118,9 @@ $tp="r10"; $j="r11"; $i="r12"; # non-volatile registers +$c1="r19"; +$n1="r20"; +$a1="r21"; $nap_d="r22"; # interleaved ap and np in double format $a0="r23"; # ap[0] $t0="r24"; # temporary registers @@ -180,8 +191,8 @@ $T3a="f30"; $T3b="f31"; # . . # +-------------------------------+ # . . -# -12*size_t +-------------------------------+ -# | 10 saved gpr, r22-r31 | +# -13*size_t +-------------------------------+ +# | 13 saved gpr, r19-r31 | # . . # . . # -12*8 +-------------------------------+ @@ -215,6 +226,9 @@ $code=<<___; mr $i,$sp $STUX $sp,$sp,$tp ; alloca + $PUSH r19,`-12*8-13*$SIZE_T`($i) + $PUSH r20,`-12*8-12*$SIZE_T`($i) + $PUSH r21,`-12*8-11*$SIZE_T`($i) $PUSH r22,`-12*8-10*$SIZE_T`($i) $PUSH r23,`-12*8-9*$SIZE_T`($i) $PUSH r24,`-12*8-8*$SIZE_T`($i) @@ -237,40 +251,26 @@ $code=<<___; stfd f29,`-3*8`($i) stfd f30,`-2*8`($i) stfd f31,`-1*8`($i) -___ -$code.=<<___ if ($SIZE_T==8); - ld $a0,0($ap) ; pull ap[0] value - ld $n0,0($n0) ; pull n0[0] value - ld $t3,0($bp) ; bp[0] -___ -$code.=<<___ if ($SIZE_T==4); - mr $t1,$n0 - lwz $a0,0($ap) ; pull ap[0,1] value - lwz $t0,4($ap) - lwz $n0,0($t1) ; pull n0[0,1] value - lwz $t1,4($t1) - lwz $t3,0($bp) ; bp[0,1] - lwz $t2,4($bp) - insrdi $a0,$t0,32,0 - insrdi $n0,$t1,32,0 - insrdi $t3,$t2,32,0 -___ -$code.=<<___; + addi $tp,$sp,`$FRAME+$TRANSFER+8+64` li $i,-64 add $nap_d,$tp,$num and $nap_d,$nap_d,$i ; align to 64 bytes - - mulld $t7,$a0,$t3 ; ap[0]*bp[0] ; nap_d is off by 1, because it's used with stfdu/lfdu addi $nap_d,$nap_d,-8 srwi $j,$num,`3+1` ; counter register, num/2 - mulld $t7,$t7,$n0 ; tp[0]*n0 addi $j,$j,-1 addi $tp,$sp,`$FRAME+$TRANSFER-8` li $carry,0 mtctr $j +___ + +$code.=<<___ if ($SIZE_T==8); + ld $a0,0($ap) ; pull ap[0] value + ld $t3,0($bp) ; bp[0] + ld $n0,0($n0) ; pull n0[0] value + mulld $t7,$a0,$t3 ; ap[0]*bp[0] ; transfer bp[0] to FPU as 4x16-bit values extrdi $t0,$t3,16,48 extrdi $t1,$t3,16,32 @@ -280,6 +280,8 @@ $code.=<<___; std $t1,`$FRAME+8`($sp) std $t2,`$FRAME+16`($sp) std $t3,`$FRAME+24`($sp) + + mulld $t7,$t7,$n0 ; tp[0]*n0 ; transfer (ap[0]*bp[0])*n0 to FPU as 4x16-bit values extrdi $t4,$t7,16,48 extrdi $t5,$t7,16,32 @@ -289,21 +291,61 @@ $code.=<<___; std $t5,`$FRAME+40`($sp) std $t6,`$FRAME+48`($sp) std $t7,`$FRAME+56`($sp) -___ -$code.=<<___ if ($SIZE_T==8); - lwz $t0,4($ap) ; load a[j] as 32-bit word pair - lwz $t1,0($ap) - lwz $t2,12($ap) ; load a[j+1] as 32-bit word pair + + extrdi $t0,$a0,32,32 ; lwz $t0,4($ap) + extrdi $t1,$a0,32,0 ; lwz $t1,0($ap) + lwz $t2,12($ap) ; load a[1] as 32-bit word pair lwz $t3,8($ap) - lwz $t4,4($np) ; load n[j] as 32-bit word pair + lwz $t4,4($np) ; load n[0] as 32-bit word pair lwz $t5,0($np) - lwz $t6,12($np) ; load n[j+1] as 32-bit word pair + lwz $t6,12($np) ; load n[1] as 32-bit word pair lwz $t7,8($np) ___ $code.=<<___ if ($SIZE_T==4); - lwz $t0,0($ap) ; load a[j..j+3] as 32-bit word pairs - lwz $t1,4($ap) - lwz $t2,8($ap) + lwz $a0,0($ap) ; pull ap[0,1] value + mr $n1,$n0 + lwz $a1,4($ap) + li $c1,0 + lwz $t1,0($bp) ; bp[0,1] + lwz $t3,4($bp) + lwz $n0,0($n1) ; pull n0[0,1] value + lwz $n1,4($n1) + + mullw $t4,$a0,$t1 ; mulld ap[0]*bp[0] + mulhwu $t5,$a0,$t1 + mullw $t6,$a1,$t1 + mullw $t7,$a0,$t3 + add $t5,$t5,$t6 + add $t5,$t5,$t7 + ; transfer bp[0] to FPU as 4x16-bit values + extrwi $t0,$t1,16,16 + extrwi $t1,$t1,16,0 + extrwi $t2,$t3,16,16 + extrwi $t3,$t3,16,0 + std $t0,`$FRAME+0`($sp) ; yes, std in 32-bit build + std $t1,`$FRAME+8`($sp) + std $t2,`$FRAME+16`($sp) + std $t3,`$FRAME+24`($sp) + + mullw $t0,$t4,$n0 ; mulld tp[0]*n0 + mulhwu $t1,$t4,$n0 + mullw $t2,$t5,$n0 + mullw $t3,$t4,$n1 + add $t1,$t1,$t2 + add $t1,$t1,$t3 + ; transfer (ap[0]*bp[0])*n0 to FPU as 4x16-bit values + extrwi $t4,$t0,16,16 + extrwi $t5,$t0,16,0 + extrwi $t6,$t1,16,16 + extrwi $t7,$t1,16,0 + std $t4,`$FRAME+32`($sp) ; yes, std in 32-bit build + std $t5,`$FRAME+40`($sp) + std $t6,`$FRAME+48`($sp) + std $t7,`$FRAME+56`($sp) + + mr $t0,$a0 ; lwz $t0,0($ap) + mr $t1,$a1 ; lwz $t1,4($ap) + lwz $t2,8($ap) ; load a[j..j+3] as 32-bit word pairs lwz $t3,12($ap) lwz $t4,0($np) ; load n[j..j+3] as 32-bit word pairs lwz $t5,4($np) @@ -319,7 +361,7 @@ $code.=<<___; lfd $nb,`$FRAME+40`($sp) lfd $nc,`$FRAME+48`($sp) lfd $nd,`$FRAME+56`($sp) - std $t0,`$FRAME+64`($sp) + std $t0,`$FRAME+64`($sp) ; yes, std even in 32-bit build std $t1,`$FRAME+72`($sp) std $t2,`$FRAME+80`($sp) std $t3,`$FRAME+88`($sp) @@ -441,7 +483,7 @@ $code.=<<___ if ($SIZE_T==4); lwz $t7,12($np) ___ $code.=<<___; - std $t0,`$FRAME+64`($sp) + std $t0,`$FRAME+64`($sp) ; yes, std even in 32-bit build std $t1,`$FRAME+72`($sp) std $t2,`$FRAME+80`($sp) std $t3,`$FRAME+88`($sp) @@ -449,6 +491,9 @@ $code.=<<___; std $t5,`$FRAME+104`($sp) std $t6,`$FRAME+112`($sp) std $t7,`$FRAME+120`($sp) +___ +if ($SIZE_T==8 or $flavour =~ /osx/) { +$code.=<<___; ld $t0,`$FRAME+0`($sp) ld $t1,`$FRAME+8`($sp) ld $t2,`$FRAME+16`($sp) @@ -457,6 +502,20 @@ $code.=<<___; ld $t5,`$FRAME+40`($sp) ld $t6,`$FRAME+48`($sp) ld $t7,`$FRAME+56`($sp) +___ +} else { +$code.=<<___; + lwz $t1,`$FRAME+0`($sp) + lwz $t0,`$FRAME+4`($sp) + lwz $t3,`$FRAME+8`($sp) + lwz $t2,`$FRAME+12`($sp) + lwz $t5,`$FRAME+16`($sp) + lwz $t4,`$FRAME+20`($sp) + lwz $t7,`$FRAME+24`($sp) + lwz $t6,`$FRAME+28`($sp) +___ +} +$code.=<<___; lfd $A0,`$FRAME+64`($sp) lfd $A1,`$FRAME+72`($sp) lfd $A2,`$FRAME+80`($sp) @@ -488,7 +547,9 @@ $code.=<<___; fmadd $T0b,$A0,$bb,$dotb stfd $A2,24($nap_d) ; save a[j+1] in double format stfd $A3,32($nap_d) - +___ +if ($SIZE_T==8 or $flavour =~ /osx/) { +$code.=<<___; fmadd $T1a,$A0,$bc,$T1a fmadd $T1b,$A0,$bd,$T1b fmadd $T2a,$A1,$bc,$T2a @@ -561,11 +622,123 @@ $code.=<<___; stfd $T3b,`$FRAME+56`($sp) std $t0,8($tp) ; tp[j-1] stdu $t4,16($tp) ; tp[j] +___ +} else { +$code.=<<___; + fmadd $T1a,$A0,$bc,$T1a + fmadd $T1b,$A0,$bd,$T1b + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + fmadd $T2a,$A1,$bc,$T2a + fmadd $T2b,$A1,$bd,$T2b + stfd $N0,40($nap_d) ; save n[j] in double format + stfd $N1,48($nap_d) + srwi $c1,$t1,16 + insrwi $carry,$t1,16,0 + fmadd $T3a,$A2,$bc,$T3a + fmadd $T3b,$A2,$bd,$T3b + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + fmul $dota,$A3,$bc + fmul $dotb,$A3,$bd + stfd $N2,56($nap_d) ; save n[j+1] in double format + stfdu $N3,64($nap_d) + insrwi $t0,$t2,16,0 ; 0..31 bits + srwi $c1,$t3,16 + insrwi $carry,$t3,16,0 + + fmadd $T1a,$N1,$na,$T1a + fmadd $T1b,$N1,$nb,$T1b + lwz $t3,`$FRAME+32`($sp) ; permuted $t1 + lwz $t2,`$FRAME+36`($sp) ; permuted $t0 + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + fmadd $T2a,$N2,$na,$T2a + fmadd $T2b,$N2,$nb,$T2b + srwi $c1,$t5,16 + insrwi $carry,$t5,16,0 + fmadd $T3a,$N3,$na,$T3a + fmadd $T3b,$N3,$nb,$T3b + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + fmadd $T0a,$N0,$na,$T0a + fmadd $T0b,$N0,$nb,$T0b + insrwi $t4,$t6,16,0 ; 32..63 bits + srwi $c1,$t7,16 + insrwi $carry,$t7,16,0 + + fmadd $T1a,$N0,$nc,$T1a + fmadd $T1b,$N0,$nd,$T1b + lwz $t7,`$FRAME+40`($sp) ; permuted $t3 + lwz $t6,`$FRAME+44`($sp) ; permuted $t2 + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + fmadd $T2a,$N1,$nc,$T2a + fmadd $T2b,$N1,$nd,$T2b + stw $t0,12($tp) ; tp[j-1] + stw $t4,8($tp) + srwi $c1,$t3,16 + insrwi $carry,$t3,16,0 + fmadd $T3a,$N2,$nc,$T3a + fmadd $T3b,$N2,$nd,$T3b + lwz $t1,`$FRAME+48`($sp) ; permuted $t5 + lwz $t0,`$FRAME+52`($sp) ; permuted $t4 + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + fmadd $dota,$N3,$nc,$dota + fmadd $dotb,$N3,$nd,$dotb + insrwi $t2,$t6,16,0 ; 64..95 bits + srwi $c1,$t7,16 + insrwi $carry,$t7,16,0 + + fctid $T0a,$T0a + fctid $T0b,$T0b + lwz $t5,`$FRAME+56`($sp) ; permuted $t7 + lwz $t4,`$FRAME+60`($sp) ; permuted $t6 + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + fctid $T1a,$T1a + fctid $T1b,$T1b + srwi $c1,$t1,16 + insrwi $carry,$t1,16,0 + fctid $T2a,$T2a + fctid $T2b,$T2b + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + fctid $T3a,$T3a + fctid $T3b,$T3b + insrwi $t0,$t4,16,0 ; 96..127 bits + srwi $c1,$t5,16 + insrwi $carry,$t5,16,0 + + stfd $T0a,`$FRAME+0`($sp) + stfd $T0b,`$FRAME+8`($sp) + stfd $T1a,`$FRAME+16`($sp) + stfd $T1b,`$FRAME+24`($sp) + stfd $T2a,`$FRAME+32`($sp) + stfd $T2b,`$FRAME+40`($sp) + stfd $T3a,`$FRAME+48`($sp) + stfd $T3b,`$FRAME+56`($sp) + stw $t2,20($tp) ; tp[j] + stwu $t0,16($tp) +___ +} +$code.=<<___; bdnz- L1st fctid $dota,$dota fctid $dotb,$dotb - +___ +if ($SIZE_T==8 or $flavour =~ /osx/) { +$code.=<<___; ld $t0,`$FRAME+0`($sp) ld $t1,`$FRAME+8`($sp) ld $t2,`$FRAME+16`($sp) @@ -611,33 +784,117 @@ $code.=<<___; insrdi $t6,$t7,48,0 srdi $ovf,$t7,48 std $t6,8($tp) ; tp[num-1] +___ +} else { +$code.=<<___; + lwz $t1,`$FRAME+0`($sp) + lwz $t0,`$FRAME+4`($sp) + lwz $t3,`$FRAME+8`($sp) + lwz $t2,`$FRAME+12`($sp) + lwz $t5,`$FRAME+16`($sp) + lwz $t4,`$FRAME+20`($sp) + lwz $t7,`$FRAME+24`($sp) + lwz $t6,`$FRAME+28`($sp) + stfd $dota,`$FRAME+64`($sp) + stfd $dotb,`$FRAME+72`($sp) + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + insrwi $carry,$t1,16,0 + srwi $c1,$t1,16 + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + insrwi $t0,$t2,16,0 ; 0..31 bits + insrwi $carry,$t3,16,0 + srwi $c1,$t3,16 + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + insrwi $carry,$t5,16,0 + srwi $c1,$t5,16 + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + insrwi $t4,$t6,16,0 ; 32..63 bits + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + stw $t0,12($tp) ; tp[j-1] + stw $t4,8($tp) + + lwz $t3,`$FRAME+32`($sp) ; permuted $t1 + lwz $t2,`$FRAME+36`($sp) ; permuted $t0 + lwz $t7,`$FRAME+40`($sp) ; permuted $t3 + lwz $t6,`$FRAME+44`($sp) ; permuted $t2 + lwz $t1,`$FRAME+48`($sp) ; permuted $t5 + lwz $t0,`$FRAME+52`($sp) ; permuted $t4 + lwz $t5,`$FRAME+56`($sp) ; permuted $t7 + lwz $t4,`$FRAME+60`($sp) ; permuted $t6 + + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + insrwi $carry,$t3,16,0 + srwi $c1,$t3,16 + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + insrwi $t2,$t6,16,0 ; 64..95 bits + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + insrwi $carry,$t1,16,0 + srwi $c1,$t1,16 + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + insrwi $t0,$t4,16,0 ; 96..127 bits + insrwi $carry,$t5,16,0 + srwi $c1,$t5,16 + stw $t2,20($tp) ; tp[j] + stwu $t0,16($tp) + + lwz $t7,`$FRAME+64`($sp) + lwz $t6,`$FRAME+68`($sp) + lwz $t5,`$FRAME+72`($sp) + lwz $t4,`$FRAME+76`($sp) + + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + + insrwi $t6,$t4,16,0 + srwi $t4,$t4,16 + insrwi $t4,$t5,16,0 + srwi $ovf,$t5,16 + stw $t6,12($tp) ; tp[num-1] + stw $t4,8($tp) +___ +} +$code.=<<___; slwi $t7,$num,2 subf $nap_d,$t7,$nap_d ; rewind pointer li $i,8 ; i=1 .align 5 Louter: -___ -$code.=<<___ if ($SIZE_T==8); - ldx $t3,$bp,$i ; bp[i] -___ -$code.=<<___ if ($SIZE_T==4); - add $t0,$bp,$i - lwz $t3,0($t0) ; bp[i,i+1] - lwz $t0,4($t0) - insrdi $t3,$t0,32,0 -___ -$code.=<<___; - ld $t6,`$FRAME+$TRANSFER+8`($sp) ; tp[0] - mulld $t7,$a0,$t3 ; ap[0]*bp[i] - addi $tp,$sp,`$FRAME+$TRANSFER` - add $t7,$t7,$t6 ; ap[0]*bp[i]+tp[0] li $carry,0 - mulld $t7,$t7,$n0 ; tp[0]*n0 mtctr $j +___ +$code.=<<___ if ($SIZE_T==8); + ldx $t3,$bp,$i ; bp[i] + ld $t6,`$FRAME+$TRANSFER+8`($sp) ; tp[0] + mulld $t7,$a0,$t3 ; ap[0]*bp[i] + add $t7,$t7,$t6 ; ap[0]*bp[i]+tp[0] ; transfer bp[i] to FPU as 4x16-bit values extrdi $t0,$t3,16,48 extrdi $t1,$t3,16,32 @@ -647,6 +904,8 @@ $code.=<<___; std $t1,`$FRAME+8`($sp) std $t2,`$FRAME+16`($sp) std $t3,`$FRAME+24`($sp) + + mulld $t7,$t7,$n0 ; tp[0]*n0 ; transfer (ap[0]*bp[i]+tp[0])*n0 to FPU as 4x16-bit values extrdi $t4,$t7,16,48 extrdi $t5,$t7,16,32 @@ -656,7 +915,50 @@ $code.=<<___; std $t5,`$FRAME+40`($sp) std $t6,`$FRAME+48`($sp) std $t7,`$FRAME+56`($sp) +___ +$code.=<<___ if ($SIZE_T==4); + add $t0,$bp,$i + li $c1,0 + lwz $t1,0($t0) ; bp[i,i+1] + lwz $t3,4($t0) + + mullw $t4,$a0,$t1 ; ap[0]*bp[i] + lwz $t0,`$FRAME+$TRANSFER+8+4`($sp) ; tp[0] + mulhwu $t5,$a0,$t1 + lwz $t2,`$FRAME+$TRANSFER+8`($sp) ; tp[0] + mullw $t6,$a1,$t1 + mullw $t7,$a0,$t3 + add $t5,$t5,$t6 + add $t5,$t5,$t7 + addc $t4,$t4,$t0 ; ap[0]*bp[i]+tp[0] + adde $t5,$t5,$t2 + ; transfer bp[i] to FPU as 4x16-bit values + extrwi $t0,$t1,16,16 + extrwi $t1,$t1,16,0 + extrwi $t2,$t3,16,16 + extrwi $t3,$t3,16,0 + std $t0,`$FRAME+0`($sp) ; yes, std in 32-bit build + std $t1,`$FRAME+8`($sp) + std $t2,`$FRAME+16`($sp) + std $t3,`$FRAME+24`($sp) + mullw $t0,$t4,$n0 ; mulld tp[0]*n0 + mulhwu $t1,$t4,$n0 + mullw $t2,$t5,$n0 + mullw $t3,$t4,$n1 + add $t1,$t1,$t2 + add $t1,$t1,$t3 + ; transfer (ap[0]*bp[i]+tp[0])*n0 to FPU as 4x16-bit values + extrwi $t4,$t0,16,16 + extrwi $t5,$t0,16,0 + extrwi $t6,$t1,16,16 + extrwi $t7,$t1,16,0 + std $t4,`$FRAME+32`($sp) ; yes, std in 32-bit build + std $t5,`$FRAME+40`($sp) + std $t6,`$FRAME+48`($sp) + std $t7,`$FRAME+56`($sp) +___ +$code.=<<___; lfd $A0,8($nap_d) ; load a[j] in double format lfd $A1,16($nap_d) lfd $A2,24($nap_d) ; load a[j+1] in double format @@ -769,7 +1071,9 @@ Linner: fmul $dotb,$A3,$bd lfd $A2,24($nap_d) ; load a[j+1] in double format lfd $A3,32($nap_d) - +___ +if ($SIZE_T==8 or $flavour =~ /osx/) { +$code.=<<___; fmadd $T1a,$N1,$na,$T1a fmadd $T1b,$N1,$nb,$T1b ld $t0,`$FRAME+0`($sp) @@ -856,10 +1160,131 @@ $code.=<<___; addze $carry,$carry std $t3,-16($tp) ; tp[j-1] std $t5,-8($tp) ; tp[j] +___ +} else { +$code.=<<___; + fmadd $T1a,$N1,$na,$T1a + fmadd $T1b,$N1,$nb,$T1b + lwz $t1,`$FRAME+0`($sp) + lwz $t0,`$FRAME+4`($sp) + fmadd $T2a,$N2,$na,$T2a + fmadd $T2b,$N2,$nb,$T2b + lwz $t3,`$FRAME+8`($sp) + lwz $t2,`$FRAME+12`($sp) + fmadd $T3a,$N3,$na,$T3a + fmadd $T3b,$N3,$nb,$T3b + lwz $t5,`$FRAME+16`($sp) + lwz $t4,`$FRAME+20`($sp) + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + fmadd $T0a,$N0,$na,$T0a + fmadd $T0b,$N0,$nb,$T0b + lwz $t7,`$FRAME+24`($sp) + lwz $t6,`$FRAME+28`($sp) + srwi $c1,$t1,16 + insrwi $carry,$t1,16,0 + + fmadd $T1a,$N0,$nc,$T1a + fmadd $T1b,$N0,$nd,$T1b + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + fmadd $T2a,$N1,$nc,$T2a + fmadd $T2b,$N1,$nd,$T2b + insrwi $t0,$t2,16,0 ; 0..31 bits + srwi $c1,$t3,16 + insrwi $carry,$t3,16,0 + fmadd $T3a,$N2,$nc,$T3a + fmadd $T3b,$N2,$nd,$T3b + lwz $t2,12($tp) ; tp[j] + lwz $t3,8($tp) + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + fmadd $dota,$N3,$nc,$dota + fmadd $dotb,$N3,$nd,$dotb + srwi $c1,$t5,16 + insrwi $carry,$t5,16,0 + + fctid $T0a,$T0a + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + fctid $T0b,$T0b + insrwi $t4,$t6,16,0 ; 32..63 bits + srwi $c1,$t7,16 + insrwi $carry,$t7,16,0 + fctid $T1a,$T1a + addc $t0,$t0,$t2 + adde $t4,$t4,$t3 + lwz $t3,`$FRAME+32`($sp) ; permuted $t1 + lwz $t2,`$FRAME+36`($sp) ; permuted $t0 + fctid $T1b,$T1b + addze $carry,$carry + addze $c1,$c1 + stw $t0,4($tp) ; tp[j-1] + stw $t4,0($tp) + fctid $T2a,$T2a + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + lwz $t7,`$FRAME+40`($sp) ; permuted $t3 + lwz $t6,`$FRAME+44`($sp) ; permuted $t2 + fctid $T2b,$T2b + srwi $c1,$t3,16 + insrwi $carry,$t3,16,0 + lwz $t1,`$FRAME+48`($sp) ; permuted $t5 + lwz $t0,`$FRAME+52`($sp) ; permuted $t4 + fctid $T3a,$T3a + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + lwz $t5,`$FRAME+56`($sp) ; permuted $t7 + lwz $t4,`$FRAME+60`($sp) ; permuted $t6 + fctid $T3b,$T3b + + insrwi $t2,$t6,16,0 ; 64..95 bits + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + lwz $t6,20($tp) + lwzu $t7,16($tp) + addc $t0,$t0,$carry + stfd $T0a,`$FRAME+0`($sp) + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + stfd $T0b,`$FRAME+8`($sp) + insrwi $carry,$t1,16,0 + srwi $c1,$t1,16 + addc $t4,$t4,$carry + stfd $T1a,`$FRAME+16`($sp) + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + insrwi $t0,$t4,16,0 ; 96..127 bits + stfd $T1b,`$FRAME+24`($sp) + insrwi $carry,$t5,16,0 + srwi $c1,$t5,16 + + addc $t2,$t2,$t6 + stfd $T2a,`$FRAME+32`($sp) + adde $t0,$t0,$t7 + stfd $T2b,`$FRAME+40`($sp) + addze $carry,$carry + stfd $T3a,`$FRAME+48`($sp) + addze $c1,$c1 + stfd $T3b,`$FRAME+56`($sp) + stw $t2,-4($tp) ; tp[j] + stw $t0,-8($tp) +___ +} +$code.=<<___; bdnz- Linner fctid $dota,$dota fctid $dotb,$dotb +___ +if ($SIZE_T==8 or $flavour =~ /osx/) { +$code.=<<___; ld $t0,`$FRAME+0`($sp) ld $t1,`$FRAME+8`($sp) ld $t2,`$FRAME+16`($sp) @@ -926,7 +1351,116 @@ $code.=<<___; insrdi $t6,$t7,48,0 srdi $ovf,$t7,48 std $t6,0($tp) ; tp[num-1] +___ +} else { +$code.=<<___; + lwz $t1,`$FRAME+0`($sp) + lwz $t0,`$FRAME+4`($sp) + lwz $t3,`$FRAME+8`($sp) + lwz $t2,`$FRAME+12`($sp) + lwz $t5,`$FRAME+16`($sp) + lwz $t4,`$FRAME+20`($sp) + lwz $t7,`$FRAME+24`($sp) + lwz $t6,`$FRAME+28`($sp) + stfd $dota,`$FRAME+64`($sp) + stfd $dotb,`$FRAME+72`($sp) + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + insrwi $carry,$t1,16,0 + srwi $c1,$t1,16 + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + insrwi $t0,$t2,16,0 ; 0..31 bits + lwz $t2,12($tp) ; tp[j] + insrwi $carry,$t3,16,0 + srwi $c1,$t3,16 + lwz $t3,8($tp) + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + insrwi $carry,$t5,16,0 + srwi $c1,$t5,16 + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + insrwi $t4,$t6,16,0 ; 32..63 bits + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + + addc $t0,$t0,$t2 + adde $t4,$t4,$t3 + addze $carry,$carry + addze $c1,$c1 + stw $t0,4($tp) ; tp[j-1] + stw $t4,0($tp) + + lwz $t3,`$FRAME+32`($sp) ; permuted $t1 + lwz $t2,`$FRAME+36`($sp) ; permuted $t0 + lwz $t7,`$FRAME+40`($sp) ; permuted $t3 + lwz $t6,`$FRAME+44`($sp) ; permuted $t2 + lwz $t1,`$FRAME+48`($sp) ; permuted $t5 + lwz $t0,`$FRAME+52`($sp) ; permuted $t4 + lwz $t5,`$FRAME+56`($sp) ; permuted $t7 + lwz $t4,`$FRAME+60`($sp) ; permuted $t6 + + addc $t2,$t2,$carry + adde $t3,$t3,$c1 + srwi $carry,$t2,16 + insrwi $carry,$t3,16,0 + srwi $c1,$t3,16 + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + srwi $carry,$t6,16 + insrwi $t2,$t6,16,0 ; 64..95 bits + lwz $t6,20($tp) + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + lwzu $t7,16($tp) + addc $t0,$t0,$carry + adde $t1,$t1,$c1 + srwi $carry,$t0,16 + insrwi $carry,$t1,16,0 + srwi $c1,$t1,16 + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + srwi $carry,$t4,16 + insrwi $t0,$t4,16,0 ; 96..127 bits + insrwi $carry,$t5,16,0 + srwi $c1,$t5,16 + + addc $t2,$t2,$t6 + adde $t0,$t0,$t7 + lwz $t7,`$FRAME+64`($sp) + lwz $t6,`$FRAME+68`($sp) + addze $carry,$carry + addze $c1,$c1 + lwz $t5,`$FRAME+72`($sp) + lwz $t4,`$FRAME+76`($sp) + + addc $t6,$t6,$carry + adde $t7,$t7,$c1 + stw $t2,-4($tp) ; tp[j] + stw $t0,-8($tp) + addc $t6,$t6,$ovf + addze $t7,$t7 + srwi $carry,$t6,16 + insrwi $carry,$t7,16,0 + srwi $c1,$t7,16 + addc $t4,$t4,$carry + adde $t5,$t5,$c1 + + insrwi $t6,$t4,16,0 + srwi $t4,$t4,16 + insrwi $t4,$t5,16,0 + srwi $ovf,$t5,16 + stw $t6,4($tp) ; tp[num-1] + stw $t4,0($tp) +___ +} +$code.=<<___; slwi $t7,$num,2 addi $i,$i,8 subf $nap_d,$t7,$nap_d ; rewind pointer @@ -994,14 +1528,14 @@ $code.=<<___ if ($SIZE_T==4); mtctr $j .align 4 -Lsub: ld $t0,8($tp) ; load tp[j..j+3] in 64-bit word order - ldu $t2,16($tp) +Lsub: lwz $t0,12($tp) ; load tp[j..j+3] in 64-bit word order + lwz $t1,8($tp) + lwz $t2,20($tp) + lwzu $t3,16($tp) lwz $t4,4($np) ; load np[j..j+3] in 32-bit word order lwz $t5,8($np) lwz $t6,12($np) lwzu $t7,16($np) - extrdi $t1,$t0,32,0 - extrdi $t3,$t2,32,0 subfe $t4,$t4,$t0 ; tp[j]-np[j] stw $t0,4($ap) ; save tp[j..j+3] in 32-bit word order subfe $t5,$t5,$t1 ; tp[j+1]-np[j+1] @@ -1052,6 +1586,9 @@ ___ $code.=<<___; $POP $i,0($sp) li r3,1 ; signal "handled" + $POP r19,`-12*8-13*$SIZE_T`($i) + $POP r20,`-12*8-12*$SIZE_T`($i) + $POP r21,`-12*8-11*$SIZE_T`($i) $POP r22,`-12*8-10*$SIZE_T`($i) $POP r23,`-12*8-9*$SIZE_T`($i) $POP r24,`-12*8-8*$SIZE_T`($i) @@ -1077,8 +1614,9 @@ $code.=<<___; mr $sp,$i blr .long 0 - .byte 0,12,4,0,0x8c,10,6,0 + .byte 0,12,4,0,0x8c,13,6,0 .long 0 +.size .$fname,.-.$fname .asciz "Montgomery Multiplication for PPC64, CRYPTOGAMS by <appro\@openssl.org>" ___ diff --git a/openssl/crypto/bn/asm/rsaz-avx2.pl b/openssl/crypto/bn/asm/rsaz-avx2.pl new file mode 100755 index 000000000..3b6ccf83d --- /dev/null +++ b/openssl/crypto/bn/asm/rsaz-avx2.pl @@ -0,0 +1,1898 @@ +#!/usr/bin/env perl + +############################################################################## +# # +# Copyright (c) 2012, Intel Corporation # +# # +# All rights reserved. # +# # +# Redistribution and use in source and binary forms, with or without # +# modification, are permitted provided that the following conditions are # +# met: # +# # +# * Redistributions of source code must retain the above copyright # +# notice, this list of conditions and the following disclaimer. # +# # +# * Redistributions in binary form must reproduce the above copyright # +# notice, this list of conditions and the following disclaimer in the # +# documentation and/or other materials provided with the # +# distribution. # +# # +# * Neither the name of the Intel Corporation nor the names of its # +# contributors may be used to endorse or promote products derived from # +# this software without specific prior written permission. # +# # +# # +# THIS SOFTWARE IS PROVIDED BY INTEL CORPORATION ""AS IS"" AND ANY # +# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # +# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR # +# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL CORPORATION OR # +# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, # +# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, # +# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR # +# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF # +# LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING # +# NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS # +# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. # +# # +############################################################################## +# Developers and authors: # +# Shay Gueron (1, 2), and Vlad Krasnov (1) # +# (1) Intel Corporation, Israel Development Center, Haifa, Israel # +# (2) University of Haifa, Israel # +############################################################################## +# Reference: # +# [1] S. Gueron, V. Krasnov: "Software Implementation of Modular # +# Exponentiation, Using Advanced Vector Instructions Architectures", # +# F. Ozbudak and F. Rodriguez-Henriquez (Eds.): WAIFI 2012, LNCS 7369, # +# pp. 119?135, 2012. Springer-Verlag Berlin Heidelberg 2012 # +# [2] S. Gueron: "Efficient Software Implementations of Modular # +# Exponentiation", Journal of Cryptographic Engineering 2:31-43 (2012). # +# [3] S. Gueron, V. Krasnov: "Speeding up Big-numbers Squaring",IEEE # +# Proceedings of 9th International Conference on Information Technology: # +# New Generations (ITNG 2012), pp.821-823 (2012) # +# [4] S. Gueron, V. Krasnov: "[PATCH] Efficient and side channel analysis # +# resistant 1024-bit modular exponentiation, for optimizing RSA2048 # +# on AVX2 capable x86_64 platforms", # +# http://rt.openssl.org/Ticket/Display.html?id=2850&user=guest&pass=guest# +############################################################################## +# +# +13% improvement over original submission by <appro@openssl.org> +# +# rsa2048 sign/sec OpenSSL 1.0.1 scalar(*) this +# 2.3GHz Haswell 621 765/+23% 1113/+79% +# 2.3GHz Broadwell(**) 688 1200(***)/+74% 1120/+63% +# +# (*) if system doesn't support AVX2, for reference purposes; +# (**) scaled to 2.3GHz to simplify comparison; +# (***) scalar AD*X code is faster than AVX2 and is preferred code +# path for Broadwell; + +$flavour = shift; +$output = shift; +if ($flavour =~ /\./) { $output = $flavour; undef $flavour; } + +$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/); + +$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; +( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or +( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or +die "can't locate x86_64-xlate.pl"; + +if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1` + =~ /GNU assembler version ([2-9]\.[0-9]+)/) { + $avx = ($1>=2.19) + ($1>=2.22); + $addx = ($1>=2.23); +} + +if (!$avx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) && + `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) { + $avx = ($1>=2.09) + ($1>=2.10); + $addx = ($1>=2.10); +} + +if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) && + `ml64 2>&1` =~ /Version ([0-9]+)\./) { + $avx = ($1>=10) + ($1>=11); + $addx = ($1>=11); +} + +if (!$avx && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) { + my $ver = $2 + $3/100.0; # 3.1->3.01, 3.10->3.10 + $avx = ($ver>=3.0) + ($ver>=3.01); + $addx = ($ver>=3.03); +} + +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT = *OUT; + +if ($avx>1) {{{ +{ # void AMS_WW( +my $rp="%rdi"; # BN_ULONG *rp, +my $ap="%rsi"; # const BN_ULONG *ap, +my $np="%rdx"; # const BN_ULONG *np, +my $n0="%ecx"; # const BN_ULONG n0, +my $rep="%r8d"; # int repeat); + +# The registers that hold the accumulated redundant result +# The AMM works on 1024 bit operands, and redundant word size is 29 +# Therefore: ceil(1024/29)/4 = 9 +my $ACC0="%ymm0"; +my $ACC1="%ymm1"; +my $ACC2="%ymm2"; +my $ACC3="%ymm3"; +my $ACC4="%ymm4"; +my $ACC5="%ymm5"; +my $ACC6="%ymm6"; +my $ACC7="%ymm7"; +my $ACC8="%ymm8"; +my $ACC9="%ymm9"; +# Registers that hold the broadcasted words of bp, currently used +my $B1="%ymm10"; +my $B2="%ymm11"; +# Registers that hold the broadcasted words of Y, currently used +my $Y1="%ymm12"; +my $Y2="%ymm13"; +# Helper registers +my $TEMP1="%ymm14"; +my $AND_MASK="%ymm15"; +# alu registers that hold the first words of the ACC +my $r0="%r9"; +my $r1="%r10"; +my $r2="%r11"; +my $r3="%r12"; + +my $i="%r14d"; # loop counter +my $tmp = "%r15"; + +my $FrameSize=32*18+32*8; # place for A^2 and 2*A + +my $aap=$r0; +my $tp0="%rbx"; +my $tp1=$r3; +my $tpa=$tmp; + +$np="%r13"; # reassigned argument + +$code.=<<___; +.text + +.globl rsaz_1024_sqr_avx2 +.type rsaz_1024_sqr_avx2,\@function,5 +.align 64 +rsaz_1024_sqr_avx2: # 702 cycles, 14% faster than rsaz_1024_mul_avx2 + lea (%rsp), %rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + vzeroupper +___ +$code.=<<___ if ($win64); + lea -0xa8(%rsp),%rsp + vmovaps %xmm6,-0xd8(%rax) + vmovaps %xmm7,-0xc8(%rax) + vmovaps %xmm8,-0xb8(%rax) + vmovaps %xmm9,-0xa8(%rax) + vmovaps %xmm10,-0x98(%rax) + vmovaps %xmm11,-0x88(%rax) + vmovaps %xmm12,-0x78(%rax) + vmovaps %xmm13,-0x68(%rax) + vmovaps %xmm14,-0x58(%rax) + vmovaps %xmm15,-0x48(%rax) +.Lsqr_1024_body: +___ +$code.=<<___; + mov %rax,%rbp + mov %rdx, $np # reassigned argument + sub \$$FrameSize, %rsp + mov $np, $tmp + sub \$-128, $rp # size optimization + sub \$-128, $ap + sub \$-128, $np + + and \$4095, $tmp # see if $np crosses page + add \$32*10, $tmp + shr \$12, $tmp + vpxor $ACC9,$ACC9,$ACC9 + jz .Lsqr_1024_no_n_copy + + # unaligned 256-bit load that crosses page boundary can + # cause >2x performance degradation here, so if $np does + # cross page boundary, copy it to stack and make sure stack + # frame doesn't... + sub \$32*10,%rsp + vmovdqu 32*0-128($np), $ACC0 + and \$-2048, %rsp + vmovdqu 32*1-128($np), $ACC1 + vmovdqu 32*2-128($np), $ACC2 + vmovdqu 32*3-128($np), $ACC3 + vmovdqu 32*4-128($np), $ACC4 + vmovdqu 32*5-128($np), $ACC5 + vmovdqu 32*6-128($np), $ACC6 + vmovdqu 32*7-128($np), $ACC7 + vmovdqu 32*8-128($np), $ACC8 + lea $FrameSize+128(%rsp),$np + vmovdqu $ACC0, 32*0-128($np) + vmovdqu $ACC1, 32*1-128($np) + vmovdqu $ACC2, 32*2-128($np) + vmovdqu $ACC3, 32*3-128($np) + vmovdqu $ACC4, 32*4-128($np) + vmovdqu $ACC5, 32*5-128($np) + vmovdqu $ACC6, 32*6-128($np) + vmovdqu $ACC7, 32*7-128($np) + vmovdqu $ACC8, 32*8-128($np) + vmovdqu $ACC9, 32*9-128($np) # $ACC9 is zero + +.Lsqr_1024_no_n_copy: + and \$-1024, %rsp + + vmovdqu 32*1-128($ap), $ACC1 + vmovdqu 32*2-128($ap), $ACC2 + vmovdqu 32*3-128($ap), $ACC3 + vmovdqu 32*4-128($ap), $ACC4 + vmovdqu 32*5-128($ap), $ACC5 + vmovdqu 32*6-128($ap), $ACC6 + vmovdqu 32*7-128($ap), $ACC7 + vmovdqu 32*8-128($ap), $ACC8 + + lea 192(%rsp), $tp0 # 64+128=192 + vpbroadcastq .Land_mask(%rip), $AND_MASK + jmp .LOOP_GRANDE_SQR_1024 + +.align 32 +.LOOP_GRANDE_SQR_1024: + lea 32*18+128(%rsp), $aap # size optimization + lea 448(%rsp), $tp1 # 64+128+256=448 + + # the squaring is performed as described in Variant B of + # "Speeding up Big-Number Squaring", so start by calculating + # the A*2=A+A vector + vpaddq $ACC1, $ACC1, $ACC1 + vpbroadcastq 32*0-128($ap), $B1 + vpaddq $ACC2, $ACC2, $ACC2 + vmovdqa $ACC1, 32*0-128($aap) + vpaddq $ACC3, $ACC3, $ACC3 + vmovdqa $ACC2, 32*1-128($aap) + vpaddq $ACC4, $ACC4, $ACC4 + vmovdqa $ACC3, 32*2-128($aap) + vpaddq $ACC5, $ACC5, $ACC5 + vmovdqa $ACC4, 32*3-128($aap) + vpaddq $ACC6, $ACC6, $ACC6 + vmovdqa $ACC5, 32*4-128($aap) + vpaddq $ACC7, $ACC7, $ACC7 + vmovdqa $ACC6, 32*5-128($aap) + vpaddq $ACC8, $ACC8, $ACC8 + vmovdqa $ACC7, 32*6-128($aap) + vpxor $ACC9, $ACC9, $ACC9 + vmovdqa $ACC8, 32*7-128($aap) + + vpmuludq 32*0-128($ap), $B1, $ACC0 + vpbroadcastq 32*1-128($ap), $B2 + vmovdqu $ACC9, 32*9-192($tp0) # zero upper half + vpmuludq $B1, $ACC1, $ACC1 + vmovdqu $ACC9, 32*10-448($tp1) + vpmuludq $B1, $ACC2, $ACC2 + vmovdqu $ACC9, 32*11-448($tp1) + vpmuludq $B1, $ACC3, $ACC3 + vmovdqu $ACC9, 32*12-448($tp1) + vpmuludq $B1, $ACC4, $ACC4 + vmovdqu $ACC9, 32*13-448($tp1) + vpmuludq $B1, $ACC5, $ACC5 + vmovdqu $ACC9, 32*14-448($tp1) + vpmuludq $B1, $ACC6, $ACC6 + vmovdqu $ACC9, 32*15-448($tp1) + vpmuludq $B1, $ACC7, $ACC7 + vmovdqu $ACC9, 32*16-448($tp1) + vpmuludq $B1, $ACC8, $ACC8 + vpbroadcastq 32*2-128($ap), $B1 + vmovdqu $ACC9, 32*17-448($tp1) + + mov $ap, $tpa + mov \$4, $i + jmp .Lsqr_entry_1024 +___ +$TEMP0=$Y1; +$TEMP2=$Y2; +$code.=<<___; +.align 32 +.LOOP_SQR_1024: + vpbroadcastq 32*1-128($tpa), $B2 + vpmuludq 32*0-128($ap), $B1, $ACC0 + vpaddq 32*0-192($tp0), $ACC0, $ACC0 + vpmuludq 32*0-128($aap), $B1, $ACC1 + vpaddq 32*1-192($tp0), $ACC1, $ACC1 + vpmuludq 32*1-128($aap), $B1, $ACC2 + vpaddq 32*2-192($tp0), $ACC2, $ACC2 + vpmuludq 32*2-128($aap), $B1, $ACC3 + vpaddq 32*3-192($tp0), $ACC3, $ACC3 + vpmuludq 32*3-128($aap), $B1, $ACC4 + vpaddq 32*4-192($tp0), $ACC4, $ACC4 + vpmuludq 32*4-128($aap), $B1, $ACC5 + vpaddq 32*5-192($tp0), $ACC5, $ACC5 + vpmuludq 32*5-128($aap), $B1, $ACC6 + vpaddq 32*6-192($tp0), $ACC6, $ACC6 + vpmuludq 32*6-128($aap), $B1, $ACC7 + vpaddq 32*7-192($tp0), $ACC7, $ACC7 + vpmuludq 32*7-128($aap), $B1, $ACC8 + vpbroadcastq 32*2-128($tpa), $B1 + vpaddq 32*8-192($tp0), $ACC8, $ACC8 +.Lsqr_entry_1024: + vmovdqu $ACC0, 32*0-192($tp0) + vmovdqu $ACC1, 32*1-192($tp0) + + vpmuludq 32*1-128($ap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC2, $ACC2 + vpmuludq 32*1-128($aap), $B2, $TEMP1 + vpaddq $TEMP1, $ACC3, $ACC3 + vpmuludq 32*2-128($aap), $B2, $TEMP2 + vpaddq $TEMP2, $ACC4, $ACC4 + vpmuludq 32*3-128($aap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC5, $ACC5 + vpmuludq 32*4-128($aap), $B2, $TEMP1 + vpaddq $TEMP1, $ACC6, $ACC6 + vpmuludq 32*5-128($aap), $B2, $TEMP2 + vpaddq $TEMP2, $ACC7, $ACC7 + vpmuludq 32*6-128($aap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC8, $ACC8 + vpmuludq 32*7-128($aap), $B2, $ACC0 + vpbroadcastq 32*3-128($tpa), $B2 + vpaddq 32*9-192($tp0), $ACC0, $ACC0 + + vmovdqu $ACC2, 32*2-192($tp0) + vmovdqu $ACC3, 32*3-192($tp0) + + vpmuludq 32*2-128($ap), $B1, $TEMP2 + vpaddq $TEMP2, $ACC4, $ACC4 + vpmuludq 32*2-128($aap), $B1, $TEMP0 + vpaddq $TEMP0, $ACC5, $ACC5 + vpmuludq 32*3-128($aap), $B1, $TEMP1 + vpaddq $TEMP1, $ACC6, $ACC6 + vpmuludq 32*4-128($aap), $B1, $TEMP2 + vpaddq $TEMP2, $ACC7, $ACC7 + vpmuludq 32*5-128($aap), $B1, $TEMP0 + vpaddq $TEMP0, $ACC8, $ACC8 + vpmuludq 32*6-128($aap), $B1, $TEMP1 + vpaddq $TEMP1, $ACC0, $ACC0 + vpmuludq 32*7-128($aap), $B1, $ACC1 + vpbroadcastq 32*4-128($tpa), $B1 + vpaddq 32*10-448($tp1), $ACC1, $ACC1 + + vmovdqu $ACC4, 32*4-192($tp0) + vmovdqu $ACC5, 32*5-192($tp0) + + vpmuludq 32*3-128($ap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC6, $ACC6 + vpmuludq 32*3-128($aap), $B2, $TEMP1 + vpaddq $TEMP1, $ACC7, $ACC7 + vpmuludq 32*4-128($aap), $B2, $TEMP2 + vpaddq $TEMP2, $ACC8, $ACC8 + vpmuludq 32*5-128($aap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC0, $ACC0 + vpmuludq 32*6-128($aap), $B2, $TEMP1 + vpaddq $TEMP1, $ACC1, $ACC1 + vpmuludq 32*7-128($aap), $B2, $ACC2 + vpbroadcastq 32*5-128($tpa), $B2 + vpaddq 32*11-448($tp1), $ACC2, $ACC2 + + vmovdqu $ACC6, 32*6-192($tp0) + vmovdqu $ACC7, 32*7-192($tp0) + + vpmuludq 32*4-128($ap), $B1, $TEMP0 + vpaddq $TEMP0, $ACC8, $ACC8 + vpmuludq 32*4-128($aap), $B1, $TEMP1 + vpaddq $TEMP1, $ACC0, $ACC0 + vpmuludq 32*5-128($aap), $B1, $TEMP2 + vpaddq $TEMP2, $ACC1, $ACC1 + vpmuludq 32*6-128($aap), $B1, $TEMP0 + vpaddq $TEMP0, $ACC2, $ACC2 + vpmuludq 32*7-128($aap), $B1, $ACC3 + vpbroadcastq 32*6-128($tpa), $B1 + vpaddq 32*12-448($tp1), $ACC3, $ACC3 + + vmovdqu $ACC8, 32*8-192($tp0) + vmovdqu $ACC0, 32*9-192($tp0) + lea 8($tp0), $tp0 + + vpmuludq 32*5-128($ap), $B2, $TEMP2 + vpaddq $TEMP2, $ACC1, $ACC1 + vpmuludq 32*5-128($aap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC2, $ACC2 + vpmuludq 32*6-128($aap), $B2, $TEMP1 + vpaddq $TEMP1, $ACC3, $ACC3 + vpmuludq 32*7-128($aap), $B2, $ACC4 + vpbroadcastq 32*7-128($tpa), $B2 + vpaddq 32*13-448($tp1), $ACC4, $ACC4 + + vmovdqu $ACC1, 32*10-448($tp1) + vmovdqu $ACC2, 32*11-448($tp1) + + vpmuludq 32*6-128($ap), $B1, $TEMP0 + vpaddq $TEMP0, $ACC3, $ACC3 + vpmuludq 32*6-128($aap), $B1, $TEMP1 + vpbroadcastq 32*8-128($tpa), $ACC0 # borrow $ACC0 for $B1 + vpaddq $TEMP1, $ACC4, $ACC4 + vpmuludq 32*7-128($aap), $B1, $ACC5 + vpbroadcastq 32*0+8-128($tpa), $B1 # for next iteration + vpaddq 32*14-448($tp1), $ACC5, $ACC5 + + vmovdqu $ACC3, 32*12-448($tp1) + vmovdqu $ACC4, 32*13-448($tp1) + lea 8($tpa), $tpa + + vpmuludq 32*7-128($ap), $B2, $TEMP0 + vpaddq $TEMP0, $ACC5, $ACC5 + vpmuludq 32*7-128($aap), $B2, $ACC6 + vpaddq 32*15-448($tp1), $ACC6, $ACC6 + + vpmuludq 32*8-128($ap), $ACC0, $ACC7 + vmovdqu $ACC5, 32*14-448($tp1) + vpaddq 32*16-448($tp1), $ACC7, $ACC7 + vmovdqu $ACC6, 32*15-448($tp1) + vmovdqu $ACC7, 32*16-448($tp1) + lea 8($tp1), $tp1 + + dec $i + jnz .LOOP_SQR_1024 +___ +$ZERO = $ACC9; +$TEMP0 = $B1; +$TEMP2 = $B2; +$TEMP3 = $Y1; +$TEMP4 = $Y2; +$code.=<<___; + #we need to fix indexes 32-39 to avoid overflow + vmovdqu 32*8(%rsp), $ACC8 # 32*8-192($tp0), + vmovdqu 32*9(%rsp), $ACC1 # 32*9-192($tp0) + vmovdqu 32*10(%rsp), $ACC2 # 32*10-192($tp0) + lea 192(%rsp), $tp0 # 64+128=192 + + vpsrlq \$29, $ACC8, $TEMP1 + vpand $AND_MASK, $ACC8, $ACC8 + vpsrlq \$29, $ACC1, $TEMP2 + vpand $AND_MASK, $ACC1, $ACC1 + + vpermq \$0x93, $TEMP1, $TEMP1 + vpxor $ZERO, $ZERO, $ZERO + vpermq \$0x93, $TEMP2, $TEMP2 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC8, $ACC8 + vpblendd \$3, $TEMP2, $ZERO, $TEMP2 + vpaddq $TEMP1, $ACC1, $ACC1 + vpaddq $TEMP2, $ACC2, $ACC2 + vmovdqu $ACC1, 32*9-192($tp0) + vmovdqu $ACC2, 32*10-192($tp0) + + mov (%rsp), %rax + mov 8(%rsp), $r1 + mov 16(%rsp), $r2 + mov 24(%rsp), $r3 + vmovdqu 32*1(%rsp), $ACC1 + vmovdqu 32*2-192($tp0), $ACC2 + vmovdqu 32*3-192($tp0), $ACC3 + vmovdqu 32*4-192($tp0), $ACC4 + vmovdqu 32*5-192($tp0), $ACC5 + vmovdqu 32*6-192($tp0), $ACC6 + vmovdqu 32*7-192($tp0), $ACC7 + + mov %rax, $r0 + imull $n0, %eax + and \$0x1fffffff, %eax + vmovd %eax, $Y1 + + mov %rax, %rdx + imulq -128($np), %rax + vpbroadcastq $Y1, $Y1 + add %rax, $r0 + mov %rdx, %rax + imulq 8-128($np), %rax + shr \$29, $r0 + add %rax, $r1 + mov %rdx, %rax + imulq 16-128($np), %rax + add $r0, $r1 + add %rax, $r2 + imulq 24-128($np), %rdx + add %rdx, $r3 + + mov $r1, %rax + imull $n0, %eax + and \$0x1fffffff, %eax + + mov \$9, $i + jmp .LOOP_REDUCE_1024 + +.align 32 +.LOOP_REDUCE_1024: + vmovd %eax, $Y2 + vpbroadcastq $Y2, $Y2 + + vpmuludq 32*1-128($np), $Y1, $TEMP0 + mov %rax, %rdx + imulq -128($np), %rax + vpaddq $TEMP0, $ACC1, $ACC1 + add %rax, $r1 + vpmuludq 32*2-128($np), $Y1, $TEMP1 + mov %rdx, %rax + imulq 8-128($np), %rax + vpaddq $TEMP1, $ACC2, $ACC2 + vpmuludq 32*3-128($np), $Y1, $TEMP2 + .byte 0x67 + add %rax, $r2 + .byte 0x67 + mov %rdx, %rax + imulq 16-128($np), %rax + shr \$29, $r1 + vpaddq $TEMP2, $ACC3, $ACC3 + vpmuludq 32*4-128($np), $Y1, $TEMP0 + add %rax, $r3 + add $r1, $r2 + vpaddq $TEMP0, $ACC4, $ACC4 + vpmuludq 32*5-128($np), $Y1, $TEMP1 + mov $r2, %rax + imull $n0, %eax + vpaddq $TEMP1, $ACC5, $ACC5 + vpmuludq 32*6-128($np), $Y1, $TEMP2 + and \$0x1fffffff, %eax + vpaddq $TEMP2, $ACC6, $ACC6 + vpmuludq 32*7-128($np), $Y1, $TEMP0 + vpaddq $TEMP0, $ACC7, $ACC7 + vpmuludq 32*8-128($np), $Y1, $TEMP1 + vmovd %eax, $Y1 + #vmovdqu 32*1-8-128($np), $TEMP2 # moved below + vpaddq $TEMP1, $ACC8, $ACC8 + #vmovdqu 32*2-8-128($np), $TEMP0 # moved below + vpbroadcastq $Y1, $Y1 + + vpmuludq 32*1-8-128($np), $Y2, $TEMP2 # see above + vmovdqu 32*3-8-128($np), $TEMP1 + mov %rax, %rdx + imulq -128($np), %rax + vpaddq $TEMP2, $ACC1, $ACC1 + vpmuludq 32*2-8-128($np), $Y2, $TEMP0 # see above + vmovdqu 32*4-8-128($np), $TEMP2 + add %rax, $r2 + mov %rdx, %rax + imulq 8-128($np), %rax + vpaddq $TEMP0, $ACC2, $ACC2 + add $r3, %rax + shr \$29, $r2 + vpmuludq $Y2, $TEMP1, $TEMP1 + vmovdqu 32*5-8-128($np), $TEMP0 + add $r2, %rax + vpaddq $TEMP1, $ACC3, $ACC3 + vpmuludq $Y2, $TEMP2, $TEMP2 + vmovdqu 32*6-8-128($np), $TEMP1 + .byte 0x67 + mov %rax, $r3 + imull $n0, %eax + vpaddq $TEMP2, $ACC4, $ACC4 + vpmuludq $Y2, $TEMP0, $TEMP0 + .byte 0xc4,0x41,0x7e,0x6f,0x9d,0x58,0x00,0x00,0x00 # vmovdqu 32*7-8-128($np), $TEMP2 + and \$0x1fffffff, %eax + vpaddq $TEMP0, $ACC5, $ACC5 + vpmuludq $Y2, $TEMP1, $TEMP1 + vmovdqu 32*8-8-128($np), $TEMP0 + vpaddq $TEMP1, $ACC6, $ACC6 + vpmuludq $Y2, $TEMP2, $TEMP2 + vmovdqu 32*9-8-128($np), $ACC9 + vmovd %eax, $ACC0 # borrow ACC0 for Y2 + imulq -128($np), %rax + vpaddq $TEMP2, $ACC7, $ACC7 + vpmuludq $Y2, $TEMP0, $TEMP0 + vmovdqu 32*1-16-128($np), $TEMP1 + vpbroadcastq $ACC0, $ACC0 + vpaddq $TEMP0, $ACC8, $ACC8 + vpmuludq $Y2, $ACC9, $ACC9 + vmovdqu 32*2-16-128($np), $TEMP2 + add %rax, $r3 + +___ +($ACC0,$Y2)=($Y2,$ACC0); +$code.=<<___; + vmovdqu 32*1-24-128($np), $ACC0 + vpmuludq $Y1, $TEMP1, $TEMP1 + vmovdqu 32*3-16-128($np), $TEMP0 + vpaddq $TEMP1, $ACC1, $ACC1 + vpmuludq $Y2, $ACC0, $ACC0 + vpmuludq $Y1, $TEMP2, $TEMP2 + .byte 0xc4,0x41,0x7e,0x6f,0xb5,0xf0,0xff,0xff,0xff # vmovdqu 32*4-16-128($np), $TEMP1 + vpaddq $ACC1, $ACC0, $ACC0 + vpaddq $TEMP2, $ACC2, $ACC2 + vpmuludq $Y1, $TEMP0, $TEMP0 + vmovdqu 32*5-16-128($np), $TEMP2 + .byte 0x67 + vmovq $ACC0, %rax + vmovdqu $ACC0, (%rsp) # transfer $r0-$r3 + vpaddq $TEMP0, $ACC3, $ACC3 + vpmuludq $Y1, $TEMP1, $TEMP1 + vmovdqu 32*6-16-128($np), $TEMP0 + vpaddq $TEMP1, $ACC4, $ACC4 + vpmuludq $Y1, $TEMP2, $TEMP2 + vmovdqu 32*7-16-128($np), $TEMP1 + vpaddq $TEMP2, $ACC5, $ACC5 + vpmuludq $Y1, $TEMP0, $TEMP0 + vmovdqu 32*8-16-128($np), $TEMP2 + vpaddq $TEMP0, $ACC6, $ACC6 + vpmuludq $Y1, $TEMP1, $TEMP1 + shr \$29, $r3 + vmovdqu 32*9-16-128($np), $TEMP0 + add $r3, %rax + vpaddq $TEMP1, $ACC7, $ACC7 + vpmuludq $Y1, $TEMP2, $TEMP2 + #vmovdqu 32*2-24-128($np), $TEMP1 # moved below + mov %rax, $r0 + imull $n0, %eax + vpaddq $TEMP2, $ACC8, $ACC8 + vpmuludq $Y1, $TEMP0, $TEMP0 + and \$0x1fffffff, %eax + vmovd %eax, $Y1 + vmovdqu 32*3-24-128($np), $TEMP2 + .byte 0x67 + vpaddq $TEMP0, $ACC9, $ACC9 + vpbroadcastq $Y1, $Y1 + + vpmuludq 32*2-24-128($np), $Y2, $TEMP1 # see above + vmovdqu 32*4-24-128($np), $TEMP0 + mov %rax, %rdx + imulq -128($np), %rax + mov 8(%rsp), $r1 + vpaddq $TEMP1, $ACC2, $ACC1 + vpmuludq $Y2, $TEMP2, $TEMP2 + vmovdqu 32*5-24-128($np), $TEMP1 + add %rax, $r0 + mov %rdx, %rax + imulq 8-128($np), %rax + .byte 0x67 + shr \$29, $r0 + mov 16(%rsp), $r2 + vpaddq $TEMP2, $ACC3, $ACC2 + vpmuludq $Y2, $TEMP0, $TEMP0 + vmovdqu 32*6-24-128($np), $TEMP2 + add %rax, $r1 + mov %rdx, %rax + imulq 16-128($np), %rax + vpaddq $TEMP0, $ACC4, $ACC3 + vpmuludq $Y2, $TEMP1, $TEMP1 + vmovdqu 32*7-24-128($np), $TEMP0 + imulq 24-128($np), %rdx # future $r3 + add %rax, $r2 + lea ($r0,$r1), %rax + vpaddq $TEMP1, $ACC5, $ACC4 + vpmuludq $Y2, $TEMP2, $TEMP2 + vmovdqu 32*8-24-128($np), $TEMP1 + mov %rax, $r1 + imull $n0, %eax + vpmuludq $Y2, $TEMP0, $TEMP0 + vpaddq $TEMP2, $ACC6, $ACC5 + vmovdqu 32*9-24-128($np), $TEMP2 + and \$0x1fffffff, %eax + vpaddq $TEMP0, $ACC7, $ACC6 + vpmuludq $Y2, $TEMP1, $TEMP1 + add 24(%rsp), %rdx + vpaddq $TEMP1, $ACC8, $ACC7 + vpmuludq $Y2, $TEMP2, $TEMP2 + vpaddq $TEMP2, $ACC9, $ACC8 + vmovq $r3, $ACC9 + mov %rdx, $r3 + + dec $i + jnz .LOOP_REDUCE_1024 +___ +($ACC0,$Y2)=($Y2,$ACC0); +$code.=<<___; + lea 448(%rsp), $tp1 # size optimization + vpaddq $ACC9, $Y2, $ACC0 + vpxor $ZERO, $ZERO, $ZERO + + vpaddq 32*9-192($tp0), $ACC0, $ACC0 + vpaddq 32*10-448($tp1), $ACC1, $ACC1 + vpaddq 32*11-448($tp1), $ACC2, $ACC2 + vpaddq 32*12-448($tp1), $ACC3, $ACC3 + vpaddq 32*13-448($tp1), $ACC4, $ACC4 + vpaddq 32*14-448($tp1), $ACC5, $ACC5 + vpaddq 32*15-448($tp1), $ACC6, $ACC6 + vpaddq 32*16-448($tp1), $ACC7, $ACC7 + vpaddq 32*17-448($tp1), $ACC8, $ACC8 + + vpsrlq \$29, $ACC0, $TEMP1 + vpand $AND_MASK, $ACC0, $ACC0 + vpsrlq \$29, $ACC1, $TEMP2 + vpand $AND_MASK, $ACC1, $ACC1 + vpsrlq \$29, $ACC2, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC2, $ACC2 + vpsrlq \$29, $ACC3, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC3, $ACC3 + vpermq \$0x93, $TEMP3, $TEMP3 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP4, $TEMP4 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC0, $ACC0 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC1, $ACC1 + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC2, $ACC2 + vpblendd \$3, $TEMP4, $ZERO, $TEMP4 + vpaddq $TEMP3, $ACC3, $ACC3 + vpaddq $TEMP4, $ACC4, $ACC4 + + vpsrlq \$29, $ACC0, $TEMP1 + vpand $AND_MASK, $ACC0, $ACC0 + vpsrlq \$29, $ACC1, $TEMP2 + vpand $AND_MASK, $ACC1, $ACC1 + vpsrlq \$29, $ACC2, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC2, $ACC2 + vpsrlq \$29, $ACC3, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC3, $ACC3 + vpermq \$0x93, $TEMP3, $TEMP3 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP4, $TEMP4 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC0, $ACC0 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC1, $ACC1 + vmovdqu $ACC0, 32*0-128($rp) + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC2, $ACC2 + vmovdqu $ACC1, 32*1-128($rp) + vpblendd \$3, $TEMP4, $ZERO, $TEMP4 + vpaddq $TEMP3, $ACC3, $ACC3 + vmovdqu $ACC2, 32*2-128($rp) + vpaddq $TEMP4, $ACC4, $ACC4 + vmovdqu $ACC3, 32*3-128($rp) +___ +$TEMP5=$ACC0; +$code.=<<___; + vpsrlq \$29, $ACC4, $TEMP1 + vpand $AND_MASK, $ACC4, $ACC4 + vpsrlq \$29, $ACC5, $TEMP2 + vpand $AND_MASK, $ACC5, $ACC5 + vpsrlq \$29, $ACC6, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC6, $ACC6 + vpsrlq \$29, $ACC7, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC7, $ACC7 + vpsrlq \$29, $ACC8, $TEMP5 + vpermq \$0x93, $TEMP3, $TEMP3 + vpand $AND_MASK, $ACC8, $ACC8 + vpermq \$0x93, $TEMP4, $TEMP4 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP5, $TEMP5 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC4, $ACC4 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC5, $ACC5 + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC6, $ACC6 + vpblendd \$3, $TEMP4, $TEMP5, $TEMP4 + vpaddq $TEMP3, $ACC7, $ACC7 + vpaddq $TEMP4, $ACC8, $ACC8 + + vpsrlq \$29, $ACC4, $TEMP1 + vpand $AND_MASK, $ACC4, $ACC4 + vpsrlq \$29, $ACC5, $TEMP2 + vpand $AND_MASK, $ACC5, $ACC5 + vpsrlq \$29, $ACC6, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC6, $ACC6 + vpsrlq \$29, $ACC7, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC7, $ACC7 + vpsrlq \$29, $ACC8, $TEMP5 + vpermq \$0x93, $TEMP3, $TEMP3 + vpand $AND_MASK, $ACC8, $ACC8 + vpermq \$0x93, $TEMP4, $TEMP4 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP5, $TEMP5 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC4, $ACC4 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC5, $ACC5 + vmovdqu $ACC4, 32*4-128($rp) + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC6, $ACC6 + vmovdqu $ACC5, 32*5-128($rp) + vpblendd \$3, $TEMP4, $TEMP5, $TEMP4 + vpaddq $TEMP3, $ACC7, $ACC7 + vmovdqu $ACC6, 32*6-128($rp) + vpaddq $TEMP4, $ACC8, $ACC8 + vmovdqu $ACC7, 32*7-128($rp) + vmovdqu $ACC8, 32*8-128($rp) + + mov $rp, $ap + dec $rep + jne .LOOP_GRANDE_SQR_1024 + + vzeroall + mov %rbp, %rax +___ +$code.=<<___ if ($win64); + movaps -0xd8(%rax),%xmm6 + movaps -0xc8(%rax),%xmm7 + movaps -0xb8(%rax),%xmm8 + movaps -0xa8(%rax),%xmm9 + movaps -0x98(%rax),%xmm10 + movaps -0x88(%rax),%xmm11 + movaps -0x78(%rax),%xmm12 + movaps -0x68(%rax),%xmm13 + movaps -0x58(%rax),%xmm14 + movaps -0x48(%rax),%xmm15 +___ +$code.=<<___; + mov -48(%rax),%r15 + mov -40(%rax),%r14 + mov -32(%rax),%r13 + mov -24(%rax),%r12 + mov -16(%rax),%rbp + mov -8(%rax),%rbx + lea (%rax),%rsp # restore %rsp +.Lsqr_1024_epilogue: + ret +.size rsaz_1024_sqr_avx2,.-rsaz_1024_sqr_avx2 +___ +} + +{ # void AMM_WW( +my $rp="%rdi"; # BN_ULONG *rp, +my $ap="%rsi"; # const BN_ULONG *ap, +my $bp="%rdx"; # const BN_ULONG *bp, +my $np="%rcx"; # const BN_ULONG *np, +my $n0="%r8d"; # unsigned int n0); + +# The registers that hold the accumulated redundant result +# The AMM works on 1024 bit operands, and redundant word size is 29 +# Therefore: ceil(1024/29)/4 = 9 +my $ACC0="%ymm0"; +my $ACC1="%ymm1"; +my $ACC2="%ymm2"; +my $ACC3="%ymm3"; +my $ACC4="%ymm4"; +my $ACC5="%ymm5"; +my $ACC6="%ymm6"; +my $ACC7="%ymm7"; +my $ACC8="%ymm8"; +my $ACC9="%ymm9"; + +# Registers that hold the broadcasted words of multiplier, currently used +my $Bi="%ymm10"; +my $Yi="%ymm11"; + +# Helper registers +my $TEMP0=$ACC0; +my $TEMP1="%ymm12"; +my $TEMP2="%ymm13"; +my $ZERO="%ymm14"; +my $AND_MASK="%ymm15"; + +# alu registers that hold the first words of the ACC +my $r0="%r9"; +my $r1="%r10"; +my $r2="%r11"; +my $r3="%r12"; + +my $i="%r14d"; +my $tmp="%r15"; + +$bp="%r13"; # reassigned argument + +$code.=<<___; +.globl rsaz_1024_mul_avx2 +.type rsaz_1024_mul_avx2,\@function,5 +.align 64 +rsaz_1024_mul_avx2: + lea (%rsp), %rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 +___ +$code.=<<___ if ($win64); + vzeroupper + lea -0xa8(%rsp),%rsp + vmovaps %xmm6,-0xd8(%rax) + vmovaps %xmm7,-0xc8(%rax) + vmovaps %xmm8,-0xb8(%rax) + vmovaps %xmm9,-0xa8(%rax) + vmovaps %xmm10,-0x98(%rax) + vmovaps %xmm11,-0x88(%rax) + vmovaps %xmm12,-0x78(%rax) + vmovaps %xmm13,-0x68(%rax) + vmovaps %xmm14,-0x58(%rax) + vmovaps %xmm15,-0x48(%rax) +.Lmul_1024_body: +___ +$code.=<<___; + mov %rax,%rbp + vzeroall + mov %rdx, $bp # reassigned argument + sub \$64,%rsp + + # unaligned 256-bit load that crosses page boundary can + # cause severe performance degradation here, so if $ap does + # cross page boundary, swap it with $bp [meaning that caller + # is advised to lay down $ap and $bp next to each other, so + # that only one can cross page boundary]. + .byte 0x67,0x67 + mov $ap, $tmp + and \$4095, $tmp + add \$32*10, $tmp + shr \$12, $tmp + mov $ap, $tmp + cmovnz $bp, $ap + cmovnz $tmp, $bp + + mov $np, $tmp + sub \$-128,$ap # size optimization + sub \$-128,$np + sub \$-128,$rp + + and \$4095, $tmp # see if $np crosses page + add \$32*10, $tmp + .byte 0x67,0x67 + shr \$12, $tmp + jz .Lmul_1024_no_n_copy + + # unaligned 256-bit load that crosses page boundary can + # cause severe performance degradation here, so if $np does + # cross page boundary, copy it to stack and make sure stack + # frame doesn't... + sub \$32*10,%rsp + vmovdqu 32*0-128($np), $ACC0 + and \$-512, %rsp + vmovdqu 32*1-128($np), $ACC1 + vmovdqu 32*2-128($np), $ACC2 + vmovdqu 32*3-128($np), $ACC3 + vmovdqu 32*4-128($np), $ACC4 + vmovdqu 32*5-128($np), $ACC5 + vmovdqu 32*6-128($np), $ACC6 + vmovdqu 32*7-128($np), $ACC7 + vmovdqu 32*8-128($np), $ACC8 + lea 64+128(%rsp),$np + vmovdqu $ACC0, 32*0-128($np) + vpxor $ACC0, $ACC0, $ACC0 + vmovdqu $ACC1, 32*1-128($np) + vpxor $ACC1, $ACC1, $ACC1 + vmovdqu $ACC2, 32*2-128($np) + vpxor $ACC2, $ACC2, $ACC2 + vmovdqu $ACC3, 32*3-128($np) + vpxor $ACC3, $ACC3, $ACC3 + vmovdqu $ACC4, 32*4-128($np) + vpxor $ACC4, $ACC4, $ACC4 + vmovdqu $ACC5, 32*5-128($np) + vpxor $ACC5, $ACC5, $ACC5 + vmovdqu $ACC6, 32*6-128($np) + vpxor $ACC6, $ACC6, $ACC6 + vmovdqu $ACC7, 32*7-128($np) + vpxor $ACC7, $ACC7, $ACC7 + vmovdqu $ACC8, 32*8-128($np) + vmovdqa $ACC0, $ACC8 + vmovdqu $ACC9, 32*9-128($np) # $ACC9 is zero after vzeroall +.Lmul_1024_no_n_copy: + and \$-64,%rsp + + mov ($bp), %rbx + vpbroadcastq ($bp), $Bi + vmovdqu $ACC0, (%rsp) # clear top of stack + xor $r0, $r0 + .byte 0x67 + xor $r1, $r1 + xor $r2, $r2 + xor $r3, $r3 + + vmovdqu .Land_mask(%rip), $AND_MASK + mov \$9, $i + vmovdqu $ACC9, 32*9-128($rp) # $ACC9 is zero after vzeroall + jmp .Loop_mul_1024 + +.align 32 +.Loop_mul_1024: + vpsrlq \$29, $ACC3, $ACC9 # correct $ACC3(*) + mov %rbx, %rax + imulq -128($ap), %rax + add $r0, %rax + mov %rbx, $r1 + imulq 8-128($ap), $r1 + add 8(%rsp), $r1 + + mov %rax, $r0 + imull $n0, %eax + and \$0x1fffffff, %eax + + mov %rbx, $r2 + imulq 16-128($ap), $r2 + add 16(%rsp), $r2 + + mov %rbx, $r3 + imulq 24-128($ap), $r3 + add 24(%rsp), $r3 + vpmuludq 32*1-128($ap),$Bi,$TEMP0 + vmovd %eax, $Yi + vpaddq $TEMP0,$ACC1,$ACC1 + vpmuludq 32*2-128($ap),$Bi,$TEMP1 + vpbroadcastq $Yi, $Yi + vpaddq $TEMP1,$ACC2,$ACC2 + vpmuludq 32*3-128($ap),$Bi,$TEMP2 + vpand $AND_MASK, $ACC3, $ACC3 # correct $ACC3 + vpaddq $TEMP2,$ACC3,$ACC3 + vpmuludq 32*4-128($ap),$Bi,$TEMP0 + vpaddq $TEMP0,$ACC4,$ACC4 + vpmuludq 32*5-128($ap),$Bi,$TEMP1 + vpaddq $TEMP1,$ACC5,$ACC5 + vpmuludq 32*6-128($ap),$Bi,$TEMP2 + vpaddq $TEMP2,$ACC6,$ACC6 + vpmuludq 32*7-128($ap),$Bi,$TEMP0 + vpermq \$0x93, $ACC9, $ACC9 # correct $ACC3 + vpaddq $TEMP0,$ACC7,$ACC7 + vpmuludq 32*8-128($ap),$Bi,$TEMP1 + vpbroadcastq 8($bp), $Bi + vpaddq $TEMP1,$ACC8,$ACC8 + + mov %rax,%rdx + imulq -128($np),%rax + add %rax,$r0 + mov %rdx,%rax + imulq 8-128($np),%rax + add %rax,$r1 + mov %rdx,%rax + imulq 16-128($np),%rax + add %rax,$r2 + shr \$29, $r0 + imulq 24-128($np),%rdx + add %rdx,$r3 + add $r0, $r1 + + vpmuludq 32*1-128($np),$Yi,$TEMP2 + vmovq $Bi, %rbx + vpaddq $TEMP2,$ACC1,$ACC1 + vpmuludq 32*2-128($np),$Yi,$TEMP0 + vpaddq $TEMP0,$ACC2,$ACC2 + vpmuludq 32*3-128($np),$Yi,$TEMP1 + vpaddq $TEMP1,$ACC3,$ACC3 + vpmuludq 32*4-128($np),$Yi,$TEMP2 + vpaddq $TEMP2,$ACC4,$ACC4 + vpmuludq 32*5-128($np),$Yi,$TEMP0 + vpaddq $TEMP0,$ACC5,$ACC5 + vpmuludq 32*6-128($np),$Yi,$TEMP1 + vpaddq $TEMP1,$ACC6,$ACC6 + vpmuludq 32*7-128($np),$Yi,$TEMP2 + vpblendd \$3, $ZERO, $ACC9, $ACC9 # correct $ACC3 + vpaddq $TEMP2,$ACC7,$ACC7 + vpmuludq 32*8-128($np),$Yi,$TEMP0 + vpaddq $ACC9, $ACC3, $ACC3 # correct $ACC3 + vpaddq $TEMP0,$ACC8,$ACC8 + + mov %rbx, %rax + imulq -128($ap),%rax + add %rax,$r1 + vmovdqu -8+32*1-128($ap),$TEMP1 + mov %rbx, %rax + imulq 8-128($ap),%rax + add %rax,$r2 + vmovdqu -8+32*2-128($ap),$TEMP2 + + mov $r1, %rax + imull $n0, %eax + and \$0x1fffffff, %eax + + imulq 16-128($ap),%rbx + add %rbx,$r3 + vpmuludq $Bi,$TEMP1,$TEMP1 + vmovd %eax, $Yi + vmovdqu -8+32*3-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC1,$ACC1 + vpmuludq $Bi,$TEMP2,$TEMP2 + vpbroadcastq $Yi, $Yi + vmovdqu -8+32*4-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC2,$ACC2 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovdqu -8+32*5-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC3,$ACC3 + vpmuludq $Bi,$TEMP1,$TEMP1 + vmovdqu -8+32*6-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC4,$ACC4 + vpmuludq $Bi,$TEMP2,$TEMP2 + vmovdqu -8+32*7-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC5,$ACC5 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovdqu -8+32*8-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC6,$ACC6 + vpmuludq $Bi,$TEMP1,$TEMP1 + vmovdqu -8+32*9-128($ap),$ACC9 + vpaddq $TEMP1,$ACC7,$ACC7 + vpmuludq $Bi,$TEMP2,$TEMP2 + vpaddq $TEMP2,$ACC8,$ACC8 + vpmuludq $Bi,$ACC9,$ACC9 + vpbroadcastq 16($bp), $Bi + + mov %rax,%rdx + imulq -128($np),%rax + add %rax,$r1 + vmovdqu -8+32*1-128($np),$TEMP0 + mov %rdx,%rax + imulq 8-128($np),%rax + add %rax,$r2 + vmovdqu -8+32*2-128($np),$TEMP1 + shr \$29, $r1 + imulq 16-128($np),%rdx + add %rdx,$r3 + add $r1, $r2 + + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovq $Bi, %rbx + vmovdqu -8+32*3-128($np),$TEMP2 + vpaddq $TEMP0,$ACC1,$ACC1 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu -8+32*4-128($np),$TEMP0 + vpaddq $TEMP1,$ACC2,$ACC2 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -8+32*5-128($np),$TEMP1 + vpaddq $TEMP2,$ACC3,$ACC3 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovdqu -8+32*6-128($np),$TEMP2 + vpaddq $TEMP0,$ACC4,$ACC4 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu -8+32*7-128($np),$TEMP0 + vpaddq $TEMP1,$ACC5,$ACC5 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -8+32*8-128($np),$TEMP1 + vpaddq $TEMP2,$ACC6,$ACC6 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovdqu -8+32*9-128($np),$TEMP2 + vpaddq $TEMP0,$ACC7,$ACC7 + vpmuludq $Yi,$TEMP1,$TEMP1 + vpaddq $TEMP1,$ACC8,$ACC8 + vpmuludq $Yi,$TEMP2,$TEMP2 + vpaddq $TEMP2,$ACC9,$ACC9 + + vmovdqu -16+32*1-128($ap),$TEMP0 + mov %rbx,%rax + imulq -128($ap),%rax + add $r2,%rax + + vmovdqu -16+32*2-128($ap),$TEMP1 + mov %rax,$r2 + imull $n0, %eax + and \$0x1fffffff, %eax + + imulq 8-128($ap),%rbx + add %rbx,$r3 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovd %eax, $Yi + vmovdqu -16+32*3-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC1,$ACC1 + vpmuludq $Bi,$TEMP1,$TEMP1 + vpbroadcastq $Yi, $Yi + vmovdqu -16+32*4-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC2,$ACC2 + vpmuludq $Bi,$TEMP2,$TEMP2 + vmovdqu -16+32*5-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC3,$ACC3 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovdqu -16+32*6-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC4,$ACC4 + vpmuludq $Bi,$TEMP1,$TEMP1 + vmovdqu -16+32*7-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC5,$ACC5 + vpmuludq $Bi,$TEMP2,$TEMP2 + vmovdqu -16+32*8-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC6,$ACC6 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovdqu -16+32*9-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC7,$ACC7 + vpmuludq $Bi,$TEMP1,$TEMP1 + vpaddq $TEMP1,$ACC8,$ACC8 + vpmuludq $Bi,$TEMP2,$TEMP2 + vpbroadcastq 24($bp), $Bi + vpaddq $TEMP2,$ACC9,$ACC9 + + vmovdqu -16+32*1-128($np),$TEMP0 + mov %rax,%rdx + imulq -128($np),%rax + add %rax,$r2 + vmovdqu -16+32*2-128($np),$TEMP1 + imulq 8-128($np),%rdx + add %rdx,$r3 + shr \$29, $r2 + + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovq $Bi, %rbx + vmovdqu -16+32*3-128($np),$TEMP2 + vpaddq $TEMP0,$ACC1,$ACC1 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu -16+32*4-128($np),$TEMP0 + vpaddq $TEMP1,$ACC2,$ACC2 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -16+32*5-128($np),$TEMP1 + vpaddq $TEMP2,$ACC3,$ACC3 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovdqu -16+32*6-128($np),$TEMP2 + vpaddq $TEMP0,$ACC4,$ACC4 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu -16+32*7-128($np),$TEMP0 + vpaddq $TEMP1,$ACC5,$ACC5 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -16+32*8-128($np),$TEMP1 + vpaddq $TEMP2,$ACC6,$ACC6 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovdqu -16+32*9-128($np),$TEMP2 + vpaddq $TEMP0,$ACC7,$ACC7 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu -24+32*1-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC8,$ACC8 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -24+32*2-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC9,$ACC9 + + add $r2, $r3 + imulq -128($ap),%rbx + add %rbx,$r3 + + mov $r3, %rax + imull $n0, %eax + and \$0x1fffffff, %eax + + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovd %eax, $Yi + vmovdqu -24+32*3-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC1,$ACC1 + vpmuludq $Bi,$TEMP1,$TEMP1 + vpbroadcastq $Yi, $Yi + vmovdqu -24+32*4-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC2,$ACC2 + vpmuludq $Bi,$TEMP2,$TEMP2 + vmovdqu -24+32*5-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC3,$ACC3 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovdqu -24+32*6-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC4,$ACC4 + vpmuludq $Bi,$TEMP1,$TEMP1 + vmovdqu -24+32*7-128($ap),$TEMP0 + vpaddq $TEMP1,$ACC5,$ACC5 + vpmuludq $Bi,$TEMP2,$TEMP2 + vmovdqu -24+32*8-128($ap),$TEMP1 + vpaddq $TEMP2,$ACC6,$ACC6 + vpmuludq $Bi,$TEMP0,$TEMP0 + vmovdqu -24+32*9-128($ap),$TEMP2 + vpaddq $TEMP0,$ACC7,$ACC7 + vpmuludq $Bi,$TEMP1,$TEMP1 + vpaddq $TEMP1,$ACC8,$ACC8 + vpmuludq $Bi,$TEMP2,$TEMP2 + vpbroadcastq 32($bp), $Bi + vpaddq $TEMP2,$ACC9,$ACC9 + add \$32, $bp # $bp++ + + vmovdqu -24+32*1-128($np),$TEMP0 + imulq -128($np),%rax + add %rax,$r3 + shr \$29, $r3 + + vmovdqu -24+32*2-128($np),$TEMP1 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovq $Bi, %rbx + vmovdqu -24+32*3-128($np),$TEMP2 + vpaddq $TEMP0,$ACC1,$ACC0 # $ACC0==$TEMP0 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu $ACC0, (%rsp) # transfer $r0-$r3 + vpaddq $TEMP1,$ACC2,$ACC1 + vmovdqu -24+32*4-128($np),$TEMP0 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -24+32*5-128($np),$TEMP1 + vpaddq $TEMP2,$ACC3,$ACC2 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovdqu -24+32*6-128($np),$TEMP2 + vpaddq $TEMP0,$ACC4,$ACC3 + vpmuludq $Yi,$TEMP1,$TEMP1 + vmovdqu -24+32*7-128($np),$TEMP0 + vpaddq $TEMP1,$ACC5,$ACC4 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovdqu -24+32*8-128($np),$TEMP1 + vpaddq $TEMP2,$ACC6,$ACC5 + vpmuludq $Yi,$TEMP0,$TEMP0 + vmovdqu -24+32*9-128($np),$TEMP2 + mov $r3, $r0 + vpaddq $TEMP0,$ACC7,$ACC6 + vpmuludq $Yi,$TEMP1,$TEMP1 + add (%rsp), $r0 + vpaddq $TEMP1,$ACC8,$ACC7 + vpmuludq $Yi,$TEMP2,$TEMP2 + vmovq $r3, $TEMP1 + vpaddq $TEMP2,$ACC9,$ACC8 + + dec $i + jnz .Loop_mul_1024 +___ + +# (*) Original implementation was correcting ACC1-ACC3 for overflow +# after 7 loop runs, or after 28 iterations, or 56 additions. +# But as we underutilize resources, it's possible to correct in +# each iteration with marginal performance loss. But then, as +# we do it in each iteration, we can correct less digits, and +# avoid performance penalties completely. Also note that we +# correct only three digits out of four. This works because +# most significant digit is subjected to less additions. + +$TEMP0 = $ACC9; +$TEMP3 = $Bi; +$TEMP4 = $Yi; +$code.=<<___; + vpermq \$0, $AND_MASK, $AND_MASK + vpaddq (%rsp), $TEMP1, $ACC0 + + vpsrlq \$29, $ACC0, $TEMP1 + vpand $AND_MASK, $ACC0, $ACC0 + vpsrlq \$29, $ACC1, $TEMP2 + vpand $AND_MASK, $ACC1, $ACC1 + vpsrlq \$29, $ACC2, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC2, $ACC2 + vpsrlq \$29, $ACC3, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC3, $ACC3 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP3, $TEMP3 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpermq \$0x93, $TEMP4, $TEMP4 + vpaddq $TEMP0, $ACC0, $ACC0 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC1, $ACC1 + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC2, $ACC2 + vpblendd \$3, $TEMP4, $ZERO, $TEMP4 + vpaddq $TEMP3, $ACC3, $ACC3 + vpaddq $TEMP4, $ACC4, $ACC4 + + vpsrlq \$29, $ACC0, $TEMP1 + vpand $AND_MASK, $ACC0, $ACC0 + vpsrlq \$29, $ACC1, $TEMP2 + vpand $AND_MASK, $ACC1, $ACC1 + vpsrlq \$29, $ACC2, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC2, $ACC2 + vpsrlq \$29, $ACC3, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC3, $ACC3 + vpermq \$0x93, $TEMP3, $TEMP3 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP4, $TEMP4 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC0, $ACC0 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC1, $ACC1 + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC2, $ACC2 + vpblendd \$3, $TEMP4, $ZERO, $TEMP4 + vpaddq $TEMP3, $ACC3, $ACC3 + vpaddq $TEMP4, $ACC4, $ACC4 + + vmovdqu $ACC0, 0-128($rp) + vmovdqu $ACC1, 32-128($rp) + vmovdqu $ACC2, 64-128($rp) + vmovdqu $ACC3, 96-128($rp) +___ + +$TEMP5=$ACC0; +$code.=<<___; + vpsrlq \$29, $ACC4, $TEMP1 + vpand $AND_MASK, $ACC4, $ACC4 + vpsrlq \$29, $ACC5, $TEMP2 + vpand $AND_MASK, $ACC5, $ACC5 + vpsrlq \$29, $ACC6, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC6, $ACC6 + vpsrlq \$29, $ACC7, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC7, $ACC7 + vpsrlq \$29, $ACC8, $TEMP5 + vpermq \$0x93, $TEMP3, $TEMP3 + vpand $AND_MASK, $ACC8, $ACC8 + vpermq \$0x93, $TEMP4, $TEMP4 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP5, $TEMP5 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC4, $ACC4 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC5, $ACC5 + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC6, $ACC6 + vpblendd \$3, $TEMP4, $TEMP5, $TEMP4 + vpaddq $TEMP3, $ACC7, $ACC7 + vpaddq $TEMP4, $ACC8, $ACC8 + + vpsrlq \$29, $ACC4, $TEMP1 + vpand $AND_MASK, $ACC4, $ACC4 + vpsrlq \$29, $ACC5, $TEMP2 + vpand $AND_MASK, $ACC5, $ACC5 + vpsrlq \$29, $ACC6, $TEMP3 + vpermq \$0x93, $TEMP1, $TEMP1 + vpand $AND_MASK, $ACC6, $ACC6 + vpsrlq \$29, $ACC7, $TEMP4 + vpermq \$0x93, $TEMP2, $TEMP2 + vpand $AND_MASK, $ACC7, $ACC7 + vpsrlq \$29, $ACC8, $TEMP5 + vpermq \$0x93, $TEMP3, $TEMP3 + vpand $AND_MASK, $ACC8, $ACC8 + vpermq \$0x93, $TEMP4, $TEMP4 + + vpblendd \$3, $ZERO, $TEMP1, $TEMP0 + vpermq \$0x93, $TEMP5, $TEMP5 + vpblendd \$3, $TEMP1, $TEMP2, $TEMP1 + vpaddq $TEMP0, $ACC4, $ACC4 + vpblendd \$3, $TEMP2, $TEMP3, $TEMP2 + vpaddq $TEMP1, $ACC5, $ACC5 + vpblendd \$3, $TEMP3, $TEMP4, $TEMP3 + vpaddq $TEMP2, $ACC6, $ACC6 + vpblendd \$3, $TEMP4, $TEMP5, $TEMP4 + vpaddq $TEMP3, $ACC7, $ACC7 + vpaddq $TEMP4, $ACC8, $ACC8 + + vmovdqu $ACC4, 128-128($rp) + vmovdqu $ACC5, 160-128($rp) + vmovdqu $ACC6, 192-128($rp) + vmovdqu $ACC7, 224-128($rp) + vmovdqu $ACC8, 256-128($rp) + vzeroupper + + mov %rbp, %rax +___ +$code.=<<___ if ($win64); + movaps -0xd8(%rax),%xmm6 + movaps -0xc8(%rax),%xmm7 + movaps -0xb8(%rax),%xmm8 + movaps -0xa8(%rax),%xmm9 + movaps -0x98(%rax),%xmm10 + movaps -0x88(%rax),%xmm11 + movaps -0x78(%rax),%xmm12 + movaps -0x68(%rax),%xmm13 + movaps -0x58(%rax),%xmm14 + movaps -0x48(%rax),%xmm15 +___ +$code.=<<___; + mov -48(%rax),%r15 + mov -40(%rax),%r14 + mov -32(%rax),%r13 + mov -24(%rax),%r12 + mov -16(%rax),%rbp + mov -8(%rax),%rbx + lea (%rax),%rsp # restore %rsp +.Lmul_1024_epilogue: + ret +.size rsaz_1024_mul_avx2,.-rsaz_1024_mul_avx2 +___ +} +{ +my ($out,$inp) = $win64 ? ("%rcx","%rdx") : ("%rdi","%rsi"); +my @T = map("%r$_",(8..11)); + +$code.=<<___; +.globl rsaz_1024_red2norm_avx2 +.type rsaz_1024_red2norm_avx2,\@abi-omnipotent +.align 32 +rsaz_1024_red2norm_avx2: + sub \$-128,$inp # size optimization + xor %rax,%rax +___ + +for ($j=0,$i=0; $i<16; $i++) { + my $k=0; + while (29*$j<64*($i+1)) { # load data till boundary + $code.=" mov `8*$j-128`($inp), @T[0]\n"; + $j++; $k++; push(@T,shift(@T)); + } + $l=$k; + while ($k>1) { # shift loaded data but last value + $code.=" shl \$`29*($j-$k)`,@T[-$k]\n"; + $k--; + } + $code.=<<___; # shift last value + mov @T[-1], @T[0] + shl \$`29*($j-1)`, @T[-1] + shr \$`-29*($j-1)`, @T[0] +___ + while ($l) { # accumulate all values + $code.=" add @T[-$l], %rax\n"; + $l--; + } + $code.=<<___; + adc \$0, @T[0] # consume eventual carry + mov %rax, 8*$i($out) + mov @T[0], %rax +___ + push(@T,shift(@T)); +} +$code.=<<___; + ret +.size rsaz_1024_red2norm_avx2,.-rsaz_1024_red2norm_avx2 + +.globl rsaz_1024_norm2red_avx2 +.type rsaz_1024_norm2red_avx2,\@abi-omnipotent +.align 32 +rsaz_1024_norm2red_avx2: + sub \$-128,$out # size optimization + mov ($inp),@T[0] + mov \$0x1fffffff,%eax +___ +for ($j=0,$i=0; $i<16; $i++) { + $code.=" mov `8*($i+1)`($inp),@T[1]\n" if ($i<15); + $code.=" xor @T[1],@T[1]\n" if ($i==15); + my $k=1; + while (29*($j+1)<64*($i+1)) { + $code.=<<___; + mov @T[0],@T[-$k] + shr \$`29*$j`,@T[-$k] + and %rax,@T[-$k] # &0x1fffffff + mov @T[-$k],`8*$j-128`($out) +___ + $j++; $k++; + } + $code.=<<___; + shrd \$`29*$j`,@T[1],@T[0] + and %rax,@T[0] + mov @T[0],`8*$j-128`($out) +___ + $j++; + push(@T,shift(@T)); +} +$code.=<<___; + mov @T[0],`8*$j-128`($out) # zero + mov @T[0],`8*($j+1)-128`($out) + mov @T[0],`8*($j+2)-128`($out) + mov @T[0],`8*($j+3)-128`($out) + ret +.size rsaz_1024_norm2red_avx2,.-rsaz_1024_norm2red_avx2 +___ +} +{ +my ($out,$inp,$power) = $win64 ? ("%rcx","%rdx","%r8d") : ("%rdi","%rsi","%edx"); + +$code.=<<___; +.globl rsaz_1024_scatter5_avx2 +.type rsaz_1024_scatter5_avx2,\@abi-omnipotent +.align 32 +rsaz_1024_scatter5_avx2: + vzeroupper + vmovdqu .Lscatter_permd(%rip),%ymm5 + shl \$4,$power + lea ($out,$power),$out + mov \$9,%eax + jmp .Loop_scatter_1024 + +.align 32 +.Loop_scatter_1024: + vmovdqu ($inp),%ymm0 + lea 32($inp),$inp + vpermd %ymm0,%ymm5,%ymm0 + vmovdqu %xmm0,($out) + lea 16*32($out),$out + dec %eax + jnz .Loop_scatter_1024 + + vzeroupper + ret +.size rsaz_1024_scatter5_avx2,.-rsaz_1024_scatter5_avx2 + +.globl rsaz_1024_gather5_avx2 +.type rsaz_1024_gather5_avx2,\@abi-omnipotent +.align 32 +rsaz_1024_gather5_avx2: +___ +$code.=<<___ if ($win64); + lea -0x88(%rsp),%rax + vzeroupper +.LSEH_begin_rsaz_1024_gather5: + # I can't trust assembler to use specific encoding:-( + .byte 0x48,0x8d,0x60,0xe0 #lea -0x20(%rax),%rsp + .byte 0xc5,0xf8,0x29,0x70,0xe0 #vmovaps %xmm6,-0x20(%rax) + .byte 0xc5,0xf8,0x29,0x78,0xf0 #vmovaps %xmm7,-0x10(%rax) + .byte 0xc5,0x78,0x29,0x40,0x00 #vmovaps %xmm8,0(%rax) + .byte 0xc5,0x78,0x29,0x48,0x10 #vmovaps %xmm9,0x10(%rax) + .byte 0xc5,0x78,0x29,0x50,0x20 #vmovaps %xmm10,0x20(%rax) + .byte 0xc5,0x78,0x29,0x58,0x30 #vmovaps %xmm11,0x30(%rax) + .byte 0xc5,0x78,0x29,0x60,0x40 #vmovaps %xmm12,0x40(%rax) + .byte 0xc5,0x78,0x29,0x68,0x50 #vmovaps %xmm13,0x50(%rax) + .byte 0xc5,0x78,0x29,0x70,0x60 #vmovaps %xmm14,0x60(%rax) + .byte 0xc5,0x78,0x29,0x78,0x70 #vmovaps %xmm15,0x70(%rax) +___ +$code.=<<___; + lea .Lgather_table(%rip),%r11 + mov $power,%eax + and \$3,$power + shr \$2,%eax # cache line number + shl \$4,$power # offset within cache line + + vmovdqu -32(%r11),%ymm7 # .Lgather_permd + vpbroadcastb 8(%r11,%rax), %xmm8 + vpbroadcastb 7(%r11,%rax), %xmm9 + vpbroadcastb 6(%r11,%rax), %xmm10 + vpbroadcastb 5(%r11,%rax), %xmm11 + vpbroadcastb 4(%r11,%rax), %xmm12 + vpbroadcastb 3(%r11,%rax), %xmm13 + vpbroadcastb 2(%r11,%rax), %xmm14 + vpbroadcastb 1(%r11,%rax), %xmm15 + + lea 64($inp,$power),$inp + mov \$64,%r11 # size optimization + mov \$9,%eax + jmp .Loop_gather_1024 + +.align 32 +.Loop_gather_1024: + vpand -64($inp), %xmm8,%xmm0 + vpand ($inp), %xmm9,%xmm1 + vpand 64($inp), %xmm10,%xmm2 + vpand ($inp,%r11,2), %xmm11,%xmm3 + vpor %xmm0,%xmm1,%xmm1 + vpand 64($inp,%r11,2), %xmm12,%xmm4 + vpor %xmm2,%xmm3,%xmm3 + vpand ($inp,%r11,4), %xmm13,%xmm5 + vpor %xmm1,%xmm3,%xmm3 + vpand 64($inp,%r11,4), %xmm14,%xmm6 + vpor %xmm4,%xmm5,%xmm5 + vpand -128($inp,%r11,8), %xmm15,%xmm2 + lea ($inp,%r11,8),$inp + vpor %xmm3,%xmm5,%xmm5 + vpor %xmm2,%xmm6,%xmm6 + vpor %xmm5,%xmm6,%xmm6 + vpermd %ymm6,%ymm7,%ymm6 + vmovdqu %ymm6,($out) + lea 32($out),$out + dec %eax + jnz .Loop_gather_1024 + + vpxor %ymm0,%ymm0,%ymm0 + vmovdqu %ymm0,($out) + vzeroupper +___ +$code.=<<___ if ($win64); + movaps (%rsp),%xmm6 + movaps 0x10(%rsp),%xmm7 + movaps 0x20(%rsp),%xmm8 + movaps 0x30(%rsp),%xmm9 + movaps 0x40(%rsp),%xmm10 + movaps 0x50(%rsp),%xmm11 + movaps 0x60(%rsp),%xmm12 + movaps 0x70(%rsp),%xmm13 + movaps 0x80(%rsp),%xmm14 + movaps 0x90(%rsp),%xmm15 + lea 0xa8(%rsp),%rsp +.LSEH_end_rsaz_1024_gather5: +___ +$code.=<<___; + ret +.size rsaz_1024_gather5_avx2,.-rsaz_1024_gather5_avx2 +___ +} + +$code.=<<___; +.extern OPENSSL_ia32cap_P +.globl rsaz_avx2_eligible +.type rsaz_avx2_eligible,\@abi-omnipotent +.align 32 +rsaz_avx2_eligible: + mov OPENSSL_ia32cap_P+8(%rip),%eax +___ +$code.=<<___ if ($addx); + mov \$`1<<8|1<<19`,%ecx + mov \$0,%edx + and %eax,%ecx + cmp \$`1<<8|1<<19`,%ecx # check for BMI2+AD*X + cmove %edx,%eax +___ +$code.=<<___; + and \$`1<<5`,%eax + shr \$5,%eax + ret +.size rsaz_avx2_eligible,.-rsaz_avx2_eligible + +.align 64 +.Land_mask: + .quad 0x1fffffff,0x1fffffff,0x1fffffff,-1 +.Lscatter_permd: + .long 0,2,4,6,7,7,7,7 +.Lgather_permd: + .long 0,7,1,7,2,7,3,7 +.Lgather_table: + .byte 0,0,0,0,0,0,0,0, 0xff,0,0,0,0,0,0,0 +.align 64 +___ + +if ($win64) { +$rec="%rcx"; +$frame="%rdx"; +$context="%r8"; +$disp="%r9"; + +$code.=<<___ +.extern __imp_RtlVirtualUnwind +.type rsaz_se_handler,\@abi-omnipotent +.align 16 +rsaz_se_handler: + push %rsi + push %rdi + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + pushfq + sub \$64,%rsp + + mov 120($context),%rax # pull context->Rax + mov 248($context),%rbx # pull context->Rip + + mov 8($disp),%rsi # disp->ImageBase + mov 56($disp),%r11 # disp->HandlerData + + mov 0(%r11),%r10d # HandlerData[0] + lea (%rsi,%r10),%r10 # prologue label + cmp %r10,%rbx # context->Rip<prologue label + jb .Lcommon_seh_tail + + mov 152($context),%rax # pull context->Rsp + + mov 4(%r11),%r10d # HandlerData[1] + lea (%rsi,%r10),%r10 # epilogue label + cmp %r10,%rbx # context->Rip>=epilogue label + jae .Lcommon_seh_tail + + mov 160($context),%rax # pull context->Rbp + + mov -48(%rax),%r15 + mov -40(%rax),%r14 + mov -32(%rax),%r13 + mov -24(%rax),%r12 + mov -16(%rax),%rbp + mov -8(%rax),%rbx + mov %r15,240($context) + mov %r14,232($context) + mov %r13,224($context) + mov %r12,216($context) + mov %rbp,160($context) + mov %rbx,144($context) + + lea -0xd8(%rax),%rsi # %xmm save area + lea 512($context),%rdi # & context.Xmm6 + mov \$20,%ecx # 10*sizeof(%xmm0)/sizeof(%rax) + .long 0xa548f3fc # cld; rep movsq + +.Lcommon_seh_tail: + mov 8(%rax),%rdi + mov 16(%rax),%rsi + mov %rax,152($context) # restore context->Rsp + mov %rsi,168($context) # restore context->Rsi + mov %rdi,176($context) # restore context->Rdi + + mov 40($disp),%rdi # disp->ContextRecord + mov $context,%rsi # context + mov \$154,%ecx # sizeof(CONTEXT) + .long 0xa548f3fc # cld; rep movsq + + mov $disp,%rsi + xor %rcx,%rcx # arg1, UNW_FLAG_NHANDLER + mov 8(%rsi),%rdx # arg2, disp->ImageBase + mov 0(%rsi),%r8 # arg3, disp->ControlPc + mov 16(%rsi),%r9 # arg4, disp->FunctionEntry + mov 40(%rsi),%r10 # disp->ContextRecord + lea 56(%rsi),%r11 # &disp->HandlerData + lea 24(%rsi),%r12 # &disp->EstablisherFrame + mov %r10,32(%rsp) # arg5 + mov %r11,40(%rsp) # arg6 + mov %r12,48(%rsp) # arg7 + mov %rcx,56(%rsp) # arg8, (NULL) + call *__imp_RtlVirtualUnwind(%rip) + + mov \$1,%eax # ExceptionContinueSearch + add \$64,%rsp + popfq + pop %r15 + pop %r14 + pop %r13 + pop %r12 + pop %rbp + pop %rbx + pop %rdi + pop %rsi + ret +.size rsaz_se_handler,.-rsaz_se_handler + +.section .pdata +.align 4 + .rva .LSEH_begin_rsaz_1024_sqr_avx2 + .rva .LSEH_end_rsaz_1024_sqr_avx2 + .rva .LSEH_info_rsaz_1024_sqr_avx2 + + .rva .LSEH_begin_rsaz_1024_mul_avx2 + .rva .LSEH_end_rsaz_1024_mul_avx2 + .rva .LSEH_info_rsaz_1024_mul_avx2 + + .rva .LSEH_begin_rsaz_1024_gather5 + .rva .LSEH_end_rsaz_1024_gather5 + .rva .LSEH_info_rsaz_1024_gather5 +.section .xdata +.align 8 +.LSEH_info_rsaz_1024_sqr_avx2: + .byte 9,0,0,0 + .rva rsaz_se_handler + .rva .Lsqr_1024_body,.Lsqr_1024_epilogue +.LSEH_info_rsaz_1024_mul_avx2: + .byte 9,0,0,0 + .rva rsaz_se_handler + .rva .Lmul_1024_body,.Lmul_1024_epilogue +.LSEH_info_rsaz_1024_gather5: + .byte 0x01,0x33,0x16,0x00 + .byte 0x36,0xf8,0x09,0x00 #vmovaps 0x90(rsp),xmm15 + .byte 0x31,0xe8,0x08,0x00 #vmovaps 0x80(rsp),xmm14 + .byte 0x2c,0xd8,0x07,0x00 #vmovaps 0x70(rsp),xmm13 + .byte 0x27,0xc8,0x06,0x00 #vmovaps 0x60(rsp),xmm12 + .byte 0x22,0xb8,0x05,0x00 #vmovaps 0x50(rsp),xmm11 + .byte 0x1d,0xa8,0x04,0x00 #vmovaps 0x40(rsp),xmm10 + .byte 0x18,0x98,0x03,0x00 #vmovaps 0x30(rsp),xmm9 + .byte 0x13,0x88,0x02,0x00 #vmovaps 0x20(rsp),xmm8 + .byte 0x0e,0x78,0x01,0x00 #vmovaps 0x10(rsp),xmm7 + .byte 0x09,0x68,0x00,0x00 #vmovaps 0x00(rsp),xmm6 + .byte 0x04,0x01,0x15,0x00 #sub rsp,0xa8 +___ +} + +foreach (split("\n",$code)) { + s/\`([^\`]*)\`/eval($1)/ge; + + s/\b(sh[rl]d?\s+\$)(-?[0-9]+)/$1.$2%64/ge or + + s/\b(vmov[dq])\b(.+)%ymm([0-9]+)/$1$2%xmm$3/go or + s/\b(vmovdqu)\b(.+)%x%ymm([0-9]+)/$1$2%xmm$3/go or + s/\b(vpinsr[qd])\b(.+)%ymm([0-9]+)/$1$2%xmm$3/go or + s/\b(vpextr[qd])\b(.+)%ymm([0-9]+)/$1$2%xmm$3/go or + s/\b(vpbroadcast[qd]\s+)%ymm([0-9]+)/$1%xmm$2/go; + print $_,"\n"; +} + +}}} else {{{ +print <<___; # assembler is too old +.text + +.globl rsaz_avx2_eligible +.type rsaz_avx2_eligible,\@abi-omnipotent +rsaz_avx2_eligible: + xor %eax,%eax + ret +.size rsaz_avx2_eligible,.-rsaz_avx2_eligible + +.globl rsaz_1024_sqr_avx2 +.globl rsaz_1024_mul_avx2 +.globl rsaz_1024_norm2red_avx2 +.globl rsaz_1024_red2norm_avx2 +.globl rsaz_1024_scatter5_avx2 +.globl rsaz_1024_gather5_avx2 +.type rsaz_1024_sqr_avx2,\@abi-omnipotent +rsaz_1024_sqr_avx2: +rsaz_1024_mul_avx2: +rsaz_1024_norm2red_avx2: +rsaz_1024_red2norm_avx2: +rsaz_1024_scatter5_avx2: +rsaz_1024_gather5_avx2: + .byte 0x0f,0x0b # ud2 + ret +.size rsaz_1024_sqr_avx2,.-rsaz_1024_sqr_avx2 +___ +}}} + +close STDOUT; diff --git a/openssl/crypto/bn/asm/rsaz-x86_64.pl b/openssl/crypto/bn/asm/rsaz-x86_64.pl new file mode 100755 index 000000000..3bd45dbac --- /dev/null +++ b/openssl/crypto/bn/asm/rsaz-x86_64.pl @@ -0,0 +1,2144 @@ +#!/usr/bin/env perl + +############################################################################## +# # +# Copyright (c) 2012, Intel Corporation # +# # +# All rights reserved. # +# # +# Redistribution and use in source and binary forms, with or without # +# modification, are permitted provided that the following conditions are # +# met: # +# # +# * Redistributions of source code must retain the above copyright # +# notice, this list of conditions and the following disclaimer. # +# # +# * Redistributions in binary form must reproduce the above copyright # +# notice, this list of conditions and the following disclaimer in the # +# documentation and/or other materials provided with the # +# distribution. # +# # +# * Neither the name of the Intel Corporation nor the names of its # +# contributors may be used to endorse or promote products derived from # +# this software without specific prior written permission. # +# # +# # +# THIS SOFTWARE IS PROVIDED BY INTEL CORPORATION ""AS IS"" AND ANY # +# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # +# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR # +# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL CORPORATION OR # +# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, # +# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, # +# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR # +# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF # +# LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING # +# NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS # +# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. # +# # +############################################################################## +# Developers and authors: # +# Shay Gueron (1, 2), and Vlad Krasnov (1) # +# (1) Intel Architecture Group, Microprocessor and Chipset Development, # +# Israel Development Center, Haifa, Israel # +# (2) University of Haifa # +############################################################################## +# Reference: # +# [1] S. Gueron, "Efficient Software Implementations of Modular # +# Exponentiation", http://eprint.iacr.org/2011/239 # +# [2] S. Gueron, V. Krasnov. "Speeding up Big-Numbers Squaring". # +# IEEE Proceedings of 9th International Conference on Information # +# Technology: New Generations (ITNG 2012), 821-823 (2012). # +# [3] S. Gueron, Efficient Software Implementations of Modular Exponentiation# +# Journal of Cryptographic Engineering 2:31-43 (2012). # +# [4] S. Gueron, V. Krasnov: "[PATCH] Efficient and side channel analysis # +# resistant 512-bit and 1024-bit modular exponentiation for optimizing # +# RSA1024 and RSA2048 on x86_64 platforms", # +# http://rt.openssl.org/Ticket/Display.html?id=2582&user=guest&pass=guest# +############################################################################## + +# While original submission covers 512- and 1024-bit exponentiation, +# this module is limited to 512-bit version only (and as such +# accelerates RSA1024 sign). This is because improvement for longer +# keys is not high enough to justify the effort, highest measured +# was ~5% on Westmere. [This is relative to OpenSSL 1.0.2, upcoming +# for the moment of this writing!] Nor does this module implement +# "monolithic" complete exponentiation jumbo-subroutine, but adheres +# to more modular mixture of C and assembly. And it's optimized even +# for processors other than Intel Core family (see table below for +# improvement coefficients). +# <appro@openssl.org> +# +# RSA1024 sign/sec this/original |this/rsax(*) this/fips(*) +# ----------------+--------------------------- +# Opteron +13% |+5% +20% +# Bulldozer -0% |-1% +10% +# P4 +11% |+7% +8% +# Westmere +5% |+14% +17% +# Sandy Bridge +2% |+12% +29% +# Ivy Bridge +1% |+11% +35% +# Haswell(**) -0% |+12% +39% +# Atom +13% |+11% +4% +# VIA Nano +70% |+9% +25% +# +# (*) rsax engine and fips numbers are presented for reference +# purposes; +# (**) MULX was attempted, but found to give only marginal improvement; + +$flavour = shift; +$output = shift; +if ($flavour =~ /\./) { $output = $flavour; undef $flavour; } + +$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/); + +$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; +( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or +( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or +die "can't locate x86_64-xlate.pl"; + +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; + +if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1` + =~ /GNU assembler version ([2-9]\.[0-9]+)/) { + $addx = ($1>=2.23); +} + +if (!$addx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) && + `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) { + $addx = ($1>=2.10); +} + +if (!$addx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) && + `ml64 2>&1` =~ /Version ([0-9]+)\./) { + $addx = ($1>=12); +} + +if (!$addx && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9])\.([0-9]+)/) { + my $ver = $2 + $3/100.0; # 3.1->3.01, 3.10->3.10 + $addx = ($ver>=3.03); +} + +($out, $inp, $mod) = ("%rdi", "%rsi", "%rbp"); # common internal API +{ +my ($out,$inp,$mod,$n0,$times) = ("%rdi","%rsi","%rdx","%rcx","%r8d"); + +$code.=<<___; +.text + +.extern OPENSSL_ia32cap_P + +.globl rsaz_512_sqr +.type rsaz_512_sqr,\@function,5 +.align 32 +rsaz_512_sqr: # 25-29% faster than rsaz_512_mul + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + subq \$128+24, %rsp +.Lsqr_body: + movq $mod, %rbp # common argument + movq ($inp), %rdx + movq 8($inp), %rax + movq $n0, 128(%rsp) +___ +$code.=<<___ if ($addx); + movl \$0x80100,%r11d + andl OPENSSL_ia32cap_P+8(%rip),%r11d + cmpl \$0x80100,%r11d # check for MULX and ADO/CX + je .Loop_sqrx +___ +$code.=<<___; + jmp .Loop_sqr + +.align 32 +.Loop_sqr: + movl $times,128+8(%rsp) +#first iteration + movq %rdx, %rbx + mulq %rdx + movq %rax, %r8 + movq 16($inp), %rax + movq %rdx, %r9 + + mulq %rbx + addq %rax, %r9 + movq 24($inp), %rax + movq %rdx, %r10 + adcq \$0, %r10 + + mulq %rbx + addq %rax, %r10 + movq 32($inp), %rax + movq %rdx, %r11 + adcq \$0, %r11 + + mulq %rbx + addq %rax, %r11 + movq 40($inp), %rax + movq %rdx, %r12 + adcq \$0, %r12 + + mulq %rbx + addq %rax, %r12 + movq 48($inp), %rax + movq %rdx, %r13 + adcq \$0, %r13 + + mulq %rbx + addq %rax, %r13 + movq 56($inp), %rax + movq %rdx, %r14 + adcq \$0, %r14 + + mulq %rbx + addq %rax, %r14 + movq %rbx, %rax + movq %rdx, %r15 + adcq \$0, %r15 + + addq %r8, %r8 #shlq \$1, %r8 + movq %r9, %rcx + adcq %r9, %r9 #shld \$1, %r8, %r9 + + mulq %rax + movq %rax, (%rsp) + addq %rdx, %r8 + adcq \$0, %r9 + + movq %r8, 8(%rsp) + shrq \$63, %rcx + +#second iteration + movq 8($inp), %r8 + movq 16($inp), %rax + mulq %r8 + addq %rax, %r10 + movq 24($inp), %rax + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r8 + addq %rax, %r11 + movq 32($inp), %rax + adcq \$0, %rdx + addq %rbx, %r11 + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r8 + addq %rax, %r12 + movq 40($inp), %rax + adcq \$0, %rdx + addq %rbx, %r12 + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r8 + addq %rax, %r13 + movq 48($inp), %rax + adcq \$0, %rdx + addq %rbx, %r13 + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r8 + addq %rax, %r14 + movq 56($inp), %rax + adcq \$0, %rdx + addq %rbx, %r14 + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r8 + addq %rax, %r15 + movq %r8, %rax + adcq \$0, %rdx + addq %rbx, %r15 + movq %rdx, %r8 + movq %r10, %rdx + adcq \$0, %r8 + + add %rdx, %rdx + lea (%rcx,%r10,2), %r10 #shld \$1, %rcx, %r10 + movq %r11, %rbx + adcq %r11, %r11 #shld \$1, %r10, %r11 + + mulq %rax + addq %rax, %r9 + adcq %rdx, %r10 + adcq \$0, %r11 + + movq %r9, 16(%rsp) + movq %r10, 24(%rsp) + shrq \$63, %rbx + +#third iteration + movq 16($inp), %r9 + movq 24($inp), %rax + mulq %r9 + addq %rax, %r12 + movq 32($inp), %rax + movq %rdx, %rcx + adcq \$0, %rcx + + mulq %r9 + addq %rax, %r13 + movq 40($inp), %rax + adcq \$0, %rdx + addq %rcx, %r13 + movq %rdx, %rcx + adcq \$0, %rcx + + mulq %r9 + addq %rax, %r14 + movq 48($inp), %rax + adcq \$0, %rdx + addq %rcx, %r14 + movq %rdx, %rcx + adcq \$0, %rcx + + mulq %r9 + movq %r12, %r10 + lea (%rbx,%r12,2), %r12 #shld \$1, %rbx, %r12 + addq %rax, %r15 + movq 56($inp), %rax + adcq \$0, %rdx + addq %rcx, %r15 + movq %rdx, %rcx + adcq \$0, %rcx + + mulq %r9 + shrq \$63, %r10 + addq %rax, %r8 + movq %r9, %rax + adcq \$0, %rdx + addq %rcx, %r8 + movq %rdx, %r9 + adcq \$0, %r9 + + movq %r13, %rcx + leaq (%r10,%r13,2), %r13 #shld \$1, %r12, %r13 + + mulq %rax + addq %rax, %r11 + adcq %rdx, %r12 + adcq \$0, %r13 + + movq %r11, 32(%rsp) + movq %r12, 40(%rsp) + shrq \$63, %rcx + +#fourth iteration + movq 24($inp), %r10 + movq 32($inp), %rax + mulq %r10 + addq %rax, %r14 + movq 40($inp), %rax + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r10 + addq %rax, %r15 + movq 48($inp), %rax + adcq \$0, %rdx + addq %rbx, %r15 + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r10 + movq %r14, %r12 + leaq (%rcx,%r14,2), %r14 #shld \$1, %rcx, %r14 + addq %rax, %r8 + movq 56($inp), %rax + adcq \$0, %rdx + addq %rbx, %r8 + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r10 + shrq \$63, %r12 + addq %rax, %r9 + movq %r10, %rax + adcq \$0, %rdx + addq %rbx, %r9 + movq %rdx, %r10 + adcq \$0, %r10 + + movq %r15, %rbx + leaq (%r12,%r15,2),%r15 #shld \$1, %r14, %r15 + + mulq %rax + addq %rax, %r13 + adcq %rdx, %r14 + adcq \$0, %r15 + + movq %r13, 48(%rsp) + movq %r14, 56(%rsp) + shrq \$63, %rbx + +#fifth iteration + movq 32($inp), %r11 + movq 40($inp), %rax + mulq %r11 + addq %rax, %r8 + movq 48($inp), %rax + movq %rdx, %rcx + adcq \$0, %rcx + + mulq %r11 + addq %rax, %r9 + movq 56($inp), %rax + adcq \$0, %rdx + movq %r8, %r12 + leaq (%rbx,%r8,2), %r8 #shld \$1, %rbx, %r8 + addq %rcx, %r9 + movq %rdx, %rcx + adcq \$0, %rcx + + mulq %r11 + shrq \$63, %r12 + addq %rax, %r10 + movq %r11, %rax + adcq \$0, %rdx + addq %rcx, %r10 + movq %rdx, %r11 + adcq \$0, %r11 + + movq %r9, %rcx + leaq (%r12,%r9,2), %r9 #shld \$1, %r8, %r9 + + mulq %rax + addq %rax, %r15 + adcq %rdx, %r8 + adcq \$0, %r9 + + movq %r15, 64(%rsp) + movq %r8, 72(%rsp) + shrq \$63, %rcx + +#sixth iteration + movq 40($inp), %r12 + movq 48($inp), %rax + mulq %r12 + addq %rax, %r10 + movq 56($inp), %rax + movq %rdx, %rbx + adcq \$0, %rbx + + mulq %r12 + addq %rax, %r11 + movq %r12, %rax + movq %r10, %r15 + leaq (%rcx,%r10,2), %r10 #shld \$1, %rcx, %r10 + adcq \$0, %rdx + shrq \$63, %r15 + addq %rbx, %r11 + movq %rdx, %r12 + adcq \$0, %r12 + + movq %r11, %rbx + leaq (%r15,%r11,2), %r11 #shld \$1, %r10, %r11 + + mulq %rax + addq %rax, %r9 + adcq %rdx, %r10 + adcq \$0, %r11 + + movq %r9, 80(%rsp) + movq %r10, 88(%rsp) + +#seventh iteration + movq 48($inp), %r13 + movq 56($inp), %rax + mulq %r13 + addq %rax, %r12 + movq %r13, %rax + movq %rdx, %r13 + adcq \$0, %r13 + + xorq %r14, %r14 + shlq \$1, %rbx + adcq %r12, %r12 #shld \$1, %rbx, %r12 + adcq %r13, %r13 #shld \$1, %r12, %r13 + adcq %r14, %r14 #shld \$1, %r13, %r14 + + mulq %rax + addq %rax, %r11 + adcq %rdx, %r12 + adcq \$0, %r13 + + movq %r11, 96(%rsp) + movq %r12, 104(%rsp) + +#eighth iteration + movq 56($inp), %rax + mulq %rax + addq %rax, %r13 + adcq \$0, %rdx + + addq %rdx, %r14 + + movq %r13, 112(%rsp) + movq %r14, 120(%rsp) + + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reduce + + addq 64(%rsp), %r8 + adcq 72(%rsp), %r9 + adcq 80(%rsp), %r10 + adcq 88(%rsp), %r11 + adcq 96(%rsp), %r12 + adcq 104(%rsp), %r13 + adcq 112(%rsp), %r14 + adcq 120(%rsp), %r15 + sbbq %rcx, %rcx + + call __rsaz_512_subtract + + movq %r8, %rdx + movq %r9, %rax + movl 128+8(%rsp), $times + movq $out, $inp + + decl $times + jnz .Loop_sqr +___ +if ($addx) { +$code.=<<___; + jmp .Lsqr_tail + +.align 32 +.Loop_sqrx: + movl $times,128+8(%rsp) + movq $out, %xmm0 # off-load + movq %rbp, %xmm1 # off-load +#first iteration + mulx %rax, %r8, %r9 + + mulx 16($inp), %rcx, %r10 + xor %rbp, %rbp # cf=0, of=0 + + mulx 24($inp), %rax, %r11 + adcx %rcx, %r9 + + mulx 32($inp), %rcx, %r12 + adcx %rax, %r10 + + mulx 40($inp), %rax, %r13 + adcx %rcx, %r11 + + .byte 0xc4,0x62,0xf3,0xf6,0xb6,0x30,0x00,0x00,0x00 # mulx 48($inp), %rcx, %r14 + adcx %rax, %r12 + adcx %rcx, %r13 + + .byte 0xc4,0x62,0xfb,0xf6,0xbe,0x38,0x00,0x00,0x00 # mulx 56($inp), %rax, %r15 + adcx %rax, %r14 + adcx %rbp, %r15 # %rbp is 0 + + mov %r9, %rcx + shld \$1, %r8, %r9 + shl \$1, %r8 + + xor %ebp, %ebp + mulx %rdx, %rax, %rdx + adcx %rdx, %r8 + mov 8($inp), %rdx + adcx %rbp, %r9 + + mov %rax, (%rsp) + mov %r8, 8(%rsp) + +#second iteration + mulx 16($inp), %rax, %rbx + adox %rax, %r10 + adcx %rbx, %r11 + + .byte 0xc4,0x62,0xc3,0xf6,0x86,0x18,0x00,0x00,0x00 # mulx 24($inp), $out, %r8 + adox $out, %r11 + adcx %r8, %r12 + + mulx 32($inp), %rax, %rbx + adox %rax, %r12 + adcx %rbx, %r13 + + mulx 40($inp), $out, %r8 + adox $out, %r13 + adcx %r8, %r14 + + .byte 0xc4,0xe2,0xfb,0xf6,0x9e,0x30,0x00,0x00,0x00 # mulx 48($inp), %rax, %rbx + adox %rax, %r14 + adcx %rbx, %r15 + + .byte 0xc4,0x62,0xc3,0xf6,0x86,0x38,0x00,0x00,0x00 # mulx 56($inp), $out, %r8 + adox $out, %r15 + adcx %rbp, %r8 + adox %rbp, %r8 + + mov %r11, %rbx + shld \$1, %r10, %r11 + shld \$1, %rcx, %r10 + + xor %ebp,%ebp + mulx %rdx, %rax, %rcx + mov 16($inp), %rdx + adcx %rax, %r9 + adcx %rcx, %r10 + adcx %rbp, %r11 + + mov %r9, 16(%rsp) + .byte 0x4c,0x89,0x94,0x24,0x18,0x00,0x00,0x00 # mov %r10, 24(%rsp) + +#third iteration + .byte 0xc4,0x62,0xc3,0xf6,0x8e,0x18,0x00,0x00,0x00 # mulx 24($inp), $out, %r9 + adox $out, %r12 + adcx %r9, %r13 + + mulx 32($inp), %rax, %rcx + adox %rax, %r13 + adcx %rcx, %r14 + + mulx 40($inp), $out, %r9 + adox $out, %r14 + adcx %r9, %r15 + + .byte 0xc4,0xe2,0xfb,0xf6,0x8e,0x30,0x00,0x00,0x00 # mulx 48($inp), %rax, %rcx + adox %rax, %r15 + adcx %rcx, %r8 + + .byte 0xc4,0x62,0xc3,0xf6,0x8e,0x38,0x00,0x00,0x00 # mulx 56($inp), $out, %r9 + adox $out, %r8 + adcx %rbp, %r9 + adox %rbp, %r9 + + mov %r13, %rcx + shld \$1, %r12, %r13 + shld \$1, %rbx, %r12 + + xor %ebp, %ebp + mulx %rdx, %rax, %rdx + adcx %rax, %r11 + adcx %rdx, %r12 + mov 24($inp), %rdx + adcx %rbp, %r13 + + mov %r11, 32(%rsp) + .byte 0x4c,0x89,0xa4,0x24,0x28,0x00,0x00,0x00 # mov %r12, 40(%rsp) + +#fourth iteration + .byte 0xc4,0xe2,0xfb,0xf6,0x9e,0x20,0x00,0x00,0x00 # mulx 32($inp), %rax, %rbx + adox %rax, %r14 + adcx %rbx, %r15 + + mulx 40($inp), $out, %r10 + adox $out, %r15 + adcx %r10, %r8 + + mulx 48($inp), %rax, %rbx + adox %rax, %r8 + adcx %rbx, %r9 + + mulx 56($inp), $out, %r10 + adox $out, %r9 + adcx %rbp, %r10 + adox %rbp, %r10 + + .byte 0x66 + mov %r15, %rbx + shld \$1, %r14, %r15 + shld \$1, %rcx, %r14 + + xor %ebp, %ebp + mulx %rdx, %rax, %rdx + adcx %rax, %r13 + adcx %rdx, %r14 + mov 32($inp), %rdx + adcx %rbp, %r15 + + mov %r13, 48(%rsp) + mov %r14, 56(%rsp) + +#fifth iteration + .byte 0xc4,0x62,0xc3,0xf6,0x9e,0x28,0x00,0x00,0x00 # mulx 40($inp), $out, %r11 + adox $out, %r8 + adcx %r11, %r9 + + mulx 48($inp), %rax, %rcx + adox %rax, %r9 + adcx %rcx, %r10 + + mulx 56($inp), $out, %r11 + adox $out, %r10 + adcx %rbp, %r11 + adox %rbp, %r11 + + mov %r9, %rcx + shld \$1, %r8, %r9 + shld \$1, %rbx, %r8 + + xor %ebp, %ebp + mulx %rdx, %rax, %rdx + adcx %rax, %r15 + adcx %rdx, %r8 + mov 40($inp), %rdx + adcx %rbp, %r9 + + mov %r15, 64(%rsp) + mov %r8, 72(%rsp) + +#sixth iteration + .byte 0xc4,0xe2,0xfb,0xf6,0x9e,0x30,0x00,0x00,0x00 # mulx 48($inp), %rax, %rbx + adox %rax, %r10 + adcx %rbx, %r11 + + .byte 0xc4,0x62,0xc3,0xf6,0xa6,0x38,0x00,0x00,0x00 # mulx 56($inp), $out, %r12 + adox $out, %r11 + adcx %rbp, %r12 + adox %rbp, %r12 + + mov %r11, %rbx + shld \$1, %r10, %r11 + shld \$1, %rcx, %r10 + + xor %ebp, %ebp + mulx %rdx, %rax, %rdx + adcx %rax, %r9 + adcx %rdx, %r10 + mov 48($inp), %rdx + adcx %rbp, %r11 + + mov %r9, 80(%rsp) + mov %r10, 88(%rsp) + +#seventh iteration + .byte 0xc4,0x62,0xfb,0xf6,0xae,0x38,0x00,0x00,0x00 # mulx 56($inp), %rax, %r13 + adox %rax, %r12 + adox %rbp, %r13 + + xor %r14, %r14 + shld \$1, %r13, %r14 + shld \$1, %r12, %r13 + shld \$1, %rbx, %r12 + + xor %ebp, %ebp + mulx %rdx, %rax, %rdx + adcx %rax, %r11 + adcx %rdx, %r12 + mov 56($inp), %rdx + adcx %rbp, %r13 + + .byte 0x4c,0x89,0x9c,0x24,0x60,0x00,0x00,0x00 # mov %r11, 96(%rsp) + .byte 0x4c,0x89,0xa4,0x24,0x68,0x00,0x00,0x00 # mov %r12, 104(%rsp) + +#eighth iteration + mulx %rdx, %rax, %rdx + adox %rax, %r13 + adox %rbp, %rdx + + .byte 0x66 + add %rdx, %r14 + + movq %r13, 112(%rsp) + movq %r14, 120(%rsp) + movq %xmm0, $out + movq %xmm1, %rbp + + movq 128(%rsp), %rdx # pull $n0 + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reducex + + addq 64(%rsp), %r8 + adcq 72(%rsp), %r9 + adcq 80(%rsp), %r10 + adcq 88(%rsp), %r11 + adcq 96(%rsp), %r12 + adcq 104(%rsp), %r13 + adcq 112(%rsp), %r14 + adcq 120(%rsp), %r15 + sbbq %rcx, %rcx + + call __rsaz_512_subtract + + movq %r8, %rdx + movq %r9, %rax + movl 128+8(%rsp), $times + movq $out, $inp + + decl $times + jnz .Loop_sqrx + +.Lsqr_tail: +___ +} +$code.=<<___; + + leaq 128+24+48(%rsp), %rax + movq -48(%rax), %r15 + movq -40(%rax), %r14 + movq -32(%rax), %r13 + movq -24(%rax), %r12 + movq -16(%rax), %rbp + movq -8(%rax), %rbx + leaq (%rax), %rsp +.Lsqr_epilogue: + ret +.size rsaz_512_sqr,.-rsaz_512_sqr +___ +} +{ +my ($out,$ap,$bp,$mod,$n0) = ("%rdi","%rsi","%rdx","%rcx","%r8"); +$code.=<<___; +.globl rsaz_512_mul +.type rsaz_512_mul,\@function,5 +.align 32 +rsaz_512_mul: + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + subq \$128+24, %rsp +.Lmul_body: + movq $out, %xmm0 # off-load arguments + movq $mod, %xmm1 + movq $n0, 128(%rsp) +___ +$code.=<<___ if ($addx); + movl \$0x80100,%r11d + andl OPENSSL_ia32cap_P+8(%rip),%r11d + cmpl \$0x80100,%r11d # check for MULX and ADO/CX + je .Lmulx +___ +$code.=<<___; + movq ($bp), %rbx # pass b[0] + movq $bp, %rbp # pass argument + call __rsaz_512_mul + + movq %xmm0, $out + movq %xmm1, %rbp + + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reduce +___ +$code.=<<___ if ($addx); + jmp .Lmul_tail + +.align 32 +.Lmulx: + movq $bp, %rbp # pass argument + movq ($bp), %rdx # pass b[0] + call __rsaz_512_mulx + + movq %xmm0, $out + movq %xmm1, %rbp + + movq 128(%rsp), %rdx # pull $n0 + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reducex +.Lmul_tail: +___ +$code.=<<___; + addq 64(%rsp), %r8 + adcq 72(%rsp), %r9 + adcq 80(%rsp), %r10 + adcq 88(%rsp), %r11 + adcq 96(%rsp), %r12 + adcq 104(%rsp), %r13 + adcq 112(%rsp), %r14 + adcq 120(%rsp), %r15 + sbbq %rcx, %rcx + + call __rsaz_512_subtract + + leaq 128+24+48(%rsp), %rax + movq -48(%rax), %r15 + movq -40(%rax), %r14 + movq -32(%rax), %r13 + movq -24(%rax), %r12 + movq -16(%rax), %rbp + movq -8(%rax), %rbx + leaq (%rax), %rsp +.Lmul_epilogue: + ret +.size rsaz_512_mul,.-rsaz_512_mul +___ +} +{ +my ($out,$ap,$bp,$mod,$n0,$pwr) = ("%rdi","%rsi","%rdx","%rcx","%r8","%r9d"); +$code.=<<___; +.globl rsaz_512_mul_gather4 +.type rsaz_512_mul_gather4,\@function,6 +.align 32 +rsaz_512_mul_gather4: + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + mov $pwr, $pwr + subq \$128+24, %rsp +.Lmul_gather4_body: +___ +$code.=<<___ if ($addx); + movl \$0x80100,%r11d + andl OPENSSL_ia32cap_P+8(%rip),%r11d + cmpl \$0x80100,%r11d # check for MULX and ADO/CX + je .Lmulx_gather +___ +$code.=<<___; + movl 64($bp,$pwr,4), %eax + movq $out, %xmm0 # off-load arguments + movl ($bp,$pwr,4), %ebx + movq $mod, %xmm1 + movq $n0, 128(%rsp) + + shlq \$32, %rax + or %rax, %rbx + movq ($ap), %rax + movq 8($ap), %rcx + leaq 128($bp,$pwr,4), %rbp + mulq %rbx # 0 iteration + movq %rax, (%rsp) + movq %rcx, %rax + movq %rdx, %r8 + + mulq %rbx + movd (%rbp), %xmm4 + addq %rax, %r8 + movq 16($ap), %rax + movq %rdx, %r9 + adcq \$0, %r9 + + mulq %rbx + movd 64(%rbp), %xmm5 + addq %rax, %r9 + movq 24($ap), %rax + movq %rdx, %r10 + adcq \$0, %r10 + + mulq %rbx + pslldq \$4, %xmm5 + addq %rax, %r10 + movq 32($ap), %rax + movq %rdx, %r11 + adcq \$0, %r11 + + mulq %rbx + por %xmm5, %xmm4 + addq %rax, %r11 + movq 40($ap), %rax + movq %rdx, %r12 + adcq \$0, %r12 + + mulq %rbx + addq %rax, %r12 + movq 48($ap), %rax + movq %rdx, %r13 + adcq \$0, %r13 + + mulq %rbx + leaq 128(%rbp), %rbp + addq %rax, %r13 + movq 56($ap), %rax + movq %rdx, %r14 + adcq \$0, %r14 + + mulq %rbx + movq %xmm4, %rbx + addq %rax, %r14 + movq ($ap), %rax + movq %rdx, %r15 + adcq \$0, %r15 + + leaq 8(%rsp), %rdi + movl \$7, %ecx + jmp .Loop_mul_gather + +.align 32 +.Loop_mul_gather: + mulq %rbx + addq %rax, %r8 + movq 8($ap), %rax + movq %r8, (%rdi) + movq %rdx, %r8 + adcq \$0, %r8 + + mulq %rbx + movd (%rbp), %xmm4 + addq %rax, %r9 + movq 16($ap), %rax + adcq \$0, %rdx + addq %r9, %r8 + movq %rdx, %r9 + adcq \$0, %r9 + + mulq %rbx + movd 64(%rbp), %xmm5 + addq %rax, %r10 + movq 24($ap), %rax + adcq \$0, %rdx + addq %r10, %r9 + movq %rdx, %r10 + adcq \$0, %r10 + + mulq %rbx + pslldq \$4, %xmm5 + addq %rax, %r11 + movq 32($ap), %rax + adcq \$0, %rdx + addq %r11, %r10 + movq %rdx, %r11 + adcq \$0, %r11 + + mulq %rbx + por %xmm5, %xmm4 + addq %rax, %r12 + movq 40($ap), %rax + adcq \$0, %rdx + addq %r12, %r11 + movq %rdx, %r12 + adcq \$0, %r12 + + mulq %rbx + addq %rax, %r13 + movq 48($ap), %rax + adcq \$0, %rdx + addq %r13, %r12 + movq %rdx, %r13 + adcq \$0, %r13 + + mulq %rbx + addq %rax, %r14 + movq 56($ap), %rax + adcq \$0, %rdx + addq %r14, %r13 + movq %rdx, %r14 + adcq \$0, %r14 + + mulq %rbx + movq %xmm4, %rbx + addq %rax, %r15 + movq ($ap), %rax + adcq \$0, %rdx + addq %r15, %r14 + movq %rdx, %r15 + adcq \$0, %r15 + + leaq 128(%rbp), %rbp + leaq 8(%rdi), %rdi + + decl %ecx + jnz .Loop_mul_gather + + movq %r8, (%rdi) + movq %r9, 8(%rdi) + movq %r10, 16(%rdi) + movq %r11, 24(%rdi) + movq %r12, 32(%rdi) + movq %r13, 40(%rdi) + movq %r14, 48(%rdi) + movq %r15, 56(%rdi) + + movq %xmm0, $out + movq %xmm1, %rbp + + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reduce +___ +$code.=<<___ if ($addx); + jmp .Lmul_gather_tail + +.align 32 +.Lmulx_gather: + mov 64($bp,$pwr,4), %eax + movq $out, %xmm0 # off-load arguments + lea 128($bp,$pwr,4), %rbp + mov ($bp,$pwr,4), %edx + movq $mod, %xmm1 + mov $n0, 128(%rsp) + + shl \$32, %rax + or %rax, %rdx + mulx ($ap), %rbx, %r8 # 0 iteration + mov %rbx, (%rsp) + xor %edi, %edi # cf=0, of=0 + + mulx 8($ap), %rax, %r9 + movd (%rbp), %xmm4 + + mulx 16($ap), %rbx, %r10 + movd 64(%rbp), %xmm5 + adcx %rax, %r8 + + mulx 24($ap), %rax, %r11 + pslldq \$4, %xmm5 + adcx %rbx, %r9 + + mulx 32($ap), %rbx, %r12 + por %xmm5, %xmm4 + adcx %rax, %r10 + + mulx 40($ap), %rax, %r13 + adcx %rbx, %r11 + + mulx 48($ap), %rbx, %r14 + lea 128(%rbp), %rbp + adcx %rax, %r12 + + mulx 56($ap), %rax, %r15 + movq %xmm4, %rdx + adcx %rbx, %r13 + adcx %rax, %r14 + mov %r8, %rbx + adcx %rdi, %r15 # %rdi is 0 + + mov \$-7, %rcx + jmp .Loop_mulx_gather + +.align 32 +.Loop_mulx_gather: + mulx ($ap), %rax, %r8 + adcx %rax, %rbx + adox %r9, %r8 + + mulx 8($ap), %rax, %r9 + .byte 0x66,0x0f,0x6e,0xa5,0x00,0x00,0x00,0x00 # movd (%rbp), %xmm4 + adcx %rax, %r8 + adox %r10, %r9 + + mulx 16($ap), %rax, %r10 + movd 64(%rbp), %xmm5 + lea 128(%rbp), %rbp + adcx %rax, %r9 + adox %r11, %r10 + + .byte 0xc4,0x62,0xfb,0xf6,0x9e,0x18,0x00,0x00,0x00 # mulx 24($ap), %rax, %r11 + pslldq \$4, %xmm5 + por %xmm5, %xmm4 + adcx %rax, %r10 + adox %r12, %r11 + + mulx 32($ap), %rax, %r12 + adcx %rax, %r11 + adox %r13, %r12 + + mulx 40($ap), %rax, %r13 + adcx %rax, %r12 + adox %r14, %r13 + + .byte 0xc4,0x62,0xfb,0xf6,0xb6,0x30,0x00,0x00,0x00 # mulx 48($ap), %rax, %r14 + adcx %rax, %r13 + adox %r15, %r14 + + mulx 56($ap), %rax, %r15 + movq %xmm4, %rdx + mov %rbx, 64(%rsp,%rcx,8) + adcx %rax, %r14 + adox %rdi, %r15 + mov %r8, %rbx + adcx %rdi, %r15 # cf=0 + + inc %rcx # of=0 + jnz .Loop_mulx_gather + + mov %r8, 64(%rsp) + mov %r9, 64+8(%rsp) + mov %r10, 64+16(%rsp) + mov %r11, 64+24(%rsp) + mov %r12, 64+32(%rsp) + mov %r13, 64+40(%rsp) + mov %r14, 64+48(%rsp) + mov %r15, 64+56(%rsp) + + movq %xmm0, $out + movq %xmm1, %rbp + + mov 128(%rsp), %rdx # pull $n0 + mov (%rsp), %r8 + mov 8(%rsp), %r9 + mov 16(%rsp), %r10 + mov 24(%rsp), %r11 + mov 32(%rsp), %r12 + mov 40(%rsp), %r13 + mov 48(%rsp), %r14 + mov 56(%rsp), %r15 + + call __rsaz_512_reducex + +.Lmul_gather_tail: +___ +$code.=<<___; + addq 64(%rsp), %r8 + adcq 72(%rsp), %r9 + adcq 80(%rsp), %r10 + adcq 88(%rsp), %r11 + adcq 96(%rsp), %r12 + adcq 104(%rsp), %r13 + adcq 112(%rsp), %r14 + adcq 120(%rsp), %r15 + sbbq %rcx, %rcx + + call __rsaz_512_subtract + + leaq 128+24+48(%rsp), %rax + movq -48(%rax), %r15 + movq -40(%rax), %r14 + movq -32(%rax), %r13 + movq -24(%rax), %r12 + movq -16(%rax), %rbp + movq -8(%rax), %rbx + leaq (%rax), %rsp +.Lmul_gather4_epilogue: + ret +.size rsaz_512_mul_gather4,.-rsaz_512_mul_gather4 +___ +} +{ +my ($out,$ap,$mod,$n0,$tbl,$pwr) = ("%rdi","%rsi","%rdx","%rcx","%r8","%r9d"); +$code.=<<___; +.globl rsaz_512_mul_scatter4 +.type rsaz_512_mul_scatter4,\@function,6 +.align 32 +rsaz_512_mul_scatter4: + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + mov $pwr, $pwr + subq \$128+24, %rsp +.Lmul_scatter4_body: + leaq ($tbl,$pwr,4), $tbl + movq $out, %xmm0 # off-load arguments + movq $mod, %xmm1 + movq $tbl, %xmm2 + movq $n0, 128(%rsp) + + movq $out, %rbp +___ +$code.=<<___ if ($addx); + movl \$0x80100,%r11d + andl OPENSSL_ia32cap_P+8(%rip),%r11d + cmpl \$0x80100,%r11d # check for MULX and ADO/CX + je .Lmulx_scatter +___ +$code.=<<___; + movq ($out),%rbx # pass b[0] + call __rsaz_512_mul + + movq %xmm0, $out + movq %xmm1, %rbp + + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reduce +___ +$code.=<<___ if ($addx); + jmp .Lmul_scatter_tail + +.align 32 +.Lmulx_scatter: + movq ($out), %rdx # pass b[0] + call __rsaz_512_mulx + + movq %xmm0, $out + movq %xmm1, %rbp + + movq 128(%rsp), %rdx # pull $n0 + movq (%rsp), %r8 + movq 8(%rsp), %r9 + movq 16(%rsp), %r10 + movq 24(%rsp), %r11 + movq 32(%rsp), %r12 + movq 40(%rsp), %r13 + movq 48(%rsp), %r14 + movq 56(%rsp), %r15 + + call __rsaz_512_reducex + +.Lmul_scatter_tail: +___ +$code.=<<___; + addq 64(%rsp), %r8 + adcq 72(%rsp), %r9 + adcq 80(%rsp), %r10 + adcq 88(%rsp), %r11 + adcq 96(%rsp), %r12 + adcq 104(%rsp), %r13 + adcq 112(%rsp), %r14 + adcq 120(%rsp), %r15 + movq %xmm2, $inp + sbbq %rcx, %rcx + + call __rsaz_512_subtract + + movl %r8d, 64*0($inp) # scatter + shrq \$32, %r8 + movl %r9d, 64*2($inp) + shrq \$32, %r9 + movl %r10d, 64*4($inp) + shrq \$32, %r10 + movl %r11d, 64*6($inp) + shrq \$32, %r11 + movl %r12d, 64*8($inp) + shrq \$32, %r12 + movl %r13d, 64*10($inp) + shrq \$32, %r13 + movl %r14d, 64*12($inp) + shrq \$32, %r14 + movl %r15d, 64*14($inp) + shrq \$32, %r15 + movl %r8d, 64*1($inp) + movl %r9d, 64*3($inp) + movl %r10d, 64*5($inp) + movl %r11d, 64*7($inp) + movl %r12d, 64*9($inp) + movl %r13d, 64*11($inp) + movl %r14d, 64*13($inp) + movl %r15d, 64*15($inp) + + leaq 128+24+48(%rsp), %rax + movq -48(%rax), %r15 + movq -40(%rax), %r14 + movq -32(%rax), %r13 + movq -24(%rax), %r12 + movq -16(%rax), %rbp + movq -8(%rax), %rbx + leaq (%rax), %rsp +.Lmul_scatter4_epilogue: + ret +.size rsaz_512_mul_scatter4,.-rsaz_512_mul_scatter4 +___ +} +{ +my ($out,$inp,$mod,$n0) = ("%rdi","%rsi","%rdx","%rcx"); +$code.=<<___; +.globl rsaz_512_mul_by_one +.type rsaz_512_mul_by_one,\@function,4 +.align 32 +rsaz_512_mul_by_one: + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + + subq \$128+24, %rsp +.Lmul_by_one_body: +___ +$code.=<<___ if ($addx); + movl OPENSSL_ia32cap_P+8(%rip),%eax +___ +$code.=<<___; + movq $mod, %rbp # reassign argument + movq $n0, 128(%rsp) + + movq ($inp), %r8 + pxor %xmm0, %xmm0 + movq 8($inp), %r9 + movq 16($inp), %r10 + movq 24($inp), %r11 + movq 32($inp), %r12 + movq 40($inp), %r13 + movq 48($inp), %r14 + movq 56($inp), %r15 + + movdqa %xmm0, (%rsp) + movdqa %xmm0, 16(%rsp) + movdqa %xmm0, 32(%rsp) + movdqa %xmm0, 48(%rsp) + movdqa %xmm0, 64(%rsp) + movdqa %xmm0, 80(%rsp) + movdqa %xmm0, 96(%rsp) +___ +$code.=<<___ if ($addx); + andl \$0x80100,%eax + cmpl \$0x80100,%eax # check for MULX and ADO/CX + je .Lby_one_callx +___ +$code.=<<___; + call __rsaz_512_reduce +___ +$code.=<<___ if ($addx); + jmp .Lby_one_tail +.align 32 +.Lby_one_callx: + movq 128(%rsp), %rdx # pull $n0 + call __rsaz_512_reducex +.Lby_one_tail: +___ +$code.=<<___; + movq %r8, ($out) + movq %r9, 8($out) + movq %r10, 16($out) + movq %r11, 24($out) + movq %r12, 32($out) + movq %r13, 40($out) + movq %r14, 48($out) + movq %r15, 56($out) + + leaq 128+24+48(%rsp), %rax + movq -48(%rax), %r15 + movq -40(%rax), %r14 + movq -32(%rax), %r13 + movq -24(%rax), %r12 + movq -16(%rax), %rbp + movq -8(%rax), %rbx + leaq (%rax), %rsp +.Lmul_by_one_epilogue: + ret +.size rsaz_512_mul_by_one,.-rsaz_512_mul_by_one +___ +} +{ # __rsaz_512_reduce + # + # input: %r8-%r15, %rbp - mod, 128(%rsp) - n0 + # output: %r8-%r15 + # clobbers: everything except %rbp and %rdi +$code.=<<___; +.type __rsaz_512_reduce,\@abi-omnipotent +.align 32 +__rsaz_512_reduce: + movq %r8, %rbx + imulq 128+8(%rsp), %rbx + movq 0(%rbp), %rax + movl \$8, %ecx + jmp .Lreduction_loop + +.align 32 +.Lreduction_loop: + mulq %rbx + movq 8(%rbp), %rax + negq %r8 + movq %rdx, %r8 + adcq \$0, %r8 + + mulq %rbx + addq %rax, %r9 + movq 16(%rbp), %rax + adcq \$0, %rdx + addq %r9, %r8 + movq %rdx, %r9 + adcq \$0, %r9 + + mulq %rbx + addq %rax, %r10 + movq 24(%rbp), %rax + adcq \$0, %rdx + addq %r10, %r9 + movq %rdx, %r10 + adcq \$0, %r10 + + mulq %rbx + addq %rax, %r11 + movq 32(%rbp), %rax + adcq \$0, %rdx + addq %r11, %r10 + movq 128+8(%rsp), %rsi + #movq %rdx, %r11 + #adcq \$0, %r11 + adcq \$0, %rdx + movq %rdx, %r11 + + mulq %rbx + addq %rax, %r12 + movq 40(%rbp), %rax + adcq \$0, %rdx + imulq %r8, %rsi + addq %r12, %r11 + movq %rdx, %r12 + adcq \$0, %r12 + + mulq %rbx + addq %rax, %r13 + movq 48(%rbp), %rax + adcq \$0, %rdx + addq %r13, %r12 + movq %rdx, %r13 + adcq \$0, %r13 + + mulq %rbx + addq %rax, %r14 + movq 56(%rbp), %rax + adcq \$0, %rdx + addq %r14, %r13 + movq %rdx, %r14 + adcq \$0, %r14 + + mulq %rbx + movq %rsi, %rbx + addq %rax, %r15 + movq 0(%rbp), %rax + adcq \$0, %rdx + addq %r15, %r14 + movq %rdx, %r15 + adcq \$0, %r15 + + decl %ecx + jne .Lreduction_loop + + ret +.size __rsaz_512_reduce,.-__rsaz_512_reduce +___ +} +if ($addx) { + # __rsaz_512_reducex + # + # input: %r8-%r15, %rbp - mod, 128(%rsp) - n0 + # output: %r8-%r15 + # clobbers: everything except %rbp and %rdi +$code.=<<___; +.type __rsaz_512_reducex,\@abi-omnipotent +.align 32 +__rsaz_512_reducex: + #movq 128+8(%rsp), %rdx # pull $n0 + imulq %r8, %rdx + xorq %rsi, %rsi # cf=0,of=0 + movl \$8, %ecx + jmp .Lreduction_loopx + +.align 32 +.Lreduction_loopx: + mov %r8, %rbx + mulx 0(%rbp), %rax, %r8 + adcx %rbx, %rax + adox %r9, %r8 + + mulx 8(%rbp), %rax, %r9 + adcx %rax, %r8 + adox %r10, %r9 + + mulx 16(%rbp), %rbx, %r10 + adcx %rbx, %r9 + adox %r11, %r10 + + mulx 24(%rbp), %rbx, %r11 + adcx %rbx, %r10 + adox %r12, %r11 + + .byte 0xc4,0x62,0xe3,0xf6,0xa5,0x20,0x00,0x00,0x00 # mulx 32(%rbp), %rbx, %r12 + mov %rdx, %rax + mov %r8, %rdx + adcx %rbx, %r11 + adox %r13, %r12 + + mulx 128+8(%rsp), %rbx, %rdx + mov %rax, %rdx + + mulx 40(%rbp), %rax, %r13 + adcx %rax, %r12 + adox %r14, %r13 + + .byte 0xc4,0x62,0xfb,0xf6,0xb5,0x30,0x00,0x00,0x00 # mulx 48(%rbp), %rax, %r14 + adcx %rax, %r13 + adox %r15, %r14 + + mulx 56(%rbp), %rax, %r15 + mov %rbx, %rdx + adcx %rax, %r14 + adox %rsi, %r15 # %rsi is 0 + adcx %rsi, %r15 # cf=0 + + decl %ecx # of=0 + jne .Lreduction_loopx + + ret +.size __rsaz_512_reducex,.-__rsaz_512_reducex +___ +} +{ # __rsaz_512_subtract + # input: %r8-%r15, %rdi - $out, %rbp - $mod, %rcx - mask + # output: + # clobbers: everything but %rdi, %rsi and %rbp +$code.=<<___; +.type __rsaz_512_subtract,\@abi-omnipotent +.align 32 +__rsaz_512_subtract: + movq %r8, ($out) + movq %r9, 8($out) + movq %r10, 16($out) + movq %r11, 24($out) + movq %r12, 32($out) + movq %r13, 40($out) + movq %r14, 48($out) + movq %r15, 56($out) + + movq 0($mod), %r8 + movq 8($mod), %r9 + negq %r8 + notq %r9 + andq %rcx, %r8 + movq 16($mod), %r10 + andq %rcx, %r9 + notq %r10 + movq 24($mod), %r11 + andq %rcx, %r10 + notq %r11 + movq 32($mod), %r12 + andq %rcx, %r11 + notq %r12 + movq 40($mod), %r13 + andq %rcx, %r12 + notq %r13 + movq 48($mod), %r14 + andq %rcx, %r13 + notq %r14 + movq 56($mod), %r15 + andq %rcx, %r14 + notq %r15 + andq %rcx, %r15 + + addq ($out), %r8 + adcq 8($out), %r9 + adcq 16($out), %r10 + adcq 24($out), %r11 + adcq 32($out), %r12 + adcq 40($out), %r13 + adcq 48($out), %r14 + adcq 56($out), %r15 + + movq %r8, ($out) + movq %r9, 8($out) + movq %r10, 16($out) + movq %r11, 24($out) + movq %r12, 32($out) + movq %r13, 40($out) + movq %r14, 48($out) + movq %r15, 56($out) + + ret +.size __rsaz_512_subtract,.-__rsaz_512_subtract +___ +} +{ # __rsaz_512_mul + # + # input: %rsi - ap, %rbp - bp + # ouput: + # clobbers: everything +my ($ap,$bp) = ("%rsi","%rbp"); +$code.=<<___; +.type __rsaz_512_mul,\@abi-omnipotent +.align 32 +__rsaz_512_mul: + leaq 8(%rsp), %rdi + + movq ($ap), %rax + mulq %rbx + movq %rax, (%rdi) + movq 8($ap), %rax + movq %rdx, %r8 + + mulq %rbx + addq %rax, %r8 + movq 16($ap), %rax + movq %rdx, %r9 + adcq \$0, %r9 + + mulq %rbx + addq %rax, %r9 + movq 24($ap), %rax + movq %rdx, %r10 + adcq \$0, %r10 + + mulq %rbx + addq %rax, %r10 + movq 32($ap), %rax + movq %rdx, %r11 + adcq \$0, %r11 + + mulq %rbx + addq %rax, %r11 + movq 40($ap), %rax + movq %rdx, %r12 + adcq \$0, %r12 + + mulq %rbx + addq %rax, %r12 + movq 48($ap), %rax + movq %rdx, %r13 + adcq \$0, %r13 + + mulq %rbx + addq %rax, %r13 + movq 56($ap), %rax + movq %rdx, %r14 + adcq \$0, %r14 + + mulq %rbx + addq %rax, %r14 + movq ($ap), %rax + movq %rdx, %r15 + adcq \$0, %r15 + + leaq 8($bp), $bp + leaq 8(%rdi), %rdi + + movl \$7, %ecx + jmp .Loop_mul + +.align 32 +.Loop_mul: + movq ($bp), %rbx + mulq %rbx + addq %rax, %r8 + movq 8($ap), %rax + movq %r8, (%rdi) + movq %rdx, %r8 + adcq \$0, %r8 + + mulq %rbx + addq %rax, %r9 + movq 16($ap), %rax + adcq \$0, %rdx + addq %r9, %r8 + movq %rdx, %r9 + adcq \$0, %r9 + + mulq %rbx + addq %rax, %r10 + movq 24($ap), %rax + adcq \$0, %rdx + addq %r10, %r9 + movq %rdx, %r10 + adcq \$0, %r10 + + mulq %rbx + addq %rax, %r11 + movq 32($ap), %rax + adcq \$0, %rdx + addq %r11, %r10 + movq %rdx, %r11 + adcq \$0, %r11 + + mulq %rbx + addq %rax, %r12 + movq 40($ap), %rax + adcq \$0, %rdx + addq %r12, %r11 + movq %rdx, %r12 + adcq \$0, %r12 + + mulq %rbx + addq %rax, %r13 + movq 48($ap), %rax + adcq \$0, %rdx + addq %r13, %r12 + movq %rdx, %r13 + adcq \$0, %r13 + + mulq %rbx + addq %rax, %r14 + movq 56($ap), %rax + adcq \$0, %rdx + addq %r14, %r13 + movq %rdx, %r14 + leaq 8($bp), $bp + adcq \$0, %r14 + + mulq %rbx + addq %rax, %r15 + movq ($ap), %rax + adcq \$0, %rdx + addq %r15, %r14 + movq %rdx, %r15 + adcq \$0, %r15 + + leaq 8(%rdi), %rdi + + decl %ecx + jnz .Loop_mul + + movq %r8, (%rdi) + movq %r9, 8(%rdi) + movq %r10, 16(%rdi) + movq %r11, 24(%rdi) + movq %r12, 32(%rdi) + movq %r13, 40(%rdi) + movq %r14, 48(%rdi) + movq %r15, 56(%rdi) + + ret +.size __rsaz_512_mul,.-__rsaz_512_mul +___ +} +if ($addx) { + # __rsaz_512_mulx + # + # input: %rsi - ap, %rbp - bp + # ouput: + # clobbers: everything +my ($ap,$bp,$zero) = ("%rsi","%rbp","%rdi"); +$code.=<<___; +.type __rsaz_512_mulx,\@abi-omnipotent +.align 32 +__rsaz_512_mulx: + mulx ($ap), %rbx, %r8 # initial %rdx preloaded by caller + mov \$-6, %rcx + + mulx 8($ap), %rax, %r9 + movq %rbx, 8(%rsp) + + mulx 16($ap), %rbx, %r10 + adc %rax, %r8 + + mulx 24($ap), %rax, %r11 + adc %rbx, %r9 + + mulx 32($ap), %rbx, %r12 + adc %rax, %r10 + + mulx 40($ap), %rax, %r13 + adc %rbx, %r11 + + mulx 48($ap), %rbx, %r14 + adc %rax, %r12 + + mulx 56($ap), %rax, %r15 + mov 8($bp), %rdx + adc %rbx, %r13 + adc %rax, %r14 + adc \$0, %r15 + + xor $zero, $zero # cf=0,of=0 + jmp .Loop_mulx + +.align 32 +.Loop_mulx: + movq %r8, %rbx + mulx ($ap), %rax, %r8 + adcx %rax, %rbx + adox %r9, %r8 + + mulx 8($ap), %rax, %r9 + adcx %rax, %r8 + adox %r10, %r9 + + mulx 16($ap), %rax, %r10 + adcx %rax, %r9 + adox %r11, %r10 + + mulx 24($ap), %rax, %r11 + adcx %rax, %r10 + adox %r12, %r11 + + .byte 0x3e,0xc4,0x62,0xfb,0xf6,0xa6,0x20,0x00,0x00,0x00 # mulx 32($ap), %rax, %r12 + adcx %rax, %r11 + adox %r13, %r12 + + mulx 40($ap), %rax, %r13 + adcx %rax, %r12 + adox %r14, %r13 + + mulx 48($ap), %rax, %r14 + adcx %rax, %r13 + adox %r15, %r14 + + mulx 56($ap), %rax, %r15 + movq 64($bp,%rcx,8), %rdx + movq %rbx, 8+64-8(%rsp,%rcx,8) + adcx %rax, %r14 + adox $zero, %r15 + adcx $zero, %r15 # cf=0 + + inc %rcx # of=0 + jnz .Loop_mulx + + movq %r8, %rbx + mulx ($ap), %rax, %r8 + adcx %rax, %rbx + adox %r9, %r8 + + .byte 0xc4,0x62,0xfb,0xf6,0x8e,0x08,0x00,0x00,0x00 # mulx 8($ap), %rax, %r9 + adcx %rax, %r8 + adox %r10, %r9 + + .byte 0xc4,0x62,0xfb,0xf6,0x96,0x10,0x00,0x00,0x00 # mulx 16($ap), %rax, %r10 + adcx %rax, %r9 + adox %r11, %r10 + + mulx 24($ap), %rax, %r11 + adcx %rax, %r10 + adox %r12, %r11 + + mulx 32($ap), %rax, %r12 + adcx %rax, %r11 + adox %r13, %r12 + + mulx 40($ap), %rax, %r13 + adcx %rax, %r12 + adox %r14, %r13 + + .byte 0xc4,0x62,0xfb,0xf6,0xb6,0x30,0x00,0x00,0x00 # mulx 48($ap), %rax, %r14 + adcx %rax, %r13 + adox %r15, %r14 + + .byte 0xc4,0x62,0xfb,0xf6,0xbe,0x38,0x00,0x00,0x00 # mulx 56($ap), %rax, %r15 + adcx %rax, %r14 + adox $zero, %r15 + adcx $zero, %r15 + + mov %rbx, 8+64-8(%rsp) + mov %r8, 8+64(%rsp) + mov %r9, 8+64+8(%rsp) + mov %r10, 8+64+16(%rsp) + mov %r11, 8+64+24(%rsp) + mov %r12, 8+64+32(%rsp) + mov %r13, 8+64+40(%rsp) + mov %r14, 8+64+48(%rsp) + mov %r15, 8+64+56(%rsp) + + ret +.size __rsaz_512_mulx,.-__rsaz_512_mulx +___ +} +{ +my ($out,$inp,$power)= $win64 ? ("%rcx","%rdx","%r8d") : ("%rdi","%rsi","%edx"); +$code.=<<___; +.globl rsaz_512_scatter4 +.type rsaz_512_scatter4,\@abi-omnipotent +.align 16 +rsaz_512_scatter4: + leaq ($out,$power,4), $out + movl \$8, %r9d + jmp .Loop_scatter +.align 16 +.Loop_scatter: + movq ($inp), %rax + leaq 8($inp), $inp + movl %eax, ($out) + shrq \$32, %rax + movl %eax, 64($out) + leaq 128($out), $out + decl %r9d + jnz .Loop_scatter + ret +.size rsaz_512_scatter4,.-rsaz_512_scatter4 + +.globl rsaz_512_gather4 +.type rsaz_512_gather4,\@abi-omnipotent +.align 16 +rsaz_512_gather4: + leaq ($inp,$power,4), $inp + movl \$8, %r9d + jmp .Loop_gather +.align 16 +.Loop_gather: + movl ($inp), %eax + movl 64($inp), %r8d + leaq 128($inp), $inp + shlq \$32, %r8 + or %r8, %rax + movq %rax, ($out) + leaq 8($out), $out + decl %r9d + jnz .Loop_gather + ret +.size rsaz_512_gather4,.-rsaz_512_gather4 +___ +} + +# EXCEPTION_DISPOSITION handler (EXCEPTION_RECORD *rec,ULONG64 frame, +# CONTEXT *context,DISPATCHER_CONTEXT *disp) +if ($win64) { +$rec="%rcx"; +$frame="%rdx"; +$context="%r8"; +$disp="%r9"; + +$code.=<<___; +.extern __imp_RtlVirtualUnwind +.type se_handler,\@abi-omnipotent +.align 16 +se_handler: + push %rsi + push %rdi + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 + pushfq + sub \$64,%rsp + + mov 120($context),%rax # pull context->Rax + mov 248($context),%rbx # pull context->Rip + + mov 8($disp),%rsi # disp->ImageBase + mov 56($disp),%r11 # disp->HandlerData + + mov 0(%r11),%r10d # HandlerData[0] + lea (%rsi,%r10),%r10 # end of prologue label + cmp %r10,%rbx # context->Rip<end of prologue label + jb .Lcommon_seh_tail + + mov 152($context),%rax # pull context->Rsp + + mov 4(%r11),%r10d # HandlerData[1] + lea (%rsi,%r10),%r10 # epilogue label + cmp %r10,%rbx # context->Rip>=epilogue label + jae .Lcommon_seh_tail + + lea 128+24+48(%rax),%rax + + mov -8(%rax),%rbx + mov -16(%rax),%rbp + mov -24(%rax),%r12 + mov -32(%rax),%r13 + mov -40(%rax),%r14 + mov -48(%rax),%r15 + mov %rbx,144($context) # restore context->Rbx + mov %rbp,160($context) # restore context->Rbp + mov %r12,216($context) # restore context->R12 + mov %r13,224($context) # restore context->R13 + mov %r14,232($context) # restore context->R14 + mov %r15,240($context) # restore context->R15 + +.Lcommon_seh_tail: + mov 8(%rax),%rdi + mov 16(%rax),%rsi + mov %rax,152($context) # restore context->Rsp + mov %rsi,168($context) # restore context->Rsi + mov %rdi,176($context) # restore context->Rdi + + mov 40($disp),%rdi # disp->ContextRecord + mov $context,%rsi # context + mov \$154,%ecx # sizeof(CONTEXT) + .long 0xa548f3fc # cld; rep movsq + + mov $disp,%rsi + xor %rcx,%rcx # arg1, UNW_FLAG_NHANDLER + mov 8(%rsi),%rdx # arg2, disp->ImageBase + mov 0(%rsi),%r8 # arg3, disp->ControlPc + mov 16(%rsi),%r9 # arg4, disp->FunctionEntry + mov 40(%rsi),%r10 # disp->ContextRecord + lea 56(%rsi),%r11 # &disp->HandlerData + lea 24(%rsi),%r12 # &disp->EstablisherFrame + mov %r10,32(%rsp) # arg5 + mov %r11,40(%rsp) # arg6 + mov %r12,48(%rsp) # arg7 + mov %rcx,56(%rsp) # arg8, (NULL) + call *__imp_RtlVirtualUnwind(%rip) + + mov \$1,%eax # ExceptionContinueSearch + add \$64,%rsp + popfq + pop %r15 + pop %r14 + pop %r13 + pop %r12 + pop %rbp + pop %rbx + pop %rdi + pop %rsi + ret +.size sqr_handler,.-sqr_handler + +.section .pdata +.align 4 + .rva .LSEH_begin_rsaz_512_sqr + .rva .LSEH_end_rsaz_512_sqr + .rva .LSEH_info_rsaz_512_sqr + + .rva .LSEH_begin_rsaz_512_mul + .rva .LSEH_end_rsaz_512_mul + .rva .LSEH_info_rsaz_512_mul + + .rva .LSEH_begin_rsaz_512_mul_gather4 + .rva .LSEH_end_rsaz_512_mul_gather4 + .rva .LSEH_info_rsaz_512_mul_gather4 + + .rva .LSEH_begin_rsaz_512_mul_scatter4 + .rva .LSEH_end_rsaz_512_mul_scatter4 + .rva .LSEH_info_rsaz_512_mul_scatter4 + + .rva .LSEH_begin_rsaz_512_mul_by_one + .rva .LSEH_end_rsaz_512_mul_by_one + .rva .LSEH_info_rsaz_512_mul_by_one + +.section .xdata +.align 8 +.LSEH_info_rsaz_512_sqr: + .byte 9,0,0,0 + .rva se_handler + .rva .Lsqr_body,.Lsqr_epilogue # HandlerData[] +.LSEH_info_rsaz_512_mul: + .byte 9,0,0,0 + .rva se_handler + .rva .Lmul_body,.Lmul_epilogue # HandlerData[] +.LSEH_info_rsaz_512_mul_gather4: + .byte 9,0,0,0 + .rva se_handler + .rva .Lmul_gather4_body,.Lmul_gather4_epilogue # HandlerData[] +.LSEH_info_rsaz_512_mul_scatter4: + .byte 9,0,0,0 + .rva se_handler + .rva .Lmul_scatter4_body,.Lmul_scatter4_epilogue # HandlerData[] +.LSEH_info_rsaz_512_mul_by_one: + .byte 9,0,0,0 + .rva se_handler + .rva .Lmul_by_one_body,.Lmul_by_one_epilogue # HandlerData[] +___ +} + +$code =~ s/\`([^\`]*)\`/eval $1/gem; +print $code; +close STDOUT; diff --git a/openssl/crypto/bn/asm/sparct4-mont.pl b/openssl/crypto/bn/asm/sparct4-mont.pl new file mode 100755 index 000000000..71b45002a --- /dev/null +++ b/openssl/crypto/bn/asm/sparct4-mont.pl @@ -0,0 +1,1222 @@ +#!/usr/bin/env perl + +# ==================================================================== +# Written by David S. Miller <davem@devemloft.net> and Andy Polyakov +# <appro@openssl.org>. The module is licensed under 2-clause BSD +# license. November 2012. All rights reserved. +# ==================================================================== + +###################################################################### +# Montgomery squaring-n-multiplication module for SPARC T4. +# +# The module consists of three parts: +# +# 1) collection of "single-op" subroutines that perform single +# operation, Montgomery squaring or multiplication, on 512-, +# 1024-, 1536- and 2048-bit operands; +# 2) collection of "multi-op" subroutines that perform 5 squaring and +# 1 multiplication operations on operands of above lengths; +# 3) fall-back and helper VIS3 subroutines. +# +# RSA sign is dominated by multi-op subroutine, while RSA verify and +# DSA - by single-op. Special note about 4096-bit RSA verify result. +# Operands are too long for dedicated hardware and it's handled by +# VIS3 code, which is why you don't see any improvement. It's surely +# possible to improve it [by deploying 'mpmul' instruction], maybe in +# the future... +# +# Performance improvement. +# +# 64-bit process, VIS3: +# sign verify sign/s verify/s +# rsa 1024 bits 0.000628s 0.000028s 1592.4 35434.4 +# rsa 2048 bits 0.003282s 0.000106s 304.7 9438.3 +# rsa 4096 bits 0.025866s 0.000340s 38.7 2940.9 +# dsa 1024 bits 0.000301s 0.000332s 3323.7 3013.9 +# dsa 2048 bits 0.001056s 0.001233s 946.9 810.8 +# +# 64-bit process, this module: +# sign verify sign/s verify/s +# rsa 1024 bits 0.000256s 0.000016s 3904.4 61411.9 +# rsa 2048 bits 0.000946s 0.000029s 1056.8 34292.7 +# rsa 4096 bits 0.005061s 0.000340s 197.6 2940.5 +# dsa 1024 bits 0.000176s 0.000195s 5674.7 5130.5 +# dsa 2048 bits 0.000296s 0.000354s 3383.2 2827.6 +# +###################################################################### +# 32-bit process, VIS3: +# sign verify sign/s verify/s +# rsa 1024 bits 0.000665s 0.000028s 1504.8 35233.3 +# rsa 2048 bits 0.003349s 0.000106s 298.6 9433.4 +# rsa 4096 bits 0.025959s 0.000341s 38.5 2934.8 +# dsa 1024 bits 0.000320s 0.000341s 3123.3 2929.6 +# dsa 2048 bits 0.001101s 0.001260s 908.2 793.4 +# +# 32-bit process, this module: +# sign verify sign/s verify/s +# rsa 1024 bits 0.000301s 0.000017s 3317.1 60240.0 +# rsa 2048 bits 0.001034s 0.000030s 966.9 33812.7 +# rsa 4096 bits 0.005244s 0.000341s 190.7 2935.4 +# dsa 1024 bits 0.000201s 0.000205s 4976.1 4879.2 +# dsa 2048 bits 0.000328s 0.000360s 3051.1 2774.2 +# +# 32-bit code is prone to performance degradation as interrupt rate +# dispatched to CPU executing the code grows. This is because in +# standard process of handling interrupt in 32-bit process context +# upper halves of most integer registers used as input or output are +# zeroed. This renders result invalid, and operation has to be re-run. +# If CPU is "bothered" with timer interrupts only, the penalty is +# hardly measurable. But in order to mitigate this problem for higher +# interrupt rates contemporary Linux kernel recognizes biased stack +# even in 32-bit process context and preserves full register contents. +# See http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=517ffce4e1a03aea979fe3a18a3dd1761a24fafb +# for details. + +$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; +push(@INC,"${dir}","${dir}../../perlasm"); +require "sparcv9_modes.pl"; + +$code.=<<___; +#include "sparc_arch.h" + +#ifdef __arch64__ +.register %g2,#scratch +.register %g3,#scratch +#endif + +.section ".text",#alloc,#execinstr + +#ifdef __PIC__ +SPARC_PIC_THUNK(%g1) +#endif +___ + +######################################################################## +# Register layout for mont[mul|sqr] instructions. +# For details see "Oracle SPARC Architecture 2011" manual at +# http://www.oracle.com/technetwork/server-storage/sun-sparc-enterprise/documentation/. +# +my @R=map("%f".2*$_,(0..11,30,31,12..29)); +my @N=(map("%l$_",(0..7)),map("%o$_",(0..5))); @N=(@N,@N,@N[0..3]); +my @A=(@N[0..13],@R[14..31]); +my @B=(map("%i$_",(0..5)),map("%l$_",(0..7))); @B=(@B,@B,map("%o$_",(0..3))); + +######################################################################## +# int bn_mul_mont_t4_$NUM(u64 *rp,const u64 *ap,const u64 *bp, +# const u64 *np,const BN_ULONG *n0); +# +sub generate_bn_mul_mont_t4() { +my $NUM=shift; +my ($rp,$ap,$bp,$np,$sentinel)=map("%g$_",(1..5)); + +$code.=<<___; +.globl bn_mul_mont_t4_$NUM +.align 32 +bn_mul_mont_t4_$NUM: +#ifdef __arch64__ + mov 0,$sentinel + mov -128,%g4 +#elif defined(SPARCV9_64BIT_STACK) + SPARC_LOAD_ADDRESS_LEAF(OPENSSL_sparcv9cap_P,%g1,%g5) + ld [%g1+0],%g1 ! OPENSSL_sparcv9_P[0] + mov -2047,%g4 + and %g1,SPARCV9_64BIT_STACK,%g1 + movrz %g1,0,%g4 + mov -1,$sentinel + add %g4,-128,%g4 +#else + mov -1,$sentinel + mov -128,%g4 +#endif + sllx $sentinel,32,$sentinel + save %sp,%g4,%sp +#ifndef __arch64__ + save %sp,-128,%sp ! warm it up + save %sp,-128,%sp + save %sp,-128,%sp + save %sp,-128,%sp + save %sp,-128,%sp + save %sp,-128,%sp + restore + restore + restore + restore + restore + restore +#endif + and %sp,1,%g4 + or $sentinel,%fp,%fp + or %g4,$sentinel,$sentinel + + ! copy arguments to global registers + mov %i0,$rp + mov %i1,$ap + mov %i2,$bp + mov %i3,$np + ld [%i4+0],%f1 ! load *n0 + ld [%i4+4],%f0 + fsrc2 %f0,%f60 +___ + +# load ap[$NUM] ######################################################## +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for($i=0; $i<14 && $i<$NUM; $i++) { +my $lo=$i<13?@A[$i+1]:"%o7"; +$code.=<<___; + ld [$ap+$i*8+0],$lo + ld [$ap+$i*8+4],@A[$i] + sllx @A[$i],32,@A[$i] + or $lo,@A[$i],@A[$i] +___ +} +for(; $i<$NUM; $i++) { +my ($hi,$lo)=("%f".2*($i%4),"%f".(2*($i%4)+1)); +$code.=<<___; + ld [$ap+$i*8+0],$lo + ld [$ap+$i*8+4],$hi + fsrc2 $hi,@A[$i] +___ +} +# load np[$NUM] ######################################################## +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for($i=0; $i<14 && $i<$NUM; $i++) { +my $lo=$i<13?@N[$i+1]:"%o7"; +$code.=<<___; + ld [$np+$i*8+0],$lo + ld [$np+$i*8+4],@N[$i] + sllx @N[$i],32,@N[$i] + or $lo,@N[$i],@N[$i] +___ +} +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for(; $i<28 && $i<$NUM; $i++) { +my $lo=$i<27?@N[$i+1]:"%o7"; +$code.=<<___; + ld [$np+$i*8+0],$lo + ld [$np+$i*8+4],@N[$i] + sllx @N[$i],32,@N[$i] + or $lo,@N[$i],@N[$i] +___ +} +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for(; $i<$NUM; $i++) { +my $lo=($i<$NUM-1)?@N[$i+1]:"%o7"; +$code.=<<___; + ld [$np+$i*8+0],$lo + ld [$np+$i*8+4],@N[$i] + sllx @N[$i],32,@N[$i] + or $lo,@N[$i],@N[$i] +___ +} +$code.=<<___; + cmp $ap,$bp + be SIZE_T_CC,.Lmsquare_$NUM + nop +___ + +# load bp[$NUM] ######################################################## +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for($i=0; $i<14 && $i<$NUM; $i++) { +my $lo=$i<13?@B[$i+1]:"%o7"; +$code.=<<___; + ld [$bp+$i*8+0],$lo + ld [$bp+$i*8+4],@B[$i] + sllx @B[$i],32,@B[$i] + or $lo,@B[$i],@B[$i] +___ +} +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for(; $i<$NUM; $i++) { +my $lo=($i<$NUM-1)?@B[$i+1]:"%o7"; +$code.=<<___; + ld [$bp+$i*8+0],$lo + ld [$bp+$i*8+4],@B[$i] + sllx @B[$i],32,@B[$i] + or $lo,@B[$i],@B[$i] +___ +} +# magic ################################################################ +$code.=<<___; + .word 0x81b02920+$NUM-1 ! montmul $NUM-1 +.Lmresume_$NUM: + fbu,pn %fcc3,.Lmabort_$NUM +#ifndef __arch64__ + and %fp,$sentinel,$sentinel + brz,pn $sentinel,.Lmabort_$NUM +#endif + nop +#ifdef __arch64__ + restore + restore + restore + restore + restore +#else + restore; and %fp,$sentinel,$sentinel + restore; and %fp,$sentinel,$sentinel + restore; and %fp,$sentinel,$sentinel + restore; and %fp,$sentinel,$sentinel + brz,pn $sentinel,.Lmabort1_$NUM + restore +#endif +___ + +# save tp[$NUM] ######################################################## +for($i=0; $i<14 && $i<$NUM; $i++) { +$code.=<<___; + movxtod @A[$i],@R[$i] +___ +} +$code.=<<___; +#ifdef __arch64__ + restore +#else + and %fp,$sentinel,$sentinel + restore + and $sentinel,1,%o7 + and %fp,$sentinel,$sentinel + srl %fp,0,%fp ! just in case? + or %o7,$sentinel,$sentinel + brz,a,pn $sentinel,.Lmdone_$NUM + mov 0,%i0 ! return failure +#endif +___ +for($i=0; $i<12 && $i<$NUM; $i++) { +@R[$i] =~ /%f([0-9]+)/; +my $lo = "%f".($1+1); +$code.=<<___; + st $lo,[$rp+$i*8+0] + st @R[$i],[$rp+$i*8+4] +___ +} +for(; $i<$NUM; $i++) { +my ($hi,$lo)=("%f".2*($i%4),"%f".(2*($i%4)+1)); +$code.=<<___; + fsrc2 @R[$i],$hi + st $lo,[$rp+$i*8+0] + st $hi,[$rp+$i*8+4] +___ +} +$code.=<<___; + mov 1,%i0 ! return success +.Lmdone_$NUM: + ret + restore + +.Lmabort_$NUM: + restore + restore + restore + restore + restore +.Lmabort1_$NUM: + restore + + mov 0,%i0 ! return failure + ret + restore + +.align 32 +.Lmsquare_$NUM: + save %sp,-128,%sp; or $sentinel,%fp,%fp + save %sp,-128,%sp; or $sentinel,%fp,%fp + .word 0x81b02940+$NUM-1 ! montsqr $NUM-1 + ba .Lmresume_$NUM + nop +.type bn_mul_mont_t4_$NUM, #function +.size bn_mul_mont_t4_$NUM, .-bn_mul_mont_t4_$NUM +___ +} + +for ($i=8;$i<=32;$i+=8) { + &generate_bn_mul_mont_t4($i); +} + +######################################################################## +# +sub load_ccr { +my ($ptbl,$pwr,$ccr,$skip_wr)=@_; +$code.=<<___; + srl $pwr, 2, %o4 + and $pwr, 3, %o5 + and %o4, 7, %o4 + sll %o5, 3, %o5 ! offset within first cache line + add %o5, $ptbl, $ptbl ! of the pwrtbl + or %g0, 1, %o5 + sll %o5, %o4, $ccr +___ +$code.=<<___ if (!$skip_wr); + wr $ccr, %g0, %ccr +___ +} +sub load_b_pair { +my ($pwrtbl,$B0,$B1)=@_; + +$code.=<<___; + ldx [$pwrtbl+0*32], $B0 + ldx [$pwrtbl+8*32], $B1 + ldx [$pwrtbl+1*32], %o4 + ldx [$pwrtbl+9*32], %o5 + movvs %icc, %o4, $B0 + ldx [$pwrtbl+2*32], %o4 + movvs %icc, %o5, $B1 + ldx [$pwrtbl+10*32],%o5 + move %icc, %o4, $B0 + ldx [$pwrtbl+3*32], %o4 + move %icc, %o5, $B1 + ldx [$pwrtbl+11*32],%o5 + movneg %icc, %o4, $B0 + ldx [$pwrtbl+4*32], %o4 + movneg %icc, %o5, $B1 + ldx [$pwrtbl+12*32],%o5 + movcs %xcc, %o4, $B0 + ldx [$pwrtbl+5*32],%o4 + movcs %xcc, %o5, $B1 + ldx [$pwrtbl+13*32],%o5 + movvs %xcc, %o4, $B0 + ldx [$pwrtbl+6*32], %o4 + movvs %xcc, %o5, $B1 + ldx [$pwrtbl+14*32],%o5 + move %xcc, %o4, $B0 + ldx [$pwrtbl+7*32], %o4 + move %xcc, %o5, $B1 + ldx [$pwrtbl+15*32],%o5 + movneg %xcc, %o4, $B0 + add $pwrtbl,16*32, $pwrtbl + movneg %xcc, %o5, $B1 +___ +} +sub load_b { +my ($pwrtbl,$Bi)=@_; + +$code.=<<___; + ldx [$pwrtbl+0*32], $Bi + ldx [$pwrtbl+1*32], %o4 + ldx [$pwrtbl+2*32], %o5 + movvs %icc, %o4, $Bi + ldx [$pwrtbl+3*32], %o4 + move %icc, %o5, $Bi + ldx [$pwrtbl+4*32], %o5 + movneg %icc, %o4, $Bi + ldx [$pwrtbl+5*32], %o4 + movcs %xcc, %o5, $Bi + ldx [$pwrtbl+6*32], %o5 + movvs %xcc, %o4, $Bi + ldx [$pwrtbl+7*32], %o4 + move %xcc, %o5, $Bi + add $pwrtbl,8*32, $pwrtbl + movneg %xcc, %o4, $Bi +___ +} + +######################################################################## +# int bn_pwr5_mont_t4_$NUM(u64 *tp,const u64 *np,const BN_ULONG *n0, +# const u64 *pwrtbl,int pwr,int stride); +# +sub generate_bn_pwr5_mont_t4() { +my $NUM=shift; +my ($tp,$np,$pwrtbl,$pwr,$sentinel)=map("%g$_",(1..5)); + +$code.=<<___; +.globl bn_pwr5_mont_t4_$NUM +.align 32 +bn_pwr5_mont_t4_$NUM: +#ifdef __arch64__ + mov 0,$sentinel + mov -128,%g4 +#elif defined(SPARCV9_64BIT_STACK) + SPARC_LOAD_ADDRESS_LEAF(OPENSSL_sparcv9cap_P,%g1,%g5) + ld [%g1+0],%g1 ! OPENSSL_sparcv9_P[0] + mov -2047,%g4 + and %g1,SPARCV9_64BIT_STACK,%g1 + movrz %g1,0,%g4 + mov -1,$sentinel + add %g4,-128,%g4 +#else + mov -1,$sentinel + mov -128,%g4 +#endif + sllx $sentinel,32,$sentinel + save %sp,%g4,%sp +#ifndef __arch64__ + save %sp,-128,%sp ! warm it up + save %sp,-128,%sp + save %sp,-128,%sp + save %sp,-128,%sp + save %sp,-128,%sp + save %sp,-128,%sp + restore + restore + restore + restore + restore + restore +#endif + and %sp,1,%g4 + or $sentinel,%fp,%fp + or %g4,$sentinel,$sentinel + + ! copy arguments to global registers + mov %i0,$tp + mov %i1,$np + ld [%i2+0],%f1 ! load *n0 + ld [%i2+4],%f0 + mov %i3,$pwrtbl + srl %i4,%g0,%i4 ! pack last arguments + sllx %i5,32,$pwr + or %i4,$pwr,$pwr + fsrc2 %f0,%f60 +___ + +# load tp[$NUM] ######################################################## +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for($i=0; $i<14 && $i<$NUM; $i++) { +$code.=<<___; + ldx [$tp+$i*8],@A[$i] +___ +} +for(; $i<$NUM; $i++) { +$code.=<<___; + ldd [$tp+$i*8],@A[$i] +___ +} +# load np[$NUM] ######################################################## +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for($i=0; $i<14 && $i<$NUM; $i++) { +$code.=<<___; + ldx [$np+$i*8],@N[$i] +___ +} +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for(; $i<28 && $i<$NUM; $i++) { +$code.=<<___; + ldx [$np+$i*8],@N[$i] +___ +} +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for(; $i<$NUM; $i++) { +$code.=<<___; + ldx [$np+$i*8],@N[$i] +___ +} +# load pwrtbl[pwr] ######################################################## +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp + + srlx $pwr, 32, %o4 ! unpack $pwr + srl $pwr, %g0, %o5 + sub %o4, 5, %o4 + mov $pwrtbl, %o7 + sllx %o4, 32, $pwr ! re-pack $pwr + or %o5, $pwr, $pwr + srl %o5, %o4, %o5 +___ + &load_ccr("%o7","%o5","%o4"); +$code.=<<___; + b .Lstride_$NUM + nop +.align 16 +.Lstride_$NUM: +___ +for($i=0; $i<14 && $i<$NUM; $i+=2) { + &load_b_pair("%o7",@B[$i],@B[$i+1]); +} +$code.=<<___; + save %sp,-128,%sp; or $sentinel,%fp,%fp +___ +for(; $i<$NUM; $i+=2) { + &load_b_pair("%i7",@B[$i],@B[$i+1]); +} +$code.=<<___; + srax $pwr, 32, %o4 ! unpack $pwr + srl $pwr, %g0, %o5 + sub %o4, 5, %o4 + mov $pwrtbl, %i7 + sllx %o4, 32, $pwr ! re-pack $pwr + or %o5, $pwr, $pwr + srl %o5, %o4, %o5 +___ + &load_ccr("%i7","%o5","%o4",1); + +# magic ################################################################ +for($i=0; $i<5; $i++) { +$code.=<<___; + .word 0x81b02940+$NUM-1 ! montsqr $NUM-1 + fbu,pn %fcc3,.Labort_$NUM +#ifndef __arch64__ + and %fp,$sentinel,$sentinel + brz,pn $sentinel,.Labort_$NUM +#endif + nop +___ +} +$code.=<<___; + wr %o4, %g0, %ccr + .word 0x81b02920+$NUM-1 ! montmul $NUM-1 + fbu,pn %fcc3,.Labort_$NUM +#ifndef __arch64__ + and %fp,$sentinel,$sentinel + brz,pn $sentinel,.Labort_$NUM +#endif + + srax $pwr, 32, %o4 +#ifdef __arch64__ + brgez %o4,.Lstride_$NUM + restore + restore + restore + restore + restore +#else + brgez %o4,.Lstride_$NUM + restore; and %fp,$sentinel,$sentinel + restore; and %fp,$sentinel,$sentinel + restore; and %fp,$sentinel,$sentinel + restore; and %fp,$sentinel,$sentinel + brz,pn $sentinel,.Labort1_$NUM + restore +#endif +___ + +# save tp[$NUM] ######################################################## +for($i=0; $i<14 && $i<$NUM; $i++) { +$code.=<<___; + movxtod @A[$i],@R[$i] +___ +} +$code.=<<___; +#ifdef __arch64__ + restore +#else + and %fp,$sentinel,$sentinel + restore + and $sentinel,1,%o7 + and %fp,$sentinel,$sentinel + srl %fp,0,%fp ! just in case? + or %o7,$sentinel,$sentinel + brz,a,pn $sentinel,.Ldone_$NUM + mov 0,%i0 ! return failure +#endif +___ +for($i=0; $i<$NUM; $i++) { +$code.=<<___; + std @R[$i],[$tp+$i*8] +___ +} +$code.=<<___; + mov 1,%i0 ! return success +.Ldone_$NUM: + ret + restore + +.Labort_$NUM: + restore + restore + restore + restore + restore +.Labort1_$NUM: + restore + + mov 0,%i0 ! return failure + ret + restore +.type bn_pwr5_mont_t4_$NUM, #function +.size bn_pwr5_mont_t4_$NUM, .-bn_pwr5_mont_t4_$NUM +___ +} + +for ($i=8;$i<=32;$i+=8) { + &generate_bn_pwr5_mont_t4($i); +} + +{ +######################################################################## +# Fall-back subroutines +# +# copy of bn_mul_mont_vis3 adjusted for vectors of 64-bit values +# +($n0,$m0,$m1,$lo0,$hi0, $lo1,$hi1,$aj,$alo,$nj,$nlo,$tj)= + (map("%g$_",(1..5)),map("%o$_",(0..5,7))); + +# int bn_mul_mont( +$rp="%o0"; # u64 *rp, +$ap="%o1"; # const u64 *ap, +$bp="%o2"; # const u64 *bp, +$np="%o3"; # const u64 *np, +$n0p="%o4"; # const BN_ULONG *n0, +$num="%o5"; # int num); # caller ensures that num is >=3 +$code.=<<___; +.globl bn_mul_mont_t4 +.align 32 +bn_mul_mont_t4: + add %sp, STACK_BIAS, %g4 ! real top of stack + sll $num, 3, $num ! size in bytes + add $num, 63, %g1 + andn %g1, 63, %g1 ! buffer size rounded up to 64 bytes + sub %g4, %g1, %g1 + andn %g1, 63, %g1 ! align at 64 byte + sub %g1, STACK_FRAME, %g1 ! new top of stack + sub %g1, %g4, %g1 + + save %sp, %g1, %sp +___ +# +-------------------------------+<----- %sp +# . . +# +-------------------------------+<----- aligned at 64 bytes +# | __int64 tmp[0] | +# +-------------------------------+ +# . . +# . . +# +-------------------------------+<----- aligned at 64 bytes +# . . +($rp,$ap,$bp,$np,$n0p,$num)=map("%i$_",(0..5)); +($t0,$t1,$t2,$t3,$cnt,$tp,$bufsz)=map("%l$_",(0..7)); +($ovf,$i)=($t0,$t1); +$code.=<<___; + ld [$n0p+0], $t0 ! pull n0[0..1] value + ld [$n0p+4], $t1 + add %sp, STACK_BIAS+STACK_FRAME, $tp + ldx [$bp+0], $m0 ! m0=bp[0] + sllx $t1, 32, $n0 + add $bp, 8, $bp + or $t0, $n0, $n0 + + ldx [$ap+0], $aj ! ap[0] + + mulx $aj, $m0, $lo0 ! ap[0]*bp[0] + umulxhi $aj, $m0, $hi0 + + ldx [$ap+8], $aj ! ap[1] + add $ap, 16, $ap + ldx [$np+0], $nj ! np[0] + + mulx $lo0, $n0, $m1 ! "tp[0]"*n0 + + mulx $aj, $m0, $alo ! ap[1]*bp[0] + umulxhi $aj, $m0, $aj ! ahi=aj + + mulx $nj, $m1, $lo1 ! np[0]*m1 + umulxhi $nj, $m1, $hi1 + + ldx [$np+8], $nj ! np[1] + + addcc $lo0, $lo1, $lo1 + add $np, 16, $np + addxc %g0, $hi1, $hi1 + + mulx $nj, $m1, $nlo ! np[1]*m1 + umulxhi $nj, $m1, $nj ! nhi=nj + + ba .L1st + sub $num, 24, $cnt ! cnt=num-3 + +.align 16 +.L1st: + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 + + ldx [$ap+0], $aj ! ap[j] + addcc $nlo, $hi1, $lo1 + add $ap, 8, $ap + addxc $nj, %g0, $hi1 ! nhi=nj + + ldx [$np+0], $nj ! np[j] + mulx $aj, $m0, $alo ! ap[j]*bp[0] + add $np, 8, $np + umulxhi $aj, $m0, $aj ! ahi=aj + + mulx $nj, $m1, $nlo ! np[j]*m1 + addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0] + umulxhi $nj, $m1, $nj ! nhi=nj + addxc %g0, $hi1, $hi1 + stxa $lo1, [$tp]0xe2 ! tp[j-1] + add $tp, 8, $tp ! tp++ + + brnz,pt $cnt, .L1st + sub $cnt, 8, $cnt ! j-- +!.L1st + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 ! ahi=aj + + addcc $nlo, $hi1, $lo1 + addxc $nj, %g0, $hi1 + addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0] + addxc %g0, $hi1, $hi1 + stxa $lo1, [$tp]0xe2 ! tp[j-1] + add $tp, 8, $tp + + addcc $hi0, $hi1, $hi1 + addxc %g0, %g0, $ovf ! upmost overflow bit + stxa $hi1, [$tp]0xe2 + add $tp, 8, $tp + + ba .Louter + sub $num, 16, $i ! i=num-2 + +.align 16 +.Louter: + ldx [$bp+0], $m0 ! m0=bp[i] + add $bp, 8, $bp + + sub $ap, $num, $ap ! rewind + sub $np, $num, $np + sub $tp, $num, $tp + + ldx [$ap+0], $aj ! ap[0] + ldx [$np+0], $nj ! np[0] + + mulx $aj, $m0, $lo0 ! ap[0]*bp[i] + ldx [$tp], $tj ! tp[0] + umulxhi $aj, $m0, $hi0 + ldx [$ap+8], $aj ! ap[1] + addcc $lo0, $tj, $lo0 ! ap[0]*bp[i]+tp[0] + mulx $aj, $m0, $alo ! ap[1]*bp[i] + addxc %g0, $hi0, $hi0 + mulx $lo0, $n0, $m1 ! tp[0]*n0 + umulxhi $aj, $m0, $aj ! ahi=aj + mulx $nj, $m1, $lo1 ! np[0]*m1 + add $ap, 16, $ap + umulxhi $nj, $m1, $hi1 + ldx [$np+8], $nj ! np[1] + add $np, 16, $np + addcc $lo1, $lo0, $lo1 + mulx $nj, $m1, $nlo ! np[1]*m1 + addxc %g0, $hi1, $hi1 + umulxhi $nj, $m1, $nj ! nhi=nj + + ba .Linner + sub $num, 24, $cnt ! cnt=num-3 +.align 16 +.Linner: + addcc $alo, $hi0, $lo0 + ldx [$tp+8], $tj ! tp[j] + addxc $aj, %g0, $hi0 ! ahi=aj + ldx [$ap+0], $aj ! ap[j] + add $ap, 8, $ap + addcc $nlo, $hi1, $lo1 + mulx $aj, $m0, $alo ! ap[j]*bp[i] + addxc $nj, %g0, $hi1 ! nhi=nj + ldx [$np+0], $nj ! np[j] + add $np, 8, $np + umulxhi $aj, $m0, $aj ! ahi=aj + addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j] + mulx $nj, $m1, $nlo ! np[j]*m1 + addxc %g0, $hi0, $hi0 + umulxhi $nj, $m1, $nj ! nhi=nj + addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j] + addxc %g0, $hi1, $hi1 + stx $lo1, [$tp] ! tp[j-1] + add $tp, 8, $tp + brnz,pt $cnt, .Linner + sub $cnt, 8, $cnt +!.Linner + ldx [$tp+8], $tj ! tp[j] + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 ! ahi=aj + addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j] + addxc %g0, $hi0, $hi0 + + addcc $nlo, $hi1, $lo1 + addxc $nj, %g0, $hi1 ! nhi=nj + addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j] + addxc %g0, $hi1, $hi1 + stx $lo1, [$tp] ! tp[j-1] + + subcc %g0, $ovf, %g0 ! move upmost overflow to CCR.xcc + addxccc $hi1, $hi0, $hi1 + addxc %g0, %g0, $ovf + stx $hi1, [$tp+8] + add $tp, 16, $tp + + brnz,pt $i, .Louter + sub $i, 8, $i + + sub $ap, $num, $ap ! rewind + sub $np, $num, $np + sub $tp, $num, $tp + ba .Lsub + subcc $num, 8, $cnt ! cnt=num-1 and clear CCR.xcc + +.align 16 +.Lsub: + ldx [$tp], $tj + add $tp, 8, $tp + ldx [$np+0], $nj + add $np, 8, $np + subccc $tj, $nj, $t2 ! tp[j]-np[j] + srlx $tj, 32, $tj + srlx $nj, 32, $nj + subccc $tj, $nj, $t3 + add $rp, 8, $rp + st $t2, [$rp-4] ! reverse order + st $t3, [$rp-8] + brnz,pt $cnt, .Lsub + sub $cnt, 8, $cnt + + sub $np, $num, $np ! rewind + sub $tp, $num, $tp + sub $rp, $num, $rp + + subc $ovf, %g0, $ovf ! handle upmost overflow bit + and $tp, $ovf, $ap + andn $rp, $ovf, $np + or $np, $ap, $ap ! ap=borrow?tp:rp + ba .Lcopy + sub $num, 8, $cnt + +.align 16 +.Lcopy: ! copy or in-place refresh + ldx [$ap+0], $t2 + add $ap, 8, $ap + stx %g0, [$tp] ! zap + add $tp, 8, $tp + stx $t2, [$rp+0] + add $rp, 8, $rp + brnz $cnt, .Lcopy + sub $cnt, 8, $cnt + + mov 1, %o0 + ret + restore +.type bn_mul_mont_t4, #function +.size bn_mul_mont_t4, .-bn_mul_mont_t4 +___ + +# int bn_mul_mont_gather5( +$rp="%o0"; # u64 *rp, +$ap="%o1"; # const u64 *ap, +$bp="%o2"; # const u64 *pwrtbl, +$np="%o3"; # const u64 *np, +$n0p="%o4"; # const BN_ULONG *n0, +$num="%o5"; # int num, # caller ensures that num is >=3 + # int power); +$code.=<<___; +.globl bn_mul_mont_gather5_t4 +.align 32 +bn_mul_mont_gather5_t4: + add %sp, STACK_BIAS, %g4 ! real top of stack + sll $num, 3, $num ! size in bytes + add $num, 63, %g1 + andn %g1, 63, %g1 ! buffer size rounded up to 64 bytes + sub %g4, %g1, %g1 + andn %g1, 63, %g1 ! align at 64 byte + sub %g1, STACK_FRAME, %g1 ! new top of stack + sub %g1, %g4, %g1 + LDPTR [%sp+STACK_7thARG], %g4 ! load power, 7th argument + + save %sp, %g1, %sp +___ +# +-------------------------------+<----- %sp +# . . +# +-------------------------------+<----- aligned at 64 bytes +# | __int64 tmp[0] | +# +-------------------------------+ +# . . +# . . +# +-------------------------------+<----- aligned at 64 bytes +# . . +($rp,$ap,$bp,$np,$n0p,$num)=map("%i$_",(0..5)); +($t0,$t1,$t2,$t3,$cnt,$tp,$bufsz,$ccr)=map("%l$_",(0..7)); +($ovf,$i)=($t0,$t1); + &load_ccr($bp,"%g4",$ccr); + &load_b($bp,$m0,"%o7"); # m0=bp[0] + +$code.=<<___; + ld [$n0p+0], $t0 ! pull n0[0..1] value + ld [$n0p+4], $t1 + add %sp, STACK_BIAS+STACK_FRAME, $tp + sllx $t1, 32, $n0 + or $t0, $n0, $n0 + + ldx [$ap+0], $aj ! ap[0] + + mulx $aj, $m0, $lo0 ! ap[0]*bp[0] + umulxhi $aj, $m0, $hi0 + + ldx [$ap+8], $aj ! ap[1] + add $ap, 16, $ap + ldx [$np+0], $nj ! np[0] + + mulx $lo0, $n0, $m1 ! "tp[0]"*n0 + + mulx $aj, $m0, $alo ! ap[1]*bp[0] + umulxhi $aj, $m0, $aj ! ahi=aj + + mulx $nj, $m1, $lo1 ! np[0]*m1 + umulxhi $nj, $m1, $hi1 + + ldx [$np+8], $nj ! np[1] + + addcc $lo0, $lo1, $lo1 + add $np, 16, $np + addxc %g0, $hi1, $hi1 + + mulx $nj, $m1, $nlo ! np[1]*m1 + umulxhi $nj, $m1, $nj ! nhi=nj + + ba .L1st_g5 + sub $num, 24, $cnt ! cnt=num-3 + +.align 16 +.L1st_g5: + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 + + ldx [$ap+0], $aj ! ap[j] + addcc $nlo, $hi1, $lo1 + add $ap, 8, $ap + addxc $nj, %g0, $hi1 ! nhi=nj + + ldx [$np+0], $nj ! np[j] + mulx $aj, $m0, $alo ! ap[j]*bp[0] + add $np, 8, $np + umulxhi $aj, $m0, $aj ! ahi=aj + + mulx $nj, $m1, $nlo ! np[j]*m1 + addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0] + umulxhi $nj, $m1, $nj ! nhi=nj + addxc %g0, $hi1, $hi1 + stxa $lo1, [$tp]0xe2 ! tp[j-1] + add $tp, 8, $tp ! tp++ + + brnz,pt $cnt, .L1st_g5 + sub $cnt, 8, $cnt ! j-- +!.L1st_g5 + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 ! ahi=aj + + addcc $nlo, $hi1, $lo1 + addxc $nj, %g0, $hi1 + addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0] + addxc %g0, $hi1, $hi1 + stxa $lo1, [$tp]0xe2 ! tp[j-1] + add $tp, 8, $tp + + addcc $hi0, $hi1, $hi1 + addxc %g0, %g0, $ovf ! upmost overflow bit + stxa $hi1, [$tp]0xe2 + add $tp, 8, $tp + + ba .Louter_g5 + sub $num, 16, $i ! i=num-2 + +.align 16 +.Louter_g5: + wr $ccr, %g0, %ccr +___ + &load_b($bp,$m0); # m0=bp[i] +$code.=<<___; + sub $ap, $num, $ap ! rewind + sub $np, $num, $np + sub $tp, $num, $tp + + ldx [$ap+0], $aj ! ap[0] + ldx [$np+0], $nj ! np[0] + + mulx $aj, $m0, $lo0 ! ap[0]*bp[i] + ldx [$tp], $tj ! tp[0] + umulxhi $aj, $m0, $hi0 + ldx [$ap+8], $aj ! ap[1] + addcc $lo0, $tj, $lo0 ! ap[0]*bp[i]+tp[0] + mulx $aj, $m0, $alo ! ap[1]*bp[i] + addxc %g0, $hi0, $hi0 + mulx $lo0, $n0, $m1 ! tp[0]*n0 + umulxhi $aj, $m0, $aj ! ahi=aj + mulx $nj, $m1, $lo1 ! np[0]*m1 + add $ap, 16, $ap + umulxhi $nj, $m1, $hi1 + ldx [$np+8], $nj ! np[1] + add $np, 16, $np + addcc $lo1, $lo0, $lo1 + mulx $nj, $m1, $nlo ! np[1]*m1 + addxc %g0, $hi1, $hi1 + umulxhi $nj, $m1, $nj ! nhi=nj + + ba .Linner_g5 + sub $num, 24, $cnt ! cnt=num-3 +.align 16 +.Linner_g5: + addcc $alo, $hi0, $lo0 + ldx [$tp+8], $tj ! tp[j] + addxc $aj, %g0, $hi0 ! ahi=aj + ldx [$ap+0], $aj ! ap[j] + add $ap, 8, $ap + addcc $nlo, $hi1, $lo1 + mulx $aj, $m0, $alo ! ap[j]*bp[i] + addxc $nj, %g0, $hi1 ! nhi=nj + ldx [$np+0], $nj ! np[j] + add $np, 8, $np + umulxhi $aj, $m0, $aj ! ahi=aj + addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j] + mulx $nj, $m1, $nlo ! np[j]*m1 + addxc %g0, $hi0, $hi0 + umulxhi $nj, $m1, $nj ! nhi=nj + addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j] + addxc %g0, $hi1, $hi1 + stx $lo1, [$tp] ! tp[j-1] + add $tp, 8, $tp + brnz,pt $cnt, .Linner_g5 + sub $cnt, 8, $cnt +!.Linner_g5 + ldx [$tp+8], $tj ! tp[j] + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 ! ahi=aj + addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j] + addxc %g0, $hi0, $hi0 + + addcc $nlo, $hi1, $lo1 + addxc $nj, %g0, $hi1 ! nhi=nj + addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j] + addxc %g0, $hi1, $hi1 + stx $lo1, [$tp] ! tp[j-1] + + subcc %g0, $ovf, %g0 ! move upmost overflow to CCR.xcc + addxccc $hi1, $hi0, $hi1 + addxc %g0, %g0, $ovf + stx $hi1, [$tp+8] + add $tp, 16, $tp + + brnz,pt $i, .Louter_g5 + sub $i, 8, $i + + sub $ap, $num, $ap ! rewind + sub $np, $num, $np + sub $tp, $num, $tp + ba .Lsub_g5 + subcc $num, 8, $cnt ! cnt=num-1 and clear CCR.xcc + +.align 16 +.Lsub_g5: + ldx [$tp], $tj + add $tp, 8, $tp + ldx [$np+0], $nj + add $np, 8, $np + subccc $tj, $nj, $t2 ! tp[j]-np[j] + srlx $tj, 32, $tj + srlx $nj, 32, $nj + subccc $tj, $nj, $t3 + add $rp, 8, $rp + st $t2, [$rp-4] ! reverse order + st $t3, [$rp-8] + brnz,pt $cnt, .Lsub_g5 + sub $cnt, 8, $cnt + + sub $np, $num, $np ! rewind + sub $tp, $num, $tp + sub $rp, $num, $rp + + subc $ovf, %g0, $ovf ! handle upmost overflow bit + and $tp, $ovf, $ap + andn $rp, $ovf, $np + or $np, $ap, $ap ! ap=borrow?tp:rp + ba .Lcopy_g5 + sub $num, 8, $cnt + +.align 16 +.Lcopy_g5: ! copy or in-place refresh + ldx [$ap+0], $t2 + add $ap, 8, $ap + stx %g0, [$tp] ! zap + add $tp, 8, $tp + stx $t2, [$rp+0] + add $rp, 8, $rp + brnz $cnt, .Lcopy_g5 + sub $cnt, 8, $cnt + + mov 1, %o0 + ret + restore +.type bn_mul_mont_gather5_t4, #function +.size bn_mul_mont_gather5_t4, .-bn_mul_mont_gather5_t4 +___ +} + +$code.=<<___; +.globl bn_flip_t4 +.align 32 +bn_flip_t4: +.Loop_flip: + ld [%o1+0], %o4 + sub %o2, 1, %o2 + ld [%o1+4], %o5 + add %o1, 8, %o1 + st %o5, [%o0+0] + st %o4, [%o0+4] + brnz %o2, .Loop_flip + add %o0, 8, %o0 + retl + nop +.type bn_flip_t4, #function +.size bn_flip_t4, .-bn_flip_t4 + +.globl bn_flip_n_scatter5_t4 +.align 32 +bn_flip_n_scatter5_t4: + sll %o3, 3, %o3 + srl %o1, 1, %o1 + add %o3, %o2, %o2 ! &pwrtbl[pwr] + sub %o1, 1, %o1 +.Loop_flip_n_scatter5: + ld [%o0+0], %o4 ! inp[i] + ld [%o0+4], %o5 + add %o0, 8, %o0 + sllx %o5, 32, %o5 + or %o4, %o5, %o5 + stx %o5, [%o2] + add %o2, 32*8, %o2 + brnz %o1, .Loop_flip_n_scatter5 + sub %o1, 1, %o1 + retl + nop +.type bn_flip_n_scatter5_t4, #function +.size bn_flip_n_scatter5_t4, .-bn_flip_n_scatter5_t4 + +.globl bn_gather5_t4 +.align 32 +bn_gather5_t4: +___ + &load_ccr("%o2","%o3","%g1"); +$code.=<<___; + sub %o1, 1, %o1 +.Loop_gather5: +___ + &load_b("%o2","%g1"); +$code.=<<___; + stx %g1, [%o0] + add %o0, 8, %o0 + brnz %o1, .Loop_gather5 + sub %o1, 1, %o1 + + retl + nop +.type bn_gather5_t4, #function +.size bn_gather5_t4, .-bn_gather5_t4 + +.asciz "Montgomery Multiplication for SPARC T4, David S. Miller, Andy Polyakov" +.align 4 +___ + +&emit_assembler(); + +close STDOUT; diff --git a/openssl/crypto/bn/asm/sparcv9-gf2m.pl b/openssl/crypto/bn/asm/sparcv9-gf2m.pl new file mode 100755 index 000000000..ab94cd917 --- /dev/null +++ b/openssl/crypto/bn/asm/sparcv9-gf2m.pl @@ -0,0 +1,190 @@ +#!/usr/bin/env perl +# +# ==================================================================== +# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL +# project. The module is, however, dual licensed under OpenSSL and +# CRYPTOGAMS licenses depending on where you obtain it. For further +# details see http://www.openssl.org/~appro/cryptogams/. +# ==================================================================== +# +# October 2012 +# +# The module implements bn_GF2m_mul_2x2 polynomial multiplication used +# in bn_gf2m.c. It's kind of low-hanging mechanical port from C for +# the time being... Except that it has two code paths: one suitable +# for all SPARCv9 processors and one for VIS3-capable ones. Former +# delivers ~25-45% more, more for longer keys, heaviest DH and DSA +# verify operations on venerable UltraSPARC II. On T4 VIS3 code is +# ~100-230% faster than gcc-generated code and ~35-90% faster than +# the pure SPARCv9 code path. + +$locals=16*8; + +$tab="%l0"; + +@T=("%g2","%g3"); +@i=("%g4","%g5"); + +($a1,$a2,$a4,$a8,$a12,$a48)=map("%o$_",(0..5)); +($lo,$hi,$b)=("%g1",$a8,"%o7"); $a=$lo; + +$code.=<<___; +#include <sparc_arch.h> + +#ifdef __arch64__ +.register %g2,#scratch +.register %g3,#scratch +#endif + +#ifdef __PIC__ +SPARC_PIC_THUNK(%g1) +#endif + +.globl bn_GF2m_mul_2x2 +.align 16 +bn_GF2m_mul_2x2: + SPARC_LOAD_ADDRESS_LEAF(OPENSSL_sparcv9cap_P,%g1,%g5) + ld [%g1+0],%g1 ! OPENSSL_sparcv9cap_P[0] + + andcc %g1, SPARCV9_VIS3, %g0 + bz,pn %icc,.Lsoftware + nop + + sllx %o1, 32, %o1 + sllx %o3, 32, %o3 + or %o2, %o1, %o1 + or %o4, %o3, %o3 + .word 0x95b262ab ! xmulx %o1, %o3, %o2 + .word 0x99b262cb ! xmulxhi %o1, %o3, %o4 + srlx %o2, 32, %o1 ! 13 cycles later + st %o2, [%o0+0] + st %o1, [%o0+4] + srlx %o4, 32, %o3 + st %o4, [%o0+8] + retl + st %o3, [%o0+12] + +.align 16 +.Lsoftware: + save %sp,-STACK_FRAME-$locals,%sp + + sllx %i1,32,$a + mov -1,$a12 + sllx %i3,32,$b + or %i2,$a,$a + srlx $a12,1,$a48 ! 0x7fff... + or %i4,$b,$b + srlx $a12,2,$a12 ! 0x3fff... + add %sp,STACK_BIAS+STACK_FRAME,$tab + + sllx $a,2,$a4 + mov $a,$a1 + sllx $a,1,$a2 + + srax $a4,63,@i[1] ! broadcast 61st bit + and $a48,$a4,$a4 ! (a<<2)&0x7fff... + srlx $a48,2,$a48 + srax $a2,63,@i[0] ! broadcast 62nd bit + and $a12,$a2,$a2 ! (a<<1)&0x3fff... + srax $a1,63,$lo ! broadcast 63rd bit + and $a48,$a1,$a1 ! (a<<0)&0x1fff... + + sllx $a1,3,$a8 + and $b,$lo,$lo + and $b,@i[0],@i[0] + and $b,@i[1],@i[1] + + stx %g0,[$tab+0*8] ! tab[0]=0 + xor $a1,$a2,$a12 + stx $a1,[$tab+1*8] ! tab[1]=a1 + stx $a2,[$tab+2*8] ! tab[2]=a2 + xor $a4,$a8,$a48 + stx $a12,[$tab+3*8] ! tab[3]=a1^a2 + xor $a4,$a1,$a1 + + stx $a4,[$tab+4*8] ! tab[4]=a4 + xor $a4,$a2,$a2 + stx $a1,[$tab+5*8] ! tab[5]=a1^a4 + xor $a4,$a12,$a12 + stx $a2,[$tab+6*8] ! tab[6]=a2^a4 + xor $a48,$a1,$a1 + stx $a12,[$tab+7*8] ! tab[7]=a1^a2^a4 + xor $a48,$a2,$a2 + + stx $a8,[$tab+8*8] ! tab[8]=a8 + xor $a48,$a12,$a12 + stx $a1,[$tab+9*8] ! tab[9]=a1^a8 + xor $a4,$a1,$a1 + stx $a2,[$tab+10*8] ! tab[10]=a2^a8 + xor $a4,$a2,$a2 + stx $a12,[$tab+11*8] ! tab[11]=a1^a2^a8 + + xor $a4,$a12,$a12 + stx $a48,[$tab+12*8] ! tab[12]=a4^a8 + srlx $lo,1,$hi + stx $a1,[$tab+13*8] ! tab[13]=a1^a4^a8 + sllx $lo,63,$lo + stx $a2,[$tab+14*8] ! tab[14]=a2^a4^a8 + srlx @i[0],2,@T[0] + stx $a12,[$tab+15*8] ! tab[15]=a1^a2^a4^a8 + + sllx @i[0],62,$a1 + sllx $b,3,@i[0] + srlx @i[1],3,@T[1] + and @i[0],`0xf<<3`,@i[0] + sllx @i[1],61,$a2 + ldx [$tab+@i[0]],@i[0] + srlx $b,4-3,@i[1] + xor @T[0],$hi,$hi + and @i[1],`0xf<<3`,@i[1] + xor $a1,$lo,$lo + ldx [$tab+@i[1]],@i[1] + xor @T[1],$hi,$hi + + xor @i[0],$lo,$lo + srlx $b,8-3,@i[0] + xor $a2,$lo,$lo + and @i[0],`0xf<<3`,@i[0] +___ +for($n=1;$n<14;$n++) { +$code.=<<___; + sllx @i[1],`$n*4`,@T[0] + ldx [$tab+@i[0]],@i[0] + srlx @i[1],`64-$n*4`,@T[1] + xor @T[0],$lo,$lo + srlx $b,`($n+2)*4`-3,@i[1] + xor @T[1],$hi,$hi + and @i[1],`0xf<<3`,@i[1] +___ + push(@i,shift(@i)); push(@T,shift(@T)); +} +$code.=<<___; + sllx @i[1],`$n*4`,@T[0] + ldx [$tab+@i[0]],@i[0] + srlx @i[1],`64-$n*4`,@T[1] + xor @T[0],$lo,$lo + + sllx @i[0],`($n+1)*4`,@T[0] + xor @T[1],$hi,$hi + srlx @i[0],`64-($n+1)*4`,@T[1] + xor @T[0],$lo,$lo + xor @T[1],$hi,$hi + + srlx $lo,32,%i1 + st $lo,[%i0+0] + st %i1,[%i0+4] + srlx $hi,32,%i2 + st $hi,[%i0+8] + st %i2,[%i0+12] + + ret + restore +.type bn_GF2m_mul_2x2,#function +.size bn_GF2m_mul_2x2,.-bn_GF2m_mul_2x2 +.asciz "GF(2^m) Multiplication for SPARCv9, CRYPTOGAMS by <appro\@openssl.org>" +.align 4 +___ + +$code =~ s/\`([^\`]*)\`/eval($1)/gem; +print $code; +close STDOUT; diff --git a/openssl/crypto/bn/asm/vis3-mont.pl b/openssl/crypto/bn/asm/vis3-mont.pl new file mode 100755 index 000000000..a1357de0e --- /dev/null +++ b/openssl/crypto/bn/asm/vis3-mont.pl @@ -0,0 +1,373 @@ +#!/usr/bin/env perl + +# ==================================================================== +# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL +# project. The module is, however, dual licensed under OpenSSL and +# CRYPTOGAMS licenses depending on where you obtain it. For further +# details see http://www.openssl.org/~appro/cryptogams/. +# ==================================================================== + +# October 2012. +# +# SPARCv9 VIS3 Montgomery multiplicaion procedure suitable for T3 and +# onward. There are three new instructions used here: umulxhi, +# addxc[cc] and initializing store. On T3 RSA private key operations +# are 1.54/1.87/2.11/2.26 times faster for 512/1024/2048/4096-bit key +# lengths. This is without dedicated squaring procedure. On T4 +# corresponding coefficients are 1.47/2.10/2.80/2.90x, which is mostly +# for reference purposes, because T4 has dedicated Montgomery +# multiplication and squaring *instructions* that deliver even more. + +$bits=32; +for (@ARGV) { $bits=64 if (/\-m64/ || /\-xarch\=v9/); } +if ($bits==64) { $bias=2047; $frame=192; } +else { $bias=0; $frame=112; } + +$code.=<<___ if ($bits==64); +.register %g2,#scratch +.register %g3,#scratch +___ +$code.=<<___; +.section ".text",#alloc,#execinstr +___ + +($n0,$m0,$m1,$lo0,$hi0, $lo1,$hi1,$aj,$alo,$nj,$nlo,$tj)= + (map("%g$_",(1..5)),map("%o$_",(0..5,7))); + +# int bn_mul_mont( +$rp="%o0"; # BN_ULONG *rp, +$ap="%o1"; # const BN_ULONG *ap, +$bp="%o2"; # const BN_ULONG *bp, +$np="%o3"; # const BN_ULONG *np, +$n0p="%o4"; # const BN_ULONG *n0, +$num="%o5"; # int num); # caller ensures that num is even + # and >=6 +$code.=<<___; +.globl bn_mul_mont_vis3 +.align 32 +bn_mul_mont_vis3: + add %sp, $bias, %g4 ! real top of stack + sll $num, 2, $num ! size in bytes + add $num, 63, %g5 + andn %g5, 63, %g5 ! buffer size rounded up to 64 bytes + add %g5, %g5, %g1 + add %g5, %g1, %g1 ! 3*buffer size + sub %g4, %g1, %g1 + andn %g1, 63, %g1 ! align at 64 byte + sub %g1, $frame, %g1 ! new top of stack + sub %g1, %g4, %g1 + + save %sp, %g1, %sp +___ + +# +-------------------------------+<----- %sp +# . . +# +-------------------------------+<----- aligned at 64 bytes +# | __int64 tmp[0] | +# +-------------------------------+ +# . . +# . . +# +-------------------------------+<----- aligned at 64 bytes +# | __int64 ap[1..0] | converted ap[] +# +-------------------------------+ +# | __int64 np[1..0] | converted np[] +# +-------------------------------+ +# | __int64 ap[3..2] | +# . . +# . . +# +-------------------------------+ +($rp,$ap,$bp,$np,$n0p,$num)=map("%i$_",(0..5)); +($t0,$t1,$t2,$t3,$cnt,$tp,$bufsz,$anp)=map("%l$_",(0..7)); +($ovf,$i)=($t0,$t1); +$code.=<<___; + ld [$n0p+0], $t0 ! pull n0[0..1] value + add %sp, $bias+$frame, $tp + ld [$n0p+4], $t1 + add $tp, %g5, $anp + ld [$bp+0], $t2 ! m0=bp[0] + sllx $t1, 32, $n0 + ld [$bp+4], $t3 + or $t0, $n0, $n0 + add $bp, 8, $bp + + ld [$ap+0], $t0 ! ap[0] + sllx $t3, 32, $m0 + ld [$ap+4], $t1 + or $t2, $m0, $m0 + + ld [$ap+8], $t2 ! ap[1] + sllx $t1, 32, $aj + ld [$ap+12], $t3 + or $t0, $aj, $aj + add $ap, 16, $ap + stxa $aj, [$anp]0xe2 ! converted ap[0] + + mulx $aj, $m0, $lo0 ! ap[0]*bp[0] + umulxhi $aj, $m0, $hi0 + + ld [$np+0], $t0 ! np[0] + sllx $t3, 32, $aj + ld [$np+4], $t1 + or $t2, $aj, $aj + + ld [$np+8], $t2 ! np[1] + sllx $t1, 32, $nj + ld [$np+12], $t3 + or $t0, $nj, $nj + add $np, 16, $np + stx $nj, [$anp+8] ! converted np[0] + + mulx $lo0, $n0, $m1 ! "tp[0]"*n0 + stx $aj, [$anp+16] ! converted ap[1] + + mulx $aj, $m0, $alo ! ap[1]*bp[0] + umulxhi $aj, $m0, $aj ! ahi=aj + + mulx $nj, $m1, $lo1 ! np[0]*m1 + umulxhi $nj, $m1, $hi1 + + sllx $t3, 32, $nj + or $t2, $nj, $nj + stx $nj, [$anp+24] ! converted np[1] + add $anp, 32, $anp + + addcc $lo0, $lo1, $lo1 + addxc %g0, $hi1, $hi1 + + mulx $nj, $m1, $nlo ! np[1]*m1 + umulxhi $nj, $m1, $nj ! nhi=nj + + ba .L1st + sub $num, 24, $cnt ! cnt=num-3 + +.align 16 +.L1st: + ld [$ap+0], $t0 ! ap[j] + addcc $alo, $hi0, $lo0 + ld [$ap+4], $t1 + addxc $aj, %g0, $hi0 + + sllx $t1, 32, $aj + add $ap, 8, $ap + or $t0, $aj, $aj + stxa $aj, [$anp]0xe2 ! converted ap[j] + + ld [$np+0], $t2 ! np[j] + addcc $nlo, $hi1, $lo1 + ld [$np+4], $t3 + addxc $nj, %g0, $hi1 ! nhi=nj + + sllx $t3, 32, $nj + add $np, 8, $np + mulx $aj, $m0, $alo ! ap[j]*bp[0] + or $t2, $nj, $nj + umulxhi $aj, $m0, $aj ! ahi=aj + stx $nj, [$anp+8] ! converted np[j] + add $anp, 16, $anp ! anp++ + + mulx $nj, $m1, $nlo ! np[j]*m1 + addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0] + umulxhi $nj, $m1, $nj ! nhi=nj + addxc %g0, $hi1, $hi1 + stxa $lo1, [$tp]0xe2 ! tp[j-1] + add $tp, 8, $tp ! tp++ + + brnz,pt $cnt, .L1st + sub $cnt, 8, $cnt ! j-- +!.L1st + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 ! ahi=aj + + addcc $nlo, $hi1, $lo1 + addxc $nj, %g0, $hi1 + addcc $lo0, $lo1, $lo1 ! np[j]*m1+ap[j]*bp[0] + addxc %g0, $hi1, $hi1 + stxa $lo1, [$tp]0xe2 ! tp[j-1] + add $tp, 8, $tp + + addcc $hi0, $hi1, $hi1 + addxc %g0, %g0, $ovf ! upmost overflow bit + stxa $hi1, [$tp]0xe2 + add $tp, 8, $tp + + ba .Louter + sub $num, 16, $i ! i=num-2 + +.align 16 +.Louter: + ld [$bp+0], $t2 ! m0=bp[i] + ld [$bp+4], $t3 + + sub $anp, $num, $anp ! rewind + sub $tp, $num, $tp + sub $anp, $num, $anp + + add $bp, 8, $bp + sllx $t3, 32, $m0 + ldx [$anp+0], $aj ! ap[0] + or $t2, $m0, $m0 + ldx [$anp+8], $nj ! np[0] + + mulx $aj, $m0, $lo0 ! ap[0]*bp[i] + ldx [$tp], $tj ! tp[0] + umulxhi $aj, $m0, $hi0 + ldx [$anp+16], $aj ! ap[1] + addcc $lo0, $tj, $lo0 ! ap[0]*bp[i]+tp[0] + mulx $aj, $m0, $alo ! ap[1]*bp[i] + addxc %g0, $hi0, $hi0 + mulx $lo0, $n0, $m1 ! tp[0]*n0 + umulxhi $aj, $m0, $aj ! ahi=aj + mulx $nj, $m1, $lo1 ! np[0]*m1 + umulxhi $nj, $m1, $hi1 + ldx [$anp+24], $nj ! np[1] + add $anp, 32, $anp + addcc $lo1, $lo0, $lo1 + mulx $nj, $m1, $nlo ! np[1]*m1 + addxc %g0, $hi1, $hi1 + umulxhi $nj, $m1, $nj ! nhi=nj + + ba .Linner + sub $num, 24, $cnt ! cnt=num-3 +.align 16 +.Linner: + addcc $alo, $hi0, $lo0 + ldx [$tp+8], $tj ! tp[j] + addxc $aj, %g0, $hi0 ! ahi=aj + ldx [$anp+0], $aj ! ap[j] + addcc $nlo, $hi1, $lo1 + mulx $aj, $m0, $alo ! ap[j]*bp[i] + addxc $nj, %g0, $hi1 ! nhi=nj + ldx [$anp+8], $nj ! np[j] + add $anp, 16, $anp + umulxhi $aj, $m0, $aj ! ahi=aj + addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j] + mulx $nj, $m1, $nlo ! np[j]*m1 + addxc %g0, $hi0, $hi0 + umulxhi $nj, $m1, $nj ! nhi=nj + addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j] + addxc %g0, $hi1, $hi1 + stx $lo1, [$tp] ! tp[j-1] + add $tp, 8, $tp + brnz,pt $cnt, .Linner + sub $cnt, 8, $cnt +!.Linner + ldx [$tp+8], $tj ! tp[j] + addcc $alo, $hi0, $lo0 + addxc $aj, %g0, $hi0 ! ahi=aj + addcc $lo0, $tj, $lo0 ! ap[j]*bp[i]+tp[j] + addxc %g0, $hi0, $hi0 + + addcc $nlo, $hi1, $lo1 + addxc $nj, %g0, $hi1 ! nhi=nj + addcc $lo1, $lo0, $lo1 ! np[j]*m1+ap[j]*bp[i]+tp[j] + addxc %g0, $hi1, $hi1 + stx $lo1, [$tp] ! tp[j-1] + + subcc %g0, $ovf, %g0 ! move upmost overflow to CCR.xcc + addxccc $hi1, $hi0, $hi1 + addxc %g0, %g0, $ovf + stx $hi1, [$tp+8] + add $tp, 16, $tp + + brnz,pt $i, .Louter + sub $i, 8, $i + + sub $anp, $num, $anp ! rewind + sub $tp, $num, $tp + sub $anp, $num, $anp + ba .Lsub + subcc $num, 8, $cnt ! cnt=num-1 and clear CCR.xcc + +.align 16 +.Lsub: + ldx [$tp], $tj + add $tp, 8, $tp + ldx [$anp+8], $nj + add $anp, 16, $anp + subccc $tj, $nj, $t2 ! tp[j]-np[j] + srlx $tj, 32, $tj + srlx $nj, 32, $nj + subccc $tj, $nj, $t3 + add $rp, 8, $rp + st $t2, [$rp-4] ! reverse order + st $t3, [$rp-8] + brnz,pt $cnt, .Lsub + sub $cnt, 8, $cnt + + sub $anp, $num, $anp ! rewind + sub $tp, $num, $tp + sub $anp, $num, $anp + sub $rp, $num, $rp + + subc $ovf, %g0, $ovf ! handle upmost overflow bit + and $tp, $ovf, $ap + andn $rp, $ovf, $np + or $np, $ap, $ap ! ap=borrow?tp:rp + ba .Lcopy + sub $num, 8, $cnt + +.align 16 +.Lcopy: ! copy or in-place refresh + ld [$ap+0], $t2 + ld [$ap+4], $t3 + add $ap, 8, $ap + stx %g0, [$tp] ! zap + add $tp, 8, $tp + stx %g0, [$anp] ! zap + stx %g0, [$anp+8] + add $anp, 16, $anp + st $t3, [$rp+0] ! flip order + st $t2, [$rp+4] + add $rp, 8, $rp + brnz $cnt, .Lcopy + sub $cnt, 8, $cnt + + mov 1, %o0 + ret + restore +.type bn_mul_mont_vis3, #function +.size bn_mul_mont_vis3, .-bn_mul_mont_vis3 +.asciz "Montgomery Multiplication for SPARCv9 VIS3, CRYPTOGAMS by <appro\@openssl.org>" +.align 4 +___ + +# Purpose of these subroutines is to explicitly encode VIS instructions, +# so that one can compile the module without having to specify VIS +# extentions on compiler command line, e.g. -xarch=v9 vs. -xarch=v9a. +# Idea is to reserve for option to produce "universal" binary and let +# programmer detect if current CPU is VIS capable at run-time. +sub unvis3 { +my ($mnemonic,$rs1,$rs2,$rd)=@_; +my %bias = ( "g" => 0, "o" => 8, "l" => 16, "i" => 24 ); +my ($ref,$opf); +my %visopf = ( "addxc" => 0x011, + "addxccc" => 0x013, + "umulxhi" => 0x016 ); + + $ref = "$mnemonic\t$rs1,$rs2,$rd"; + + if ($opf=$visopf{$mnemonic}) { + foreach ($rs1,$rs2,$rd) { + return $ref if (!/%([goli])([0-9])/); + $_=$bias{$1}+$2; + } + + return sprintf ".word\t0x%08x !%s", + 0x81b00000|$rd<<25|$rs1<<14|$opf<<5|$rs2, + $ref; + } else { + return $ref; + } +} + +foreach (split("\n",$code)) { + s/\`([^\`]*)\`/eval $1/ge; + + s/\b(umulxhi|addxc[c]{0,2})\s+(%[goli][0-7]),\s*(%[goli][0-7]),\s*(%[goli][0-7])/ + &unvis3($1,$2,$3,$4) + /ge; + + print $_,"\n"; +} + +close STDOUT; diff --git a/openssl/crypto/bn/asm/x86_64-gcc.c b/openssl/crypto/bn/asm/x86_64-gcc.c index 31476abeb..d5488866e 100644 --- a/openssl/crypto/bn/asm/x86_64-gcc.c +++ b/openssl/crypto/bn/asm/x86_64-gcc.c @@ -1,8 +1,8 @@ #include "../bn_lcl.h" #if !(defined(__GNUC__) && __GNUC__>=2) -# include "../bn_asm.c" /* kind of dirty hack for Sun Studio */ +# include "../bn_asm.c" /* kind of dirty hack for Sun Studio */ #else -/* +/*- * x86_64 BIGNUM accelerator version 0.1, December 2002. * * Implemented by Andy Polyakov <appro@fy.chalmers.se> for the OpenSSL @@ -28,579 +28,611 @@ * Q. How much faster does it get? * A. 'apps/openssl speed rsa dsa' output with no-asm: * - * sign verify sign/s verify/s - * rsa 512 bits 0.0006s 0.0001s 1683.8 18456.2 - * rsa 1024 bits 0.0028s 0.0002s 356.0 6407.0 - * rsa 2048 bits 0.0172s 0.0005s 58.0 1957.8 - * rsa 4096 bits 0.1155s 0.0018s 8.7 555.6 - * sign verify sign/s verify/s - * dsa 512 bits 0.0005s 0.0006s 2100.8 1768.3 - * dsa 1024 bits 0.0014s 0.0018s 692.3 559.2 - * dsa 2048 bits 0.0049s 0.0061s 204.7 165.0 + * sign verify sign/s verify/s + * rsa 512 bits 0.0006s 0.0001s 1683.8 18456.2 + * rsa 1024 bits 0.0028s 0.0002s 356.0 6407.0 + * rsa 2048 bits 0.0172s 0.0005s 58.0 1957.8 + * rsa 4096 bits 0.1155s 0.0018s 8.7 555.6 + * sign verify sign/s verify/s + * dsa 512 bits 0.0005s 0.0006s 2100.8 1768.3 + * dsa 1024 bits 0.0014s 0.0018s 692.3 559.2 + * dsa 2048 bits 0.0049s 0.0061s 204.7 165.0 * * 'apps/openssl speed rsa dsa' output with this module: * - * sign verify sign/s verify/s - * rsa 512 bits 0.0004s 0.0000s 2767.1 33297.9 - * rsa 1024 bits 0.0012s 0.0001s 867.4 14674.7 - * rsa 2048 bits 0.0061s 0.0002s 164.0 5270.0 - * rsa 4096 bits 0.0384s 0.0006s 26.1 1650.8 - * sign verify sign/s verify/s - * dsa 512 bits 0.0002s 0.0003s 4442.2 3786.3 - * dsa 1024 bits 0.0005s 0.0007s 1835.1 1497.4 - * dsa 2048 bits 0.0016s 0.0020s 620.4 504.6 + * sign verify sign/s verify/s + * rsa 512 bits 0.0004s 0.0000s 2767.1 33297.9 + * rsa 1024 bits 0.0012s 0.0001s 867.4 14674.7 + * rsa 2048 bits 0.0061s 0.0002s 164.0 5270.0 + * rsa 4096 bits 0.0384s 0.0006s 26.1 1650.8 + * sign verify sign/s verify/s + * dsa 512 bits 0.0002s 0.0003s 4442.2 3786.3 + * dsa 1024 bits 0.0005s 0.0007s 1835.1 1497.4 + * dsa 2048 bits 0.0016s 0.0020s 620.4 504.6 * * For the reference. IA-32 assembler implementation performs * very much like 64-bit code compiled with no-asm on the same * machine. */ -#ifdef _WIN64 -#define BN_ULONG unsigned long long -#else -#define BN_ULONG unsigned long -#endif +# if defined(_WIN64) || !defined(__LP64__) +# define BN_ULONG unsigned long long +# else +# define BN_ULONG unsigned long +# endif -#undef mul -#undef mul_add -#undef sqr +# undef mul +# undef mul_add -/* - * "m"(a), "+m"(r) is the way to favor DirectPath µ-code; - * "g"(0) let the compiler to decide where does it - * want to keep the value of zero; +/*- + * "m"(a), "+m"(r) is the way to favor DirectPath µ-code; + * "g"(0) let the compiler to decide where does it + * want to keep the value of zero; */ -#define mul_add(r,a,word,carry) do { \ - register BN_ULONG high,low; \ - asm ("mulq %3" \ - : "=a"(low),"=d"(high) \ - : "a"(word),"m"(a) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(carry),"+d"(high)\ - : "a"(low),"g"(0) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+m"(r),"+d"(high) \ - : "r"(carry),"g"(0) \ - : "cc"); \ - carry=high; \ - } while (0) - -#define mul(r,a,word,carry) do { \ - register BN_ULONG high,low; \ - asm ("mulq %3" \ - : "=a"(low),"=d"(high) \ - : "a"(word),"g"(a) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(carry),"+d"(high)\ - : "a"(low),"g"(0) \ - : "cc"); \ - (r)=carry, carry=high; \ - } while (0) - -#define sqr(r0,r1,a) \ - asm ("mulq %2" \ - : "=a"(r0),"=d"(r1) \ - : "a"(a) \ - : "cc"); - -BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) - { - BN_ULONG c1=0; - - if (num <= 0) return(c1); - - while (num&~3) - { - mul_add(rp[0],ap[0],w,c1); - mul_add(rp[1],ap[1],w,c1); - mul_add(rp[2],ap[2],w,c1); - mul_add(rp[3],ap[3],w,c1); - ap+=4; rp+=4; num-=4; - } - if (num) - { - mul_add(rp[0],ap[0],w,c1); if (--num==0) return c1; - mul_add(rp[1],ap[1],w,c1); if (--num==0) return c1; - mul_add(rp[2],ap[2],w,c1); return c1; - } - - return(c1); - } +# define mul_add(r,a,word,carry) do { \ + register BN_ULONG high,low; \ + asm ("mulq %3" \ + : "=a"(low),"=d"(high) \ + : "a"(word),"m"(a) \ + : "cc"); \ + asm ("addq %2,%0; adcq %3,%1" \ + : "+r"(carry),"+d"(high)\ + : "a"(low),"g"(0) \ + : "cc"); \ + asm ("addq %2,%0; adcq %3,%1" \ + : "+m"(r),"+d"(high) \ + : "r"(carry),"g"(0) \ + : "cc"); \ + carry=high; \ + } while (0) + +# define mul(r,a,word,carry) do { \ + register BN_ULONG high,low; \ + asm ("mulq %3" \ + : "=a"(low),"=d"(high) \ + : "a"(word),"g"(a) \ + : "cc"); \ + asm ("addq %2,%0; adcq %3,%1" \ + : "+r"(carry),"+d"(high)\ + : "a"(low),"g"(0) \ + : "cc"); \ + (r)=carry, carry=high; \ + } while (0) +# undef sqr +# define sqr(r0,r1,a) \ + asm ("mulq %2" \ + : "=a"(r0),"=d"(r1) \ + : "a"(a) \ + : "cc"); + +BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, + BN_ULONG w) +{ + BN_ULONG c1 = 0; + + if (num <= 0) + return (c1); + + while (num & ~3) { + mul_add(rp[0], ap[0], w, c1); + mul_add(rp[1], ap[1], w, c1); + mul_add(rp[2], ap[2], w, c1); + mul_add(rp[3], ap[3], w, c1); + ap += 4; + rp += 4; + num -= 4; + } + if (num) { + mul_add(rp[0], ap[0], w, c1); + if (--num == 0) + return c1; + mul_add(rp[1], ap[1], w, c1); + if (--num == 0) + return c1; + mul_add(rp[2], ap[2], w, c1); + return c1; + } + + return (c1); +} BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) - { - BN_ULONG c1=0; - - if (num <= 0) return(c1); - - while (num&~3) - { - mul(rp[0],ap[0],w,c1); - mul(rp[1],ap[1],w,c1); - mul(rp[2],ap[2],w,c1); - mul(rp[3],ap[3],w,c1); - ap+=4; rp+=4; num-=4; - } - if (num) - { - mul(rp[0],ap[0],w,c1); if (--num == 0) return c1; - mul(rp[1],ap[1],w,c1); if (--num == 0) return c1; - mul(rp[2],ap[2],w,c1); - } - return(c1); - } +{ + BN_ULONG c1 = 0; + + if (num <= 0) + return (c1); + + while (num & ~3) { + mul(rp[0], ap[0], w, c1); + mul(rp[1], ap[1], w, c1); + mul(rp[2], ap[2], w, c1); + mul(rp[3], ap[3], w, c1); + ap += 4; + rp += 4; + num -= 4; + } + if (num) { + mul(rp[0], ap[0], w, c1); + if (--num == 0) + return c1; + mul(rp[1], ap[1], w, c1); + if (--num == 0) + return c1; + mul(rp[2], ap[2], w, c1); + } + return (c1); +} void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n) - { - if (n <= 0) return; - - while (n&~3) - { - sqr(r[0],r[1],a[0]); - sqr(r[2],r[3],a[1]); - sqr(r[4],r[5],a[2]); - sqr(r[6],r[7],a[3]); - a+=4; r+=8; n-=4; - } - if (n) - { - sqr(r[0],r[1],a[0]); if (--n == 0) return; - sqr(r[2],r[3],a[1]); if (--n == 0) return; - sqr(r[4],r[5],a[2]); - } - } +{ + if (n <= 0) + return; + + while (n & ~3) { + sqr(r[0], r[1], a[0]); + sqr(r[2], r[3], a[1]); + sqr(r[4], r[5], a[2]); + sqr(r[6], r[7], a[3]); + a += 4; + r += 8; + n -= 4; + } + if (n) { + sqr(r[0], r[1], a[0]); + if (--n == 0) + return; + sqr(r[2], r[3], a[1]); + if (--n == 0) + return; + sqr(r[4], r[5], a[2]); + } +} BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) -{ BN_ULONG ret,waste; +{ + BN_ULONG ret, waste; - asm ("divq %4" - : "=a"(ret),"=d"(waste) - : "a"(l),"d"(h),"g"(d) - : "cc"); + asm("divq %4":"=a"(ret), "=d"(waste) + : "a"(l), "d"(h), "g"(d) + : "cc"); - return ret; + return ret; } -BN_ULONG bn_add_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n) -{ BN_ULONG ret=0,i=0; - - if (n <= 0) return 0; - - asm volatile ( - " subq %2,%2 \n" - ".p2align 4 \n" - "1: movq (%4,%2,8),%0 \n" - " adcq (%5,%2,8),%0 \n" - " movq %0,(%3,%2,8) \n" - " leaq 1(%2),%2 \n" - " loop 1b \n" - " sbbq %0,%0 \n" - : "=&a"(ret),"+c"(n),"=&r"(i) - : "r"(rp),"r"(ap),"r"(bp) - : "cc", "memory" - ); - - return ret&1; +BN_ULONG bn_add_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + int n) +{ + BN_ULONG ret; + size_t i = 0; + + if (n <= 0) + return 0; + + asm volatile (" subq %0,%0 \n" /* clear carry */ + " jmp 1f \n" + ".p2align 4 \n" + "1: movq (%4,%2,8),%0 \n" + " adcq (%5,%2,8),%0 \n" + " movq %0,(%3,%2,8) \n" + " lea 1(%2),%2 \n" + " loop 1b \n" + " sbbq %0,%0 \n":"=&r" (ret), "+c"(n), + "+r"(i) + :"r"(rp), "r"(ap), "r"(bp) + :"cc", "memory"); + + return ret & 1; } -#ifndef SIMICS -BN_ULONG bn_sub_words (BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int n) -{ BN_ULONG ret=0,i=0; - - if (n <= 0) return 0; - - asm volatile ( - " subq %2,%2 \n" - ".p2align 4 \n" - "1: movq (%4,%2,8),%0 \n" - " sbbq (%5,%2,8),%0 \n" - " movq %0,(%3,%2,8) \n" - " leaq 1(%2),%2 \n" - " loop 1b \n" - " sbbq %0,%0 \n" - : "=&a"(ret),"+c"(n),"=&r"(i) - : "r"(rp),"r"(ap),"r"(bp) - : "cc", "memory" - ); - - return ret&1; +# ifndef SIMICS +BN_ULONG bn_sub_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + int n) +{ + BN_ULONG ret; + size_t i = 0; + + if (n <= 0) + return 0; + + asm volatile (" subq %0,%0 \n" /* clear borrow */ + " jmp 1f \n" + ".p2align 4 \n" + "1: movq (%4,%2,8),%0 \n" + " sbbq (%5,%2,8),%0 \n" + " movq %0,(%3,%2,8) \n" + " lea 1(%2),%2 \n" + " loop 1b \n" + " sbbq %0,%0 \n":"=&r" (ret), "+c"(n), + "+r"(i) + :"r"(rp), "r"(ap), "r"(bp) + :"cc", "memory"); + + return ret & 1; } -#else +# else /* Simics 1.4<7 has buggy sbbq:-( */ -#define BN_MASK2 0xffffffffffffffffL +# define BN_MASK2 0xffffffffffffffffL BN_ULONG bn_sub_words(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) - { - BN_ULONG t1,t2; - int c=0; - - if (n <= 0) return((BN_ULONG)0); - - for (;;) - { - t1=a[0]; t2=b[0]; - r[0]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - if (--n <= 0) break; - - t1=a[1]; t2=b[1]; - r[1]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - if (--n <= 0) break; - - t1=a[2]; t2=b[2]; - r[2]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - if (--n <= 0) break; - - t1=a[3]; t2=b[3]; - r[3]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - if (--n <= 0) break; - - a+=4; - b+=4; - r+=4; - } - return(c); - } -#endif +{ + BN_ULONG t1, t2; + int c = 0; + + if (n <= 0) + return ((BN_ULONG)0); + + for (;;) { + t1 = a[0]; + t2 = b[0]; + r[0] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + if (--n <= 0) + break; + + t1 = a[1]; + t2 = b[1]; + r[1] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + if (--n <= 0) + break; + + t1 = a[2]; + t2 = b[2]; + r[2] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + if (--n <= 0) + break; + + t1 = a[3]; + t2 = b[3]; + r[3] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + if (--n <= 0) + break; + + a += 4; + b += 4; + r += 4; + } + return (c); +} +# endif /* mul_add_c(a,b,c0,c1,c2) -- c+=a*b for three word number c=(c2,c1,c0) */ /* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */ /* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */ -/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */ +/* + * sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number + * c=(c2,c1,c0) + */ -#if 0 +/* + * Keep in mind that carrying into high part of multiplication result + * can not overflow, because it cannot be all-ones. + */ +# if 0 /* original macros are kept for reference purposes */ -#define mul_add_c(a,b,c0,c1,c2) { \ - BN_ULONG ta=(a),tb=(b); \ - t1 = ta * tb; \ - t2 = BN_UMULT_HIGH(ta,tb); \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define mul_add_c2(a,b,c0,c1,c2) { \ - BN_ULONG ta=(a),tb=(b),t0; \ - t1 = BN_UMULT_HIGH(ta,tb); \ - t0 = ta * tb; \ - t2 = t1+t1; c2 += (t2<t1)?1:0; \ - t1 = t0+t0; t2 += (t1<t0)?1:0; \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } -#else -#define mul_add_c(a,b,c0,c1,c2) do { \ - asm ("mulq %3" \ - : "=a"(t1),"=d"(t2) \ - : "a"(a),"m"(b) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(c0),"+d"(t2) \ - : "a"(t1),"g"(0) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(c1),"+r"(c2) \ - : "d"(t2),"g"(0) \ - : "cc"); \ - } while (0) - -#define sqr_add_c(a,i,c0,c1,c2) do { \ - asm ("mulq %2" \ - : "=a"(t1),"=d"(t2) \ - : "a"(a[i]) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(c0),"+d"(t2) \ - : "a"(t1),"g"(0) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(c1),"+r"(c2) \ - : "d"(t2),"g"(0) \ - : "cc"); \ - } while (0) - -#define mul_add_c2(a,b,c0,c1,c2) do { \ - asm ("mulq %3" \ - : "=a"(t1),"=d"(t2) \ - : "a"(a),"m"(b) \ - : "cc"); \ - asm ("addq %0,%0; adcq %2,%1" \ - : "+d"(t2),"+r"(c2) \ - : "g"(0) \ - : "cc"); \ - asm ("addq %0,%0; adcq %2,%1" \ - : "+a"(t1),"+d"(t2) \ - : "g"(0) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(c0),"+d"(t2) \ - : "a"(t1),"g"(0) \ - : "cc"); \ - asm ("addq %2,%0; adcq %3,%1" \ - : "+r"(c1),"+r"(c2) \ - : "d"(t2),"g"(0) \ - : "cc"); \ - } while (0) -#endif - -#define sqr_add_c2(a,i,j,c0,c1,c2) \ - mul_add_c2((a)[i],(a)[j],c0,c1,c2) +# define mul_add_c(a,b,c0,c1,c2) do { \ + BN_ULONG ta = (a), tb = (b); \ + BN_ULONG lo, hi; \ + BN_UMULT_LOHI(lo,hi,ta,tb); \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define mul_add_c2(a,b,c0,c1,c2) do { \ + BN_ULONG ta = (a), tb = (b); \ + BN_ULONG lo, hi, tt; \ + BN_UMULT_LOHI(lo,hi,ta,tb); \ + c0 += lo; tt = hi+((c0<lo)?1:0); \ + c1 += tt; c2 += (c1<tt)?1:0; \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define sqr_add_c(a,i,c0,c1,c2) do { \ + BN_ULONG ta = (a)[i]; \ + BN_ULONG lo, hi; \ + BN_UMULT_LOHI(lo,hi,ta,ta); \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) +# else +# define mul_add_c(a,b,c0,c1,c2) do { \ + BN_ULONG t1,t2; \ + asm ("mulq %3" \ + : "=a"(t1),"=d"(t2) \ + : "a"(a),"m"(b) \ + : "cc"); \ + asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \ + : "+r"(c0),"+r"(c1),"+r"(c2) \ + : "r"(t1),"r"(t2),"g"(0) \ + : "cc"); \ + } while (0) + +# define sqr_add_c(a,i,c0,c1,c2) do { \ + BN_ULONG t1,t2; \ + asm ("mulq %2" \ + : "=a"(t1),"=d"(t2) \ + : "a"(a[i]) \ + : "cc"); \ + asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \ + : "+r"(c0),"+r"(c1),"+r"(c2) \ + : "r"(t1),"r"(t2),"g"(0) \ + : "cc"); \ + } while (0) + +# define mul_add_c2(a,b,c0,c1,c2) do { \ + BN_ULONG t1,t2; \ + asm ("mulq %3" \ + : "=a"(t1),"=d"(t2) \ + : "a"(a),"m"(b) \ + : "cc"); \ + asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \ + : "+r"(c0),"+r"(c1),"+r"(c2) \ + : "r"(t1),"r"(t2),"g"(0) \ + : "cc"); \ + asm ("addq %3,%0; adcq %4,%1; adcq %5,%2" \ + : "+r"(c0),"+r"(c1),"+r"(c2) \ + : "r"(t1),"r"(t2),"g"(0) \ + : "cc"); \ + } while (0) +# endif + +# define sqr_add_c2(a,i,j,c0,c1,c2) \ + mul_add_c2((a)[i],(a)[j],c0,c1,c2) void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) - { - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - mul_add_c(a[0],b[0],c1,c2,c3); - r[0]=c1; - c1=0; - mul_add_c(a[0],b[1],c2,c3,c1); - mul_add_c(a[1],b[0],c2,c3,c1); - r[1]=c2; - c2=0; - mul_add_c(a[2],b[0],c3,c1,c2); - mul_add_c(a[1],b[1],c3,c1,c2); - mul_add_c(a[0],b[2],c3,c1,c2); - r[2]=c3; - c3=0; - mul_add_c(a[0],b[3],c1,c2,c3); - mul_add_c(a[1],b[2],c1,c2,c3); - mul_add_c(a[2],b[1],c1,c2,c3); - mul_add_c(a[3],b[0],c1,c2,c3); - r[3]=c1; - c1=0; - mul_add_c(a[4],b[0],c2,c3,c1); - mul_add_c(a[3],b[1],c2,c3,c1); - mul_add_c(a[2],b[2],c2,c3,c1); - mul_add_c(a[1],b[3],c2,c3,c1); - mul_add_c(a[0],b[4],c2,c3,c1); - r[4]=c2; - c2=0; - mul_add_c(a[0],b[5],c3,c1,c2); - mul_add_c(a[1],b[4],c3,c1,c2); - mul_add_c(a[2],b[3],c3,c1,c2); - mul_add_c(a[3],b[2],c3,c1,c2); - mul_add_c(a[4],b[1],c3,c1,c2); - mul_add_c(a[5],b[0],c3,c1,c2); - r[5]=c3; - c3=0; - mul_add_c(a[6],b[0],c1,c2,c3); - mul_add_c(a[5],b[1],c1,c2,c3); - mul_add_c(a[4],b[2],c1,c2,c3); - mul_add_c(a[3],b[3],c1,c2,c3); - mul_add_c(a[2],b[4],c1,c2,c3); - mul_add_c(a[1],b[5],c1,c2,c3); - mul_add_c(a[0],b[6],c1,c2,c3); - r[6]=c1; - c1=0; - mul_add_c(a[0],b[7],c2,c3,c1); - mul_add_c(a[1],b[6],c2,c3,c1); - mul_add_c(a[2],b[5],c2,c3,c1); - mul_add_c(a[3],b[4],c2,c3,c1); - mul_add_c(a[4],b[3],c2,c3,c1); - mul_add_c(a[5],b[2],c2,c3,c1); - mul_add_c(a[6],b[1],c2,c3,c1); - mul_add_c(a[7],b[0],c2,c3,c1); - r[7]=c2; - c2=0; - mul_add_c(a[7],b[1],c3,c1,c2); - mul_add_c(a[6],b[2],c3,c1,c2); - mul_add_c(a[5],b[3],c3,c1,c2); - mul_add_c(a[4],b[4],c3,c1,c2); - mul_add_c(a[3],b[5],c3,c1,c2); - mul_add_c(a[2],b[6],c3,c1,c2); - mul_add_c(a[1],b[7],c3,c1,c2); - r[8]=c3; - c3=0; - mul_add_c(a[2],b[7],c1,c2,c3); - mul_add_c(a[3],b[6],c1,c2,c3); - mul_add_c(a[4],b[5],c1,c2,c3); - mul_add_c(a[5],b[4],c1,c2,c3); - mul_add_c(a[6],b[3],c1,c2,c3); - mul_add_c(a[7],b[2],c1,c2,c3); - r[9]=c1; - c1=0; - mul_add_c(a[7],b[3],c2,c3,c1); - mul_add_c(a[6],b[4],c2,c3,c1); - mul_add_c(a[5],b[5],c2,c3,c1); - mul_add_c(a[4],b[6],c2,c3,c1); - mul_add_c(a[3],b[7],c2,c3,c1); - r[10]=c2; - c2=0; - mul_add_c(a[4],b[7],c3,c1,c2); - mul_add_c(a[5],b[6],c3,c1,c2); - mul_add_c(a[6],b[5],c3,c1,c2); - mul_add_c(a[7],b[4],c3,c1,c2); - r[11]=c3; - c3=0; - mul_add_c(a[7],b[5],c1,c2,c3); - mul_add_c(a[6],b[6],c1,c2,c3); - mul_add_c(a[5],b[7],c1,c2,c3); - r[12]=c1; - c1=0; - mul_add_c(a[6],b[7],c2,c3,c1); - mul_add_c(a[7],b[6],c2,c3,c1); - r[13]=c2; - c2=0; - mul_add_c(a[7],b[7],c3,c1,c2); - r[14]=c3; - r[15]=c1; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + mul_add_c(a[0], b[0], c1, c2, c3); + r[0] = c1; + c1 = 0; + mul_add_c(a[0], b[1], c2, c3, c1); + mul_add_c(a[1], b[0], c2, c3, c1); + r[1] = c2; + c2 = 0; + mul_add_c(a[2], b[0], c3, c1, c2); + mul_add_c(a[1], b[1], c3, c1, c2); + mul_add_c(a[0], b[2], c3, c1, c2); + r[2] = c3; + c3 = 0; + mul_add_c(a[0], b[3], c1, c2, c3); + mul_add_c(a[1], b[2], c1, c2, c3); + mul_add_c(a[2], b[1], c1, c2, c3); + mul_add_c(a[3], b[0], c1, c2, c3); + r[3] = c1; + c1 = 0; + mul_add_c(a[4], b[0], c2, c3, c1); + mul_add_c(a[3], b[1], c2, c3, c1); + mul_add_c(a[2], b[2], c2, c3, c1); + mul_add_c(a[1], b[3], c2, c3, c1); + mul_add_c(a[0], b[4], c2, c3, c1); + r[4] = c2; + c2 = 0; + mul_add_c(a[0], b[5], c3, c1, c2); + mul_add_c(a[1], b[4], c3, c1, c2); + mul_add_c(a[2], b[3], c3, c1, c2); + mul_add_c(a[3], b[2], c3, c1, c2); + mul_add_c(a[4], b[1], c3, c1, c2); + mul_add_c(a[5], b[0], c3, c1, c2); + r[5] = c3; + c3 = 0; + mul_add_c(a[6], b[0], c1, c2, c3); + mul_add_c(a[5], b[1], c1, c2, c3); + mul_add_c(a[4], b[2], c1, c2, c3); + mul_add_c(a[3], b[3], c1, c2, c3); + mul_add_c(a[2], b[4], c1, c2, c3); + mul_add_c(a[1], b[5], c1, c2, c3); + mul_add_c(a[0], b[6], c1, c2, c3); + r[6] = c1; + c1 = 0; + mul_add_c(a[0], b[7], c2, c3, c1); + mul_add_c(a[1], b[6], c2, c3, c1); + mul_add_c(a[2], b[5], c2, c3, c1); + mul_add_c(a[3], b[4], c2, c3, c1); + mul_add_c(a[4], b[3], c2, c3, c1); + mul_add_c(a[5], b[2], c2, c3, c1); + mul_add_c(a[6], b[1], c2, c3, c1); + mul_add_c(a[7], b[0], c2, c3, c1); + r[7] = c2; + c2 = 0; + mul_add_c(a[7], b[1], c3, c1, c2); + mul_add_c(a[6], b[2], c3, c1, c2); + mul_add_c(a[5], b[3], c3, c1, c2); + mul_add_c(a[4], b[4], c3, c1, c2); + mul_add_c(a[3], b[5], c3, c1, c2); + mul_add_c(a[2], b[6], c3, c1, c2); + mul_add_c(a[1], b[7], c3, c1, c2); + r[8] = c3; + c3 = 0; + mul_add_c(a[2], b[7], c1, c2, c3); + mul_add_c(a[3], b[6], c1, c2, c3); + mul_add_c(a[4], b[5], c1, c2, c3); + mul_add_c(a[5], b[4], c1, c2, c3); + mul_add_c(a[6], b[3], c1, c2, c3); + mul_add_c(a[7], b[2], c1, c2, c3); + r[9] = c1; + c1 = 0; + mul_add_c(a[7], b[3], c2, c3, c1); + mul_add_c(a[6], b[4], c2, c3, c1); + mul_add_c(a[5], b[5], c2, c3, c1); + mul_add_c(a[4], b[6], c2, c3, c1); + mul_add_c(a[3], b[7], c2, c3, c1); + r[10] = c2; + c2 = 0; + mul_add_c(a[4], b[7], c3, c1, c2); + mul_add_c(a[5], b[6], c3, c1, c2); + mul_add_c(a[6], b[5], c3, c1, c2); + mul_add_c(a[7], b[4], c3, c1, c2); + r[11] = c3; + c3 = 0; + mul_add_c(a[7], b[5], c1, c2, c3); + mul_add_c(a[6], b[6], c1, c2, c3); + mul_add_c(a[5], b[7], c1, c2, c3); + r[12] = c1; + c1 = 0; + mul_add_c(a[6], b[7], c2, c3, c1); + mul_add_c(a[7], b[6], c2, c3, c1); + r[13] = c2; + c2 = 0; + mul_add_c(a[7], b[7], c3, c1, c2); + r[14] = c3; + r[15] = c1; +} void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) - { - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - mul_add_c(a[0],b[0],c1,c2,c3); - r[0]=c1; - c1=0; - mul_add_c(a[0],b[1],c2,c3,c1); - mul_add_c(a[1],b[0],c2,c3,c1); - r[1]=c2; - c2=0; - mul_add_c(a[2],b[0],c3,c1,c2); - mul_add_c(a[1],b[1],c3,c1,c2); - mul_add_c(a[0],b[2],c3,c1,c2); - r[2]=c3; - c3=0; - mul_add_c(a[0],b[3],c1,c2,c3); - mul_add_c(a[1],b[2],c1,c2,c3); - mul_add_c(a[2],b[1],c1,c2,c3); - mul_add_c(a[3],b[0],c1,c2,c3); - r[3]=c1; - c1=0; - mul_add_c(a[3],b[1],c2,c3,c1); - mul_add_c(a[2],b[2],c2,c3,c1); - mul_add_c(a[1],b[3],c2,c3,c1); - r[4]=c2; - c2=0; - mul_add_c(a[2],b[3],c3,c1,c2); - mul_add_c(a[3],b[2],c3,c1,c2); - r[5]=c3; - c3=0; - mul_add_c(a[3],b[3],c1,c2,c3); - r[6]=c1; - r[7]=c2; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + mul_add_c(a[0], b[0], c1, c2, c3); + r[0] = c1; + c1 = 0; + mul_add_c(a[0], b[1], c2, c3, c1); + mul_add_c(a[1], b[0], c2, c3, c1); + r[1] = c2; + c2 = 0; + mul_add_c(a[2], b[0], c3, c1, c2); + mul_add_c(a[1], b[1], c3, c1, c2); + mul_add_c(a[0], b[2], c3, c1, c2); + r[2] = c3; + c3 = 0; + mul_add_c(a[0], b[3], c1, c2, c3); + mul_add_c(a[1], b[2], c1, c2, c3); + mul_add_c(a[2], b[1], c1, c2, c3); + mul_add_c(a[3], b[0], c1, c2, c3); + r[3] = c1; + c1 = 0; + mul_add_c(a[3], b[1], c2, c3, c1); + mul_add_c(a[2], b[2], c2, c3, c1); + mul_add_c(a[1], b[3], c2, c3, c1); + r[4] = c2; + c2 = 0; + mul_add_c(a[2], b[3], c3, c1, c2); + mul_add_c(a[3], b[2], c3, c1, c2); + r[5] = c3; + c3 = 0; + mul_add_c(a[3], b[3], c1, c2, c3); + r[6] = c1; + r[7] = c2; +} void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a) - { - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - sqr_add_c(a,0,c1,c2,c3); - r[0]=c1; - c1=0; - sqr_add_c2(a,1,0,c2,c3,c1); - r[1]=c2; - c2=0; - sqr_add_c(a,1,c3,c1,c2); - sqr_add_c2(a,2,0,c3,c1,c2); - r[2]=c3; - c3=0; - sqr_add_c2(a,3,0,c1,c2,c3); - sqr_add_c2(a,2,1,c1,c2,c3); - r[3]=c1; - c1=0; - sqr_add_c(a,2,c2,c3,c1); - sqr_add_c2(a,3,1,c2,c3,c1); - sqr_add_c2(a,4,0,c2,c3,c1); - r[4]=c2; - c2=0; - sqr_add_c2(a,5,0,c3,c1,c2); - sqr_add_c2(a,4,1,c3,c1,c2); - sqr_add_c2(a,3,2,c3,c1,c2); - r[5]=c3; - c3=0; - sqr_add_c(a,3,c1,c2,c3); - sqr_add_c2(a,4,2,c1,c2,c3); - sqr_add_c2(a,5,1,c1,c2,c3); - sqr_add_c2(a,6,0,c1,c2,c3); - r[6]=c1; - c1=0; - sqr_add_c2(a,7,0,c2,c3,c1); - sqr_add_c2(a,6,1,c2,c3,c1); - sqr_add_c2(a,5,2,c2,c3,c1); - sqr_add_c2(a,4,3,c2,c3,c1); - r[7]=c2; - c2=0; - sqr_add_c(a,4,c3,c1,c2); - sqr_add_c2(a,5,3,c3,c1,c2); - sqr_add_c2(a,6,2,c3,c1,c2); - sqr_add_c2(a,7,1,c3,c1,c2); - r[8]=c3; - c3=0; - sqr_add_c2(a,7,2,c1,c2,c3); - sqr_add_c2(a,6,3,c1,c2,c3); - sqr_add_c2(a,5,4,c1,c2,c3); - r[9]=c1; - c1=0; - sqr_add_c(a,5,c2,c3,c1); - sqr_add_c2(a,6,4,c2,c3,c1); - sqr_add_c2(a,7,3,c2,c3,c1); - r[10]=c2; - c2=0; - sqr_add_c2(a,7,4,c3,c1,c2); - sqr_add_c2(a,6,5,c3,c1,c2); - r[11]=c3; - c3=0; - sqr_add_c(a,6,c1,c2,c3); - sqr_add_c2(a,7,5,c1,c2,c3); - r[12]=c1; - c1=0; - sqr_add_c2(a,7,6,c2,c3,c1); - r[13]=c2; - c2=0; - sqr_add_c(a,7,c3,c1,c2); - r[14]=c3; - r[15]=c1; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + sqr_add_c(a, 0, c1, c2, c3); + r[0] = c1; + c1 = 0; + sqr_add_c2(a, 1, 0, c2, c3, c1); + r[1] = c2; + c2 = 0; + sqr_add_c(a, 1, c3, c1, c2); + sqr_add_c2(a, 2, 0, c3, c1, c2); + r[2] = c3; + c3 = 0; + sqr_add_c2(a, 3, 0, c1, c2, c3); + sqr_add_c2(a, 2, 1, c1, c2, c3); + r[3] = c1; + c1 = 0; + sqr_add_c(a, 2, c2, c3, c1); + sqr_add_c2(a, 3, 1, c2, c3, c1); + sqr_add_c2(a, 4, 0, c2, c3, c1); + r[4] = c2; + c2 = 0; + sqr_add_c2(a, 5, 0, c3, c1, c2); + sqr_add_c2(a, 4, 1, c3, c1, c2); + sqr_add_c2(a, 3, 2, c3, c1, c2); + r[5] = c3; + c3 = 0; + sqr_add_c(a, 3, c1, c2, c3); + sqr_add_c2(a, 4, 2, c1, c2, c3); + sqr_add_c2(a, 5, 1, c1, c2, c3); + sqr_add_c2(a, 6, 0, c1, c2, c3); + r[6] = c1; + c1 = 0; + sqr_add_c2(a, 7, 0, c2, c3, c1); + sqr_add_c2(a, 6, 1, c2, c3, c1); + sqr_add_c2(a, 5, 2, c2, c3, c1); + sqr_add_c2(a, 4, 3, c2, c3, c1); + r[7] = c2; + c2 = 0; + sqr_add_c(a, 4, c3, c1, c2); + sqr_add_c2(a, 5, 3, c3, c1, c2); + sqr_add_c2(a, 6, 2, c3, c1, c2); + sqr_add_c2(a, 7, 1, c3, c1, c2); + r[8] = c3; + c3 = 0; + sqr_add_c2(a, 7, 2, c1, c2, c3); + sqr_add_c2(a, 6, 3, c1, c2, c3); + sqr_add_c2(a, 5, 4, c1, c2, c3); + r[9] = c1; + c1 = 0; + sqr_add_c(a, 5, c2, c3, c1); + sqr_add_c2(a, 6, 4, c2, c3, c1); + sqr_add_c2(a, 7, 3, c2, c3, c1); + r[10] = c2; + c2 = 0; + sqr_add_c2(a, 7, 4, c3, c1, c2); + sqr_add_c2(a, 6, 5, c3, c1, c2); + r[11] = c3; + c3 = 0; + sqr_add_c(a, 6, c1, c2, c3); + sqr_add_c2(a, 7, 5, c1, c2, c3); + r[12] = c1; + c1 = 0; + sqr_add_c2(a, 7, 6, c2, c3, c1); + r[13] = c2; + c2 = 0; + sqr_add_c(a, 7, c3, c1, c2); + r[14] = c3; + r[15] = c1; +} void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a) - { - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - sqr_add_c(a,0,c1,c2,c3); - r[0]=c1; - c1=0; - sqr_add_c2(a,1,0,c2,c3,c1); - r[1]=c2; - c2=0; - sqr_add_c(a,1,c3,c1,c2); - sqr_add_c2(a,2,0,c3,c1,c2); - r[2]=c3; - c3=0; - sqr_add_c2(a,3,0,c1,c2,c3); - sqr_add_c2(a,2,1,c1,c2,c3); - r[3]=c1; - c1=0; - sqr_add_c(a,2,c2,c3,c1); - sqr_add_c2(a,3,1,c2,c3,c1); - r[4]=c2; - c2=0; - sqr_add_c2(a,3,2,c3,c1,c2); - r[5]=c3; - c3=0; - sqr_add_c(a,3,c1,c2,c3); - r[6]=c1; - r[7]=c2; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + sqr_add_c(a, 0, c1, c2, c3); + r[0] = c1; + c1 = 0; + sqr_add_c2(a, 1, 0, c2, c3, c1); + r[1] = c2; + c2 = 0; + sqr_add_c(a, 1, c3, c1, c2); + sqr_add_c2(a, 2, 0, c3, c1, c2); + r[2] = c3; + c3 = 0; + sqr_add_c2(a, 3, 0, c1, c2, c3); + sqr_add_c2(a, 2, 1, c1, c2, c3); + r[3] = c1; + c1 = 0; + sqr_add_c(a, 2, c2, c3, c1); + sqr_add_c2(a, 3, 1, c2, c3, c1); + r[4] = c2; + c2 = 0; + sqr_add_c2(a, 3, 2, c3, c1, c2); + r[5] = c3; + c3 = 0; + sqr_add_c(a, 3, c1, c2, c3); + r[6] = c1; + r[7] = c2; +} #endif diff --git a/openssl/crypto/bn/asm/x86_64-mont.pl b/openssl/crypto/bn/asm/x86_64-mont.pl index 17fb94c84..2989b58f2 100644 --- a/openssl/crypto/bn/asm/x86_64-mont.pl +++ b/openssl/crypto/bn/asm/x86_64-mont.pl @@ -29,6 +29,16 @@ # to *initial* version of this module from 2005 is ~0%/30%/40%/45% # for 512-/1024-/2048-/4096-bit RSA *sign* benchmarks respectively. +# June 2013. +# +# Optimize reduction in squaring procedure and improve 1024+-bit RSA +# sign performance by 10-16% on Intel Sandy Bridge and later +# (virtually same on non-Intel processors). + +# August 2013. +# +# Add MULX/ADOX/ADCX code path. + $flavour = shift; $output = shift; if ($flavour =~ /\./) { $output = $flavour; undef $flavour; } @@ -43,6 +53,21 @@ die "can't locate x86_64-xlate.pl"; open OUT,"| \"$^X\" $xlate $flavour $output"; *STDOUT=*OUT; +if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1` + =~ /GNU assembler version ([2-9]\.[0-9]+)/) { + $addx = ($1>=2.23); +} + +if (!$addx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) && + `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) { + $addx = ($1>=2.10); +} + +if (!$addx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) && + `ml64 2>&1` =~ /Version ([0-9]+)\./) { + $addx = ($1>=12); +} + # int bn_mul_mont( $rp="%rdi"; # BN_ULONG *rp, $ap="%rsi"; # const BN_ULONG *ap, @@ -61,6 +86,8 @@ $m1="%rbp"; $code=<<___; .text +.extern OPENSSL_ia32cap_P + .globl bn_mul_mont .type bn_mul_mont,\@function,6 .align 16 @@ -69,9 +96,16 @@ bn_mul_mont: jnz .Lmul_enter cmp \$8,${num}d jb .Lmul_enter +___ +$code.=<<___ if ($addx); + mov OPENSSL_ia32cap_P+8(%rip),%r11d +___ +$code.=<<___; cmp $ap,$bp jne .Lmul4x_enter - jmp .Lsqr4x_enter + test \$7,${num}d + jz .Lsqr8x_enter + jmp .Lmul4x_enter .align 16 .Lmul_enter: @@ -227,7 +261,7 @@ $code.=<<___; lea 1($i),$i # i++ cmp $num,$i - jl .Louter + jb .Louter xor $i,$i # i=0 and clear CF! mov (%rsp),%rax # tp[0] @@ -280,6 +314,13 @@ $code.=<<___; .align 16 bn_mul4x_mont: .Lmul4x_enter: +___ +$code.=<<___ if ($addx); + and \$0x80100,%r11d + cmp \$0x80100,%r11d + je .Lmulx4x_enter +___ +$code.=<<___; push %rbx push %rbp push %r12 @@ -401,7 +442,7 @@ $code.=<<___; mov $N[1],-32(%rsp,$j,8) # tp[j-1] mov %rdx,$N[0] cmp $num,$j - jl .L1st4x + jb .L1st4x mulq $m0 # ap[j]*bp[0] add %rax,$A[0] @@ -549,7 +590,7 @@ $code.=<<___; mov $N[1],-32(%rsp,$j,8) # tp[j-1] mov %rdx,$N[0] cmp $num,$j - jl .Linner4x + jb .Linner4x mulq $m0 # ap[j]*bp[i] add %rax,$A[0] @@ -595,7 +636,7 @@ $code.=<<___; mov $N[1],(%rsp,$j,8) # store upmost overflow bit cmp $num,$i - jl .Louter4x + jb .Louter4x ___ { my @ri=("%rax","%rdx",$m0,$m1); @@ -688,25 +729,30 @@ ___ }}} {{{ ###################################################################### -# void bn_sqr4x_mont( +# void bn_sqr8x_mont( my $rptr="%rdi"; # const BN_ULONG *rptr, my $aptr="%rsi"; # const BN_ULONG *aptr, my $bptr="%rdx"; # not used my $nptr="%rcx"; # const BN_ULONG *nptr, my $n0 ="%r8"; # const BN_ULONG *n0); -my $num ="%r9"; # int num, has to be divisible by 4 and - # not less than 8 +my $num ="%r9"; # int num, has to be divisible by 8 my ($i,$j,$tptr)=("%rbp","%rcx",$rptr); my @A0=("%r10","%r11"); my @A1=("%r12","%r13"); my ($a0,$a1,$ai)=("%r14","%r15","%rbx"); +$code.=<<___ if ($addx); +.extern bn_sqrx8x_internal # see x86_64-mont5 module +___ $code.=<<___; -.type bn_sqr4x_mont,\@function,6 -.align 16 -bn_sqr4x_mont: -.Lsqr4x_enter: +.extern bn_sqr8x_internal # see x86_64-mont5 module + +.type bn_sqr8x_mont,\@function,6 +.align 32 +bn_sqr8x_mont: +.Lsqr8x_enter: + mov %rsp,%rax push %rbx push %rbp push %r12 @@ -714,787 +760,445 @@ bn_sqr4x_mont: push %r14 push %r15 + mov ${num}d,%r10d shl \$3,${num}d # convert $num to bytes - xor %r10,%r10 - mov %rsp,%r11 # put aside %rsp - sub $num,%r10 # -$num - mov ($n0),$n0 # *n0 - lea -72(%rsp,%r10,2),%rsp # alloca(frame+2*$num) - and \$-1024,%rsp # minimize TLB usage - ############################################################## - # Stack layout - # - # +0 saved $num, used in reduction section - # +8 &t[2*$num], used in reduction section - # +32 saved $rptr - # +40 saved $nptr - # +48 saved *n0 - # +56 saved %rsp - # +64 t[2*$num] - # - mov $rptr,32(%rsp) # save $rptr - mov $nptr,40(%rsp) - mov $n0, 48(%rsp) - mov %r11, 56(%rsp) # save original %rsp -.Lsqr4x_body: + shl \$3+2,%r10 # 4*$num + neg $num + ############################################################## - # Squaring part: - # - # a) multiply-n-add everything but a[i]*a[i]; - # b) shift result of a) by 1 to the left and accumulate - # a[i]*a[i] products; + # ensure that stack frame doesn't alias with $aptr modulo + # 4096. this is done to allow memory disambiguation logic + # do its job. # - lea 32(%r10),$i # $i=-($num-32) - lea ($aptr,$num),$aptr # end of a[] buffer, ($aptr,$i)=&ap[2] - - mov $num,$j # $j=$num - - # comments apply to $num==8 case - mov -32($aptr,$i),$a0 # a[0] - lea 64(%rsp,$num,2),$tptr # end of tp[] buffer, &tp[2*$num] - mov -24($aptr,$i),%rax # a[1] - lea -32($tptr,$i),$tptr # end of tp[] window, &tp[2*$num-"$i"] - mov -16($aptr,$i),$ai # a[2] - mov %rax,$a1 - - mul $a0 # a[1]*a[0] - mov %rax,$A0[0] # a[1]*a[0] - mov $ai,%rax # a[2] - mov %rdx,$A0[1] - mov $A0[0],-24($tptr,$i) # t[1] - - xor $A0[0],$A0[0] - mul $a0 # a[2]*a[0] - add %rax,$A0[1] - mov $ai,%rax - adc %rdx,$A0[0] - mov $A0[1],-16($tptr,$i) # t[2] - - lea -16($i),$j # j=-16 - - - mov 8($aptr,$j),$ai # a[3] - mul $a1 # a[2]*a[1] - mov %rax,$A1[0] # a[2]*a[1]+t[3] - mov $ai,%rax - mov %rdx,$A1[1] - - xor $A0[1],$A0[1] - add $A1[0],$A0[0] - lea 16($j),$j - adc \$0,$A0[1] - mul $a0 # a[3]*a[0] - add %rax,$A0[0] # a[3]*a[0]+a[2]*a[1]+t[3] - mov $ai,%rax - adc %rdx,$A0[1] - mov $A0[0],-8($tptr,$j) # t[3] - jmp .Lsqr4x_1st + lea -64(%rsp,$num,4),%r11 + mov ($n0),$n0 # *n0 + sub $aptr,%r11 + and \$4095,%r11 + cmp %r11,%r10 + jb .Lsqr8x_sp_alt + sub %r11,%rsp # align with $aptr + lea -64(%rsp,$num,4),%rsp # alloca(frame+4*$num) + jmp .Lsqr8x_sp_done + +.align 32 +.Lsqr8x_sp_alt: + lea 4096-64(,$num,4),%r10 # 4096-frame-4*$num + lea -64(%rsp,$num,4),%rsp # alloca(frame+4*$num) + sub %r10,%r11 + mov \$0,%r10 + cmovc %r10,%r11 + sub %r11,%rsp +.Lsqr8x_sp_done: + and \$-64,%rsp + mov $num,%r10 + neg $num + + lea 64(%rsp,$num,2),%r11 # copy of modulus + mov $n0, 32(%rsp) + mov %rax, 40(%rsp) # save original %rsp +.Lsqr8x_body: + + mov $num,$i + movq %r11, %xmm2 # save pointer to modulus copy + shr \$3+2,$i + mov OPENSSL_ia32cap_P+8(%rip),%eax + jmp .Lsqr8x_copy_n + +.align 32 +.Lsqr8x_copy_n: + movq 8*0($nptr),%xmm0 + movq 8*1($nptr),%xmm1 + movq 8*2($nptr),%xmm3 + movq 8*3($nptr),%xmm4 + lea 8*4($nptr),$nptr + movdqa %xmm0,16*0(%r11) + movdqa %xmm1,16*1(%r11) + movdqa %xmm3,16*2(%r11) + movdqa %xmm4,16*3(%r11) + lea 16*4(%r11),%r11 + dec $i + jnz .Lsqr8x_copy_n -.align 16 -.Lsqr4x_1st: - mov ($aptr,$j),$ai # a[4] - xor $A1[0],$A1[0] - mul $a1 # a[3]*a[1] - add %rax,$A1[1] # a[3]*a[1]+t[4] - mov $ai,%rax - adc %rdx,$A1[0] - - xor $A0[0],$A0[0] - add $A1[1],$A0[1] - adc \$0,$A0[0] - mul $a0 # a[4]*a[0] - add %rax,$A0[1] # a[4]*a[0]+a[3]*a[1]+t[4] - mov $ai,%rax # a[3] - adc %rdx,$A0[0] - mov $A0[1],($tptr,$j) # t[4] - - - mov 8($aptr,$j),$ai # a[5] - xor $A1[1],$A1[1] - mul $a1 # a[4]*a[3] - add %rax,$A1[0] # a[4]*a[3]+t[5] - mov $ai,%rax - adc %rdx,$A1[1] - - xor $A0[1],$A0[1] - add $A1[0],$A0[0] - adc \$0,$A0[1] - mul $a0 # a[5]*a[2] - add %rax,$A0[0] # a[5]*a[2]+a[4]*a[3]+t[5] - mov $ai,%rax - adc %rdx,$A0[1] - mov $A0[0],8($tptr,$j) # t[5] - - mov 16($aptr,$j),$ai # a[6] - xor $A1[0],$A1[0] - mul $a1 # a[5]*a[3] - add %rax,$A1[1] # a[5]*a[3]+t[6] - mov $ai,%rax - adc %rdx,$A1[0] - - xor $A0[0],$A0[0] - add $A1[1],$A0[1] - adc \$0,$A0[0] - mul $a0 # a[6]*a[2] - add %rax,$A0[1] # a[6]*a[2]+a[5]*a[3]+t[6] - mov $ai,%rax # a[3] - adc %rdx,$A0[0] - mov $A0[1],16($tptr,$j) # t[6] - - - mov 24($aptr,$j),$ai # a[7] - xor $A1[1],$A1[1] - mul $a1 # a[6]*a[5] - add %rax,$A1[0] # a[6]*a[5]+t[7] - mov $ai,%rax - adc %rdx,$A1[1] - - xor $A0[1],$A0[1] - add $A1[0],$A0[0] - lea 32($j),$j - adc \$0,$A0[1] - mul $a0 # a[7]*a[4] - add %rax,$A0[0] # a[7]*a[4]+a[6]*a[5]+t[6] - mov $ai,%rax - adc %rdx,$A0[1] - mov $A0[0],-8($tptr,$j) # t[7] - - cmp \$0,$j - jne .Lsqr4x_1st - - xor $A1[0],$A1[0] - add $A0[1],$A1[1] - adc \$0,$A1[0] - mul $a1 # a[7]*a[5] - add %rax,$A1[1] - adc %rdx,$A1[0] - - mov $A1[1],($tptr) # t[8] - lea 16($i),$i - mov $A1[0],8($tptr) # t[9] - jmp .Lsqr4x_outer + pxor %xmm0,%xmm0 + movq $rptr,%xmm1 # save $rptr + movq %r10, %xmm3 # -$num +___ +$code.=<<___ if ($addx); + and \$0x80100,%eax + cmp \$0x80100,%eax + jne .Lsqr8x_nox -.align 16 -.Lsqr4x_outer: # comments apply to $num==6 case - mov -32($aptr,$i),$a0 # a[0] - lea 64(%rsp,$num,2),$tptr # end of tp[] buffer, &tp[2*$num] - mov -24($aptr,$i),%rax # a[1] - lea -32($tptr,$i),$tptr # end of tp[] window, &tp[2*$num-"$i"] - mov -16($aptr,$i),$ai # a[2] - mov %rax,$a1 - - mov -24($tptr,$i),$A0[0] # t[1] - xor $A0[1],$A0[1] - mul $a0 # a[1]*a[0] - add %rax,$A0[0] # a[1]*a[0]+t[1] - mov $ai,%rax # a[2] - adc %rdx,$A0[1] - mov $A0[0],-24($tptr,$i) # t[1] - - xor $A0[0],$A0[0] - add -16($tptr,$i),$A0[1] # a[2]*a[0]+t[2] - adc \$0,$A0[0] - mul $a0 # a[2]*a[0] - add %rax,$A0[1] - mov $ai,%rax - adc %rdx,$A0[0] - mov $A0[1],-16($tptr,$i) # t[2] - - lea -16($i),$j # j=-16 - xor $A1[0],$A1[0] - - - mov 8($aptr,$j),$ai # a[3] - xor $A1[1],$A1[1] - add 8($tptr,$j),$A1[0] - adc \$0,$A1[1] - mul $a1 # a[2]*a[1] - add %rax,$A1[0] # a[2]*a[1]+t[3] - mov $ai,%rax - adc %rdx,$A1[1] - - xor $A0[1],$A0[1] - add $A1[0],$A0[0] - adc \$0,$A0[1] - mul $a0 # a[3]*a[0] - add %rax,$A0[0] # a[3]*a[0]+a[2]*a[1]+t[3] - mov $ai,%rax - adc %rdx,$A0[1] - mov $A0[0],8($tptr,$j) # t[3] - - lea 16($j),$j - jmp .Lsqr4x_inner + call bn_sqrx8x_internal # see x86_64-mont5 module -.align 16 -.Lsqr4x_inner: - mov ($aptr,$j),$ai # a[4] - xor $A1[0],$A1[0] - add ($tptr,$j),$A1[1] - adc \$0,$A1[0] - mul $a1 # a[3]*a[1] - add %rax,$A1[1] # a[3]*a[1]+t[4] - mov $ai,%rax - adc %rdx,$A1[0] - - xor $A0[0],$A0[0] - add $A1[1],$A0[1] - adc \$0,$A0[0] - mul $a0 # a[4]*a[0] - add %rax,$A0[1] # a[4]*a[0]+a[3]*a[1]+t[4] - mov $ai,%rax # a[3] - adc %rdx,$A0[0] - mov $A0[1],($tptr,$j) # t[4] - - mov 8($aptr,$j),$ai # a[5] - xor $A1[1],$A1[1] - add 8($tptr,$j),$A1[0] - adc \$0,$A1[1] - mul $a1 # a[4]*a[3] - add %rax,$A1[0] # a[4]*a[3]+t[5] - mov $ai,%rax - adc %rdx,$A1[1] - - xor $A0[1],$A0[1] - add $A1[0],$A0[0] - lea 16($j),$j # j++ - adc \$0,$A0[1] - mul $a0 # a[5]*a[2] - add %rax,$A0[0] # a[5]*a[2]+a[4]*a[3]+t[5] - mov $ai,%rax - adc %rdx,$A0[1] - mov $A0[0],-8($tptr,$j) # t[5], "preloaded t[1]" below - - cmp \$0,$j - jne .Lsqr4x_inner - - xor $A1[0],$A1[0] - add $A0[1],$A1[1] - adc \$0,$A1[0] - mul $a1 # a[5]*a[3] - add %rax,$A1[1] - adc %rdx,$A1[0] - - mov $A1[1],($tptr) # t[6], "preloaded t[2]" below - mov $A1[0],8($tptr) # t[7], "preloaded t[3]" below - - add \$16,$i - jnz .Lsqr4x_outer - - # comments apply to $num==4 case - mov -32($aptr),$a0 # a[0] - lea 64(%rsp,$num,2),$tptr # end of tp[] buffer, &tp[2*$num] - mov -24($aptr),%rax # a[1] - lea -32($tptr,$i),$tptr # end of tp[] window, &tp[2*$num-"$i"] - mov -16($aptr),$ai # a[2] - mov %rax,$a1 - - xor $A0[1],$A0[1] - mul $a0 # a[1]*a[0] - add %rax,$A0[0] # a[1]*a[0]+t[1], preloaded t[1] - mov $ai,%rax # a[2] - adc %rdx,$A0[1] - mov $A0[0],-24($tptr) # t[1] - - xor $A0[0],$A0[0] - add $A1[1],$A0[1] # a[2]*a[0]+t[2], preloaded t[2] - adc \$0,$A0[0] - mul $a0 # a[2]*a[0] - add %rax,$A0[1] - mov $ai,%rax - adc %rdx,$A0[0] - mov $A0[1],-16($tptr) # t[2] - - mov -8($aptr),$ai # a[3] - mul $a1 # a[2]*a[1] - add %rax,$A1[0] # a[2]*a[1]+t[3], preloaded t[3] - mov $ai,%rax - adc \$0,%rdx - - xor $A0[1],$A0[1] - add $A1[0],$A0[0] - mov %rdx,$A1[1] - adc \$0,$A0[1] - mul $a0 # a[3]*a[0] - add %rax,$A0[0] # a[3]*a[0]+a[2]*a[1]+t[3] - mov $ai,%rax - adc %rdx,$A0[1] - mov $A0[0],-8($tptr) # t[3] - - xor $A1[0],$A1[0] - add $A0[1],$A1[1] - adc \$0,$A1[0] - mul $a1 # a[3]*a[1] - add %rax,$A1[1] - mov -16($aptr),%rax # a[2] - adc %rdx,$A1[0] - - mov $A1[1],($tptr) # t[4] - mov $A1[0],8($tptr) # t[5] - - mul $ai # a[2]*a[3] + pxor %xmm0,%xmm0 + lea 48(%rsp),%rax + lea 64(%rsp,$num,2),%rdx + shr \$3+2,$num + mov 40(%rsp),%rsi # restore %rsp + jmp .Lsqr8x_zero + +.align 32 +.Lsqr8x_nox: ___ -{ -my ($shift,$carry)=($a0,$a1); -my @S=(@A1,$ai,$n0); $code.=<<___; - add \$16,$i - xor $shift,$shift - sub $num,$i # $i=16-$num - xor $carry,$carry - - add $A1[0],%rax # t[5] - adc \$0,%rdx - mov %rax,8($tptr) # t[5] - mov %rdx,16($tptr) # t[6] - mov $carry,24($tptr) # t[7] - - mov -16($aptr,$i),%rax # a[0] - lea 64(%rsp,$num,2),$tptr - xor $A0[0],$A0[0] # t[0] - mov -24($tptr,$i,2),$A0[1] # t[1] - - lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[1] # | t[2*i]>>63 - mov -16($tptr,$i,2),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov -8($tptr,$i,2),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[0] - mov -8($aptr,$i),%rax # a[i+1] # prefetch - mov $S[0],-32($tptr,$i,2) - adc %rdx,$S[1] - - lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1 | shift - mov $S[1],-24($tptr,$i,2) - sbb $carry,$carry # mov cf,$carry - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[3] # | t[2*i]>>63 - mov 0($tptr,$i,2),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov 8($tptr,$i,2),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[2] - mov 0($aptr,$i),%rax # a[i+1] # prefetch - mov $S[2],-16($tptr,$i,2) - adc %rdx,$S[3] - lea 16($i),$i - mov $S[3],-40($tptr,$i,2) - sbb $carry,$carry # mov cf,$carry - jmp .Lsqr4x_shift_n_add + call bn_sqr8x_internal # see x86_64-mont5 module -.align 16 -.Lsqr4x_shift_n_add: - lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[1] # | t[2*i]>>63 - mov -16($tptr,$i,2),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov -8($tptr,$i,2),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[0] - mov -8($aptr,$i),%rax # a[i+1] # prefetch - mov $S[0],-32($tptr,$i,2) - adc %rdx,$S[1] - - lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1 | shift - mov $S[1],-24($tptr,$i,2) - sbb $carry,$carry # mov cf,$carry - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[3] # | t[2*i]>>63 - mov 0($tptr,$i,2),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov 8($tptr,$i,2),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[2] - mov 0($aptr,$i),%rax # a[i+1] # prefetch - mov $S[2],-16($tptr,$i,2) - adc %rdx,$S[3] - - lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift - mov $S[3],-8($tptr,$i,2) - sbb $carry,$carry # mov cf,$carry - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[1] # | t[2*i]>>63 - mov 16($tptr,$i,2),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov 24($tptr,$i,2),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[0] - mov 8($aptr,$i),%rax # a[i+1] # prefetch - mov $S[0],0($tptr,$i,2) - adc %rdx,$S[1] - - lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1 | shift - mov $S[1],8($tptr,$i,2) - sbb $carry,$carry # mov cf,$carry - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[3] # | t[2*i]>>63 - mov 32($tptr,$i,2),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov 40($tptr,$i,2),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[2] - mov 16($aptr,$i),%rax # a[i+1] # prefetch - mov $S[2],16($tptr,$i,2) - adc %rdx,$S[3] - mov $S[3],24($tptr,$i,2) - sbb $carry,$carry # mov cf,$carry - add \$32,$i - jnz .Lsqr4x_shift_n_add - - lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[1] # | t[2*i]>>63 - mov -16($tptr),$A0[0] # t[2*i+2] # prefetch - mov $A0[1],$shift # shift=t[2*i+1]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - mov -8($tptr),$A0[1] # t[2*i+2+1] # prefetch - adc %rax,$S[0] - mov -8($aptr),%rax # a[i+1] # prefetch - mov $S[0],-32($tptr) - adc %rdx,$S[1] - - lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1|shift - mov $S[1],-24($tptr) - sbb $carry,$carry # mov cf,$carry - shr \$63,$A0[0] - lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | - shr \$63,$A0[1] - or $A0[0],$S[3] # | t[2*i]>>63 - mul %rax # a[i]*a[i] - neg $carry # mov $carry,cf - adc %rax,$S[2] - adc %rdx,$S[3] - mov $S[2],-16($tptr) - mov $S[3],-8($tptr) -___ -} -############################################################## -# Montgomery reduction part, "word-by-word" algorithm. -# -{ -my ($topbit,$nptr)=("%rbp",$aptr); -my ($m0,$m1)=($a0,$a1); -my @Ni=("%rbx","%r9"); -$code.=<<___; - mov 40(%rsp),$nptr # restore $nptr - mov 48(%rsp),$n0 # restore *n0 - xor $j,$j - mov $num,0(%rsp) # save $num - sub $num,$j # $j=-$num - mov 64(%rsp),$A0[0] # t[0] # modsched # - mov $n0,$m0 # # modsched # - lea 64(%rsp,$num,2),%rax # end of t[] buffer - lea 64(%rsp,$num),$tptr # end of t[] window - mov %rax,8(%rsp) # save end of t[] buffer - lea ($nptr,$num),$nptr # end of n[] buffer - xor $topbit,$topbit # $topbit=0 - - mov 0($nptr,$j),%rax # n[0] # modsched # - mov 8($nptr,$j),$Ni[1] # n[1] # modsched # - imulq $A0[0],$m0 # m0=t[0]*n0 # modsched # - mov %rax,$Ni[0] # # modsched # - jmp .Lsqr4x_mont_outer + pxor %xmm0,%xmm0 + lea 48(%rsp),%rax + lea 64(%rsp,$num,2),%rdx + shr \$3+2,$num + mov 40(%rsp),%rsi # restore %rsp + jmp .Lsqr8x_zero + +.align 32 +.Lsqr8x_zero: + movdqa %xmm0,16*0(%rax) # wipe t + movdqa %xmm0,16*1(%rax) + movdqa %xmm0,16*2(%rax) + movdqa %xmm0,16*3(%rax) + lea 16*4(%rax),%rax + movdqa %xmm0,16*0(%rdx) # wipe n + movdqa %xmm0,16*1(%rdx) + movdqa %xmm0,16*2(%rdx) + movdqa %xmm0,16*3(%rdx) + lea 16*4(%rdx),%rdx + dec $num + jnz .Lsqr8x_zero -.align 16 -.Lsqr4x_mont_outer: - xor $A0[1],$A0[1] - mul $m0 # n[0]*m0 - add %rax,$A0[0] # n[0]*m0+t[0] - mov $Ni[1],%rax - adc %rdx,$A0[1] - mov $n0,$m1 + mov \$1,%rax + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lsqr8x_epilogue: + ret +.size bn_sqr8x_mont,.-bn_sqr8x_mont +___ +}}} + +if ($addx) {{{ +my $bp="%rdx"; # original value - xor $A0[0],$A0[0] - add 8($tptr,$j),$A0[1] - adc \$0,$A0[0] - mul $m0 # n[1]*m0 - add %rax,$A0[1] # n[1]*m0+t[1] - mov $Ni[0],%rax - adc %rdx,$A0[0] - - imulq $A0[1],$m1 - - mov 16($nptr,$j),$Ni[0] # n[2] - xor $A1[1],$A1[1] - add $A0[1],$A1[0] - adc \$0,$A1[1] - mul $m1 # n[0]*m1 - add %rax,$A1[0] # n[0]*m1+"t[1]" - mov $Ni[0],%rax - adc %rdx,$A1[1] - mov $A1[0],8($tptr,$j) # "t[1]" - - xor $A0[1],$A0[1] - add 16($tptr,$j),$A0[0] - adc \$0,$A0[1] - mul $m0 # n[2]*m0 - add %rax,$A0[0] # n[2]*m0+t[2] - mov $Ni[1],%rax - adc %rdx,$A0[1] - - mov 24($nptr,$j),$Ni[1] # n[3] - xor $A1[0],$A1[0] - add $A0[0],$A1[1] - adc \$0,$A1[0] - mul $m1 # n[1]*m1 - add %rax,$A1[1] # n[1]*m1+"t[2]" - mov $Ni[1],%rax - adc %rdx,$A1[0] - mov $A1[1],16($tptr,$j) # "t[2]" - - xor $A0[0],$A0[0] - add 24($tptr,$j),$A0[1] - lea 32($j),$j - adc \$0,$A0[0] - mul $m0 # n[3]*m0 - add %rax,$A0[1] # n[3]*m0+t[3] - mov $Ni[0],%rax - adc %rdx,$A0[0] - jmp .Lsqr4x_mont_inner +$code.=<<___; +.type bn_mulx4x_mont,\@function,6 +.align 32 +bn_mulx4x_mont: +.Lmulx4x_enter: + mov %rsp,%rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 -.align 16 -.Lsqr4x_mont_inner: - mov ($nptr,$j),$Ni[0] # n[4] - xor $A1[1],$A1[1] - add $A0[1],$A1[0] - adc \$0,$A1[1] - mul $m1 # n[2]*m1 - add %rax,$A1[0] # n[2]*m1+"t[3]" - mov $Ni[0],%rax - adc %rdx,$A1[1] - mov $A1[0],-8($tptr,$j) # "t[3]" - - xor $A0[1],$A0[1] - add ($tptr,$j),$A0[0] - adc \$0,$A0[1] - mul $m0 # n[4]*m0 - add %rax,$A0[0] # n[4]*m0+t[4] - mov $Ni[1],%rax - adc %rdx,$A0[1] - - mov 8($nptr,$j),$Ni[1] # n[5] - xor $A1[0],$A1[0] - add $A0[0],$A1[1] - adc \$0,$A1[0] - mul $m1 # n[3]*m1 - add %rax,$A1[1] # n[3]*m1+"t[4]" - mov $Ni[1],%rax - adc %rdx,$A1[0] - mov $A1[1],($tptr,$j) # "t[4]" - - xor $A0[0],$A0[0] - add 8($tptr,$j),$A0[1] - adc \$0,$A0[0] - mul $m0 # n[5]*m0 - add %rax,$A0[1] # n[5]*m0+t[5] - mov $Ni[0],%rax - adc %rdx,$A0[0] - - - mov 16($nptr,$j),$Ni[0] # n[6] - xor $A1[1],$A1[1] - add $A0[1],$A1[0] - adc \$0,$A1[1] - mul $m1 # n[4]*m1 - add %rax,$A1[0] # n[4]*m1+"t[5]" - mov $Ni[0],%rax - adc %rdx,$A1[1] - mov $A1[0],8($tptr,$j) # "t[5]" - - xor $A0[1],$A0[1] - add 16($tptr,$j),$A0[0] - adc \$0,$A0[1] - mul $m0 # n[6]*m0 - add %rax,$A0[0] # n[6]*m0+t[6] - mov $Ni[1],%rax - adc %rdx,$A0[1] - - mov 24($nptr,$j),$Ni[1] # n[7] - xor $A1[0],$A1[0] - add $A0[0],$A1[1] - adc \$0,$A1[0] - mul $m1 # n[5]*m1 - add %rax,$A1[1] # n[5]*m1+"t[6]" - mov $Ni[1],%rax - adc %rdx,$A1[0] - mov $A1[1],16($tptr,$j) # "t[6]" - - xor $A0[0],$A0[0] - add 24($tptr,$j),$A0[1] - lea 32($j),$j - adc \$0,$A0[0] - mul $m0 # n[7]*m0 - add %rax,$A0[1] # n[7]*m0+t[7] - mov $Ni[0],%rax - adc %rdx,$A0[0] - cmp \$0,$j - jne .Lsqr4x_mont_inner - - sub 0(%rsp),$j # $j=-$num # modsched # - mov $n0,$m0 # # modsched # - - xor $A1[1],$A1[1] - add $A0[1],$A1[0] - adc \$0,$A1[1] - mul $m1 # n[6]*m1 - add %rax,$A1[0] # n[6]*m1+"t[7]" - mov $Ni[1],%rax - adc %rdx,$A1[1] - mov $A1[0],-8($tptr) # "t[7]" - - xor $A0[1],$A0[1] - add ($tptr),$A0[0] # +t[8] - adc \$0,$A0[1] - mov 0($nptr,$j),$Ni[0] # n[0] # modsched # - add $topbit,$A0[0] - adc \$0,$A0[1] - - imulq 16($tptr,$j),$m0 # m0=t[0]*n0 # modsched # - xor $A1[0],$A1[0] - mov 8($nptr,$j),$Ni[1] # n[1] # modsched # - add $A0[0],$A1[1] - mov 16($tptr,$j),$A0[0] # t[0] # modsched # - adc \$0,$A1[0] - mul $m1 # n[7]*m1 - add %rax,$A1[1] # n[7]*m1+"t[8]" - mov $Ni[0],%rax # # modsched # - adc %rdx,$A1[0] - mov $A1[1],($tptr) # "t[8]" - - xor $topbit,$topbit - add 8($tptr),$A1[0] # +t[9] - adc $topbit,$topbit - add $A0[1],$A1[0] - lea 16($tptr),$tptr # "t[$num]>>128" - adc \$0,$topbit - mov $A1[0],-8($tptr) # "t[9]" - cmp 8(%rsp),$tptr # are we done? - jb .Lsqr4x_mont_outer - - mov 0(%rsp),$num # restore $num - mov $topbit,($tptr) # save $topbit + shl \$3,${num}d # convert $num to bytes + .byte 0x67 + xor %r10,%r10 + sub $num,%r10 # -$num + mov ($n0),$n0 # *n0 + lea -72(%rsp,%r10),%rsp # alloca(frame+$num+8) + lea ($bp,$num),%r10 + and \$-128,%rsp + ############################################################## + # Stack layout + # +0 num + # +8 off-loaded &b[i] + # +16 end of b[num] + # +24 saved n0 + # +32 saved rp + # +40 saved %rsp + # +48 inner counter + # +56 + # +64 tmp[num+1] + # + mov $num,0(%rsp) # save $num + shr \$5,$num + mov %r10,16(%rsp) # end of b[num] + sub \$1,$num + mov $n0, 24(%rsp) # save *n0 + mov $rp, 32(%rsp) # save $rp + mov %rax,40(%rsp) # save original %rsp + mov $num,48(%rsp) # inner counter + jmp .Lmulx4x_body + +.align 32 +.Lmulx4x_body: ___ -} -############################################################## -# Post-condition, 4x unrolled copy from bn_mul_mont -# -{ -my ($tptr,$nptr)=("%rbx",$aptr); -my @ri=("%rax","%rdx","%r10","%r11"); +my ($aptr, $bptr, $nptr, $tptr, $mi, $bi, $zero, $num)= + ("%rsi","%rdi","%rcx","%rbx","%r8","%r9","%rbp","%rax"); +my $rptr=$bptr; $code.=<<___; - mov 64(%rsp,$num),@ri[0] # tp[0] - lea 64(%rsp,$num),$tptr # upper half of t[2*$num] holds result - mov 40(%rsp),$nptr # restore $nptr - shr \$5,$num # num/4 - mov 8($tptr),@ri[1] # t[1] - xor $i,$i # i=0 and clear CF! - - mov 32(%rsp),$rptr # restore $rptr - sub 0($nptr),@ri[0] - mov 16($tptr),@ri[2] # t[2] - mov 24($tptr),@ri[3] # t[3] - sbb 8($nptr),@ri[1] - lea -1($num),$j # j=num/4-1 - jmp .Lsqr4x_sub -.align 16 -.Lsqr4x_sub: - mov @ri[0],0($rptr,$i,8) # rp[i]=tp[i]-np[i] - mov @ri[1],8($rptr,$i,8) # rp[i]=tp[i]-np[i] - sbb 16($nptr,$i,8),@ri[2] - mov 32($tptr,$i,8),@ri[0] # tp[i+1] - mov 40($tptr,$i,8),@ri[1] - sbb 24($nptr,$i,8),@ri[3] - mov @ri[2],16($rptr,$i,8) # rp[i]=tp[i]-np[i] - mov @ri[3],24($rptr,$i,8) # rp[i]=tp[i]-np[i] - sbb 32($nptr,$i,8),@ri[0] - mov 48($tptr,$i,8),@ri[2] - mov 56($tptr,$i,8),@ri[3] - sbb 40($nptr,$i,8),@ri[1] - lea 4($i),$i # i++ - dec $j # doesn't affect CF! - jnz .Lsqr4x_sub - - mov @ri[0],0($rptr,$i,8) # rp[i]=tp[i]-np[i] - mov 32($tptr,$i,8),@ri[0] # load overflow bit - sbb 16($nptr,$i,8),@ri[2] - mov @ri[1],8($rptr,$i,8) # rp[i]=tp[i]-np[i] - sbb 24($nptr,$i,8),@ri[3] - mov @ri[2],16($rptr,$i,8) # rp[i]=tp[i]-np[i] - - sbb \$0,@ri[0] # handle upmost overflow bit - mov @ri[3],24($rptr,$i,8) # rp[i]=tp[i]-np[i] - xor $i,$i # i=0 - and @ri[0],$tptr - not @ri[0] - mov $rptr,$nptr - and @ri[0],$nptr - lea -1($num),$j - or $nptr,$tptr # tp=borrow?tp:rp + lea 8($bp),$bptr + mov ($bp),%rdx # b[0], $bp==%rdx actually + lea 64+32(%rsp),$tptr + mov %rdx,$bi + + mulx 0*8($aptr),$mi,%rax # a[0]*b[0] + mulx 1*8($aptr),%r11,%r14 # a[1]*b[0] + add %rax,%r11 + mov $bptr,8(%rsp) # off-load &b[i] + mulx 2*8($aptr),%r12,%r13 # ... + adc %r14,%r12 + adc \$0,%r13 + + mov $mi,$bptr # borrow $bptr + imulq 24(%rsp),$mi # "t[0]"*n0 + xor $zero,$zero # cf=0, of=0 + + mulx 3*8($aptr),%rax,%r14 + mov $mi,%rdx + lea 4*8($aptr),$aptr + adcx %rax,%r13 + adcx $zero,%r14 # cf=0 + + mulx 0*8($nptr),%rax,%r10 + adcx %rax,$bptr # discarded + adox %r11,%r10 + mulx 1*8($nptr),%rax,%r11 + adcx %rax,%r10 + adox %r12,%r11 + .byte 0xc4,0x62,0xfb,0xf6,0xa1,0x10,0x00,0x00,0x00 # mulx 2*8($nptr),%rax,%r12 + mov 48(%rsp),$bptr # counter value + mov %r10,-4*8($tptr) + adcx %rax,%r11 + adox %r13,%r12 + mulx 3*8($nptr),%rax,%r15 + mov $bi,%rdx + mov %r11,-3*8($tptr) + adcx %rax,%r12 + adox $zero,%r15 # of=0 + lea 4*8($nptr),$nptr + mov %r12,-2*8($tptr) + + jmp .Lmulx4x_1st + +.align 32 +.Lmulx4x_1st: + adcx $zero,%r15 # cf=0, modulo-scheduled + mulx 0*8($aptr),%r10,%rax # a[4]*b[0] + adcx %r14,%r10 + mulx 1*8($aptr),%r11,%r14 # a[5]*b[0] + adcx %rax,%r11 + mulx 2*8($aptr),%r12,%rax # ... + adcx %r14,%r12 + mulx 3*8($aptr),%r13,%r14 + .byte 0x67,0x67 + mov $mi,%rdx + adcx %rax,%r13 + adcx $zero,%r14 # cf=0 + lea 4*8($aptr),$aptr + lea 4*8($tptr),$tptr + + adox %r15,%r10 + mulx 0*8($nptr),%rax,%r15 + adcx %rax,%r10 + adox %r15,%r11 + mulx 1*8($nptr),%rax,%r15 + adcx %rax,%r11 + adox %r15,%r12 + mulx 2*8($nptr),%rax,%r15 + mov %r10,-5*8($tptr) + adcx %rax,%r12 + mov %r11,-4*8($tptr) + adox %r15,%r13 + mulx 3*8($nptr),%rax,%r15 + mov $bi,%rdx + mov %r12,-3*8($tptr) + adcx %rax,%r13 + adox $zero,%r15 + lea 4*8($nptr),$nptr + mov %r13,-2*8($tptr) + + dec $bptr # of=0, pass cf + jnz .Lmulx4x_1st + + mov 0(%rsp),$num # load num + mov 8(%rsp),$bptr # re-load &b[i] + adc $zero,%r15 # modulo-scheduled + add %r15,%r14 + sbb %r15,%r15 # top-most carry + mov %r14,-1*8($tptr) + jmp .Lmulx4x_outer + +.align 32 +.Lmulx4x_outer: + mov ($bptr),%rdx # b[i] + lea 8($bptr),$bptr # b++ + sub $num,$aptr # rewind $aptr + mov %r15,($tptr) # save top-most carry + lea 64+4*8(%rsp),$tptr + sub $num,$nptr # rewind $nptr + + mulx 0*8($aptr),$mi,%r11 # a[0]*b[i] + xor %ebp,%ebp # xor $zero,$zero # cf=0, of=0 + mov %rdx,$bi + mulx 1*8($aptr),%r14,%r12 # a[1]*b[i] + adox -4*8($tptr),$mi + adcx %r14,%r11 + mulx 2*8($aptr),%r15,%r13 # ... + adox -3*8($tptr),%r11 + adcx %r15,%r12 + adox $zero,%r12 + adcx $zero,%r13 + + mov $bptr,8(%rsp) # off-load &b[i] + .byte 0x67 + mov $mi,%r15 + imulq 24(%rsp),$mi # "t[0]"*n0 + xor %ebp,%ebp # xor $zero,$zero # cf=0, of=0 + + mulx 3*8($aptr),%rax,%r14 + mov $mi,%rdx + adox -2*8($tptr),%r12 + adcx %rax,%r13 + adox -1*8($tptr),%r13 + adcx $zero,%r14 + lea 4*8($aptr),$aptr + adox $zero,%r14 + + mulx 0*8($nptr),%rax,%r10 + adcx %rax,%r15 # discarded + adox %r11,%r10 + mulx 1*8($nptr),%rax,%r11 + adcx %rax,%r10 + adox %r12,%r11 + mulx 2*8($nptr),%rax,%r12 + mov %r10,-4*8($tptr) + adcx %rax,%r11 + adox %r13,%r12 + mulx 3*8($nptr),%rax,%r15 + mov $bi,%rdx + mov %r11,-3*8($tptr) + lea 4*8($nptr),$nptr + adcx %rax,%r12 + adox $zero,%r15 # of=0 + mov 48(%rsp),$bptr # counter value + mov %r12,-2*8($tptr) + + jmp .Lmulx4x_inner + +.align 32 +.Lmulx4x_inner: + mulx 0*8($aptr),%r10,%rax # a[4]*b[i] + adcx $zero,%r15 # cf=0, modulo-scheduled + adox %r14,%r10 + mulx 1*8($aptr),%r11,%r14 # a[5]*b[i] + adcx 0*8($tptr),%r10 + adox %rax,%r11 + mulx 2*8($aptr),%r12,%rax # ... + adcx 1*8($tptr),%r11 + adox %r14,%r12 + mulx 3*8($aptr),%r13,%r14 + mov $mi,%rdx + adcx 2*8($tptr),%r12 + adox %rax,%r13 + adcx 3*8($tptr),%r13 + adox $zero,%r14 # of=0 + lea 4*8($aptr),$aptr + lea 4*8($tptr),$tptr + adcx $zero,%r14 # cf=0 + + adox %r15,%r10 + mulx 0*8($nptr),%rax,%r15 + adcx %rax,%r10 + adox %r15,%r11 + mulx 1*8($nptr),%rax,%r15 + adcx %rax,%r11 + adox %r15,%r12 + mulx 2*8($nptr),%rax,%r15 + mov %r10,-5*8($tptr) + adcx %rax,%r12 + adox %r15,%r13 + mulx 3*8($nptr),%rax,%r15 + mov $bi,%rdx + mov %r11,-4*8($tptr) + mov %r12,-3*8($tptr) + adcx %rax,%r13 + adox $zero,%r15 + lea 4*8($nptr),$nptr + mov %r13,-2*8($tptr) + + dec $bptr # of=0, pass cf + jnz .Lmulx4x_inner + + mov 0(%rsp),$num # load num + mov 8(%rsp),$bptr # re-load &b[i] + adc $zero,%r15 # modulo-scheduled + sub 0*8($tptr),$zero # pull top-most carry + adc %r15,%r14 + mov -8($nptr),$mi + sbb %r15,%r15 # top-most carry + mov %r14,-1*8($tptr) + + cmp 16(%rsp),$bptr + jne .Lmulx4x_outer + + sub %r14,$mi # compare top-most words + sbb $mi,$mi + or $mi,%r15 + + neg $num + xor %rdx,%rdx + mov 32(%rsp),$rptr # restore rp + lea 64(%rsp),$tptr pxor %xmm0,%xmm0 - lea 64(%rsp,$num,8),$nptr - movdqu ($tptr),%xmm1 - lea ($nptr,$num,8),$nptr - movdqa %xmm0,64(%rsp) # zap lower half of temporary vector - movdqa %xmm0,($nptr) # zap upper half of temporary vector - movdqu %xmm1,($rptr) - jmp .Lsqr4x_copy -.align 16 -.Lsqr4x_copy: # copy or in-place refresh - movdqu 16($tptr,$i),%xmm2 - movdqu 32($tptr,$i),%xmm1 - movdqa %xmm0,80(%rsp,$i) # zap lower half of temporary vector - movdqa %xmm0,96(%rsp,$i) # zap lower half of temporary vector - movdqa %xmm0,16($nptr,$i) # zap upper half of temporary vector - movdqa %xmm0,32($nptr,$i) # zap upper half of temporary vector - movdqu %xmm2,16($rptr,$i) - movdqu %xmm1,32($rptr,$i) - lea 32($i),$i - dec $j - jnz .Lsqr4x_copy - - movdqu 16($tptr,$i),%xmm2 - movdqa %xmm0,80(%rsp,$i) # zap lower half of temporary vector - movdqa %xmm0,16($nptr,$i) # zap upper half of temporary vector - movdqu %xmm2,16($rptr,$i) -___ -} -$code.=<<___; - mov 56(%rsp),%rsi # restore %rsp + mov 0*8($nptr,$num),%r8 + mov 1*8($nptr,$num),%r9 + neg %r8 + jmp .Lmulx4x_sub_entry + +.align 32 +.Lmulx4x_sub: + mov 0*8($nptr,$num),%r8 + mov 1*8($nptr,$num),%r9 + not %r8 +.Lmulx4x_sub_entry: + mov 2*8($nptr,$num),%r10 + not %r9 + and %r15,%r8 + mov 3*8($nptr,$num),%r11 + not %r10 + and %r15,%r9 + not %r11 + and %r15,%r10 + and %r15,%r11 + + neg %rdx # mov %rdx,%cf + adc 0*8($tptr),%r8 + adc 1*8($tptr),%r9 + movdqa %xmm0,($tptr) + adc 2*8($tptr),%r10 + adc 3*8($tptr),%r11 + movdqa %xmm0,16($tptr) + lea 4*8($tptr),$tptr + sbb %rdx,%rdx # mov %cf,%rdx + + mov %r8,0*8($rptr) + mov %r9,1*8($rptr) + mov %r10,2*8($rptr) + mov %r11,3*8($rptr) + lea 4*8($rptr),$rptr + + add \$32,$num + jnz .Lmulx4x_sub + + mov 40(%rsp),%rsi # restore %rsp mov \$1,%rax - mov 0(%rsi),%r15 - mov 8(%rsi),%r14 - mov 16(%rsi),%r13 - mov 24(%rsi),%r12 - mov 32(%rsi),%rbp - mov 40(%rsi),%rbx - lea 48(%rsi),%rsp -.Lsqr4x_epilogue: + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lmulx4x_epilogue: ret -.size bn_sqr4x_mont,.-bn_sqr4x_mont +.size bn_mulx4x_mont,.-bn_mulx4x_mont ___ }}} $code.=<<___; @@ -1581,18 +1285,22 @@ sqr_handler: mov 120($context),%rax # pull context->Rax mov 248($context),%rbx # pull context->Rip - lea .Lsqr4x_body(%rip),%r10 + mov 8($disp),%rsi # disp->ImageBase + mov 56($disp),%r11 # disp->HandlerData + + mov 0(%r11),%r10d # HandlerData[0] + lea (%rsi,%r10),%r10 # end of prologue label cmp %r10,%rbx # context->Rip<.Lsqr_body jb .Lcommon_seh_tail mov 152($context),%rax # pull context->Rsp - lea .Lsqr4x_epilogue(%rip),%r10 + mov 4(%r11),%r10d # HandlerData[1] + lea (%rsi,%r10),%r10 # epilogue label cmp %r10,%rbx # context->Rip>=.Lsqr_epilogue jae .Lcommon_seh_tail - mov 56(%rax),%rax # pull saved stack pointer - lea 48(%rax),%rax + mov 40(%rax),%rax # pull saved stack pointer mov -8(%rax),%rbx mov -16(%rax),%rbp @@ -1657,10 +1365,16 @@ sqr_handler: .rva .LSEH_end_bn_mul4x_mont .rva .LSEH_info_bn_mul4x_mont - .rva .LSEH_begin_bn_sqr4x_mont - .rva .LSEH_end_bn_sqr4x_mont - .rva .LSEH_info_bn_sqr4x_mont - + .rva .LSEH_begin_bn_sqr8x_mont + .rva .LSEH_end_bn_sqr8x_mont + .rva .LSEH_info_bn_sqr8x_mont +___ +$code.=<<___ if ($addx); + .rva .LSEH_begin_bn_mulx4x_mont + .rva .LSEH_end_bn_mulx4x_mont + .rva .LSEH_info_bn_mulx4x_mont +___ +$code.=<<___; .section .xdata .align 8 .LSEH_info_bn_mul_mont: @@ -1671,9 +1385,16 @@ sqr_handler: .byte 9,0,0,0 .rva mul_handler .rva .Lmul4x_body,.Lmul4x_epilogue # HandlerData[] -.LSEH_info_bn_sqr4x_mont: +.LSEH_info_bn_sqr8x_mont: + .byte 9,0,0,0 + .rva sqr_handler + .rva .Lsqr8x_body,.Lsqr8x_epilogue # HandlerData[] +___ +$code.=<<___ if ($addx); +.LSEH_info_bn_mulx4x_mont: .byte 9,0,0,0 .rva sqr_handler + .rva .Lmulx4x_body,.Lmulx4x_epilogue # HandlerData[] ___ } diff --git a/openssl/crypto/bn/asm/x86_64-mont5.pl b/openssl/crypto/bn/asm/x86_64-mont5.pl index dae0fe245..fa22c30b1 100644 --- a/openssl/crypto/bn/asm/x86_64-mont5.pl +++ b/openssl/crypto/bn/asm/x86_64-mont5.pl @@ -17,6 +17,13 @@ # is implemented, so that scatter-/gathering can be tuned without # bn_exp.c modifications. +# August 2013. +# +# Add MULX/AD*X code paths and additional interfaces to optimize for +# branch prediction unit. For input lengths that are multiples of 8 +# the np argument is not just modulus value, but one interleaved +# with 0. This is to optimize post-condition... + $flavour = shift; $output = shift; if ($flavour =~ /\./) { $output = $flavour; undef $flavour; } @@ -31,6 +38,21 @@ die "can't locate x86_64-xlate.pl"; open OUT,"| \"$^X\" $xlate $flavour $output"; *STDOUT=*OUT; +if (`$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1` + =~ /GNU assembler version ([2-9]\.[0-9]+)/) { + $addx = ($1>=2.23); +} + +if (!$addx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) && + `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) { + $addx = ($1>=2.10); +} + +if (!$addx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) && + `ml64 2>&1` =~ /Version ([0-9]+)\./) { + $addx = ($1>=12); +} + # int bn_mul_mont_gather5( $rp="%rdi"; # BN_ULONG *rp, $ap="%rsi"; # const BN_ULONG *ap, @@ -53,19 +75,25 @@ $m1="%rbp"; $code=<<___; .text +.extern OPENSSL_ia32cap_P + .globl bn_mul_mont_gather5 .type bn_mul_mont_gather5,\@function,6 .align 64 bn_mul_mont_gather5: - test \$3,${num}d + test \$7,${num}d jnz .Lmul_enter - cmp \$8,${num}d - jb .Lmul_enter +___ +$code.=<<___ if ($addx); + mov OPENSSL_ia32cap_P+8(%rip),%r11d +___ +$code.=<<___; jmp .Lmul4x_enter .align 16 .Lmul_enter: mov ${num}d,${num}d + mov %rsp,%rax mov `($win64?56:8)`(%rsp),%r10d # load 7th argument push %rbx push %rbp @@ -78,10 +106,8 @@ $code.=<<___ if ($win64); lea -0x28(%rsp),%rsp movaps %xmm6,(%rsp) movaps %xmm7,0x10(%rsp) -.Lmul_alloca: ___ $code.=<<___; - mov %rsp,%rax lea 2($num),%r11 neg %r11 lea (%rsp,%r11,8),%rsp # tp=alloca(8*(num+2)) @@ -287,7 +313,7 @@ $code.=<<___; lea 1($i),$i # i++ cmp $num,$i - jl .Louter + jb .Louter xor $i,$i # i=0 and clear CF! mov (%rsp),%rax # tp[0] @@ -323,18 +349,17 @@ $code.=<<___; mov \$1,%rax ___ $code.=<<___ if ($win64); - movaps (%rsi),%xmm6 - movaps 0x10(%rsi),%xmm7 - lea 0x28(%rsi),%rsi + movaps -88(%rsi),%xmm6 + movaps -72(%rsi),%xmm7 ___ $code.=<<___; - mov (%rsi),%r15 - mov 8(%rsi),%r14 - mov 16(%rsi),%r13 - mov 24(%rsi),%r12 - mov 32(%rsi),%rbp - mov 40(%rsi),%rbx - lea 48(%rsi),%rsp + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp .Lmul_epilogue: ret .size bn_mul_mont_gather5,.-bn_mul_mont_gather5 @@ -344,11 +369,18 @@ my @A=("%r10","%r11"); my @N=("%r13","%rdi"); $code.=<<___; .type bn_mul4x_mont_gather5,\@function,6 -.align 16 +.align 32 bn_mul4x_mont_gather5: .Lmul4x_enter: - mov ${num}d,${num}d - mov `($win64?56:8)`(%rsp),%r10d # load 7th argument +___ +$code.=<<___ if ($addx); + and \$0x80100,%r11d + cmp \$0x80100,%r11d + je .Lmulx4x_enter +___ +$code.=<<___; + .byte 0x67 + mov %rsp,%rax push %rbx push %rbp push %r12 @@ -360,23 +392,78 @@ $code.=<<___ if ($win64); lea -0x28(%rsp),%rsp movaps %xmm6,(%rsp) movaps %xmm7,0x10(%rsp) -.Lmul4x_alloca: ___ $code.=<<___; - mov %rsp,%rax - lea 4($num),%r11 - neg %r11 - lea (%rsp,%r11,8),%rsp # tp=alloca(8*(num+4)) - and \$-1024,%rsp # minimize TLB usage - - mov %rax,8(%rsp,$num,8) # tp[num+1]=%rsp + .byte 0x67 + mov ${num}d,%r10d + shl \$3,${num}d + shl \$3+2,%r10d # 4*$num + neg $num # -$num + + ############################################################## + # ensure that stack frame doesn't alias with $aptr+4*$num + # modulo 4096, which covers ret[num], am[num] and n[2*num] + # (see bn_exp.c). this is done to allow memory disambiguation + # logic do its magic. [excessive frame is allocated in order + # to allow bn_from_mont8x to clear it.] + # + lea -64(%rsp,$num,2),%r11 + sub $ap,%r11 + and \$4095,%r11 + cmp %r11,%r10 + jb .Lmul4xsp_alt + sub %r11,%rsp # align with $ap + lea -64(%rsp,$num,2),%rsp # alloca(128+num*8) + jmp .Lmul4xsp_done + +.align 32 +.Lmul4xsp_alt: + lea 4096-64(,$num,2),%r10 + lea -64(%rsp,$num,2),%rsp # alloca(128+num*8) + sub %r10,%r11 + mov \$0,%r10 + cmovc %r10,%r11 + sub %r11,%rsp +.Lmul4xsp_done: + and \$-64,%rsp + neg $num + + mov %rax,40(%rsp) .Lmul4x_body: - mov $rp,16(%rsp,$num,8) # tp[num+2]=$rp - mov %rdx,%r12 # reassign $bp + + call mul4x_internal + + mov 40(%rsp),%rsi # restore %rsp + mov \$1,%rax +___ +$code.=<<___ if ($win64); + movaps -88(%rsi),%xmm6 + movaps -72(%rsi),%xmm7 +___ +$code.=<<___; + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lmul4x_epilogue: + ret +.size bn_mul4x_mont_gather5,.-bn_mul4x_mont_gather5 + +.type mul4x_internal,\@abi-omnipotent +.align 32 +mul4x_internal: + shl \$5,$num + mov `($win64?56:8)`(%rax),%r10d # load 7th argument + lea 256(%rdx,$num),%r13 + shr \$5,$num # restore $num ___ $bp="%r12"; $STRIDE=2**5*8; # 5 is "window size" $N=$STRIDE/4; # should match cache line size + $tp=$i; $code.=<<___; mov %r10,%r11 shr \$`log($N/8)/log(2)`,%r10 @@ -384,459 +471,2771 @@ $code.=<<___; not %r10 lea .Lmagic_masks(%rip),%rax and \$`2**5/($N/8)-1`,%r10 # 5 is "window size" - lea 96($bp,%r11,8),$bp # pointer within 1st cache line + lea 96(%rdx,%r11,8),$bp # pointer within 1st cache line movq 0(%rax,%r10,8),%xmm4 # set of masks denoting which movq 8(%rax,%r10,8),%xmm5 # cache line contains element + add \$7,%r11 movq 16(%rax,%r10,8),%xmm6 # denoted by 7th argument movq 24(%rax,%r10,8),%xmm7 + and \$7,%r11 movq `0*$STRIDE/4-96`($bp),%xmm0 + lea $STRIDE($bp),$tp # borrow $tp movq `1*$STRIDE/4-96`($bp),%xmm1 pand %xmm4,%xmm0 movq `2*$STRIDE/4-96`($bp),%xmm2 pand %xmm5,%xmm1 movq `3*$STRIDE/4-96`($bp),%xmm3 pand %xmm6,%xmm2 + .byte 0x67 por %xmm1,%xmm0 + movq `0*$STRIDE/4-96`($tp),%xmm1 + .byte 0x67 pand %xmm7,%xmm3 + .byte 0x67 por %xmm2,%xmm0 - lea $STRIDE($bp),$bp + movq `1*$STRIDE/4-96`($tp),%xmm2 + .byte 0x67 + pand %xmm4,%xmm1 + .byte 0x67 por %xmm3,%xmm0 + movq `2*$STRIDE/4-96`($tp),%xmm3 movq %xmm0,$m0 # m0=bp[0] + movq `3*$STRIDE/4-96`($tp),%xmm0 + mov %r13,16+8(%rsp) # save end of b[num] + mov $rp, 56+8(%rsp) # save $rp + mov ($n0),$n0 # pull n0[0] value mov ($ap),%rax - - xor $i,$i # i=0 - xor $j,$j # j=0 - - movq `0*$STRIDE/4-96`($bp),%xmm0 - movq `1*$STRIDE/4-96`($bp),%xmm1 - pand %xmm4,%xmm0 - movq `2*$STRIDE/4-96`($bp),%xmm2 - pand %xmm5,%xmm1 + lea ($ap,$num),$ap # end of a[num] + neg $num mov $n0,$m1 mulq $m0 # ap[0]*bp[0] mov %rax,$A[0] mov ($np),%rax - movq `3*$STRIDE/4-96`($bp),%xmm3 - pand %xmm6,%xmm2 - por %xmm1,%xmm0 - pand %xmm7,%xmm3 + pand %xmm5,%xmm2 + pand %xmm6,%xmm3 + por %xmm2,%xmm1 imulq $A[0],$m1 # "tp[0]"*n0 + ############################################################## + # $tp is chosen so that writing to top-most element of the + # vector occurs just "above" references to powers table, + # "above" modulo cache-line size, which effectively precludes + # possibility of memory disambiguation logic failure when + # accessing the table. + # + lea 64+8(%rsp,%r11,8),$tp mov %rdx,$A[1] - por %xmm2,%xmm0 - lea $STRIDE($bp),$bp - por %xmm3,%xmm0 + pand %xmm7,%xmm0 + por %xmm3,%xmm1 + lea 2*$STRIDE($bp),$bp + por %xmm1,%xmm0 mulq $m1 # np[0]*m1 add %rax,$A[0] # discarded - mov 8($ap),%rax + mov 8($ap,$num),%rax adc \$0,%rdx mov %rdx,$N[1] mulq $m0 add %rax,$A[1] - mov 8($np),%rax + mov 16*1($np),%rax # interleaved with 0, therefore 16*n adc \$0,%rdx mov %rdx,$A[0] mulq $m1 add %rax,$N[1] - mov 16($ap),%rax + mov 16($ap,$num),%rax adc \$0,%rdx add $A[1],$N[1] - lea 4($j),$j # j++ + lea 4*8($num),$j # j=4 + lea 16*4($np),$np adc \$0,%rdx - mov $N[1],(%rsp) + mov $N[1],($tp) mov %rdx,$N[0] jmp .L1st4x -.align 16 + +.align 32 .L1st4x: mulq $m0 # ap[j]*bp[0] add %rax,$A[0] - mov -16($np,$j,8),%rax + mov -16*2($np),%rax + lea 32($tp),$tp adc \$0,%rdx mov %rdx,$A[1] mulq $m1 # np[j]*m1 add %rax,$N[0] - mov -8($ap,$j,8),%rax + mov -8($ap,$j),%rax adc \$0,%rdx add $A[0],$N[0] # np[j]*m1+ap[j]*bp[0] adc \$0,%rdx - mov $N[0],-24(%rsp,$j,8) # tp[j-1] + mov $N[0],-24($tp) # tp[j-1] mov %rdx,$N[1] mulq $m0 # ap[j]*bp[0] add %rax,$A[1] - mov -8($np,$j,8),%rax + mov -16*1($np),%rax adc \$0,%rdx mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov ($ap,$j,8),%rax + mov ($ap,$j),%rax adc \$0,%rdx add $A[1],$N[1] # np[j]*m1+ap[j]*bp[0] adc \$0,%rdx - mov $N[1],-16(%rsp,$j,8) # tp[j-1] + mov $N[1],-16($tp) # tp[j-1] mov %rdx,$N[0] mulq $m0 # ap[j]*bp[0] add %rax,$A[0] - mov ($np,$j,8),%rax + mov 16*0($np),%rax adc \$0,%rdx mov %rdx,$A[1] mulq $m1 # np[j]*m1 add %rax,$N[0] - mov 8($ap,$j,8),%rax + mov 8($ap,$j),%rax adc \$0,%rdx add $A[0],$N[0] # np[j]*m1+ap[j]*bp[0] adc \$0,%rdx - mov $N[0],-8(%rsp,$j,8) # tp[j-1] + mov $N[0],-8($tp) # tp[j-1] mov %rdx,$N[1] mulq $m0 # ap[j]*bp[0] add %rax,$A[1] - mov 8($np,$j,8),%rax + mov 16*1($np),%rax adc \$0,%rdx - lea 4($j),$j # j++ mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov -16($ap,$j,8),%rax + mov 16($ap,$j),%rax adc \$0,%rdx add $A[1],$N[1] # np[j]*m1+ap[j]*bp[0] + lea 16*4($np),$np adc \$0,%rdx - mov $N[1],-32(%rsp,$j,8) # tp[j-1] + mov $N[1],($tp) # tp[j-1] mov %rdx,$N[0] - cmp $num,$j - jl .L1st4x + + add \$32,$j # j+=4 + jnz .L1st4x mulq $m0 # ap[j]*bp[0] add %rax,$A[0] - mov -16($np,$j,8),%rax + mov -16*2($np),%rax + lea 32($tp),$tp adc \$0,%rdx mov %rdx,$A[1] mulq $m1 # np[j]*m1 add %rax,$N[0] - mov -8($ap,$j,8),%rax + mov -8($ap),%rax adc \$0,%rdx add $A[0],$N[0] # np[j]*m1+ap[j]*bp[0] adc \$0,%rdx - mov $N[0],-24(%rsp,$j,8) # tp[j-1] + mov $N[0],-24($tp) # tp[j-1] mov %rdx,$N[1] mulq $m0 # ap[j]*bp[0] add %rax,$A[1] - mov -8($np,$j,8),%rax + mov -16*1($np),%rax adc \$0,%rdx mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov ($ap),%rax # ap[0] + mov ($ap,$num),%rax # ap[0] adc \$0,%rdx add $A[1],$N[1] # np[j]*m1+ap[j]*bp[0] adc \$0,%rdx - mov $N[1],-16(%rsp,$j,8) # tp[j-1] + mov $N[1],-16($tp) # tp[j-1] mov %rdx,$N[0] movq %xmm0,$m0 # bp[1] + lea ($np,$num,2),$np # rewind $np xor $N[1],$N[1] add $A[0],$N[0] adc \$0,$N[1] - mov $N[0],-8(%rsp,$j,8) - mov $N[1],(%rsp,$j,8) # store upmost overflow bit + mov $N[0],-8($tp) - lea 1($i),$i # i++ -.align 4 -.Louter4x: - xor $j,$j # j=0 - movq `0*$STRIDE/4-96`($bp),%xmm0 - movq `1*$STRIDE/4-96`($bp),%xmm1 - pand %xmm4,%xmm0 - movq `2*$STRIDE/4-96`($bp),%xmm2 - pand %xmm5,%xmm1 + jmp .Louter4x - mov (%rsp),$A[0] +.align 32 +.Louter4x: + mov ($tp,$num),$A[0] mov $n0,$m1 mulq $m0 # ap[0]*bp[i] add %rax,$A[0] # ap[0]*bp[i]+tp[0] mov ($np),%rax adc \$0,%rdx + movq `0*$STRIDE/4-96`($bp),%xmm0 + movq `1*$STRIDE/4-96`($bp),%xmm1 + pand %xmm4,%xmm0 + movq `2*$STRIDE/4-96`($bp),%xmm2 + pand %xmm5,%xmm1 movq `3*$STRIDE/4-96`($bp),%xmm3 - pand %xmm6,%xmm2 - por %xmm1,%xmm0 - pand %xmm7,%xmm3 imulq $A[0],$m1 # tp[0]*n0 + .byte 0x67 mov %rdx,$A[1] + mov $N[1],($tp) # store upmost overflow bit + pand %xmm6,%xmm2 + por %xmm1,%xmm0 + pand %xmm7,%xmm3 por %xmm2,%xmm0 + lea ($tp,$num),$tp # rewind $tp lea $STRIDE($bp),$bp por %xmm3,%xmm0 mulq $m1 # np[0]*m1 add %rax,$A[0] # "$N[0]", discarded - mov 8($ap),%rax + mov 8($ap,$num),%rax adc \$0,%rdx mov %rdx,$N[1] mulq $m0 # ap[j]*bp[i] add %rax,$A[1] - mov 8($np),%rax + mov 16*1($np),%rax # interleaved with 0, therefore 16*n adc \$0,%rdx - add 8(%rsp),$A[1] # +tp[1] + add 8($tp),$A[1] # +tp[1] adc \$0,%rdx mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov 16($ap),%rax + mov 16($ap,$num),%rax adc \$0,%rdx add $A[1],$N[1] # np[j]*m1+ap[j]*bp[i]+tp[j] - lea 4($j),$j # j+=2 + lea 4*8($num),$j # j=4 + lea 16*4($np),$np adc \$0,%rdx mov %rdx,$N[0] jmp .Linner4x -.align 16 + +.align 32 .Linner4x: mulq $m0 # ap[j]*bp[i] add %rax,$A[0] - mov -16($np,$j,8),%rax + mov -16*2($np),%rax adc \$0,%rdx - add -16(%rsp,$j,8),$A[0] # ap[j]*bp[i]+tp[j] + add 16($tp),$A[0] # ap[j]*bp[i]+tp[j] + lea 32($tp),$tp adc \$0,%rdx mov %rdx,$A[1] mulq $m1 # np[j]*m1 add %rax,$N[0] - mov -8($ap,$j,8),%rax + mov -8($ap,$j),%rax adc \$0,%rdx add $A[0],$N[0] adc \$0,%rdx - mov $N[1],-32(%rsp,$j,8) # tp[j-1] + mov $N[1],-32($tp) # tp[j-1] mov %rdx,$N[1] mulq $m0 # ap[j]*bp[i] add %rax,$A[1] - mov -8($np,$j,8),%rax + mov -16*1($np),%rax adc \$0,%rdx - add -8(%rsp,$j,8),$A[1] + add -8($tp),$A[1] adc \$0,%rdx mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov ($ap,$j,8),%rax + mov ($ap,$j),%rax adc \$0,%rdx add $A[1],$N[1] adc \$0,%rdx - mov $N[0],-24(%rsp,$j,8) # tp[j-1] + mov $N[0],-24($tp) # tp[j-1] mov %rdx,$N[0] mulq $m0 # ap[j]*bp[i] add %rax,$A[0] - mov ($np,$j,8),%rax + mov 16*0($np),%rax adc \$0,%rdx - add (%rsp,$j,8),$A[0] # ap[j]*bp[i]+tp[j] + add ($tp),$A[0] # ap[j]*bp[i]+tp[j] adc \$0,%rdx mov %rdx,$A[1] mulq $m1 # np[j]*m1 add %rax,$N[0] - mov 8($ap,$j,8),%rax + mov 8($ap,$j),%rax adc \$0,%rdx add $A[0],$N[0] adc \$0,%rdx - mov $N[1],-16(%rsp,$j,8) # tp[j-1] + mov $N[1],-16($tp) # tp[j-1] mov %rdx,$N[1] mulq $m0 # ap[j]*bp[i] add %rax,$A[1] - mov 8($np,$j,8),%rax + mov 16*1($np),%rax adc \$0,%rdx - add 8(%rsp,$j,8),$A[1] + add 8($tp),$A[1] adc \$0,%rdx - lea 4($j),$j # j++ mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov -16($ap,$j,8),%rax + mov 16($ap,$j),%rax adc \$0,%rdx add $A[1],$N[1] + lea 16*4($np),$np adc \$0,%rdx - mov $N[0],-40(%rsp,$j,8) # tp[j-1] + mov $N[0],-8($tp) # tp[j-1] mov %rdx,$N[0] - cmp $num,$j - jl .Linner4x + + add \$32,$j # j+=4 + jnz .Linner4x mulq $m0 # ap[j]*bp[i] add %rax,$A[0] - mov -16($np,$j,8),%rax + mov -16*2($np),%rax adc \$0,%rdx - add -16(%rsp,$j,8),$A[0] # ap[j]*bp[i]+tp[j] + add 16($tp),$A[0] # ap[j]*bp[i]+tp[j] + lea 32($tp),$tp adc \$0,%rdx mov %rdx,$A[1] mulq $m1 # np[j]*m1 add %rax,$N[0] - mov -8($ap,$j,8),%rax + mov -8($ap),%rax adc \$0,%rdx add $A[0],$N[0] adc \$0,%rdx - mov $N[1],-32(%rsp,$j,8) # tp[j-1] + mov $N[1],-32($tp) # tp[j-1] mov %rdx,$N[1] mulq $m0 # ap[j]*bp[i] add %rax,$A[1] - mov -8($np,$j,8),%rax + mov $m1,%rax + mov -16*1($np),$m1 adc \$0,%rdx - add -8(%rsp,$j,8),$A[1] + add -8($tp),$A[1] adc \$0,%rdx - lea 1($i),$i # i++ mov %rdx,$A[0] mulq $m1 # np[j]*m1 add %rax,$N[1] - mov ($ap),%rax # ap[0] + mov ($ap,$num),%rax # ap[0] adc \$0,%rdx add $A[1],$N[1] adc \$0,%rdx - mov $N[0],-24(%rsp,$j,8) # tp[j-1] + mov $N[0],-24($tp) # tp[j-1] mov %rdx,$N[0] movq %xmm0,$m0 # bp[i+1] - mov $N[1],-16(%rsp,$j,8) # tp[j-1] + mov $N[1],-16($tp) # tp[j-1] + lea ($np,$num,2),$np # rewind $np xor $N[1],$N[1] add $A[0],$N[0] adc \$0,$N[1] - add (%rsp,$num,8),$N[0] # pull upmost overflow bit - adc \$0,$N[1] - mov $N[0],-8(%rsp,$j,8) - mov $N[1],(%rsp,$j,8) # store upmost overflow bit + add ($tp),$N[0] # pull upmost overflow bit + adc \$0,$N[1] # upmost overflow bit + mov $N[0],-8($tp) - cmp $num,$i - jl .Louter4x + cmp 16+8(%rsp),$bp + jb .Louter4x ___ -{ -my @ri=("%rax","%rdx",$m0,$m1); +if (1) { $code.=<<___; - mov 16(%rsp,$num,8),$rp # restore $rp - mov 0(%rsp),@ri[0] # tp[0] - pxor %xmm0,%xmm0 - mov 8(%rsp),@ri[1] # tp[1] - shr \$2,$num # num/=4 - lea (%rsp),$ap # borrow ap for tp - xor $i,$i # i=0 and clear CF! - - sub 0($np),@ri[0] - mov 16($ap),@ri[2] # tp[2] - mov 24($ap),@ri[3] # tp[3] - sbb 8($np),@ri[1] - lea -1($num),$j # j=num/4-1 + sub $N[0],$m1 # compare top-most words + adc $j,$j # $j is zero + or $j,$N[1] + xor \$1,$N[1] + lea ($tp,$num),%rbx # tptr in .sqr4x_sub + lea ($np,$N[1],8),%rbp # nptr in .sqr4x_sub + mov %r9,%rcx + sar \$3+2,%rcx # cf=0 + mov 56+8(%rsp),%rdi # rptr in .sqr4x_sub + jmp .Lsqr4x_sub +___ +} else { +my @ri=("%rax",$bp,$m0,$m1); +my $rp="%rdx"; +$code.=<<___ + xor \$1,$N[1] + lea ($tp,$num),$tp # rewind $tp + sar \$5,$num # cf=0 + lea ($np,$N[1],8),$np + mov 56+8(%rsp),$rp # restore $rp jmp .Lsub4x -.align 16 + +.align 32 .Lsub4x: - mov @ri[0],0($rp,$i,8) # rp[i]=tp[i]-np[i] - mov @ri[1],8($rp,$i,8) # rp[i]=tp[i]-np[i] - sbb 16($np,$i,8),@ri[2] - mov 32($ap,$i,8),@ri[0] # tp[i+1] - mov 40($ap,$i,8),@ri[1] - sbb 24($np,$i,8),@ri[3] - mov @ri[2],16($rp,$i,8) # rp[i]=tp[i]-np[i] - mov @ri[3],24($rp,$i,8) # rp[i]=tp[i]-np[i] - sbb 32($np,$i,8),@ri[0] - mov 48($ap,$i,8),@ri[2] - mov 56($ap,$i,8),@ri[3] - sbb 40($np,$i,8),@ri[1] - lea 4($i),$i # i++ - dec $j # doesnn't affect CF! + .byte 0x66 + mov 8*0($tp),@ri[0] + mov 8*1($tp),@ri[1] + .byte 0x66 + sbb 16*0($np),@ri[0] + mov 8*2($tp),@ri[2] + sbb 16*1($np),@ri[1] + mov 3*8($tp),@ri[3] + lea 4*8($tp),$tp + sbb 16*2($np),@ri[2] + mov @ri[0],8*0($rp) + sbb 16*3($np),@ri[3] + lea 16*4($np),$np + mov @ri[1],8*1($rp) + mov @ri[2],8*2($rp) + mov @ri[3],8*3($rp) + lea 8*4($rp),$rp + + inc $num jnz .Lsub4x - mov @ri[0],0($rp,$i,8) # rp[i]=tp[i]-np[i] - mov 32($ap,$i,8),@ri[0] # load overflow bit - sbb 16($np,$i,8),@ri[2] - mov @ri[1],8($rp,$i,8) # rp[i]=tp[i]-np[i] - sbb 24($np,$i,8),@ri[3] - mov @ri[2],16($rp,$i,8) # rp[i]=tp[i]-np[i] + ret +___ +} +$code.=<<___; +.size mul4x_internal,.-mul4x_internal +___ +}}} + {{{ +###################################################################### +# void bn_power5( +my $rptr="%rdi"; # BN_ULONG *rptr, +my $aptr="%rsi"; # const BN_ULONG *aptr, +my $bptr="%rdx"; # const void *table, +my $nptr="%rcx"; # const BN_ULONG *nptr, +my $n0 ="%r8"; # const BN_ULONG *n0); +my $num ="%r9"; # int num, has to be divisible by 8 + # int pwr + +my ($i,$j,$tptr)=("%rbp","%rcx",$rptr); +my @A0=("%r10","%r11"); +my @A1=("%r12","%r13"); +my ($a0,$a1,$ai)=("%r14","%r15","%rbx"); - sbb \$0,@ri[0] # handle upmost overflow bit - mov @ri[3],24($rp,$i,8) # rp[i]=tp[i]-np[i] - xor $i,$i # i=0 - and @ri[0],$ap - not @ri[0] - mov $rp,$np - and @ri[0],$np - lea -1($num),$j - or $np,$ap # ap=borrow?tp:rp +$code.=<<___; +.globl bn_power5 +.type bn_power5,\@function,6 +.align 32 +bn_power5: +___ +$code.=<<___ if ($addx); + mov OPENSSL_ia32cap_P+8(%rip),%r11d + and \$0x80100,%r11d + cmp \$0x80100,%r11d + je .Lpowerx5_enter +___ +$code.=<<___; + mov %rsp,%rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 +___ +$code.=<<___ if ($win64); + lea -0x28(%rsp),%rsp + movaps %xmm6,(%rsp) + movaps %xmm7,0x10(%rsp) +___ +$code.=<<___; + mov ${num}d,%r10d + shl \$3,${num}d # convert $num to bytes + shl \$3+2,%r10d # 4*$num + neg $num + mov ($n0),$n0 # *n0 + + ############################################################## + # ensure that stack frame doesn't alias with $aptr+4*$num + # modulo 4096, which covers ret[num], am[num] and n[2*num] + # (see bn_exp.c). this is done to allow memory disambiguation + # logic do its magic. + # + lea -64(%rsp,$num,2),%r11 + sub $aptr,%r11 + and \$4095,%r11 + cmp %r11,%r10 + jb .Lpwr_sp_alt + sub %r11,%rsp # align with $aptr + lea -64(%rsp,$num,2),%rsp # alloca(frame+2*$num) + jmp .Lpwr_sp_done + +.align 32 +.Lpwr_sp_alt: + lea 4096-64(,$num,2),%r10 # 4096-frame-2*$num + lea -64(%rsp,$num,2),%rsp # alloca(frame+2*$num) + sub %r10,%r11 + mov \$0,%r10 + cmovc %r10,%r11 + sub %r11,%rsp +.Lpwr_sp_done: + and \$-64,%rsp + mov $num,%r10 + neg $num + + ############################################################## + # Stack layout + # + # +0 saved $num, used in reduction section + # +8 &t[2*$num], used in reduction section + # +32 saved *n0 + # +40 saved %rsp + # +48 t[2*$num] + # + mov $n0, 32(%rsp) + mov %rax, 40(%rsp) # save original %rsp +.Lpower5_body: + movq $rptr,%xmm1 # save $rptr + movq $nptr,%xmm2 # save $nptr + movq %r10, %xmm3 # -$num + movq $bptr,%xmm4 + + call __bn_sqr8x_internal + call __bn_sqr8x_internal + call __bn_sqr8x_internal + call __bn_sqr8x_internal + call __bn_sqr8x_internal + + movq %xmm2,$nptr + movq %xmm4,$bptr + mov $aptr,$rptr + mov 40(%rsp),%rax + lea 32(%rsp),$n0 + + call mul4x_internal + + mov 40(%rsp),%rsi # restore %rsp + mov \$1,%rax + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lpower5_epilogue: + ret +.size bn_power5,.-bn_power5 + +.globl bn_sqr8x_internal +.hidden bn_sqr8x_internal +.type bn_sqr8x_internal,\@abi-omnipotent +.align 32 +bn_sqr8x_internal: +__bn_sqr8x_internal: + ############################################################## + # Squaring part: + # + # a) multiply-n-add everything but a[i]*a[i]; + # b) shift result of a) by 1 to the left and accumulate + # a[i]*a[i] products; + # + ############################################################## + # a[1]a[0] + # a[2]a[0] + # a[3]a[0] + # a[2]a[1] + # a[4]a[0] + # a[3]a[1] + # a[5]a[0] + # a[4]a[1] + # a[3]a[2] + # a[6]a[0] + # a[5]a[1] + # a[4]a[2] + # a[7]a[0] + # a[6]a[1] + # a[5]a[2] + # a[4]a[3] + # a[7]a[1] + # a[6]a[2] + # a[5]a[3] + # a[7]a[2] + # a[6]a[3] + # a[5]a[4] + # a[7]a[3] + # a[6]a[4] + # a[7]a[4] + # a[6]a[5] + # a[7]a[5] + # a[7]a[6] + # a[1]a[0] + # a[2]a[0] + # a[3]a[0] + # a[4]a[0] + # a[5]a[0] + # a[6]a[0] + # a[7]a[0] + # a[2]a[1] + # a[3]a[1] + # a[4]a[1] + # a[5]a[1] + # a[6]a[1] + # a[7]a[1] + # a[3]a[2] + # a[4]a[2] + # a[5]a[2] + # a[6]a[2] + # a[7]a[2] + # a[4]a[3] + # a[5]a[3] + # a[6]a[3] + # a[7]a[3] + # a[5]a[4] + # a[6]a[4] + # a[7]a[4] + # a[6]a[5] + # a[7]a[5] + # a[7]a[6] + # a[0]a[0] + # a[1]a[1] + # a[2]a[2] + # a[3]a[3] + # a[4]a[4] + # a[5]a[5] + # a[6]a[6] + # a[7]a[7] + + lea 32(%r10),$i # $i=-($num-32) + lea ($aptr,$num),$aptr # end of a[] buffer, ($aptr,$i)=&ap[2] + + mov $num,$j # $j=$num + + # comments apply to $num==8 case + mov -32($aptr,$i),$a0 # a[0] + lea 48+8(%rsp,$num,2),$tptr # end of tp[] buffer, &tp[2*$num] + mov -24($aptr,$i),%rax # a[1] + lea -32($tptr,$i),$tptr # end of tp[] window, &tp[2*$num-"$i"] + mov -16($aptr,$i),$ai # a[2] + mov %rax,$a1 + + mul $a0 # a[1]*a[0] + mov %rax,$A0[0] # a[1]*a[0] + mov $ai,%rax # a[2] + mov %rdx,$A0[1] + mov $A0[0],-24($tptr,$i) # t[1] + + mul $a0 # a[2]*a[0] + add %rax,$A0[1] + mov $ai,%rax + adc \$0,%rdx + mov $A0[1],-16($tptr,$i) # t[2] + mov %rdx,$A0[0] + + + mov -8($aptr,$i),$ai # a[3] + mul $a1 # a[2]*a[1] + mov %rax,$A1[0] # a[2]*a[1]+t[3] + mov $ai,%rax + mov %rdx,$A1[1] + + lea ($i),$j + mul $a0 # a[3]*a[0] + add %rax,$A0[0] # a[3]*a[0]+a[2]*a[1]+t[3] + mov $ai,%rax + mov %rdx,$A0[1] + adc \$0,$A0[1] + add $A1[0],$A0[0] + adc \$0,$A0[1] + mov $A0[0],-8($tptr,$j) # t[3] + jmp .Lsqr4x_1st + +.align 32 +.Lsqr4x_1st: + mov ($aptr,$j),$ai # a[4] + mul $a1 # a[3]*a[1] + add %rax,$A1[1] # a[3]*a[1]+t[4] + mov $ai,%rax + mov %rdx,$A1[0] + adc \$0,$A1[0] + + mul $a0 # a[4]*a[0] + add %rax,$A0[1] # a[4]*a[0]+a[3]*a[1]+t[4] + mov $ai,%rax # a[3] + mov 8($aptr,$j),$ai # a[5] + mov %rdx,$A0[0] + adc \$0,$A0[0] + add $A1[1],$A0[1] + adc \$0,$A0[0] + + + mul $a1 # a[4]*a[3] + add %rax,$A1[0] # a[4]*a[3]+t[5] + mov $ai,%rax + mov $A0[1],($tptr,$j) # t[4] + mov %rdx,$A1[1] + adc \$0,$A1[1] + + mul $a0 # a[5]*a[2] + add %rax,$A0[0] # a[5]*a[2]+a[4]*a[3]+t[5] + mov $ai,%rax + mov 16($aptr,$j),$ai # a[6] + mov %rdx,$A0[1] + adc \$0,$A0[1] + add $A1[0],$A0[0] + adc \$0,$A0[1] + + mul $a1 # a[5]*a[3] + add %rax,$A1[1] # a[5]*a[3]+t[6] + mov $ai,%rax + mov $A0[0],8($tptr,$j) # t[5] + mov %rdx,$A1[0] + adc \$0,$A1[0] + + mul $a0 # a[6]*a[2] + add %rax,$A0[1] # a[6]*a[2]+a[5]*a[3]+t[6] + mov $ai,%rax # a[3] + mov 24($aptr,$j),$ai # a[7] + mov %rdx,$A0[0] + adc \$0,$A0[0] + add $A1[1],$A0[1] + adc \$0,$A0[0] + + + mul $a1 # a[6]*a[5] + add %rax,$A1[0] # a[6]*a[5]+t[7] + mov $ai,%rax + mov $A0[1],16($tptr,$j) # t[6] + mov %rdx,$A1[1] + adc \$0,$A1[1] + lea 32($j),$j + + mul $a0 # a[7]*a[4] + add %rax,$A0[0] # a[7]*a[4]+a[6]*a[5]+t[6] + mov $ai,%rax + mov %rdx,$A0[1] + adc \$0,$A0[1] + add $A1[0],$A0[0] + adc \$0,$A0[1] + mov $A0[0],-8($tptr,$j) # t[7] + + cmp \$0,$j + jne .Lsqr4x_1st + + mul $a1 # a[7]*a[5] + add %rax,$A1[1] + lea 16($i),$i + adc \$0,%rdx + add $A0[1],$A1[1] + adc \$0,%rdx + + mov $A1[1],($tptr) # t[8] + mov %rdx,$A1[0] + mov %rdx,8($tptr) # t[9] + jmp .Lsqr4x_outer + +.align 32 +.Lsqr4x_outer: # comments apply to $num==6 case + mov -32($aptr,$i),$a0 # a[0] + lea 48+8(%rsp,$num,2),$tptr # end of tp[] buffer, &tp[2*$num] + mov -24($aptr,$i),%rax # a[1] + lea -32($tptr,$i),$tptr # end of tp[] window, &tp[2*$num-"$i"] + mov -16($aptr,$i),$ai # a[2] + mov %rax,$a1 + + mul $a0 # a[1]*a[0] + mov -24($tptr,$i),$A0[0] # t[1] + add %rax,$A0[0] # a[1]*a[0]+t[1] + mov $ai,%rax # a[2] + adc \$0,%rdx + mov $A0[0],-24($tptr,$i) # t[1] + mov %rdx,$A0[1] + + mul $a0 # a[2]*a[0] + add %rax,$A0[1] + mov $ai,%rax + adc \$0,%rdx + add -16($tptr,$i),$A0[1] # a[2]*a[0]+t[2] + mov %rdx,$A0[0] + adc \$0,$A0[0] + mov $A0[1],-16($tptr,$i) # t[2] + + xor $A1[0],$A1[0] + + mov -8($aptr,$i),$ai # a[3] + mul $a1 # a[2]*a[1] + add %rax,$A1[0] # a[2]*a[1]+t[3] + mov $ai,%rax + adc \$0,%rdx + add -8($tptr,$i),$A1[0] + mov %rdx,$A1[1] + adc \$0,$A1[1] + + mul $a0 # a[3]*a[0] + add %rax,$A0[0] # a[3]*a[0]+a[2]*a[1]+t[3] + mov $ai,%rax + adc \$0,%rdx + add $A1[0],$A0[0] + mov %rdx,$A0[1] + adc \$0,$A0[1] + mov $A0[0],-8($tptr,$i) # t[3] + + lea ($i),$j + jmp .Lsqr4x_inner + +.align 32 +.Lsqr4x_inner: + mov ($aptr,$j),$ai # a[4] + mul $a1 # a[3]*a[1] + add %rax,$A1[1] # a[3]*a[1]+t[4] + mov $ai,%rax + mov %rdx,$A1[0] + adc \$0,$A1[0] + add ($tptr,$j),$A1[1] + adc \$0,$A1[0] + + .byte 0x67 + mul $a0 # a[4]*a[0] + add %rax,$A0[1] # a[4]*a[0]+a[3]*a[1]+t[4] + mov $ai,%rax # a[3] + mov 8($aptr,$j),$ai # a[5] + mov %rdx,$A0[0] + adc \$0,$A0[0] + add $A1[1],$A0[1] + adc \$0,$A0[0] + + mul $a1 # a[4]*a[3] + add %rax,$A1[0] # a[4]*a[3]+t[5] + mov $A0[1],($tptr,$j) # t[4] + mov $ai,%rax + mov %rdx,$A1[1] + adc \$0,$A1[1] + add 8($tptr,$j),$A1[0] + lea 16($j),$j # j++ + adc \$0,$A1[1] + + mul $a0 # a[5]*a[2] + add %rax,$A0[0] # a[5]*a[2]+a[4]*a[3]+t[5] + mov $ai,%rax + adc \$0,%rdx + add $A1[0],$A0[0] + mov %rdx,$A0[1] + adc \$0,$A0[1] + mov $A0[0],-8($tptr,$j) # t[5], "preloaded t[1]" below + + cmp \$0,$j + jne .Lsqr4x_inner + + .byte 0x67 + mul $a1 # a[5]*a[3] + add %rax,$A1[1] + adc \$0,%rdx + add $A0[1],$A1[1] + adc \$0,%rdx + + mov $A1[1],($tptr) # t[6], "preloaded t[2]" below + mov %rdx,$A1[0] + mov %rdx,8($tptr) # t[7], "preloaded t[3]" below + + add \$16,$i + jnz .Lsqr4x_outer + + # comments apply to $num==4 case + mov -32($aptr),$a0 # a[0] + lea 48+8(%rsp,$num,2),$tptr # end of tp[] buffer, &tp[2*$num] + mov -24($aptr),%rax # a[1] + lea -32($tptr,$i),$tptr # end of tp[] window, &tp[2*$num-"$i"] + mov -16($aptr),$ai # a[2] + mov %rax,$a1 + + mul $a0 # a[1]*a[0] + add %rax,$A0[0] # a[1]*a[0]+t[1], preloaded t[1] + mov $ai,%rax # a[2] + mov %rdx,$A0[1] + adc \$0,$A0[1] + + mul $a0 # a[2]*a[0] + add %rax,$A0[1] + mov $ai,%rax + mov $A0[0],-24($tptr) # t[1] + mov %rdx,$A0[0] + adc \$0,$A0[0] + add $A1[1],$A0[1] # a[2]*a[0]+t[2], preloaded t[2] + mov -8($aptr),$ai # a[3] + adc \$0,$A0[0] + + mul $a1 # a[2]*a[1] + add %rax,$A1[0] # a[2]*a[1]+t[3], preloaded t[3] + mov $ai,%rax + mov $A0[1],-16($tptr) # t[2] + mov %rdx,$A1[1] + adc \$0,$A1[1] + + mul $a0 # a[3]*a[0] + add %rax,$A0[0] # a[3]*a[0]+a[2]*a[1]+t[3] + mov $ai,%rax + mov %rdx,$A0[1] + adc \$0,$A0[1] + add $A1[0],$A0[0] + adc \$0,$A0[1] + mov $A0[0],-8($tptr) # t[3] + + mul $a1 # a[3]*a[1] + add %rax,$A1[1] + mov -16($aptr),%rax # a[2] + adc \$0,%rdx + add $A0[1],$A1[1] + adc \$0,%rdx + + mov $A1[1],($tptr) # t[4] + mov %rdx,$A1[0] + mov %rdx,8($tptr) # t[5] + + mul $ai # a[2]*a[3] +___ +{ +my ($shift,$carry)=($a0,$a1); +my @S=(@A1,$ai,$n0); +$code.=<<___; + add \$16,$i + xor $shift,$shift + sub $num,$i # $i=16-$num + xor $carry,$carry + + add $A1[0],%rax # t[5] + adc \$0,%rdx + mov %rax,8($tptr) # t[5] + mov %rdx,16($tptr) # t[6] + mov $carry,24($tptr) # t[7] + + mov -16($aptr,$i),%rax # a[0] + lea 48+8(%rsp),$tptr + xor $A0[0],$A0[0] # t[0] + mov 8($tptr),$A0[1] # t[1] + + lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[1] # | t[2*i]>>63 + mov 16($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov 24($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[0] + mov -8($aptr,$i),%rax # a[i+1] # prefetch + mov $S[0],($tptr) + adc %rdx,$S[1] + + lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1 | shift + mov $S[1],8($tptr) + sbb $carry,$carry # mov cf,$carry + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[3] # | t[2*i]>>63 + mov 32($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov 40($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[2] + mov 0($aptr,$i),%rax # a[i+1] # prefetch + mov $S[2],16($tptr) + adc %rdx,$S[3] + lea 16($i),$i + mov $S[3],24($tptr) + sbb $carry,$carry # mov cf,$carry + lea 64($tptr),$tptr + jmp .Lsqr4x_shift_n_add + +.align 32 +.Lsqr4x_shift_n_add: + lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[1] # | t[2*i]>>63 + mov -16($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov -8($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[0] + mov -8($aptr,$i),%rax # a[i+1] # prefetch + mov $S[0],-32($tptr) + adc %rdx,$S[1] + + lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1 | shift + mov $S[1],-24($tptr) + sbb $carry,$carry # mov cf,$carry + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[3] # | t[2*i]>>63 + mov 0($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov 8($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[2] + mov 0($aptr,$i),%rax # a[i+1] # prefetch + mov $S[2],-16($tptr) + adc %rdx,$S[3] + + lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift + mov $S[3],-8($tptr) + sbb $carry,$carry # mov cf,$carry + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[1] # | t[2*i]>>63 + mov 16($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov 24($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[0] + mov 8($aptr,$i),%rax # a[i+1] # prefetch + mov $S[0],0($tptr) + adc %rdx,$S[1] + + lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1 | shift + mov $S[1],8($tptr) + sbb $carry,$carry # mov cf,$carry + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[3] # | t[2*i]>>63 + mov 32($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov 40($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[2] + mov 16($aptr,$i),%rax # a[i+1] # prefetch + mov $S[2],16($tptr) + adc %rdx,$S[3] + mov $S[3],24($tptr) + sbb $carry,$carry # mov cf,$carry + lea 64($tptr),$tptr + add \$32,$i + jnz .Lsqr4x_shift_n_add + + lea ($shift,$A0[0],2),$S[0] # t[2*i]<<1 | shift + .byte 0x67 + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[1] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[1] # | t[2*i]>>63 + mov -16($tptr),$A0[0] # t[2*i+2] # prefetch + mov $A0[1],$shift # shift=t[2*i+1]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + mov -8($tptr),$A0[1] # t[2*i+2+1] # prefetch + adc %rax,$S[0] + mov -8($aptr),%rax # a[i+1] # prefetch + mov $S[0],-32($tptr) + adc %rdx,$S[1] + + lea ($shift,$A0[0],2),$S[2] # t[2*i]<<1|shift + mov $S[1],-24($tptr) + sbb $carry,$carry # mov cf,$carry + shr \$63,$A0[0] + lea ($j,$A0[1],2),$S[3] # t[2*i+1]<<1 | + shr \$63,$A0[1] + or $A0[0],$S[3] # | t[2*i]>>63 + mul %rax # a[i]*a[i] + neg $carry # mov $carry,cf + adc %rax,$S[2] + adc %rdx,$S[3] + mov $S[2],-16($tptr) + mov $S[3],-8($tptr) +___ +} +###################################################################### +# Montgomery reduction part, "word-by-word" algorithm. +# +# This new path is inspired by multiple submissions from Intel, by +# Shay Gueron, Vlad Krasnov, Erdinc Ozturk, James Guilford, +# Vinodh Gopal... +{ +my ($nptr,$tptr,$carry,$m0)=("%rbp","%rdi","%rsi","%rbx"); - movdqu ($ap),%xmm1 - movdqa %xmm0,(%rsp) - movdqu %xmm1,($rp) - jmp .Lcopy4x -.align 16 -.Lcopy4x: # copy or in-place refresh - movdqu 16($ap,$i),%xmm2 - movdqu 32($ap,$i),%xmm1 - movdqa %xmm0,16(%rsp,$i) - movdqu %xmm2,16($rp,$i) - movdqa %xmm0,32(%rsp,$i) - movdqu %xmm1,32($rp,$i) - lea 32($i),$i - dec $j - jnz .Lcopy4x - - shl \$2,$num - movdqu 16($ap,$i),%xmm2 - movdqa %xmm0,16(%rsp,$i) - movdqu %xmm2,16($rp,$i) +$code.=<<___; + movq %xmm2,$nptr +sqr8x_reduction: + xor %rax,%rax + lea ($nptr,$num,2),%rcx # end of n[] + lea 48+8(%rsp,$num,2),%rdx # end of t[] buffer + mov %rcx,0+8(%rsp) + lea 48+8(%rsp,$num),$tptr # end of initial t[] window + mov %rdx,8+8(%rsp) + neg $num + jmp .L8x_reduction_loop + +.align 32 +.L8x_reduction_loop: + lea ($tptr,$num),$tptr # start of current t[] window + .byte 0x66 + mov 8*0($tptr),$m0 + mov 8*1($tptr),%r9 + mov 8*2($tptr),%r10 + mov 8*3($tptr),%r11 + mov 8*4($tptr),%r12 + mov 8*5($tptr),%r13 + mov 8*6($tptr),%r14 + mov 8*7($tptr),%r15 + mov %rax,(%rdx) # store top-most carry bit + lea 8*8($tptr),$tptr + + .byte 0x67 + mov $m0,%r8 + imulq 32+8(%rsp),$m0 # n0*a[0] + mov 16*0($nptr),%rax # n[0] + mov \$8,%ecx + jmp .L8x_reduce + +.align 32 +.L8x_reduce: + mulq $m0 + mov 16*1($nptr),%rax # n[1] + neg %r8 + mov %rdx,%r8 + adc \$0,%r8 + + mulq $m0 + add %rax,%r9 + mov 16*2($nptr),%rax + adc \$0,%rdx + add %r9,%r8 + mov $m0,48-8+8(%rsp,%rcx,8) # put aside n0*a[i] + mov %rdx,%r9 + adc \$0,%r9 + + mulq $m0 + add %rax,%r10 + mov 16*3($nptr),%rax + adc \$0,%rdx + add %r10,%r9 + mov 32+8(%rsp),$carry # pull n0, borrow $carry + mov %rdx,%r10 + adc \$0,%r10 + + mulq $m0 + add %rax,%r11 + mov 16*4($nptr),%rax + adc \$0,%rdx + imulq %r8,$carry # modulo-scheduled + add %r11,%r10 + mov %rdx,%r11 + adc \$0,%r11 + + mulq $m0 + add %rax,%r12 + mov 16*5($nptr),%rax + adc \$0,%rdx + add %r12,%r11 + mov %rdx,%r12 + adc \$0,%r12 + + mulq $m0 + add %rax,%r13 + mov 16*6($nptr),%rax + adc \$0,%rdx + add %r13,%r12 + mov %rdx,%r13 + adc \$0,%r13 + + mulq $m0 + add %rax,%r14 + mov 16*7($nptr),%rax + adc \$0,%rdx + add %r14,%r13 + mov %rdx,%r14 + adc \$0,%r14 + + mulq $m0 + mov $carry,$m0 # n0*a[i] + add %rax,%r15 + mov 16*0($nptr),%rax # n[0] + adc \$0,%rdx + add %r15,%r14 + mov %rdx,%r15 + adc \$0,%r15 + + dec %ecx + jnz .L8x_reduce + + lea 16*8($nptr),$nptr + xor %rax,%rax + mov 8+8(%rsp),%rdx # pull end of t[] + cmp 0+8(%rsp),$nptr # end of n[]? + jae .L8x_no_tail + + .byte 0x66 + add 8*0($tptr),%r8 + adc 8*1($tptr),%r9 + adc 8*2($tptr),%r10 + adc 8*3($tptr),%r11 + adc 8*4($tptr),%r12 + adc 8*5($tptr),%r13 + adc 8*6($tptr),%r14 + adc 8*7($tptr),%r15 + sbb $carry,$carry # top carry + + mov 48+56+8(%rsp),$m0 # pull n0*a[0] + mov \$8,%ecx + mov 16*0($nptr),%rax + jmp .L8x_tail + +.align 32 +.L8x_tail: + mulq $m0 + add %rax,%r8 + mov 16*1($nptr),%rax + mov %r8,($tptr) # save result + mov %rdx,%r8 + adc \$0,%r8 + + mulq $m0 + add %rax,%r9 + mov 16*2($nptr),%rax + adc \$0,%rdx + add %r9,%r8 + lea 8($tptr),$tptr # $tptr++ + mov %rdx,%r9 + adc \$0,%r9 + + mulq $m0 + add %rax,%r10 + mov 16*3($nptr),%rax + adc \$0,%rdx + add %r10,%r9 + mov %rdx,%r10 + adc \$0,%r10 + + mulq $m0 + add %rax,%r11 + mov 16*4($nptr),%rax + adc \$0,%rdx + add %r11,%r10 + mov %rdx,%r11 + adc \$0,%r11 + + mulq $m0 + add %rax,%r12 + mov 16*5($nptr),%rax + adc \$0,%rdx + add %r12,%r11 + mov %rdx,%r12 + adc \$0,%r12 + + mulq $m0 + add %rax,%r13 + mov 16*6($nptr),%rax + adc \$0,%rdx + add %r13,%r12 + mov %rdx,%r13 + adc \$0,%r13 + + mulq $m0 + add %rax,%r14 + mov 16*7($nptr),%rax + adc \$0,%rdx + add %r14,%r13 + mov %rdx,%r14 + adc \$0,%r14 + + mulq $m0 + mov 48-16+8(%rsp,%rcx,8),$m0# pull n0*a[i] + add %rax,%r15 + adc \$0,%rdx + add %r15,%r14 + mov 16*0($nptr),%rax # pull n[0] + mov %rdx,%r15 + adc \$0,%r15 + + dec %ecx + jnz .L8x_tail + + lea 16*8($nptr),$nptr + mov 8+8(%rsp),%rdx # pull end of t[] + cmp 0+8(%rsp),$nptr # end of n[]? + jae .L8x_tail_done # break out of loop + + mov 48+56+8(%rsp),$m0 # pull n0*a[0] + neg $carry + mov 8*0($nptr),%rax # pull n[0] + adc 8*0($tptr),%r8 + adc 8*1($tptr),%r9 + adc 8*2($tptr),%r10 + adc 8*3($tptr),%r11 + adc 8*4($tptr),%r12 + adc 8*5($tptr),%r13 + adc 8*6($tptr),%r14 + adc 8*7($tptr),%r15 + sbb $carry,$carry # top carry + + mov \$8,%ecx + jmp .L8x_tail + +.align 32 +.L8x_tail_done: + add (%rdx),%r8 # can this overflow? + xor %rax,%rax + + neg $carry +.L8x_no_tail: + adc 8*0($tptr),%r8 + adc 8*1($tptr),%r9 + adc 8*2($tptr),%r10 + adc 8*3($tptr),%r11 + adc 8*4($tptr),%r12 + adc 8*5($tptr),%r13 + adc 8*6($tptr),%r14 + adc 8*7($tptr),%r15 + adc \$0,%rax # top-most carry + mov -16($nptr),%rcx # np[num-1] + xor $carry,$carry + + movq %xmm2,$nptr # restore $nptr + + mov %r8,8*0($tptr) # store top 512 bits + mov %r9,8*1($tptr) + movq %xmm3,$num # $num is %r9, can't be moved upwards + mov %r10,8*2($tptr) + mov %r11,8*3($tptr) + mov %r12,8*4($tptr) + mov %r13,8*5($tptr) + mov %r14,8*6($tptr) + mov %r15,8*7($tptr) + lea 8*8($tptr),$tptr + + cmp %rdx,$tptr # end of t[]? + jb .L8x_reduction_loop +___ +} +############################################################## +# Post-condition, 4x unrolled +# +{ +my ($tptr,$nptr)=("%rbx","%rbp"); +$code.=<<___; + #xor %rsi,%rsi # %rsi was $carry above + sub %r15,%rcx # compare top-most words + lea (%rdi,$num),$tptr # %rdi was $tptr above + adc %rsi,%rsi + mov $num,%rcx + or %rsi,%rax + movq %xmm1,$rptr # restore $rptr + xor \$1,%rax + movq %xmm1,$aptr # prepare for back-to-back call + lea ($nptr,%rax,8),$nptr + sar \$3+2,%rcx # cf=0 + jmp .Lsqr4x_sub + +.align 32 +.Lsqr4x_sub: + .byte 0x66 + mov 8*0($tptr),%r12 + mov 8*1($tptr),%r13 + sbb 16*0($nptr),%r12 + mov 8*2($tptr),%r14 + sbb 16*1($nptr),%r13 + mov 8*3($tptr),%r15 + lea 8*4($tptr),$tptr + sbb 16*2($nptr),%r14 + mov %r12,8*0($rptr) + sbb 16*3($nptr),%r15 + lea 16*4($nptr),$nptr + mov %r13,8*1($rptr) + mov %r14,8*2($rptr) + mov %r15,8*3($rptr) + lea 8*4($rptr),$rptr + + inc %rcx # pass %cf + jnz .Lsqr4x_sub ___ } $code.=<<___; - mov 8(%rsp,$num,8),%rsi # restore %rsp - mov \$1,%rax + mov $num,%r10 # prepare for back-to-back call + neg $num # restore $num + ret +.size bn_sqr8x_internal,.-bn_sqr8x_internal +___ +{ +$code.=<<___; +.globl bn_from_montgomery +.type bn_from_montgomery,\@abi-omnipotent +.align 32 +bn_from_montgomery: + testl \$7,`($win64?"48(%rsp)":"%r9d")` + jz bn_from_mont8x + xor %eax,%eax + ret +.size bn_from_montgomery,.-bn_from_montgomery + +.type bn_from_mont8x,\@function,6 +.align 32 +bn_from_mont8x: + .byte 0x67 + mov %rsp,%rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 ___ $code.=<<___ if ($win64); - movaps (%rsi),%xmm6 - movaps 0x10(%rsi),%xmm7 - lea 0x28(%rsi),%rsi + lea -0x28(%rsp),%rsp + movaps %xmm6,(%rsp) + movaps %xmm7,0x10(%rsp) ___ $code.=<<___; - mov (%rsi),%r15 - mov 8(%rsi),%r14 - mov 16(%rsi),%r13 - mov 24(%rsi),%r12 - mov 32(%rsi),%rbp - mov 40(%rsi),%rbx - lea 48(%rsi),%rsp -.Lmul4x_epilogue: + .byte 0x67 + mov ${num}d,%r10d + shl \$3,${num}d # convert $num to bytes + shl \$3+2,%r10d # 4*$num + neg $num + mov ($n0),$n0 # *n0 + + ############################################################## + # ensure that stack frame doesn't alias with $aptr+4*$num + # modulo 4096, which covers ret[num], am[num] and n[2*num] + # (see bn_exp.c). this is done to allow memory disambiguation + # logic do its magic. + # + lea -64(%rsp,$num,2),%r11 + sub $aptr,%r11 + and \$4095,%r11 + cmp %r11,%r10 + jb .Lfrom_sp_alt + sub %r11,%rsp # align with $aptr + lea -64(%rsp,$num,2),%rsp # alloca(frame+2*$num) + jmp .Lfrom_sp_done + +.align 32 +.Lfrom_sp_alt: + lea 4096-64(,$num,2),%r10 # 4096-frame-2*$num + lea -64(%rsp,$num,2),%rsp # alloca(frame+2*$num) + sub %r10,%r11 + mov \$0,%r10 + cmovc %r10,%r11 + sub %r11,%rsp +.Lfrom_sp_done: + and \$-64,%rsp + mov $num,%r10 + neg $num + + ############################################################## + # Stack layout + # + # +0 saved $num, used in reduction section + # +8 &t[2*$num], used in reduction section + # +32 saved *n0 + # +40 saved %rsp + # +48 t[2*$num] + # + mov $n0, 32(%rsp) + mov %rax, 40(%rsp) # save original %rsp +.Lfrom_body: + mov $num,%r11 + lea 48(%rsp),%rax + pxor %xmm0,%xmm0 + jmp .Lmul_by_1 + +.align 32 +.Lmul_by_1: + movdqu ($aptr),%xmm1 + movdqu 16($aptr),%xmm2 + movdqu 32($aptr),%xmm3 + movdqa %xmm0,(%rax,$num) + movdqu 48($aptr),%xmm4 + movdqa %xmm0,16(%rax,$num) + .byte 0x48,0x8d,0xb6,0x40,0x00,0x00,0x00 # lea 64($aptr),$aptr + movdqa %xmm1,(%rax) + movdqa %xmm0,32(%rax,$num) + movdqa %xmm2,16(%rax) + movdqa %xmm0,48(%rax,$num) + movdqa %xmm3,32(%rax) + movdqa %xmm4,48(%rax) + lea 64(%rax),%rax + sub \$64,%r11 + jnz .Lmul_by_1 + + movq $rptr,%xmm1 + movq $nptr,%xmm2 + .byte 0x67 + mov $nptr,%rbp + movq %r10, %xmm3 # -num +___ +$code.=<<___ if ($addx); + mov OPENSSL_ia32cap_P+8(%rip),%r11d + and \$0x80100,%r11d + cmp \$0x80100,%r11d + jne .Lfrom_mont_nox + + lea (%rax,$num),$rptr + call sqrx8x_reduction + + pxor %xmm0,%xmm0 + lea 48(%rsp),%rax + mov 40(%rsp),%rsi # restore %rsp + jmp .Lfrom_mont_zero + +.align 32 +.Lfrom_mont_nox: +___ +$code.=<<___; + call sqr8x_reduction + + pxor %xmm0,%xmm0 + lea 48(%rsp),%rax + mov 40(%rsp),%rsi # restore %rsp + jmp .Lfrom_mont_zero + +.align 32 +.Lfrom_mont_zero: + movdqa %xmm0,16*0(%rax) + movdqa %xmm0,16*1(%rax) + movdqa %xmm0,16*2(%rax) + movdqa %xmm0,16*3(%rax) + lea 16*4(%rax),%rax + sub \$32,$num + jnz .Lfrom_mont_zero + + mov \$1,%rax + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lfrom_epilogue: ret -.size bn_mul4x_mont_gather5,.-bn_mul4x_mont_gather5 +.size bn_from_mont8x,.-bn_from_mont8x ___ +} }}} + +if ($addx) {{{ +my $bp="%rdx"; # restore original value + +$code.=<<___; +.type bn_mulx4x_mont_gather5,\@function,6 +.align 32 +bn_mulx4x_mont_gather5: +.Lmulx4x_enter: + .byte 0x67 + mov %rsp,%rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 +___ +$code.=<<___ if ($win64); + lea -0x28(%rsp),%rsp + movaps %xmm6,(%rsp) + movaps %xmm7,0x10(%rsp) +___ +$code.=<<___; + .byte 0x67 + mov ${num}d,%r10d + shl \$3,${num}d # convert $num to bytes + shl \$3+2,%r10d # 4*$num + neg $num # -$num + mov ($n0),$n0 # *n0 + + ############################################################## + # ensure that stack frame doesn't alias with $aptr+4*$num + # modulo 4096, which covers a[num], ret[num] and n[2*num] + # (see bn_exp.c). this is done to allow memory disambiguation + # logic do its magic. [excessive frame is allocated in order + # to allow bn_from_mont8x to clear it.] + # + lea -64(%rsp,$num,2),%r11 + sub $ap,%r11 + and \$4095,%r11 + cmp %r11,%r10 + jb .Lmulx4xsp_alt + sub %r11,%rsp # align with $aptr + lea -64(%rsp,$num,2),%rsp # alloca(frame+$num) + jmp .Lmulx4xsp_done + +.align 32 +.Lmulx4xsp_alt: + lea 4096-64(,$num,2),%r10 # 4096-frame-$num + lea -64(%rsp,$num,2),%rsp # alloca(frame+$num) + sub %r10,%r11 + mov \$0,%r10 + cmovc %r10,%r11 + sub %r11,%rsp +.Lmulx4xsp_done: + and \$-64,%rsp # ensure alignment + ############################################################## + # Stack layout + # +0 -num + # +8 off-loaded &b[i] + # +16 end of b[num] + # +24 inner counter + # +32 saved n0 + # +40 saved %rsp + # +48 + # +56 saved rp + # +64 tmp[num+1] + # + mov $n0, 32(%rsp) # save *n0 + mov %rax,40(%rsp) # save original %rsp +.Lmulx4x_body: + call mulx4x_internal + + mov 40(%rsp),%rsi # restore %rsp + mov \$1,%rax +___ +$code.=<<___ if ($win64); + movaps -88(%rsi),%xmm6 + movaps -72(%rsi),%xmm7 +___ +$code.=<<___; + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lmulx4x_epilogue: + ret +.size bn_mulx4x_mont_gather5,.-bn_mulx4x_mont_gather5 + +.type mulx4x_internal,\@abi-omnipotent +.align 32 +mulx4x_internal: + .byte 0x4c,0x89,0x8c,0x24,0x08,0x00,0x00,0x00 # mov $num,8(%rsp) # save -$num + .byte 0x67 + neg $num # restore $num + shl \$5,$num + lea 256($bp,$num),%r13 + shr \$5+5,$num + mov `($win64?56:8)`(%rax),%r10d # load 7th argument + sub \$1,$num + mov %r13,16+8(%rsp) # end of b[num] + mov $num,24+8(%rsp) # inner counter + mov $rp, 56+8(%rsp) # save $rp +___ +my ($aptr, $bptr, $nptr, $tptr, $mi, $bi, $zero, $num)= + ("%rsi","%rdi","%rcx","%rbx","%r8","%r9","%rbp","%rax"); +my $rptr=$bptr; +my $STRIDE=2**5*8; # 5 is "window size" +my $N=$STRIDE/4; # should match cache line size +$code.=<<___; + mov %r10,%r11 + shr \$`log($N/8)/log(2)`,%r10 + and \$`$N/8-1`,%r11 + not %r10 + lea .Lmagic_masks(%rip),%rax + and \$`2**5/($N/8)-1`,%r10 # 5 is "window size" + lea 96($bp,%r11,8),$bptr # pointer within 1st cache line + movq 0(%rax,%r10,8),%xmm4 # set of masks denoting which + movq 8(%rax,%r10,8),%xmm5 # cache line contains element + add \$7,%r11 + movq 16(%rax,%r10,8),%xmm6 # denoted by 7th argument + movq 24(%rax,%r10,8),%xmm7 + and \$7,%r11 + + movq `0*$STRIDE/4-96`($bptr),%xmm0 + lea $STRIDE($bptr),$tptr # borrow $tptr + movq `1*$STRIDE/4-96`($bptr),%xmm1 + pand %xmm4,%xmm0 + movq `2*$STRIDE/4-96`($bptr),%xmm2 + pand %xmm5,%xmm1 + movq `3*$STRIDE/4-96`($bptr),%xmm3 + pand %xmm6,%xmm2 + por %xmm1,%xmm0 + movq `0*$STRIDE/4-96`($tptr),%xmm1 + pand %xmm7,%xmm3 + por %xmm2,%xmm0 + movq `1*$STRIDE/4-96`($tptr),%xmm2 + por %xmm3,%xmm0 + .byte 0x67,0x67 + pand %xmm4,%xmm1 + movq `2*$STRIDE/4-96`($tptr),%xmm3 + + movq %xmm0,%rdx # bp[0] + movq `3*$STRIDE/4-96`($tptr),%xmm0 + lea 2*$STRIDE($bptr),$bptr # next &b[i] + pand %xmm5,%xmm2 + .byte 0x67,0x67 + pand %xmm6,%xmm3 + ############################################################## + # $tptr is chosen so that writing to top-most element of the + # vector occurs just "above" references to powers table, + # "above" modulo cache-line size, which effectively precludes + # possibility of memory disambiguation logic failure when + # accessing the table. + # + lea 64+8*4+8(%rsp,%r11,8),$tptr + + mov %rdx,$bi + mulx 0*8($aptr),$mi,%rax # a[0]*b[0] + mulx 1*8($aptr),%r11,%r12 # a[1]*b[0] + add %rax,%r11 + mulx 2*8($aptr),%rax,%r13 # ... + adc %rax,%r12 + adc \$0,%r13 + mulx 3*8($aptr),%rax,%r14 + + mov $mi,%r15 + imulq 32+8(%rsp),$mi # "t[0]"*n0 + xor $zero,$zero # cf=0, of=0 + mov $mi,%rdx + + por %xmm2,%xmm1 + pand %xmm7,%xmm0 + por %xmm3,%xmm1 + mov $bptr,8+8(%rsp) # off-load &b[i] + por %xmm1,%xmm0 + + .byte 0x48,0x8d,0xb6,0x20,0x00,0x00,0x00 # lea 4*8($aptr),$aptr + adcx %rax,%r13 + adcx $zero,%r14 # cf=0 + + mulx 0*16($nptr),%rax,%r10 + adcx %rax,%r15 # discarded + adox %r11,%r10 + mulx 1*16($nptr),%rax,%r11 + adcx %rax,%r10 + adox %r12,%r11 + mulx 2*16($nptr),%rax,%r12 + mov 24+8(%rsp),$bptr # counter value + .byte 0x66 + mov %r10,-8*4($tptr) + adcx %rax,%r11 + adox %r13,%r12 + mulx 3*16($nptr),%rax,%r15 + .byte 0x67,0x67 + mov $bi,%rdx + mov %r11,-8*3($tptr) + adcx %rax,%r12 + adox $zero,%r15 # of=0 + .byte 0x48,0x8d,0x89,0x40,0x00,0x00,0x00 # lea 4*16($nptr),$nptr + mov %r12,-8*2($tptr) + #jmp .Lmulx4x_1st + +.align 32 +.Lmulx4x_1st: + adcx $zero,%r15 # cf=0, modulo-scheduled + mulx 0*8($aptr),%r10,%rax # a[4]*b[0] + adcx %r14,%r10 + mulx 1*8($aptr),%r11,%r14 # a[5]*b[0] + adcx %rax,%r11 + mulx 2*8($aptr),%r12,%rax # ... + adcx %r14,%r12 + mulx 3*8($aptr),%r13,%r14 + .byte 0x67,0x67 + mov $mi,%rdx + adcx %rax,%r13 + adcx $zero,%r14 # cf=0 + lea 4*8($aptr),$aptr + lea 4*8($tptr),$tptr + + adox %r15,%r10 + mulx 0*16($nptr),%rax,%r15 + adcx %rax,%r10 + adox %r15,%r11 + mulx 1*16($nptr),%rax,%r15 + adcx %rax,%r11 + adox %r15,%r12 + mulx 2*16($nptr),%rax,%r15 + mov %r10,-5*8($tptr) + adcx %rax,%r12 + mov %r11,-4*8($tptr) + adox %r15,%r13 + mulx 3*16($nptr),%rax,%r15 + mov $bi,%rdx + mov %r12,-3*8($tptr) + adcx %rax,%r13 + adox $zero,%r15 + lea 4*16($nptr),$nptr + mov %r13,-2*8($tptr) + + dec $bptr # of=0, pass cf + jnz .Lmulx4x_1st + + mov 8(%rsp),$num # load -num + movq %xmm0,%rdx # bp[1] + adc $zero,%r15 # modulo-scheduled + lea ($aptr,$num),$aptr # rewind $aptr + add %r15,%r14 + mov 8+8(%rsp),$bptr # re-load &b[i] + adc $zero,$zero # top-most carry + mov %r14,-1*8($tptr) + jmp .Lmulx4x_outer + +.align 32 +.Lmulx4x_outer: + mov $zero,($tptr) # save top-most carry + lea 4*8($tptr,$num),$tptr # rewind $tptr + mulx 0*8($aptr),$mi,%r11 # a[0]*b[i] + xor $zero,$zero # cf=0, of=0 + mov %rdx,$bi + mulx 1*8($aptr),%r14,%r12 # a[1]*b[i] + adox -4*8($tptr),$mi # +t[0] + adcx %r14,%r11 + mulx 2*8($aptr),%r15,%r13 # ... + adox -3*8($tptr),%r11 + adcx %r15,%r12 + mulx 3*8($aptr),%rdx,%r14 + adox -2*8($tptr),%r12 + adcx %rdx,%r13 + lea ($nptr,$num,2),$nptr # rewind $nptr + lea 4*8($aptr),$aptr + adox -1*8($tptr),%r13 + adcx $zero,%r14 + adox $zero,%r14 + + .byte 0x67 + mov $mi,%r15 + imulq 32+8(%rsp),$mi # "t[0]"*n0 + + movq `0*$STRIDE/4-96`($bptr),%xmm0 + .byte 0x67,0x67 + mov $mi,%rdx + movq `1*$STRIDE/4-96`($bptr),%xmm1 + .byte 0x67 + pand %xmm4,%xmm0 + movq `2*$STRIDE/4-96`($bptr),%xmm2 + .byte 0x67 + pand %xmm5,%xmm1 + movq `3*$STRIDE/4-96`($bptr),%xmm3 + add \$$STRIDE,$bptr # next &b[i] + .byte 0x67 + pand %xmm6,%xmm2 + por %xmm1,%xmm0 + pand %xmm7,%xmm3 + xor $zero,$zero # cf=0, of=0 + mov $bptr,8+8(%rsp) # off-load &b[i] + + mulx 0*16($nptr),%rax,%r10 + adcx %rax,%r15 # discarded + adox %r11,%r10 + mulx 1*16($nptr),%rax,%r11 + adcx %rax,%r10 + adox %r12,%r11 + mulx 2*16($nptr),%rax,%r12 + adcx %rax,%r11 + adox %r13,%r12 + mulx 3*16($nptr),%rax,%r15 + mov $bi,%rdx + por %xmm2,%xmm0 + mov 24+8(%rsp),$bptr # counter value + mov %r10,-8*4($tptr) + por %xmm3,%xmm0 + adcx %rax,%r12 + mov %r11,-8*3($tptr) + adox $zero,%r15 # of=0 + mov %r12,-8*2($tptr) + lea 4*16($nptr),$nptr + jmp .Lmulx4x_inner + +.align 32 +.Lmulx4x_inner: + mulx 0*8($aptr),%r10,%rax # a[4]*b[i] + adcx $zero,%r15 # cf=0, modulo-scheduled + adox %r14,%r10 + mulx 1*8($aptr),%r11,%r14 # a[5]*b[i] + adcx 0*8($tptr),%r10 + adox %rax,%r11 + mulx 2*8($aptr),%r12,%rax # ... + adcx 1*8($tptr),%r11 + adox %r14,%r12 + mulx 3*8($aptr),%r13,%r14 + mov $mi,%rdx + adcx 2*8($tptr),%r12 + adox %rax,%r13 + adcx 3*8($tptr),%r13 + adox $zero,%r14 # of=0 + lea 4*8($aptr),$aptr + lea 4*8($tptr),$tptr + adcx $zero,%r14 # cf=0 + + adox %r15,%r10 + mulx 0*16($nptr),%rax,%r15 + adcx %rax,%r10 + adox %r15,%r11 + mulx 1*16($nptr),%rax,%r15 + adcx %rax,%r11 + adox %r15,%r12 + mulx 2*16($nptr),%rax,%r15 + mov %r10,-5*8($tptr) + adcx %rax,%r12 + adox %r15,%r13 + mov %r11,-4*8($tptr) + mulx 3*16($nptr),%rax,%r15 + mov $bi,%rdx + lea 4*16($nptr),$nptr + mov %r12,-3*8($tptr) + adcx %rax,%r13 + adox $zero,%r15 + mov %r13,-2*8($tptr) + + dec $bptr # of=0, pass cf + jnz .Lmulx4x_inner + + mov 0+8(%rsp),$num # load -num + movq %xmm0,%rdx # bp[i+1] + adc $zero,%r15 # modulo-scheduled + sub 0*8($tptr),$bptr # pull top-most carry to %cf + mov 8+8(%rsp),$bptr # re-load &b[i] + mov 16+8(%rsp),%r10 + adc %r15,%r14 + lea ($aptr,$num),$aptr # rewind $aptr + adc $zero,$zero # top-most carry + mov %r14,-1*8($tptr) + + cmp %r10,$bptr + jb .Lmulx4x_outer + + mov -16($nptr),%r10 + xor %r15,%r15 + sub %r14,%r10 # compare top-most words + adc %r15,%r15 + or %r15,$zero + xor \$1,$zero + lea ($tptr,$num),%rdi # rewind $tptr + lea ($nptr,$num,2),$nptr # rewind $nptr + .byte 0x67,0x67 + sar \$3+2,$num # cf=0 + lea ($nptr,$zero,8),%rbp + mov 56+8(%rsp),%rdx # restore rp + mov $num,%rcx + jmp .Lsqrx4x_sub # common post-condition +.size mulx4x_internal,.-mulx4x_internal +___ +} { +###################################################################### +# void bn_power5( +my $rptr="%rdi"; # BN_ULONG *rptr, +my $aptr="%rsi"; # const BN_ULONG *aptr, +my $bptr="%rdx"; # const void *table, +my $nptr="%rcx"; # const BN_ULONG *nptr, +my $n0 ="%r8"; # const BN_ULONG *n0); +my $num ="%r9"; # int num, has to be divisible by 8 + # int pwr); + +my ($i,$j,$tptr)=("%rbp","%rcx",$rptr); +my @A0=("%r10","%r11"); +my @A1=("%r12","%r13"); +my ($a0,$a1,$ai)=("%r14","%r15","%rbx"); + +$code.=<<___; +.type bn_powerx5,\@function,6 +.align 32 +bn_powerx5: +.Lpowerx5_enter: + .byte 0x67 + mov %rsp,%rax + push %rbx + push %rbp + push %r12 + push %r13 + push %r14 + push %r15 +___ +$code.=<<___ if ($win64); + lea -0x28(%rsp),%rsp + movaps %xmm6,(%rsp) + movaps %xmm7,0x10(%rsp) +___ +$code.=<<___; + .byte 0x67 + mov ${num}d,%r10d + shl \$3,${num}d # convert $num to bytes + shl \$3+2,%r10d # 4*$num + neg $num + mov ($n0),$n0 # *n0 + + ############################################################## + # ensure that stack frame doesn't alias with $aptr+4*$num + # modulo 4096, which covers ret[num], am[num] and n[2*num] + # (see bn_exp.c). this is done to allow memory disambiguation + # logic do its magic. + # + lea -64(%rsp,$num,2),%r11 + sub $aptr,%r11 + and \$4095,%r11 + cmp %r11,%r10 + jb .Lpwrx_sp_alt + sub %r11,%rsp # align with $aptr + lea -64(%rsp,$num,2),%rsp # alloca(frame+2*$num) + jmp .Lpwrx_sp_done + +.align 32 +.Lpwrx_sp_alt: + lea 4096-64(,$num,2),%r10 # 4096-frame-2*$num + lea -64(%rsp,$num,2),%rsp # alloca(frame+2*$num) + sub %r10,%r11 + mov \$0,%r10 + cmovc %r10,%r11 + sub %r11,%rsp +.Lpwrx_sp_done: + and \$-64,%rsp + mov $num,%r10 + neg $num + + ############################################################## + # Stack layout + # + # +0 saved $num, used in reduction section + # +8 &t[2*$num], used in reduction section + # +16 intermediate carry bit + # +24 top-most carry bit, used in reduction section + # +32 saved *n0 + # +40 saved %rsp + # +48 t[2*$num] + # + pxor %xmm0,%xmm0 + movq $rptr,%xmm1 # save $rptr + movq $nptr,%xmm2 # save $nptr + movq %r10, %xmm3 # -$num + movq $bptr,%xmm4 + mov $n0, 32(%rsp) + mov %rax, 40(%rsp) # save original %rsp +.Lpowerx5_body: + + call __bn_sqrx8x_internal + call __bn_sqrx8x_internal + call __bn_sqrx8x_internal + call __bn_sqrx8x_internal + call __bn_sqrx8x_internal + + mov %r10,$num # -num + mov $aptr,$rptr + movq %xmm2,$nptr + movq %xmm4,$bptr + mov 40(%rsp),%rax + + call mulx4x_internal + + mov 40(%rsp),%rsi # restore %rsp + mov \$1,%rax +___ +$code.=<<___ if ($win64); + movaps -88(%rsi),%xmm6 + movaps -72(%rsi),%xmm7 +___ +$code.=<<___; + mov -48(%rsi),%r15 + mov -40(%rsi),%r14 + mov -32(%rsi),%r13 + mov -24(%rsi),%r12 + mov -16(%rsi),%rbp + mov -8(%rsi),%rbx + lea (%rsi),%rsp +.Lpowerx5_epilogue: + ret +.size bn_powerx5,.-bn_powerx5 + +.globl bn_sqrx8x_internal +.hidden bn_sqrx8x_internal +.type bn_sqrx8x_internal,\@abi-omnipotent +.align 32 +bn_sqrx8x_internal: +__bn_sqrx8x_internal: + ################################################################## + # Squaring part: + # + # a) multiply-n-add everything but a[i]*a[i]; + # b) shift result of a) by 1 to the left and accumulate + # a[i]*a[i] products; + # + ################################################################## + # a[7]a[7]a[6]a[6]a[5]a[5]a[4]a[4]a[3]a[3]a[2]a[2]a[1]a[1]a[0]a[0] + # a[1]a[0] + # a[2]a[0] + # a[3]a[0] + # a[2]a[1] + # a[3]a[1] + # a[3]a[2] + # + # a[4]a[0] + # a[5]a[0] + # a[6]a[0] + # a[7]a[0] + # a[4]a[1] + # a[5]a[1] + # a[6]a[1] + # a[7]a[1] + # a[4]a[2] + # a[5]a[2] + # a[6]a[2] + # a[7]a[2] + # a[4]a[3] + # a[5]a[3] + # a[6]a[3] + # a[7]a[3] + # + # a[5]a[4] + # a[6]a[4] + # a[7]a[4] + # a[6]a[5] + # a[7]a[5] + # a[7]a[6] + # a[7]a[7]a[6]a[6]a[5]a[5]a[4]a[4]a[3]a[3]a[2]a[2]a[1]a[1]a[0]a[0] +___ +{ +my ($zero,$carry)=("%rbp","%rcx"); +my $aaptr=$zero; +$code.=<<___; + lea 48+8(%rsp),$tptr + lea ($aptr,$num),$aaptr + mov $num,0+8(%rsp) # save $num + mov $aaptr,8+8(%rsp) # save end of $aptr + jmp .Lsqr8x_zero_start + +.align 32 +.byte 0x66,0x66,0x66,0x2e,0x0f,0x1f,0x84,0x00,0x00,0x00,0x00,0x00 +.Lsqrx8x_zero: + .byte 0x3e + movdqa %xmm0,0*8($tptr) + movdqa %xmm0,2*8($tptr) + movdqa %xmm0,4*8($tptr) + movdqa %xmm0,6*8($tptr) +.Lsqr8x_zero_start: # aligned at 32 + movdqa %xmm0,8*8($tptr) + movdqa %xmm0,10*8($tptr) + movdqa %xmm0,12*8($tptr) + movdqa %xmm0,14*8($tptr) + lea 16*8($tptr),$tptr + sub \$64,$num + jnz .Lsqrx8x_zero + + mov 0*8($aptr),%rdx # a[0], modulo-scheduled + #xor %r9,%r9 # t[1], ex-$num, zero already + xor %r10,%r10 + xor %r11,%r11 + xor %r12,%r12 + xor %r13,%r13 + xor %r14,%r14 + xor %r15,%r15 + lea 48+8(%rsp),$tptr + xor $zero,$zero # cf=0, cf=0 + jmp .Lsqrx8x_outer_loop + +.align 32 +.Lsqrx8x_outer_loop: + mulx 1*8($aptr),%r8,%rax # a[1]*a[0] + adcx %r9,%r8 # a[1]*a[0]+=t[1] + adox %rax,%r10 + mulx 2*8($aptr),%r9,%rax # a[2]*a[0] + adcx %r10,%r9 + adox %rax,%r11 + .byte 0xc4,0xe2,0xab,0xf6,0x86,0x18,0x00,0x00,0x00 # mulx 3*8($aptr),%r10,%rax # ... + adcx %r11,%r10 + adox %rax,%r12 + .byte 0xc4,0xe2,0xa3,0xf6,0x86,0x20,0x00,0x00,0x00 # mulx 4*8($aptr),%r11,%rax + adcx %r12,%r11 + adox %rax,%r13 + mulx 5*8($aptr),%r12,%rax + adcx %r13,%r12 + adox %rax,%r14 + mulx 6*8($aptr),%r13,%rax + adcx %r14,%r13 + adox %r15,%rax + mulx 7*8($aptr),%r14,%r15 + mov 1*8($aptr),%rdx # a[1] + adcx %rax,%r14 + adox $zero,%r15 + adc 8*8($tptr),%r15 + mov %r8,1*8($tptr) # t[1] + mov %r9,2*8($tptr) # t[2] + sbb $carry,$carry # mov %cf,$carry + xor $zero,$zero # cf=0, of=0 + + + mulx 2*8($aptr),%r8,%rbx # a[2]*a[1] + mulx 3*8($aptr),%r9,%rax # a[3]*a[1] + adcx %r10,%r8 + adox %rbx,%r9 + mulx 4*8($aptr),%r10,%rbx # ... + adcx %r11,%r9 + adox %rax,%r10 + .byte 0xc4,0xe2,0xa3,0xf6,0x86,0x28,0x00,0x00,0x00 # mulx 5*8($aptr),%r11,%rax + adcx %r12,%r10 + adox %rbx,%r11 + .byte 0xc4,0xe2,0x9b,0xf6,0x9e,0x30,0x00,0x00,0x00 # mulx 6*8($aptr),%r12,%rbx + adcx %r13,%r11 + adox %r14,%r12 + .byte 0xc4,0x62,0x93,0xf6,0xb6,0x38,0x00,0x00,0x00 # mulx 7*8($aptr),%r13,%r14 + mov 2*8($aptr),%rdx # a[2] + adcx %rax,%r12 + adox %rbx,%r13 + adcx %r15,%r13 + adox $zero,%r14 # of=0 + adcx $zero,%r14 # cf=0 + + mov %r8,3*8($tptr) # t[3] + mov %r9,4*8($tptr) # t[4] + + mulx 3*8($aptr),%r8,%rbx # a[3]*a[2] + mulx 4*8($aptr),%r9,%rax # a[4]*a[2] + adcx %r10,%r8 + adox %rbx,%r9 + mulx 5*8($aptr),%r10,%rbx # ... + adcx %r11,%r9 + adox %rax,%r10 + .byte 0xc4,0xe2,0xa3,0xf6,0x86,0x30,0x00,0x00,0x00 # mulx 6*8($aptr),%r11,%rax + adcx %r12,%r10 + adox %r13,%r11 + .byte 0xc4,0x62,0x9b,0xf6,0xae,0x38,0x00,0x00,0x00 # mulx 7*8($aptr),%r12,%r13 + .byte 0x3e + mov 3*8($aptr),%rdx # a[3] + adcx %rbx,%r11 + adox %rax,%r12 + adcx %r14,%r12 + mov %r8,5*8($tptr) # t[5] + mov %r9,6*8($tptr) # t[6] + mulx 4*8($aptr),%r8,%rax # a[4]*a[3] + adox $zero,%r13 # of=0 + adcx $zero,%r13 # cf=0 + + mulx 5*8($aptr),%r9,%rbx # a[5]*a[3] + adcx %r10,%r8 + adox %rax,%r9 + mulx 6*8($aptr),%r10,%rax # ... + adcx %r11,%r9 + adox %r12,%r10 + mulx 7*8($aptr),%r11,%r12 + mov 4*8($aptr),%rdx # a[4] + mov 5*8($aptr),%r14 # a[5] + adcx %rbx,%r10 + adox %rax,%r11 + mov 6*8($aptr),%r15 # a[6] + adcx %r13,%r11 + adox $zero,%r12 # of=0 + adcx $zero,%r12 # cf=0 + + mov %r8,7*8($tptr) # t[7] + mov %r9,8*8($tptr) # t[8] + + mulx %r14,%r9,%rax # a[5]*a[4] + mov 7*8($aptr),%r8 # a[7] + adcx %r10,%r9 + mulx %r15,%r10,%rbx # a[6]*a[4] + adox %rax,%r10 + adcx %r11,%r10 + mulx %r8,%r11,%rax # a[7]*a[4] + mov %r14,%rdx # a[5] + adox %rbx,%r11 + adcx %r12,%r11 + #adox $zero,%rax # of=0 + adcx $zero,%rax # cf=0 + + mulx %r15,%r14,%rbx # a[6]*a[5] + mulx %r8,%r12,%r13 # a[7]*a[5] + mov %r15,%rdx # a[6] + lea 8*8($aptr),$aptr + adcx %r14,%r11 + adox %rbx,%r12 + adcx %rax,%r12 + adox $zero,%r13 + + .byte 0x67,0x67 + mulx %r8,%r8,%r14 # a[7]*a[6] + adcx %r8,%r13 + adcx $zero,%r14 + + cmp 8+8(%rsp),$aptr + je .Lsqrx8x_outer_break + + neg $carry # mov $carry,%cf + mov \$-8,%rcx + mov $zero,%r15 + mov 8*8($tptr),%r8 + adcx 9*8($tptr),%r9 # +=t[9] + adcx 10*8($tptr),%r10 # ... + adcx 11*8($tptr),%r11 + adc 12*8($tptr),%r12 + adc 13*8($tptr),%r13 + adc 14*8($tptr),%r14 + adc 15*8($tptr),%r15 + lea ($aptr),$aaptr + lea 2*64($tptr),$tptr + sbb %rax,%rax # mov %cf,$carry + + mov -64($aptr),%rdx # a[0] + mov %rax,16+8(%rsp) # offload $carry + mov $tptr,24+8(%rsp) + + #lea 8*8($tptr),$tptr # see 2*8*8($tptr) above + xor %eax,%eax # cf=0, of=0 + jmp .Lsqrx8x_loop + +.align 32 +.Lsqrx8x_loop: + mov %r8,%rbx + mulx 0*8($aaptr),%rax,%r8 # a[8]*a[i] + adcx %rax,%rbx # +=t[8] + adox %r9,%r8 + + mulx 1*8($aaptr),%rax,%r9 # ... + adcx %rax,%r8 + adox %r10,%r9 + + mulx 2*8($aaptr),%rax,%r10 + adcx %rax,%r9 + adox %r11,%r10 + + mulx 3*8($aaptr),%rax,%r11 + adcx %rax,%r10 + adox %r12,%r11 + + .byte 0xc4,0x62,0xfb,0xf6,0xa5,0x20,0x00,0x00,0x00 # mulx 4*8($aaptr),%rax,%r12 + adcx %rax,%r11 + adox %r13,%r12 + + mulx 5*8($aaptr),%rax,%r13 + adcx %rax,%r12 + adox %r14,%r13 + + mulx 6*8($aaptr),%rax,%r14 + mov %rbx,($tptr,%rcx,8) # store t[8+i] + mov \$0,%ebx + adcx %rax,%r13 + adox %r15,%r14 + + .byte 0xc4,0x62,0xfb,0xf6,0xbd,0x38,0x00,0x00,0x00 # mulx 7*8($aaptr),%rax,%r15 + mov 8($aptr,%rcx,8),%rdx # a[i] + adcx %rax,%r14 + adox %rbx,%r15 # %rbx is 0, of=0 + adcx %rbx,%r15 # cf=0 + + .byte 0x67 + inc %rcx # of=0 + jnz .Lsqrx8x_loop + + lea 8*8($aaptr),$aaptr + mov \$-8,%rcx + cmp 8+8(%rsp),$aaptr # done? + je .Lsqrx8x_break + + sub 16+8(%rsp),%rbx # mov 16(%rsp),%cf + .byte 0x66 + mov -64($aptr),%rdx + adcx 0*8($tptr),%r8 + adcx 1*8($tptr),%r9 + adc 2*8($tptr),%r10 + adc 3*8($tptr),%r11 + adc 4*8($tptr),%r12 + adc 5*8($tptr),%r13 + adc 6*8($tptr),%r14 + adc 7*8($tptr),%r15 + lea 8*8($tptr),$tptr + .byte 0x67 + sbb %rax,%rax # mov %cf,%rax + xor %ebx,%ebx # cf=0, of=0 + mov %rax,16+8(%rsp) # offload carry + jmp .Lsqrx8x_loop + +.align 32 +.Lsqrx8x_break: + sub 16+8(%rsp),%r8 # consume last carry + mov 24+8(%rsp),$carry # initial $tptr, borrow $carry + mov 0*8($aptr),%rdx # a[8], modulo-scheduled + xor %ebp,%ebp # xor $zero,$zero + mov %r8,0*8($tptr) + cmp $carry,$tptr # cf=0, of=0 + je .Lsqrx8x_outer_loop + + mov %r9,1*8($tptr) + mov 1*8($carry),%r9 + mov %r10,2*8($tptr) + mov 2*8($carry),%r10 + mov %r11,3*8($tptr) + mov 3*8($carry),%r11 + mov %r12,4*8($tptr) + mov 4*8($carry),%r12 + mov %r13,5*8($tptr) + mov 5*8($carry),%r13 + mov %r14,6*8($tptr) + mov 6*8($carry),%r14 + mov %r15,7*8($tptr) + mov 7*8($carry),%r15 + mov $carry,$tptr + jmp .Lsqrx8x_outer_loop + +.align 32 +.Lsqrx8x_outer_break: + mov %r9,9*8($tptr) # t[9] + movq %xmm3,%rcx # -$num + mov %r10,10*8($tptr) # ... + mov %r11,11*8($tptr) + mov %r12,12*8($tptr) + mov %r13,13*8($tptr) + mov %r14,14*8($tptr) +___ +} { +my $i="%rcx"; +$code.=<<___; + lea 48+8(%rsp),$tptr + mov ($aptr,$i),%rdx # a[0] + + mov 8($tptr),$A0[1] # t[1] + xor $A0[0],$A0[0] # t[0], of=0, cf=0 + mov 0+8(%rsp),$num # restore $num + adox $A0[1],$A0[1] + mov 16($tptr),$A1[0] # t[2] # prefetch + mov 24($tptr),$A1[1] # t[3] # prefetch + #jmp .Lsqrx4x_shift_n_add # happens to be aligned + +.align 32 +.Lsqrx4x_shift_n_add: + mulx %rdx,%rax,%rbx + adox $A1[0],$A1[0] + adcx $A0[0],%rax + .byte 0x48,0x8b,0x94,0x0e,0x08,0x00,0x00,0x00 # mov 8($aptr,$i),%rdx # a[i+1] # prefetch + .byte 0x4c,0x8b,0x97,0x20,0x00,0x00,0x00 # mov 32($tptr),$A0[0] # t[2*i+4] # prefetch + adox $A1[1],$A1[1] + adcx $A0[1],%rbx + mov 40($tptr),$A0[1] # t[2*i+4+1] # prefetch + mov %rax,0($tptr) + mov %rbx,8($tptr) + + mulx %rdx,%rax,%rbx + adox $A0[0],$A0[0] + adcx $A1[0],%rax + mov 16($aptr,$i),%rdx # a[i+2] # prefetch + mov 48($tptr),$A1[0] # t[2*i+6] # prefetch + adox $A0[1],$A0[1] + adcx $A1[1],%rbx + mov 56($tptr),$A1[1] # t[2*i+6+1] # prefetch + mov %rax,16($tptr) + mov %rbx,24($tptr) + + mulx %rdx,%rax,%rbx + adox $A1[0],$A1[0] + adcx $A0[0],%rax + mov 24($aptr,$i),%rdx # a[i+3] # prefetch + lea 32($i),$i + mov 64($tptr),$A0[0] # t[2*i+8] # prefetch + adox $A1[1],$A1[1] + adcx $A0[1],%rbx + mov 72($tptr),$A0[1] # t[2*i+8+1] # prefetch + mov %rax,32($tptr) + mov %rbx,40($tptr) + + mulx %rdx,%rax,%rbx + adox $A0[0],$A0[0] + adcx $A1[0],%rax + jrcxz .Lsqrx4x_shift_n_add_break + .byte 0x48,0x8b,0x94,0x0e,0x00,0x00,0x00,0x00 # mov 0($aptr,$i),%rdx # a[i+4] # prefetch + adox $A0[1],$A0[1] + adcx $A1[1],%rbx + mov 80($tptr),$A1[0] # t[2*i+10] # prefetch + mov 88($tptr),$A1[1] # t[2*i+10+1] # prefetch + mov %rax,48($tptr) + mov %rbx,56($tptr) + lea 64($tptr),$tptr + nop + jmp .Lsqrx4x_shift_n_add + +.align 32 +.Lsqrx4x_shift_n_add_break: + adcx $A1[1],%rbx + mov %rax,48($tptr) + mov %rbx,56($tptr) + lea 64($tptr),$tptr # end of t[] buffer +___ +} +###################################################################### +# Montgomery reduction part, "word-by-word" algorithm. +# +# This new path is inspired by multiple submissions from Intel, by +# Shay Gueron, Vlad Krasnov, Erdinc Ozturk, James Guilford, +# Vinodh Gopal... +{ +my ($nptr,$carry,$m0)=("%rbp","%rsi","%rdx"); + +$code.=<<___; + movq %xmm2,$nptr +sqrx8x_reduction: + xor %eax,%eax # initial top-most carry bit + mov 32+8(%rsp),%rbx # n0 + mov 48+8(%rsp),%rdx # "%r8", 8*0($tptr) + lea -128($nptr,$num,2),%rcx # end of n[] + #lea 48+8(%rsp,$num,2),$tptr # end of t[] buffer + mov %rcx, 0+8(%rsp) # save end of n[] + mov $tptr,8+8(%rsp) # save end of t[] + + lea 48+8(%rsp),$tptr # initial t[] window + jmp .Lsqrx8x_reduction_loop + +.align 32 +.Lsqrx8x_reduction_loop: + mov 8*1($tptr),%r9 + mov 8*2($tptr),%r10 + mov 8*3($tptr),%r11 + mov 8*4($tptr),%r12 + mov %rdx,%r8 + imulq %rbx,%rdx # n0*a[i] + mov 8*5($tptr),%r13 + mov 8*6($tptr),%r14 + mov 8*7($tptr),%r15 + mov %rax,24+8(%rsp) # store top-most carry bit + + lea 8*8($tptr),$tptr + xor $carry,$carry # cf=0,of=0 + mov \$-8,%rcx + jmp .Lsqrx8x_reduce + +.align 32 +.Lsqrx8x_reduce: + mov %r8, %rbx + mulx 16*0($nptr),%rax,%r8 # n[0] + adcx %rbx,%rax # discarded + adox %r9,%r8 + + mulx 16*1($nptr),%rbx,%r9 # n[1] + adcx %rbx,%r8 + adox %r10,%r9 + + mulx 16*2($nptr),%rbx,%r10 + adcx %rbx,%r9 + adox %r11,%r10 + + mulx 16*3($nptr),%rbx,%r11 + adcx %rbx,%r10 + adox %r12,%r11 + + .byte 0xc4,0x62,0xe3,0xf6,0xa5,0x40,0x00,0x00,0x00 # mulx 16*4($nptr),%rbx,%r12 + mov %rdx,%rax + mov %r8,%rdx + adcx %rbx,%r11 + adox %r13,%r12 + + mulx 32+8(%rsp),%rbx,%rdx # %rdx discarded + mov %rax,%rdx + mov %rax,64+48+8(%rsp,%rcx,8) # put aside n0*a[i] + + mulx 16*5($nptr),%rax,%r13 + adcx %rax,%r12 + adox %r14,%r13 + + mulx 16*6($nptr),%rax,%r14 + adcx %rax,%r13 + adox %r15,%r14 + + mulx 16*7($nptr),%rax,%r15 + mov %rbx,%rdx + adcx %rax,%r14 + adox $carry,%r15 # $carry is 0 + adcx $carry,%r15 # cf=0 + + .byte 0x67,0x67,0x67 + inc %rcx # of=0 + jnz .Lsqrx8x_reduce + + mov $carry,%rax # xor %rax,%rax + cmp 0+8(%rsp),$nptr # end of n[]? + jae .Lsqrx8x_no_tail + + mov 48+8(%rsp),%rdx # pull n0*a[0] + add 8*0($tptr),%r8 + lea 16*8($nptr),$nptr + mov \$-8,%rcx + adcx 8*1($tptr),%r9 + adcx 8*2($tptr),%r10 + adc 8*3($tptr),%r11 + adc 8*4($tptr),%r12 + adc 8*5($tptr),%r13 + adc 8*6($tptr),%r14 + adc 8*7($tptr),%r15 + lea 8*8($tptr),$tptr + sbb %rax,%rax # top carry + + xor $carry,$carry # of=0, cf=0 + mov %rax,16+8(%rsp) + jmp .Lsqrx8x_tail + +.align 32 +.Lsqrx8x_tail: + mov %r8,%rbx + mulx 16*0($nptr),%rax,%r8 + adcx %rax,%rbx + adox %r9,%r8 + + mulx 16*1($nptr),%rax,%r9 + adcx %rax,%r8 + adox %r10,%r9 + + mulx 16*2($nptr),%rax,%r10 + adcx %rax,%r9 + adox %r11,%r10 + + mulx 16*3($nptr),%rax,%r11 + adcx %rax,%r10 + adox %r12,%r11 + + .byte 0xc4,0x62,0xfb,0xf6,0xa5,0x40,0x00,0x00,0x00 # mulx 16*4($nptr),%rax,%r12 + adcx %rax,%r11 + adox %r13,%r12 + + mulx 16*5($nptr),%rax,%r13 + adcx %rax,%r12 + adox %r14,%r13 + + mulx 16*6($nptr),%rax,%r14 + adcx %rax,%r13 + adox %r15,%r14 + + mulx 16*7($nptr),%rax,%r15 + mov 72+48+8(%rsp,%rcx,8),%rdx # pull n0*a[i] + adcx %rax,%r14 + adox $carry,%r15 + mov %rbx,($tptr,%rcx,8) # save result + mov %r8,%rbx + adcx $carry,%r15 # cf=0 + + inc %rcx # of=0 + jnz .Lsqrx8x_tail + + cmp 0+8(%rsp),$nptr # end of n[]? + jae .Lsqrx8x_tail_done # break out of loop + + sub 16+8(%rsp),$carry # mov 16(%rsp),%cf + mov 48+8(%rsp),%rdx # pull n0*a[0] + lea 16*8($nptr),$nptr + adc 8*0($tptr),%r8 + adc 8*1($tptr),%r9 + adc 8*2($tptr),%r10 + adc 8*3($tptr),%r11 + adc 8*4($tptr),%r12 + adc 8*5($tptr),%r13 + adc 8*6($tptr),%r14 + adc 8*7($tptr),%r15 + lea 8*8($tptr),$tptr + sbb %rax,%rax + sub \$8,%rcx # mov \$-8,%rcx + + xor $carry,$carry # of=0, cf=0 + mov %rax,16+8(%rsp) + jmp .Lsqrx8x_tail + +.align 32 +.Lsqrx8x_tail_done: + add 24+8(%rsp),%r8 # can this overflow? + mov $carry,%rax # xor %rax,%rax + + sub 16+8(%rsp),$carry # mov 16(%rsp),%cf +.Lsqrx8x_no_tail: # %cf is 0 if jumped here + adc 8*0($tptr),%r8 + movq %xmm3,%rcx + adc 8*1($tptr),%r9 + mov 16*7($nptr),$carry + movq %xmm2,$nptr # restore $nptr + adc 8*2($tptr),%r10 + adc 8*3($tptr),%r11 + adc 8*4($tptr),%r12 + adc 8*5($tptr),%r13 + adc 8*6($tptr),%r14 + adc 8*7($tptr),%r15 + adc %rax,%rax # top-most carry + + mov 32+8(%rsp),%rbx # n0 + mov 8*8($tptr,%rcx),%rdx # modulo-scheduled "%r8" + + mov %r8,8*0($tptr) # store top 512 bits + lea 8*8($tptr),%r8 # borrow %r8 + mov %r9,8*1($tptr) + mov %r10,8*2($tptr) + mov %r11,8*3($tptr) + mov %r12,8*4($tptr) + mov %r13,8*5($tptr) + mov %r14,8*6($tptr) + mov %r15,8*7($tptr) + + lea 8*8($tptr,%rcx),$tptr # start of current t[] window + cmp 8+8(%rsp),%r8 # end of t[]? + jb .Lsqrx8x_reduction_loop +___ +} +############################################################## +# Post-condition, 4x unrolled +# +{ +my ($rptr,$nptr)=("%rdx","%rbp"); +my @ri=map("%r$_",(10..13)); +my @ni=map("%r$_",(14..15)); +$code.=<<___; + xor %rbx,%rbx + sub %r15,%rsi # compare top-most words + adc %rbx,%rbx + mov %rcx,%r10 # -$num + .byte 0x67 + or %rbx,%rax + .byte 0x67 + mov %rcx,%r9 # -$num + xor \$1,%rax + sar \$3+2,%rcx # cf=0 + #lea 48+8(%rsp,%r9),$tptr + lea ($nptr,%rax,8),$nptr + movq %xmm1,$rptr # restore $rptr + movq %xmm1,$aptr # prepare for back-to-back call + jmp .Lsqrx4x_sub + +.align 32 +.Lsqrx4x_sub: + .byte 0x66 + mov 8*0($tptr),%r12 + mov 8*1($tptr),%r13 + sbb 16*0($nptr),%r12 + mov 8*2($tptr),%r14 + sbb 16*1($nptr),%r13 + mov 8*3($tptr),%r15 + lea 8*4($tptr),$tptr + sbb 16*2($nptr),%r14 + mov %r12,8*0($rptr) + sbb 16*3($nptr),%r15 + lea 16*4($nptr),$nptr + mov %r13,8*1($rptr) + mov %r14,8*2($rptr) + mov %r15,8*3($rptr) + lea 8*4($rptr),$rptr + + inc %rcx + jnz .Lsqrx4x_sub +___ +} +$code.=<<___; + neg %r9 # restore $num + ret +.size bn_sqrx8x_internal,.-bn_sqrx8x_internal +___ +}}} { -my ($inp,$num,$tbl,$idx)=$win64?("%rcx","%rdx","%r8", "%r9") : # Win64 order - ("%rdi","%rsi","%rdx","%rcx"); # Unix order +my ($inp,$num,$tbl,$idx)=$win64?("%rcx","%edx","%r8", "%r9d") : # Win64 order + ("%rdi","%esi","%rdx","%ecx"); # Unix order my $out=$inp; my $STRIDE=2**5*8; my $N=$STRIDE/4; $code.=<<___; +.globl bn_get_bits5 +.type bn_get_bits5,\@abi-omnipotent +.align 16 +bn_get_bits5: + mov $inp,%r10 + mov $num,%ecx + shr \$3,$num + movzw (%r10,$num),%eax + and \$7,%ecx + shrl %cl,%eax + and \$31,%eax + ret +.size bn_get_bits5,.-bn_get_bits5 + .globl bn_scatter5 .type bn_scatter5,\@abi-omnipotent .align 16 @@ -868,13 +3267,13 @@ $code.=<<___ if ($win64); .byte 0x0f,0x29,0x7c,0x24,0x10 #movdqa %xmm7,0x10(%rsp) ___ $code.=<<___; - mov $idx,%r11 + mov $idx,%r11d shr \$`log($N/8)/log(2)`,$idx and \$`$N/8-1`,%r11 not $idx lea .Lmagic_masks(%rip),%rax and \$`2**5/($N/8)-1`,$idx # 5 is "window size" - lea 96($tbl,%r11,8),$tbl # pointer within 1st cache line + lea 128($tbl,%r11,8),$tbl # pointer within 1st cache line movq 0(%rax,$idx,8),%xmm4 # set of masks denoting which movq 8(%rax,$idx,8),%xmm5 # cache line contains element movq 16(%rax,$idx,8),%xmm6 # denoted by 7th argument @@ -882,15 +3281,16 @@ $code.=<<___; jmp .Lgather .align 16 .Lgather: - movq `0*$STRIDE/4-96`($tbl),%xmm0 - movq `1*$STRIDE/4-96`($tbl),%xmm1 + movq `0*$STRIDE/4-128`($tbl),%xmm0 + movq `1*$STRIDE/4-128`($tbl),%xmm1 pand %xmm4,%xmm0 - movq `2*$STRIDE/4-96`($tbl),%xmm2 + movq `2*$STRIDE/4-128`($tbl),%xmm2 pand %xmm5,%xmm1 - movq `3*$STRIDE/4-96`($tbl),%xmm3 + movq `3*$STRIDE/4-128`($tbl),%xmm3 pand %xmm6,%xmm2 por %xmm1,%xmm0 pand %xmm7,%xmm3 + .byte 0x67,0x67 por %xmm2,%xmm0 lea $STRIDE($tbl),$tbl por %xmm3,%xmm0 @@ -954,26 +3354,27 @@ mul_handler: cmp %r10,%rbx # context->Rip<end of prologue label jb .Lcommon_seh_tail - lea `40+48`(%rax),%rax - - mov 4(%r11),%r10d # HandlerData[1] - lea (%rsi,%r10),%r10 # end of alloca label - cmp %r10,%rbx # context->Rip<end of alloca label - jb .Lcommon_seh_tail - mov 152($context),%rax # pull context->Rsp - mov 8(%r11),%r10d # HandlerData[2] + mov 4(%r11),%r10d # HandlerData[1] lea (%rsi,%r10),%r10 # epilogue label cmp %r10,%rbx # context->Rip>=epilogue label jae .Lcommon_seh_tail + lea .Lmul_epilogue(%rip),%r10 + cmp %r10,%rbx + jb .Lbody_40 + mov 192($context),%r10 # pull $num mov 8(%rax,%r10,8),%rax # pull saved stack pointer + jmp .Lbody_proceed + +.Lbody_40: + mov 40(%rax),%rax # pull saved stack pointer +.Lbody_proceed: - movaps (%rax),%xmm0 - movaps 16(%rax),%xmm1 - lea `40+48`(%rax),%rax + movaps -88(%rax),%xmm0 + movaps -72(%rax),%xmm1 mov -8(%rax),%rbx mov -16(%rax),%rbp @@ -1040,6 +3441,24 @@ mul_handler: .rva .LSEH_end_bn_mul4x_mont_gather5 .rva .LSEH_info_bn_mul4x_mont_gather5 + .rva .LSEH_begin_bn_power5 + .rva .LSEH_end_bn_power5 + .rva .LSEH_info_bn_power5 + + .rva .LSEH_begin_bn_from_mont8x + .rva .LSEH_end_bn_from_mont8x + .rva .LSEH_info_bn_from_mont8x +___ +$code.=<<___ if ($addx); + .rva .LSEH_begin_bn_mulx4x_mont_gather5 + .rva .LSEH_end_bn_mulx4x_mont_gather5 + .rva .LSEH_info_bn_mulx4x_mont_gather5 + + .rva .LSEH_begin_bn_powerx5 + .rva .LSEH_end_bn_powerx5 + .rva .LSEH_info_bn_powerx5 +___ +$code.=<<___; .rva .LSEH_begin_bn_gather5 .rva .LSEH_end_bn_gather5 .rva .LSEH_info_bn_gather5 @@ -1049,12 +3468,36 @@ mul_handler: .LSEH_info_bn_mul_mont_gather5: .byte 9,0,0,0 .rva mul_handler - .rva .Lmul_alloca,.Lmul_body,.Lmul_epilogue # HandlerData[] + .rva .Lmul_body,.Lmul_epilogue # HandlerData[] .align 8 .LSEH_info_bn_mul4x_mont_gather5: .byte 9,0,0,0 .rva mul_handler - .rva .Lmul4x_alloca,.Lmul4x_body,.Lmul4x_epilogue # HandlerData[] + .rva .Lmul4x_body,.Lmul4x_epilogue # HandlerData[] +.align 8 +.LSEH_info_bn_power5: + .byte 9,0,0,0 + .rva mul_handler + .rva .Lpower5_body,.Lpower5_epilogue # HandlerData[] +.align 8 +.LSEH_info_bn_from_mont8x: + .byte 9,0,0,0 + .rva mul_handler + .rva .Lfrom_body,.Lfrom_epilogue # HandlerData[] +___ +$code.=<<___ if ($addx); +.align 8 +.LSEH_info_bn_mulx4x_mont_gather5: + .byte 9,0,0,0 + .rva mul_handler + .rva .Lmulx4x_body,.Lmulx4x_epilogue # HandlerData[] +.align 8 +.LSEH_info_bn_powerx5: + .byte 9,0,0,0 + .rva mul_handler + .rva .Lpowerx5_body,.Lpowerx5_epilogue # HandlerData[] +___ +$code.=<<___; .align 8 .LSEH_info_bn_gather5: .byte 0x01,0x0d,0x05,0x00 diff --git a/openssl/crypto/bn/bn.h b/openssl/crypto/bn/bn.h index 21a1a3fe3..78709d384 100644 --- a/openssl/crypto/bn/bn.h +++ b/openssl/crypto/bn/bn.h @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -111,191 +111,185 @@ /* ==================================================================== * Copyright 2002 Sun Microsystems, Inc. ALL RIGHTS RESERVED. * - * Portions of the attached software ("Contribution") are developed by + * Portions of the attached software ("Contribution") are developed by * SUN MICROSYSTEMS, INC., and are contributed to the OpenSSL project. * * The Contribution is licensed pursuant to the Eric Young open source * license provided above. * - * The binary polynomial arithmetic software is originally written by + * The binary polynomial arithmetic software is originally written by * Sheueling Chang Shantz and Douglas Stebila of Sun Microsystems Laboratories. * */ #ifndef HEADER_BN_H -#define HEADER_BN_H +# define HEADER_BN_H -#include <openssl/e_os2.h> -#ifndef OPENSSL_NO_FP_API -#include <stdio.h> /* FILE */ -#endif -#include <openssl/ossl_typ.h> -#include <openssl/crypto.h> +# include <openssl/e_os2.h> +# ifndef OPENSSL_NO_FP_API +# include <stdio.h> /* FILE */ +# endif +# include <openssl/ossl_typ.h> +# include <openssl/crypto.h> #ifdef __cplusplus extern "C" { #endif -/* These preprocessor symbols control various aspects of the bignum headers and - * library code. They're not defined by any "normal" configuration, as they are - * intended for development and testing purposes. NB: defining all three can be - * useful for debugging application code as well as openssl itself. - * - * BN_DEBUG - turn on various debugging alterations to the bignum code - * BN_DEBUG_RAND - uses random poisoning of unused words to trip up +/* + * These preprocessor symbols control various aspects of the bignum headers + * and library code. They're not defined by any "normal" configuration, as + * they are intended for development and testing purposes. NB: defining all + * three can be useful for debugging application code as well as openssl + * itself. BN_DEBUG - turn on various debugging alterations to the bignum + * code BN_DEBUG_RAND - uses random poisoning of unused words to trip up * mismanagement of bignum internals. You must also define BN_DEBUG. */ /* #define BN_DEBUG */ /* #define BN_DEBUG_RAND */ -#ifndef OPENSSL_SMALL_FOOTPRINT -#define BN_MUL_COMBA -#define BN_SQR_COMBA -#define BN_RECURSION -#endif +# ifndef OPENSSL_SMALL_FOOTPRINT +# define BN_MUL_COMBA +# define BN_SQR_COMBA +# define BN_RECURSION +# endif -/* This next option uses the C libraries (2 word)/(1 word) function. - * If it is not defined, I use my C version (which is slower). - * The reason for this flag is that when the particular C compiler - * library routine is used, and the library is linked with a different - * compiler, the library is missing. This mostly happens when the - * library is built with gcc and then linked using normal cc. This would - * be a common occurrence because gcc normally produces code that is - * 2 times faster than system compilers for the big number stuff. - * For machines with only one compiler (or shared libraries), this should - * be on. Again this in only really a problem on machines - * using "long long's", are 32bit, and are not using my assembler code. */ -#if defined(OPENSSL_SYS_MSDOS) || defined(OPENSSL_SYS_WINDOWS) || \ +/* + * This next option uses the C libraries (2 word)/(1 word) function. If it is + * not defined, I use my C version (which is slower). The reason for this + * flag is that when the particular C compiler library routine is used, and + * the library is linked with a different compiler, the library is missing. + * This mostly happens when the library is built with gcc and then linked + * using normal cc. This would be a common occurrence because gcc normally + * produces code that is 2 times faster than system compilers for the big + * number stuff. For machines with only one compiler (or shared libraries), + * this should be on. Again this in only really a problem on machines using + * "long long's", are 32bit, and are not using my assembler code. + */ +# if defined(OPENSSL_SYS_MSDOS) || defined(OPENSSL_SYS_WINDOWS) || \ defined(OPENSSL_SYS_WIN32) || defined(linux) -# ifndef BN_DIV2W -# define BN_DIV2W +# ifndef BN_DIV2W +# define BN_DIV2W +# endif # endif -#endif -/* assuming long is 64bit - this is the DEC Alpha - * unsigned long long is only 64 bits :-(, don't define - * BN_LLONG for the DEC Alpha */ -#ifdef SIXTY_FOUR_BIT_LONG -#define BN_ULLONG unsigned long long -#define BN_ULONG unsigned long -#define BN_LONG long -#define BN_BITS 128 -#define BN_BYTES 8 -#define BN_BITS2 64 -#define BN_BITS4 32 -#define BN_MASK (0xffffffffffffffffffffffffffffffffLL) -#define BN_MASK2 (0xffffffffffffffffL) -#define BN_MASK2l (0xffffffffL) -#define BN_MASK2h (0xffffffff00000000L) -#define BN_MASK2h1 (0xffffffff80000000L) -#define BN_TBIT (0x8000000000000000L) -#define BN_DEC_CONV (10000000000000000000UL) -#define BN_DEC_FMT1 "%lu" -#define BN_DEC_FMT2 "%019lu" -#define BN_DEC_NUM 19 -#define BN_HEX_FMT1 "%lX" -#define BN_HEX_FMT2 "%016lX" -#endif +/* + * assuming long is 64bit - this is the DEC Alpha unsigned long long is only + * 64 bits :-(, don't define BN_LLONG for the DEC Alpha + */ +# ifdef SIXTY_FOUR_BIT_LONG +# define BN_ULLONG unsigned long long +# define BN_ULONG unsigned long +# define BN_LONG long +# define BN_BITS 128 +# define BN_BYTES 8 +# define BN_BITS2 64 +# define BN_BITS4 32 +# define BN_MASK (0xffffffffffffffffffffffffffffffffLL) +# define BN_MASK2 (0xffffffffffffffffL) +# define BN_MASK2l (0xffffffffL) +# define BN_MASK2h (0xffffffff00000000L) +# define BN_MASK2h1 (0xffffffff80000000L) +# define BN_TBIT (0x8000000000000000L) +# define BN_DEC_CONV (10000000000000000000UL) +# define BN_DEC_FMT1 "%lu" +# define BN_DEC_FMT2 "%019lu" +# define BN_DEC_NUM 19 +# define BN_HEX_FMT1 "%lX" +# define BN_HEX_FMT2 "%016lX" +# endif -/* This is where the long long data type is 64 bits, but long is 32. - * For machines where there are 64bit registers, this is the mode to use. - * IRIX, on R4000 and above should use this mode, along with the relevant - * assembler code :-). Do NOT define BN_LLONG. +/* + * This is where the long long data type is 64 bits, but long is 32. For + * machines where there are 64bit registers, this is the mode to use. IRIX, + * on R4000 and above should use this mode, along with the relevant assembler + * code :-). Do NOT define BN_LLONG. */ -#ifdef SIXTY_FOUR_BIT -#undef BN_LLONG -#undef BN_ULLONG -#define BN_ULONG unsigned long long -#define BN_LONG long long -#define BN_BITS 128 -#define BN_BYTES 8 -#define BN_BITS2 64 -#define BN_BITS4 32 -#define BN_MASK2 (0xffffffffffffffffLL) -#define BN_MASK2l (0xffffffffL) -#define BN_MASK2h (0xffffffff00000000LL) -#define BN_MASK2h1 (0xffffffff80000000LL) -#define BN_TBIT (0x8000000000000000LL) -#define BN_DEC_CONV (10000000000000000000ULL) -#define BN_DEC_FMT1 "%llu" -#define BN_DEC_FMT2 "%019llu" -#define BN_DEC_NUM 19 -#define BN_HEX_FMT1 "%llX" -#define BN_HEX_FMT2 "%016llX" -#endif +# ifdef SIXTY_FOUR_BIT +# undef BN_LLONG +# undef BN_ULLONG +# define BN_ULONG unsigned long long +# define BN_LONG long long +# define BN_BITS 128 +# define BN_BYTES 8 +# define BN_BITS2 64 +# define BN_BITS4 32 +# define BN_MASK2 (0xffffffffffffffffLL) +# define BN_MASK2l (0xffffffffL) +# define BN_MASK2h (0xffffffff00000000LL) +# define BN_MASK2h1 (0xffffffff80000000LL) +# define BN_TBIT (0x8000000000000000LL) +# define BN_DEC_CONV (10000000000000000000ULL) +# define BN_DEC_FMT1 "%llu" +# define BN_DEC_FMT2 "%019llu" +# define BN_DEC_NUM 19 +# define BN_HEX_FMT1 "%llX" +# define BN_HEX_FMT2 "%016llX" +# endif -#ifdef THIRTY_TWO_BIT -#ifdef BN_LLONG -# if defined(_WIN32) && !defined(__GNUC__) -# define BN_ULLONG unsigned __int64 -# define BN_MASK (0xffffffffffffffffI64) -# else -# define BN_ULLONG unsigned long long -# define BN_MASK (0xffffffffffffffffLL) +# ifdef THIRTY_TWO_BIT +# ifdef BN_LLONG +# if defined(_WIN32) && !defined(__GNUC__) +# define BN_ULLONG unsigned __int64 +# define BN_MASK (0xffffffffffffffffI64) +# else +# define BN_ULLONG unsigned long long +# define BN_MASK (0xffffffffffffffffLL) +# endif +# endif +# define BN_ULONG unsigned int +# define BN_LONG int +# define BN_BITS 64 +# define BN_BYTES 4 +# define BN_BITS2 32 +# define BN_BITS4 16 +# define BN_MASK2 (0xffffffffL) +# define BN_MASK2l (0xffff) +# define BN_MASK2h1 (0xffff8000L) +# define BN_MASK2h (0xffff0000L) +# define BN_TBIT (0x80000000L) +# define BN_DEC_CONV (1000000000L) +# define BN_DEC_FMT1 "%u" +# define BN_DEC_FMT2 "%09u" +# define BN_DEC_NUM 9 +# define BN_HEX_FMT1 "%X" +# define BN_HEX_FMT2 "%08X" # endif -#endif -#define BN_ULONG unsigned int -#define BN_LONG int -#define BN_BITS 64 -#define BN_BYTES 4 -#define BN_BITS2 32 -#define BN_BITS4 16 -#define BN_MASK2 (0xffffffffL) -#define BN_MASK2l (0xffff) -#define BN_MASK2h1 (0xffff8000L) -#define BN_MASK2h (0xffff0000L) -#define BN_TBIT (0x80000000L) -#define BN_DEC_CONV (1000000000L) -#define BN_DEC_FMT1 "%u" -#define BN_DEC_FMT2 "%09u" -#define BN_DEC_NUM 9 -#define BN_HEX_FMT1 "%X" -#define BN_HEX_FMT2 "%08X" -#endif -/* 2011-02-22 SMS. - * In various places, a size_t variable or a type cast to size_t was - * used to perform integer-only operations on pointers. This failed on - * VMS with 64-bit pointers (CC /POINTER_SIZE = 64) because size_t is - * still only 32 bits. What's needed in these cases is an integer type - * with the same size as a pointer, which size_t is not certain to be. - * The only fix here is VMS-specific. +# define BN_DEFAULT_BITS 1280 + +# define BN_FLG_MALLOCED 0x01 +# define BN_FLG_STATIC_DATA 0x02 + +/* + * avoid leaking exponent information through timing, + * BN_mod_exp_mont() will call BN_mod_exp_mont_consttime, + * BN_div() will call BN_div_no_branch, + * BN_mod_inverse() will call BN_mod_inverse_no_branch. */ -#if defined(OPENSSL_SYS_VMS) -# if __INITIAL_POINTER_SIZE == 64 -# define PTR_SIZE_INT long long -# else /* __INITIAL_POINTER_SIZE == 64 */ -# define PTR_SIZE_INT int -# endif /* __INITIAL_POINTER_SIZE == 64 [else] */ -#else /* defined(OPENSSL_SYS_VMS) */ -# define PTR_SIZE_INT size_t -#endif /* defined(OPENSSL_SYS_VMS) [else] */ - -#define BN_DEFAULT_BITS 1280 - -#define BN_FLG_MALLOCED 0x01 -#define BN_FLG_STATIC_DATA 0x02 -#define BN_FLG_CONSTTIME 0x04 /* avoid leaking exponent information through timing, - * BN_mod_exp_mont() will call BN_mod_exp_mont_consttime, - * BN_div() will call BN_div_no_branch, - * BN_mod_inverse() will call BN_mod_inverse_no_branch. - */ - -#ifndef OPENSSL_NO_DEPRECATED -#define BN_FLG_EXP_CONSTTIME BN_FLG_CONSTTIME /* deprecated name for the flag */ - /* avoid leaking exponent information through timings - * (BN_mod_exp_mont() will call BN_mod_exp_mont_consttime) */ -#endif +# define BN_FLG_CONSTTIME 0x04 + +# ifdef OPENSSL_NO_DEPRECATED +/* deprecated name for the flag */ +# define BN_FLG_EXP_CONSTTIME BN_FLG_CONSTTIME +/* + * avoid leaking exponent information through timings + * (BN_mod_exp_mont() will call BN_mod_exp_mont_consttime) + */ +# endif -#ifndef OPENSSL_NO_DEPRECATED -#define BN_FLG_FREE 0x8000 /* used for debuging */ -#endif -#define BN_set_flags(b,n) ((b)->flags|=(n)) -#define BN_get_flags(b,n) ((b)->flags&(n)) +# ifndef OPENSSL_NO_DEPRECATED +# define BN_FLG_FREE 0x8000 + /* used for debuging */ +# endif +# define BN_set_flags(b,n) ((b)->flags|=(n)) +# define BN_get_flags(b,n) ((b)->flags&(n)) -/* get a clone of a BIGNUM with changed flags, for *temporary* use only - * (the two BIGNUMs cannot not be used in parallel!) */ -#define BN_with_flags(dest,b,n) ((dest)->d=(b)->d, \ +/* + * get a clone of a BIGNUM with changed flags, for *temporary* use only (the + * two BIGNUMs cannot not be used in parallel!) + */ +# define BN_with_flags(dest,b,n) ((dest)->d=(b)->d, \ (dest)->top=(b)->top, \ (dest)->dmax=(b)->dmax, \ (dest)->neg=(b)->neg, \ @@ -305,7 +299,7 @@ extern "C" { | (n))) /* Already declared in ossl_typ.h */ -#if 0 +# if 0 typedef struct bignum_st BIGNUM; /* Used for temp variables (declaration hidden in bn_lcl.h) */ typedef struct bignum_ctx BN_CTX; @@ -313,80 +307,81 @@ typedef struct bn_blinding_st BN_BLINDING; typedef struct bn_mont_ctx_st BN_MONT_CTX; typedef struct bn_recp_ctx_st BN_RECP_CTX; typedef struct bn_gencb_st BN_GENCB; -#endif +# endif -struct bignum_st - { - BN_ULONG *d; /* Pointer to an array of 'BN_BITS2' bit chunks. */ - int top; /* Index of last used d +1. */ - /* The next are internal book keeping for bn_expand. */ - int dmax; /* Size of the d array. */ - int neg; /* one if the number is negative */ - int flags; - }; +struct bignum_st { + BN_ULONG *d; /* Pointer to an array of 'BN_BITS2' bit + * chunks. */ + int top; /* Index of last used d +1. */ + /* The next are internal book keeping for bn_expand. */ + int dmax; /* Size of the d array. */ + int neg; /* one if the number is negative */ + int flags; +}; /* Used for montgomery multiplication */ -struct bn_mont_ctx_st - { - int ri; /* number of bits in R */ - BIGNUM RR; /* used to convert to montgomery form */ - BIGNUM N; /* The modulus */ - BIGNUM Ni; /* R*(1/R mod N) - N*Ni = 1 - * (Ni is only stored for bignum algorithm) */ - BN_ULONG n0[2];/* least significant word(s) of Ni; - (type changed with 0.9.9, was "BN_ULONG n0;" before) */ - int flags; - }; - -/* Used for reciprocal division/mod functions - * It cannot be shared between threads +struct bn_mont_ctx_st { + int ri; /* number of bits in R */ + BIGNUM RR; /* used to convert to montgomery form */ + BIGNUM N; /* The modulus */ + BIGNUM Ni; /* R*(1/R mod N) - N*Ni = 1 (Ni is only + * stored for bignum algorithm) */ + BN_ULONG n0[2]; /* least significant word(s) of Ni; (type + * changed with 0.9.9, was "BN_ULONG n0;" + * before) */ + int flags; +}; + +/* + * Used for reciprocal division/mod functions It cannot be shared between + * threads */ -struct bn_recp_ctx_st - { - BIGNUM N; /* the divisor */ - BIGNUM Nr; /* the reciprocal */ - int num_bits; - int shift; - int flags; - }; +struct bn_recp_ctx_st { + BIGNUM N; /* the divisor */ + BIGNUM Nr; /* the reciprocal */ + int num_bits; + int shift; + int flags; +}; /* Used for slow "generation" functions. */ -struct bn_gencb_st - { - unsigned int ver; /* To handle binary (in)compatibility */ - void *arg; /* callback-specific data */ - union - { - /* if(ver==1) - handles old style callbacks */ - void (*cb_1)(int, int, void *); - /* if(ver==2) - new callback style */ - int (*cb_2)(int, int, BN_GENCB *); - } cb; - }; +struct bn_gencb_st { + unsigned int ver; /* To handle binary (in)compatibility */ + void *arg; /* callback-specific data */ + union { + /* if(ver==1) - handles old style callbacks */ + void (*cb_1) (int, int, void *); + /* if(ver==2) - new callback style */ + int (*cb_2) (int, int, BN_GENCB *); + } cb; +}; /* Wrapper function to make using BN_GENCB easier, */ int BN_GENCB_call(BN_GENCB *cb, int a, int b); /* Macro to populate a BN_GENCB structure with an "old"-style callback */ -#define BN_GENCB_set_old(gencb, callback, cb_arg) { \ - BN_GENCB *tmp_gencb = (gencb); \ - tmp_gencb->ver = 1; \ - tmp_gencb->arg = (cb_arg); \ - tmp_gencb->cb.cb_1 = (callback); } +# define BN_GENCB_set_old(gencb, callback, cb_arg) { \ + BN_GENCB *tmp_gencb = (gencb); \ + tmp_gencb->ver = 1; \ + tmp_gencb->arg = (cb_arg); \ + tmp_gencb->cb.cb_1 = (callback); } /* Macro to populate a BN_GENCB structure with a "new"-style callback */ -#define BN_GENCB_set(gencb, callback, cb_arg) { \ - BN_GENCB *tmp_gencb = (gencb); \ - tmp_gencb->ver = 2; \ - tmp_gencb->arg = (cb_arg); \ - tmp_gencb->cb.cb_2 = (callback); } - -#define BN_prime_checks 0 /* default: select number of iterations - based on the size of the number */ - -/* number of Miller-Rabin iterations for an error rate of less than 2^-80 - * for random 'b'-bit input, b >= 100 (taken from table 4.4 in the Handbook - * of Applied Cryptography [Menezes, van Oorschot, Vanstone; CRC Press 1996]; - * original paper: Damgaard, Landrock, Pomerance: Average case error estimates - * for the strong probable prime test. -- Math. Comp. 61 (1993) 177-194) */ -#define BN_prime_checks_for_size(b) ((b) >= 1300 ? 2 : \ +# define BN_GENCB_set(gencb, callback, cb_arg) { \ + BN_GENCB *tmp_gencb = (gencb); \ + tmp_gencb->ver = 2; \ + tmp_gencb->arg = (cb_arg); \ + tmp_gencb->cb.cb_2 = (callback); } + +# define BN_prime_checks 0 /* default: select number of iterations based + * on the size of the number */ + +/* + * number of Miller-Rabin iterations for an error rate of less than 2^-80 for + * random 'b'-bit input, b >= 100 (taken from table 4.4 in the Handbook of + * Applied Cryptography [Menezes, van Oorschot, Vanstone; CRC Press 1996]; + * original paper: Damgaard, Landrock, Pomerance: Average case error + * estimates for the strong probable prime test. -- Math. Comp. 61 (1993) + * 177-194) + */ +# define BN_prime_checks_for_size(b) ((b) >= 1300 ? 2 : \ (b) >= 850 ? 3 : \ (b) >= 650 ? 4 : \ (b) >= 550 ? 5 : \ @@ -399,289 +394,319 @@ int BN_GENCB_call(BN_GENCB *cb, int a, int b); (b) >= 150 ? 18 : \ /* b >= 100 */ 27) -#define BN_num_bytes(a) ((BN_num_bits(a)+7)/8) +# define BN_num_bytes(a) ((BN_num_bits(a)+7)/8) /* Note that BN_abs_is_word didn't work reliably for w == 0 until 0.9.8 */ -#define BN_abs_is_word(a,w) ((((a)->top == 1) && ((a)->d[0] == (BN_ULONG)(w))) || \ - (((w) == 0) && ((a)->top == 0))) -#define BN_is_zero(a) ((a)->top == 0) -#define BN_is_one(a) (BN_abs_is_word((a),1) && !(a)->neg) -#define BN_is_word(a,w) (BN_abs_is_word((a),(w)) && (!(w) || !(a)->neg)) -#define BN_is_odd(a) (((a)->top > 0) && ((a)->d[0] & 1)) - -#define BN_one(a) (BN_set_word((a),1)) -#define BN_zero_ex(a) \ - do { \ - BIGNUM *_tmp_bn = (a); \ - _tmp_bn->top = 0; \ - _tmp_bn->neg = 0; \ - } while(0) -#ifdef OPENSSL_NO_DEPRECATED -#define BN_zero(a) BN_zero_ex(a) -#else -#define BN_zero(a) (BN_set_word((a),0)) -#endif +# define BN_abs_is_word(a,w) ((((a)->top == 1) && ((a)->d[0] == (BN_ULONG)(w))) || \ + (((w) == 0) && ((a)->top == 0))) +# define BN_is_zero(a) ((a)->top == 0) +# define BN_is_one(a) (BN_abs_is_word((a),1) && !(a)->neg) +# define BN_is_word(a,w) (BN_abs_is_word((a),(w)) && (!(w) || !(a)->neg)) +# define BN_is_odd(a) (((a)->top > 0) && ((a)->d[0] & 1)) + +# define BN_one(a) (BN_set_word((a),1)) +# define BN_zero_ex(a) \ + do { \ + BIGNUM *_tmp_bn = (a); \ + _tmp_bn->top = 0; \ + _tmp_bn->neg = 0; \ + } while(0) +# ifdef OPENSSL_NO_DEPRECATED +# define BN_zero(a) BN_zero_ex(a) +# else +# define BN_zero(a) (BN_set_word((a),0)) +# endif const BIGNUM *BN_value_one(void); -char * BN_options(void); +char *BN_options(void); BN_CTX *BN_CTX_new(void); -#ifndef OPENSSL_NO_DEPRECATED -void BN_CTX_init(BN_CTX *c); -#endif -void BN_CTX_free(BN_CTX *c); -void BN_CTX_start(BN_CTX *ctx); +# ifndef OPENSSL_NO_DEPRECATED +void BN_CTX_init(BN_CTX *c); +# endif +void BN_CTX_free(BN_CTX *c); +void BN_CTX_start(BN_CTX *ctx); BIGNUM *BN_CTX_get(BN_CTX *ctx); -void BN_CTX_end(BN_CTX *ctx); -int BN_rand(BIGNUM *rnd, int bits, int top,int bottom); -int BN_pseudo_rand(BIGNUM *rnd, int bits, int top,int bottom); -int BN_rand_range(BIGNUM *rnd, const BIGNUM *range); -int BN_pseudo_rand_range(BIGNUM *rnd, const BIGNUM *range); -int BN_num_bits(const BIGNUM *a); -int BN_num_bits_word(BN_ULONG); +void BN_CTX_end(BN_CTX *ctx); +int BN_rand(BIGNUM *rnd, int bits, int top, int bottom); +int BN_pseudo_rand(BIGNUM *rnd, int bits, int top, int bottom); +int BN_rand_range(BIGNUM *rnd, const BIGNUM *range); +int BN_pseudo_rand_range(BIGNUM *rnd, const BIGNUM *range); +int BN_num_bits(const BIGNUM *a); +int BN_num_bits_word(BN_ULONG); BIGNUM *BN_new(void); -void BN_init(BIGNUM *); -void BN_clear_free(BIGNUM *a); +void BN_init(BIGNUM *); +void BN_clear_free(BIGNUM *a); BIGNUM *BN_copy(BIGNUM *a, const BIGNUM *b); -void BN_swap(BIGNUM *a, BIGNUM *b); -BIGNUM *BN_bin2bn(const unsigned char *s,int len,BIGNUM *ret); -int BN_bn2bin(const BIGNUM *a, unsigned char *to); -BIGNUM *BN_mpi2bn(const unsigned char *s,int len,BIGNUM *ret); -int BN_bn2mpi(const BIGNUM *a, unsigned char *to); -int BN_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); -int BN_usub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); -int BN_uadd(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); -int BN_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); -int BN_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, BN_CTX *ctx); -int BN_sqr(BIGNUM *r, const BIGNUM *a,BN_CTX *ctx); +void BN_swap(BIGNUM *a, BIGNUM *b); +BIGNUM *BN_bin2bn(const unsigned char *s, int len, BIGNUM *ret); +int BN_bn2bin(const BIGNUM *a, unsigned char *to); +BIGNUM *BN_mpi2bn(const unsigned char *s, int len, BIGNUM *ret); +int BN_bn2mpi(const BIGNUM *a, unsigned char *to); +int BN_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); +int BN_usub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); +int BN_uadd(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); +int BN_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); +int BN_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, BN_CTX *ctx); +int BN_sqr(BIGNUM *r, const BIGNUM *a, BN_CTX *ctx); /** BN_set_negative sets sign of a BIGNUM * \param b pointer to the BIGNUM object - * \param n 0 if the BIGNUM b should be positive and a value != 0 otherwise + * \param n 0 if the BIGNUM b should be positive and a value != 0 otherwise */ -void BN_set_negative(BIGNUM *b, int n); +void BN_set_negative(BIGNUM *b, int n); /** BN_is_negative returns 1 if the BIGNUM is negative * \param a pointer to the BIGNUM object * \return 1 if a < 0 and 0 otherwise */ -#define BN_is_negative(a) ((a)->neg != 0) - -int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, - BN_CTX *ctx); -#define BN_mod(rem,m,d,ctx) BN_div(NULL,(rem),(m),(d),(ctx)) -int BN_nnmod(BIGNUM *r, const BIGNUM *m, const BIGNUM *d, BN_CTX *ctx); -int BN_mod_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, BN_CTX *ctx); -int BN_mod_add_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m); -int BN_mod_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, BN_CTX *ctx); -int BN_mod_sub_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m); -int BN_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const BIGNUM *m, BN_CTX *ctx); -int BN_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *m, BN_CTX *ctx); -int BN_mod_lshift1(BIGNUM *r, const BIGNUM *a, const BIGNUM *m, BN_CTX *ctx); -int BN_mod_lshift1_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *m); -int BN_mod_lshift(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m, BN_CTX *ctx); -int BN_mod_lshift_quick(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m); +# define BN_is_negative(a) ((a)->neg != 0) + +int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, + BN_CTX *ctx); +# define BN_mod(rem,m,d,ctx) BN_div(NULL,(rem),(m),(d),(ctx)) +int BN_nnmod(BIGNUM *r, const BIGNUM *m, const BIGNUM *d, BN_CTX *ctx); +int BN_mod_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, + BN_CTX *ctx); +int BN_mod_add_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *m); +int BN_mod_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, + BN_CTX *ctx); +int BN_mod_sub_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *m); +int BN_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, + BN_CTX *ctx); +int BN_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *m, BN_CTX *ctx); +int BN_mod_lshift1(BIGNUM *r, const BIGNUM *a, const BIGNUM *m, BN_CTX *ctx); +int BN_mod_lshift1_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *m); +int BN_mod_lshift(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m, + BN_CTX *ctx); +int BN_mod_lshift_quick(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m); BN_ULONG BN_mod_word(const BIGNUM *a, BN_ULONG w); BN_ULONG BN_div_word(BIGNUM *a, BN_ULONG w); -int BN_mul_word(BIGNUM *a, BN_ULONG w); -int BN_add_word(BIGNUM *a, BN_ULONG w); -int BN_sub_word(BIGNUM *a, BN_ULONG w); -int BN_set_word(BIGNUM *a, BN_ULONG w); +int BN_mul_word(BIGNUM *a, BN_ULONG w); +int BN_add_word(BIGNUM *a, BN_ULONG w); +int BN_sub_word(BIGNUM *a, BN_ULONG w); +int BN_set_word(BIGNUM *a, BN_ULONG w); BN_ULONG BN_get_word(const BIGNUM *a); -int BN_cmp(const BIGNUM *a, const BIGNUM *b); -void BN_free(BIGNUM *a); -int BN_is_bit_set(const BIGNUM *a, int n); -int BN_lshift(BIGNUM *r, const BIGNUM *a, int n); -int BN_lshift1(BIGNUM *r, const BIGNUM *a); -int BN_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p,BN_CTX *ctx); - -int BN_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m,BN_CTX *ctx); -int BN_mod_exp_mont(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx); +int BN_cmp(const BIGNUM *a, const BIGNUM *b); +void BN_free(BIGNUM *a); +int BN_is_bit_set(const BIGNUM *a, int n); +int BN_lshift(BIGNUM *r, const BIGNUM *a, int n); +int BN_lshift1(BIGNUM *r, const BIGNUM *a); +int BN_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx); + +int BN_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + const BIGNUM *m, BN_CTX *ctx); +int BN_mod_exp_mont(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx); int BN_mod_exp_mont_consttime(BIGNUM *rr, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont); -int BN_mod_exp_mont_word(BIGNUM *r, BN_ULONG a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx); -int BN_mod_exp2_mont(BIGNUM *r, const BIGNUM *a1, const BIGNUM *p1, - const BIGNUM *a2, const BIGNUM *p2,const BIGNUM *m, - BN_CTX *ctx,BN_MONT_CTX *m_ctx); -int BN_mod_exp_simple(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m,BN_CTX *ctx); - -int BN_mask_bits(BIGNUM *a,int n); -#ifndef OPENSSL_NO_FP_API -int BN_print_fp(FILE *fp, const BIGNUM *a); -#endif -#ifdef HEADER_BIO_H -int BN_print(BIO *fp, const BIGNUM *a); -#else -int BN_print(void *fp, const BIGNUM *a); -#endif -int BN_reciprocal(BIGNUM *r, const BIGNUM *m, int len, BN_CTX *ctx); -int BN_rshift(BIGNUM *r, const BIGNUM *a, int n); -int BN_rshift1(BIGNUM *r, const BIGNUM *a); -void BN_clear(BIGNUM *a); + const BIGNUM *m, BN_CTX *ctx, + BN_MONT_CTX *in_mont); +int BN_mod_exp_mont_word(BIGNUM *r, BN_ULONG a, const BIGNUM *p, + const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx); +int BN_mod_exp2_mont(BIGNUM *r, const BIGNUM *a1, const BIGNUM *p1, + const BIGNUM *a2, const BIGNUM *p2, const BIGNUM *m, + BN_CTX *ctx, BN_MONT_CTX *m_ctx); +int BN_mod_exp_simple(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + const BIGNUM *m, BN_CTX *ctx); + +int BN_mask_bits(BIGNUM *a, int n); +# ifndef OPENSSL_NO_FP_API +int BN_print_fp(FILE *fp, const BIGNUM *a); +# endif +# ifdef HEADER_BIO_H +int BN_print(BIO *fp, const BIGNUM *a); +# else +int BN_print(void *fp, const BIGNUM *a); +# endif +int BN_reciprocal(BIGNUM *r, const BIGNUM *m, int len, BN_CTX *ctx); +int BN_rshift(BIGNUM *r, const BIGNUM *a, int n); +int BN_rshift1(BIGNUM *r, const BIGNUM *a); +void BN_clear(BIGNUM *a); BIGNUM *BN_dup(const BIGNUM *a); -int BN_ucmp(const BIGNUM *a, const BIGNUM *b); -int BN_set_bit(BIGNUM *a, int n); -int BN_clear_bit(BIGNUM *a, int n); -char * BN_bn2hex(const BIGNUM *a); -char * BN_bn2dec(const BIGNUM *a); -int BN_hex2bn(BIGNUM **a, const char *str); -int BN_dec2bn(BIGNUM **a, const char *str); -int BN_asc2bn(BIGNUM **a, const char *str); -int BN_gcd(BIGNUM *r,const BIGNUM *a,const BIGNUM *b,BN_CTX *ctx); -int BN_kronecker(const BIGNUM *a,const BIGNUM *b,BN_CTX *ctx); /* returns -2 for error */ +int BN_ucmp(const BIGNUM *a, const BIGNUM *b); +int BN_set_bit(BIGNUM *a, int n); +int BN_clear_bit(BIGNUM *a, int n); +char *BN_bn2hex(const BIGNUM *a); +char *BN_bn2dec(const BIGNUM *a); +int BN_hex2bn(BIGNUM **a, const char *str); +int BN_dec2bn(BIGNUM **a, const char *str); +int BN_asc2bn(BIGNUM **a, const char *str); +int BN_gcd(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, BN_CTX *ctx); +int BN_kronecker(const BIGNUM *a, const BIGNUM *b, BN_CTX *ctx); /* returns + * -2 for + * error */ BIGNUM *BN_mod_inverse(BIGNUM *ret, - const BIGNUM *a, const BIGNUM *n,BN_CTX *ctx); + const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx); BIGNUM *BN_mod_sqrt(BIGNUM *ret, - const BIGNUM *a, const BIGNUM *n,BN_CTX *ctx); + const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx); -void BN_consttime_swap(BN_ULONG swap, BIGNUM *a, BIGNUM *b, int nwords); +void BN_consttime_swap(BN_ULONG swap, BIGNUM *a, BIGNUM *b, int nwords); /* Deprecated versions */ -#ifndef OPENSSL_NO_DEPRECATED -BIGNUM *BN_generate_prime(BIGNUM *ret,int bits,int safe, - const BIGNUM *add, const BIGNUM *rem, - void (*callback)(int,int,void *),void *cb_arg); -int BN_is_prime(const BIGNUM *p,int nchecks, - void (*callback)(int,int,void *), - BN_CTX *ctx,void *cb_arg); -int BN_is_prime_fasttest(const BIGNUM *p,int nchecks, - void (*callback)(int,int,void *),BN_CTX *ctx,void *cb_arg, - int do_trial_division); -#endif /* !defined(OPENSSL_NO_DEPRECATED) */ +# ifndef OPENSSL_NO_DEPRECATED +BIGNUM *BN_generate_prime(BIGNUM *ret, int bits, int safe, + const BIGNUM *add, const BIGNUM *rem, + void (*callback) (int, int, void *), void *cb_arg); +int BN_is_prime(const BIGNUM *p, int nchecks, + void (*callback) (int, int, void *), + BN_CTX *ctx, void *cb_arg); +int BN_is_prime_fasttest(const BIGNUM *p, int nchecks, + void (*callback) (int, int, void *), BN_CTX *ctx, + void *cb_arg, int do_trial_division); +# endif /* !defined(OPENSSL_NO_DEPRECATED) */ /* Newer versions */ -int BN_generate_prime_ex(BIGNUM *ret,int bits,int safe, const BIGNUM *add, - const BIGNUM *rem, BN_GENCB *cb); -int BN_is_prime_ex(const BIGNUM *p,int nchecks, BN_CTX *ctx, BN_GENCB *cb); -int BN_is_prime_fasttest_ex(const BIGNUM *p,int nchecks, BN_CTX *ctx, - int do_trial_division, BN_GENCB *cb); +int BN_generate_prime_ex(BIGNUM *ret, int bits, int safe, const BIGNUM *add, + const BIGNUM *rem, BN_GENCB *cb); +int BN_is_prime_ex(const BIGNUM *p, int nchecks, BN_CTX *ctx, BN_GENCB *cb); +int BN_is_prime_fasttest_ex(const BIGNUM *p, int nchecks, BN_CTX *ctx, + int do_trial_division, BN_GENCB *cb); int BN_X931_generate_Xpq(BIGNUM *Xp, BIGNUM *Xq, int nbits, BN_CTX *ctx); int BN_X931_derive_prime_ex(BIGNUM *p, BIGNUM *p1, BIGNUM *p2, - const BIGNUM *Xp, const BIGNUM *Xp1, const BIGNUM *Xp2, - const BIGNUM *e, BN_CTX *ctx, BN_GENCB *cb); -int BN_X931_generate_prime_ex(BIGNUM *p, BIGNUM *p1, BIGNUM *p2, - BIGNUM *Xp1, BIGNUM *Xp2, - const BIGNUM *Xp, - const BIGNUM *e, BN_CTX *ctx, - BN_GENCB *cb); - -BN_MONT_CTX *BN_MONT_CTX_new(void ); + const BIGNUM *Xp, const BIGNUM *Xp1, + const BIGNUM *Xp2, const BIGNUM *e, BN_CTX *ctx, + BN_GENCB *cb); +int BN_X931_generate_prime_ex(BIGNUM *p, BIGNUM *p1, BIGNUM *p2, BIGNUM *Xp1, + BIGNUM *Xp2, const BIGNUM *Xp, const BIGNUM *e, + BN_CTX *ctx, BN_GENCB *cb); + +BN_MONT_CTX *BN_MONT_CTX_new(void); void BN_MONT_CTX_init(BN_MONT_CTX *ctx); -int BN_mod_mul_montgomery(BIGNUM *r,const BIGNUM *a,const BIGNUM *b, - BN_MONT_CTX *mont, BN_CTX *ctx); -#define BN_to_montgomery(r,a,mont,ctx) BN_mod_mul_montgomery(\ - (r),(a),&((mont)->RR),(mont),(ctx)) -int BN_from_montgomery(BIGNUM *r,const BIGNUM *a, - BN_MONT_CTX *mont, BN_CTX *ctx); +int BN_mod_mul_montgomery(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + BN_MONT_CTX *mont, BN_CTX *ctx); +# define BN_to_montgomery(r,a,mont,ctx) BN_mod_mul_montgomery(\ + (r),(a),&((mont)->RR),(mont),(ctx)) +int BN_from_montgomery(BIGNUM *r, const BIGNUM *a, + BN_MONT_CTX *mont, BN_CTX *ctx); void BN_MONT_CTX_free(BN_MONT_CTX *mont); -int BN_MONT_CTX_set(BN_MONT_CTX *mont,const BIGNUM *mod,BN_CTX *ctx); -BN_MONT_CTX *BN_MONT_CTX_copy(BN_MONT_CTX *to,BN_MONT_CTX *from); +int BN_MONT_CTX_set(BN_MONT_CTX *mont, const BIGNUM *mod, BN_CTX *ctx); +BN_MONT_CTX *BN_MONT_CTX_copy(BN_MONT_CTX *to, BN_MONT_CTX *from); BN_MONT_CTX *BN_MONT_CTX_set_locked(BN_MONT_CTX **pmont, int lock, - const BIGNUM *mod, BN_CTX *ctx); + const BIGNUM *mod, BN_CTX *ctx); /* BN_BLINDING flags */ -#define BN_BLINDING_NO_UPDATE 0x00000001 -#define BN_BLINDING_NO_RECREATE 0x00000002 +# define BN_BLINDING_NO_UPDATE 0x00000001 +# define BN_BLINDING_NO_RECREATE 0x00000002 BN_BLINDING *BN_BLINDING_new(const BIGNUM *A, const BIGNUM *Ai, BIGNUM *mod); void BN_BLINDING_free(BN_BLINDING *b); -int BN_BLINDING_update(BN_BLINDING *b,BN_CTX *ctx); +int BN_BLINDING_update(BN_BLINDING *b, BN_CTX *ctx); int BN_BLINDING_convert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx); int BN_BLINDING_invert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx); int BN_BLINDING_convert_ex(BIGNUM *n, BIGNUM *r, BN_BLINDING *b, BN_CTX *); -int BN_BLINDING_invert_ex(BIGNUM *n, const BIGNUM *r, BN_BLINDING *b, BN_CTX *); -#ifndef OPENSSL_NO_DEPRECATED +int BN_BLINDING_invert_ex(BIGNUM *n, const BIGNUM *r, BN_BLINDING *b, + BN_CTX *); +# ifndef OPENSSL_NO_DEPRECATED unsigned long BN_BLINDING_get_thread_id(const BN_BLINDING *); void BN_BLINDING_set_thread_id(BN_BLINDING *, unsigned long); -#endif +# endif CRYPTO_THREADID *BN_BLINDING_thread_id(BN_BLINDING *); unsigned long BN_BLINDING_get_flags(const BN_BLINDING *); void BN_BLINDING_set_flags(BN_BLINDING *, unsigned long); BN_BLINDING *BN_BLINDING_create_param(BN_BLINDING *b, - const BIGNUM *e, BIGNUM *m, BN_CTX *ctx, - int (*bn_mod_exp)(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx), - BN_MONT_CTX *m_ctx); - -#ifndef OPENSSL_NO_DEPRECATED -void BN_set_params(int mul,int high,int low,int mont); -int BN_get_params(int which); /* 0, mul, 1 high, 2 low, 3 mont */ -#endif + const BIGNUM *e, BIGNUM *m, BN_CTX *ctx, + int (*bn_mod_exp) (BIGNUM *r, + const BIGNUM *a, + const BIGNUM *p, + const BIGNUM *m, + BN_CTX *ctx, + BN_MONT_CTX *m_ctx), + BN_MONT_CTX *m_ctx); + +# ifndef OPENSSL_NO_DEPRECATED +void BN_set_params(int mul, int high, int low, int mont); +int BN_get_params(int which); /* 0, mul, 1 high, 2 low, 3 mont */ +# endif -void BN_RECP_CTX_init(BN_RECP_CTX *recp); +void BN_RECP_CTX_init(BN_RECP_CTX *recp); BN_RECP_CTX *BN_RECP_CTX_new(void); -void BN_RECP_CTX_free(BN_RECP_CTX *recp); -int BN_RECP_CTX_set(BN_RECP_CTX *recp,const BIGNUM *rdiv,BN_CTX *ctx); -int BN_mod_mul_reciprocal(BIGNUM *r, const BIGNUM *x, const BIGNUM *y, - BN_RECP_CTX *recp,BN_CTX *ctx); -int BN_mod_exp_recp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx); -int BN_div_recp(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, - BN_RECP_CTX *recp, BN_CTX *ctx); - -#ifndef OPENSSL_NO_EC2M - -/* Functions for arithmetic over binary polynomials represented by BIGNUMs. - * +void BN_RECP_CTX_free(BN_RECP_CTX *recp); +int BN_RECP_CTX_set(BN_RECP_CTX *recp, const BIGNUM *rdiv, BN_CTX *ctx); +int BN_mod_mul_reciprocal(BIGNUM *r, const BIGNUM *x, const BIGNUM *y, + BN_RECP_CTX *recp, BN_CTX *ctx); +int BN_mod_exp_recp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + const BIGNUM *m, BN_CTX *ctx); +int BN_div_recp(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, + BN_RECP_CTX *recp, BN_CTX *ctx); + +# ifndef OPENSSL_NO_EC2M + +/* + * Functions for arithmetic over binary polynomials represented by BIGNUMs. * The BIGNUM::neg property of BIGNUMs representing binary polynomials is - * ignored. - * - * Note that input arguments are not const so that their bit arrays can - * be expanded to the appropriate size if needed. + * ignored. Note that input arguments are not const so that their bit arrays + * can be expanded to the appropriate size if needed. */ -int BN_GF2m_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); /*r = a + b*/ -#define BN_GF2m_sub(r, a, b) BN_GF2m_add(r, a, b) -int BN_GF2m_mod(BIGNUM *r, const BIGNUM *a, const BIGNUM *p); /*r=a mod p*/ -int BN_GF2m_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const BIGNUM *p, BN_CTX *ctx); /* r = (a * b) mod p */ -int BN_GF2m_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - BN_CTX *ctx); /* r = (a * a) mod p */ -int BN_GF2m_mod_inv(BIGNUM *r, const BIGNUM *b, const BIGNUM *p, - BN_CTX *ctx); /* r = (1 / b) mod p */ -int BN_GF2m_mod_div(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const BIGNUM *p, BN_CTX *ctx); /* r = (a / b) mod p */ -int BN_GF2m_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const BIGNUM *p, BN_CTX *ctx); /* r = (a ^ b) mod p */ -int BN_GF2m_mod_sqrt(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - BN_CTX *ctx); /* r = sqrt(a) mod p */ -int BN_GF2m_mod_solve_quad(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - BN_CTX *ctx); /* r^2 + r = a mod p */ -#define BN_GF2m_cmp(a, b) BN_ucmp((a), (b)) -/* Some functions allow for representation of the irreducible polynomials +/* + * r = a + b + */ +int BN_GF2m_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b); +# define BN_GF2m_sub(r, a, b) BN_GF2m_add(r, a, b) +/* + * r=a mod p + */ +int BN_GF2m_mod(BIGNUM *r, const BIGNUM *a, const BIGNUM *p); +/* r = (a * b) mod p */ +int BN_GF2m_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *p, BN_CTX *ctx); +/* r = (a * a) mod p */ +int BN_GF2m_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx); +/* r = (1 / b) mod p */ +int BN_GF2m_mod_inv(BIGNUM *r, const BIGNUM *b, const BIGNUM *p, BN_CTX *ctx); +/* r = (a / b) mod p */ +int BN_GF2m_mod_div(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *p, BN_CTX *ctx); +/* r = (a ^ b) mod p */ +int BN_GF2m_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *p, BN_CTX *ctx); +/* r = sqrt(a) mod p */ +int BN_GF2m_mod_sqrt(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + BN_CTX *ctx); +/* r^2 + r = a mod p */ +int BN_GF2m_mod_solve_quad(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + BN_CTX *ctx); +# define BN_GF2m_cmp(a, b) BN_ucmp((a), (b)) +/*- + * Some functions allow for representation of the irreducible polynomials * as an unsigned int[], say p. The irreducible f(t) is then of the form: * t^p[0] + t^p[1] + ... + t^p[k] * where m = p[0] > p[1] > ... > p[k] = 0. */ -int BN_GF2m_mod_arr(BIGNUM *r, const BIGNUM *a, const int p[]); - /* r = a mod p */ -int BN_GF2m_mod_mul_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const int p[], BN_CTX *ctx); /* r = (a * b) mod p */ -int BN_GF2m_mod_sqr_arr(BIGNUM *r, const BIGNUM *a, const int p[], - BN_CTX *ctx); /* r = (a * a) mod p */ -int BN_GF2m_mod_inv_arr(BIGNUM *r, const BIGNUM *b, const int p[], - BN_CTX *ctx); /* r = (1 / b) mod p */ -int BN_GF2m_mod_div_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const int p[], BN_CTX *ctx); /* r = (a / b) mod p */ -int BN_GF2m_mod_exp_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - const int p[], BN_CTX *ctx); /* r = (a ^ b) mod p */ -int BN_GF2m_mod_sqrt_arr(BIGNUM *r, const BIGNUM *a, - const int p[], BN_CTX *ctx); /* r = sqrt(a) mod p */ -int BN_GF2m_mod_solve_quad_arr(BIGNUM *r, const BIGNUM *a, - const int p[], BN_CTX *ctx); /* r^2 + r = a mod p */ -int BN_GF2m_poly2arr(const BIGNUM *a, int p[], int max); -int BN_GF2m_arr2poly(const int p[], BIGNUM *a); +/* r = a mod p */ +int BN_GF2m_mod_arr(BIGNUM *r, const BIGNUM *a, const int p[]); +/* r = (a * b) mod p */ +int BN_GF2m_mod_mul_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const int p[], BN_CTX *ctx); +/* r = (a * a) mod p */ +int BN_GF2m_mod_sqr_arr(BIGNUM *r, const BIGNUM *a, const int p[], + BN_CTX *ctx); +/* r = (1 / b) mod p */ +int BN_GF2m_mod_inv_arr(BIGNUM *r, const BIGNUM *b, const int p[], + BN_CTX *ctx); +/* r = (a / b) mod p */ +int BN_GF2m_mod_div_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const int p[], BN_CTX *ctx); +/* r = (a ^ b) mod p */ +int BN_GF2m_mod_exp_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const int p[], BN_CTX *ctx); +/* r = sqrt(a) mod p */ +int BN_GF2m_mod_sqrt_arr(BIGNUM *r, const BIGNUM *a, + const int p[], BN_CTX *ctx); +/* r^2 + r = a mod p */ +int BN_GF2m_mod_solve_quad_arr(BIGNUM *r, const BIGNUM *a, + const int p[], BN_CTX *ctx); +int BN_GF2m_poly2arr(const BIGNUM *a, int p[], int max); +int BN_GF2m_arr2poly(const int p[], BIGNUM *a); -#endif +# endif -/* faster mod functions for the 'NIST primes' - * 0 <= a < p^2 */ +/* + * faster mod functions for the 'NIST primes' 0 <= a < p^2 + */ int BN_nist_mod_192(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx); int BN_nist_mod_224(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx); int BN_nist_mod_256(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx); @@ -696,15 +721,16 @@ const BIGNUM *BN_get0_nist_prime_521(void); /* library internal functions */ -#define bn_expand(a,bits) ((((((bits+BN_BITS2-1))/BN_BITS2)) <= (a)->dmax)?\ - (a):bn_expand2((a),(bits+BN_BITS2-1)/BN_BITS2)) -#define bn_wexpand(a,words) (((words) <= (a)->dmax)?(a):bn_expand2((a),(words))) +# define bn_expand(a,bits) ((((((bits+BN_BITS2-1))/BN_BITS2)) <= (a)->dmax)?\ + (a):bn_expand2((a),(bits+BN_BITS2-1)/BN_BITS2)) +# define bn_wexpand(a,words) (((words) <= (a)->dmax)?(a):bn_expand2((a),(words))) BIGNUM *bn_expand2(BIGNUM *a, int words); -#ifndef OPENSSL_NO_DEPRECATED +# ifndef OPENSSL_NO_DEPRECATED BIGNUM *bn_dup_expand(const BIGNUM *a, int words); /* unused */ -#endif +# endif -/* Bignum consistency macros +/*- + * Bignum consistency macros * There is one "API" macro, bn_fix_top(), for stripping leading zeroes from * bignum data after direct manipulations on the data. There is also an * "internal" macro, bn_check_top(), for verifying that there are no leading @@ -732,86 +758,91 @@ BIGNUM *bn_dup_expand(const BIGNUM *a, int words); /* unused */ * coverage for openssl's own code. */ -#ifdef BN_DEBUG +# ifdef BN_DEBUG /* We only need assert() when debugging */ -#include <assert.h> +# include <assert.h> -#ifdef BN_DEBUG_RAND +# ifdef BN_DEBUG_RAND /* To avoid "make update" cvs wars due to BN_DEBUG, use some tricks */ -#ifndef RAND_pseudo_bytes -int RAND_pseudo_bytes(unsigned char *buf,int num); -#define BN_DEBUG_TRIX -#endif -#define bn_pollute(a) \ - do { \ - const BIGNUM *_bnum1 = (a); \ - if(_bnum1->top < _bnum1->dmax) { \ - unsigned char _tmp_char; \ - /* We cast away const without the compiler knowing, any \ - * *genuinely* constant variables that aren't mutable \ - * wouldn't be constructed with top!=dmax. */ \ - BN_ULONG *_not_const; \ - memcpy(&_not_const, &_bnum1->d, sizeof(BN_ULONG*)); \ - RAND_pseudo_bytes(&_tmp_char, 1); \ - memset((unsigned char *)(_not_const + _bnum1->top), _tmp_char, \ - (_bnum1->dmax - _bnum1->top) * sizeof(BN_ULONG)); \ - } \ - } while(0) -#ifdef BN_DEBUG_TRIX -#undef RAND_pseudo_bytes -#endif -#else -#define bn_pollute(a) -#endif -#define bn_check_top(a) \ - do { \ - const BIGNUM *_bnum2 = (a); \ - if (_bnum2 != NULL) { \ - assert((_bnum2->top == 0) || \ - (_bnum2->d[_bnum2->top - 1] != 0)); \ - bn_pollute(_bnum2); \ - } \ - } while(0) - -#define bn_fix_top(a) bn_check_top(a) - -#define bn_check_size(bn, bits) bn_wcheck_size(bn, ((bits+BN_BITS2-1))/BN_BITS2) -#define bn_wcheck_size(bn, words) \ - do { \ - const BIGNUM *_bnum2 = (bn); \ - assert(words <= (_bnum2)->dmax && words >= (_bnum2)->top); \ - } while(0) - -#else /* !BN_DEBUG */ - -#define bn_pollute(a) -#define bn_check_top(a) -#define bn_fix_top(a) bn_correct_top(a) -#define bn_check_size(bn, bits) -#define bn_wcheck_size(bn, words) +# ifndef RAND_pseudo_bytes +int RAND_pseudo_bytes(unsigned char *buf, int num); +# define BN_DEBUG_TRIX +# endif +# define bn_pollute(a) \ + do { \ + const BIGNUM *_bnum1 = (a); \ + if(_bnum1->top < _bnum1->dmax) { \ + unsigned char _tmp_char; \ + /* We cast away const without the compiler knowing, any \ + * *genuinely* constant variables that aren't mutable \ + * wouldn't be constructed with top!=dmax. */ \ + BN_ULONG *_not_const; \ + memcpy(&_not_const, &_bnum1->d, sizeof(BN_ULONG*)); \ + RAND_pseudo_bytes(&_tmp_char, 1); \ + memset((unsigned char *)(_not_const + _bnum1->top), _tmp_char, \ + (_bnum1->dmax - _bnum1->top) * sizeof(BN_ULONG)); \ + } \ + } while(0) +# ifdef BN_DEBUG_TRIX +# undef RAND_pseudo_bytes +# endif +# else +# define bn_pollute(a) +# endif +# define bn_check_top(a) \ + do { \ + const BIGNUM *_bnum2 = (a); \ + if (_bnum2 != NULL) { \ + assert((_bnum2->top == 0) || \ + (_bnum2->d[_bnum2->top - 1] != 0)); \ + bn_pollute(_bnum2); \ + } \ + } while(0) + +# define bn_fix_top(a) bn_check_top(a) + +# define bn_check_size(bn, bits) bn_wcheck_size(bn, ((bits+BN_BITS2-1))/BN_BITS2) +# define bn_wcheck_size(bn, words) \ + do { \ + const BIGNUM *_bnum2 = (bn); \ + assert((words) <= (_bnum2)->dmax && (words) >= (_bnum2)->top); \ + /* avoid unused variable warning with NDEBUG */ \ + (void)(_bnum2); \ + } while(0) + +# else /* !BN_DEBUG */ + +# define bn_pollute(a) +# define bn_check_top(a) +# define bn_fix_top(a) bn_correct_top(a) +# define bn_check_size(bn, bits) +# define bn_wcheck_size(bn, words) -#endif +# endif -#define bn_correct_top(a) \ +# define bn_correct_top(a) \ { \ BN_ULONG *ftl; \ - int tmp_top = (a)->top; \ - if (tmp_top > 0) \ - { \ - for (ftl= &((a)->d[tmp_top-1]); tmp_top > 0; tmp_top--) \ - if (*(ftl--)) break; \ - (a)->top = tmp_top; \ - } \ - bn_pollute(a); \ - } - -BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w); + int tmp_top = (a)->top; \ + if (tmp_top > 0) \ + { \ + for (ftl= &((a)->d[tmp_top-1]); tmp_top > 0; tmp_top--) \ + if (*(ftl--)) break; \ + (a)->top = tmp_top; \ + } \ + bn_pollute(a); \ + } + +BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, + BN_ULONG w); BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w); -void bn_sqr_words(BN_ULONG *rp, const BN_ULONG *ap, int num); +void bn_sqr_words(BN_ULONG *rp, const BN_ULONG *ap, int num); BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d); -BN_ULONG bn_add_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int num); -BN_ULONG bn_sub_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,int num); +BN_ULONG bn_add_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + int num); +BN_ULONG bn_sub_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + int num); /* Primes from RFC 2409 */ BIGNUM *get_rfc2409_prime_768(BIGNUM *bn); @@ -825,10 +856,11 @@ BIGNUM *get_rfc3526_prime_4096(BIGNUM *bn); BIGNUM *get_rfc3526_prime_6144(BIGNUM *bn); BIGNUM *get_rfc3526_prime_8192(BIGNUM *bn); -int BN_bntest_rand(BIGNUM *rnd, int bits, int top,int bottom); +int BN_bntest_rand(BIGNUM *rnd, int bits, int top, int bottom); /* BEGIN ERROR CODES */ -/* The following lines are auto generated by the script mkerr.pl. Any changes +/* + * The following lines are auto generated by the script mkerr.pl. Any changes * made after this point may be overwritten when the script is next run. */ void ERR_load_BN_strings(void); @@ -836,65 +868,65 @@ void ERR_load_BN_strings(void); /* Error codes for the BN functions. */ /* Function codes. */ -#define BN_F_BNRAND 127 -#define BN_F_BN_BLINDING_CONVERT_EX 100 -#define BN_F_BN_BLINDING_CREATE_PARAM 128 -#define BN_F_BN_BLINDING_INVERT_EX 101 -#define BN_F_BN_BLINDING_NEW 102 -#define BN_F_BN_BLINDING_UPDATE 103 -#define BN_F_BN_BN2DEC 104 -#define BN_F_BN_BN2HEX 105 -#define BN_F_BN_CTX_GET 116 -#define BN_F_BN_CTX_NEW 106 -#define BN_F_BN_CTX_START 129 -#define BN_F_BN_DIV 107 -#define BN_F_BN_DIV_NO_BRANCH 138 -#define BN_F_BN_DIV_RECP 130 -#define BN_F_BN_EXP 123 -#define BN_F_BN_EXPAND2 108 -#define BN_F_BN_EXPAND_INTERNAL 120 -#define BN_F_BN_GF2M_MOD 131 -#define BN_F_BN_GF2M_MOD_EXP 132 -#define BN_F_BN_GF2M_MOD_MUL 133 -#define BN_F_BN_GF2M_MOD_SOLVE_QUAD 134 -#define BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR 135 -#define BN_F_BN_GF2M_MOD_SQR 136 -#define BN_F_BN_GF2M_MOD_SQRT 137 -#define BN_F_BN_MOD_EXP2_MONT 118 -#define BN_F_BN_MOD_EXP_MONT 109 -#define BN_F_BN_MOD_EXP_MONT_CONSTTIME 124 -#define BN_F_BN_MOD_EXP_MONT_WORD 117 -#define BN_F_BN_MOD_EXP_RECP 125 -#define BN_F_BN_MOD_EXP_SIMPLE 126 -#define BN_F_BN_MOD_INVERSE 110 -#define BN_F_BN_MOD_INVERSE_NO_BRANCH 139 -#define BN_F_BN_MOD_LSHIFT_QUICK 119 -#define BN_F_BN_MOD_MUL_RECIPROCAL 111 -#define BN_F_BN_MOD_SQRT 121 -#define BN_F_BN_MPI2BN 112 -#define BN_F_BN_NEW 113 -#define BN_F_BN_RAND 114 -#define BN_F_BN_RAND_RANGE 122 -#define BN_F_BN_USUB 115 +# define BN_F_BNRAND 127 +# define BN_F_BN_BLINDING_CONVERT_EX 100 +# define BN_F_BN_BLINDING_CREATE_PARAM 128 +# define BN_F_BN_BLINDING_INVERT_EX 101 +# define BN_F_BN_BLINDING_NEW 102 +# define BN_F_BN_BLINDING_UPDATE 103 +# define BN_F_BN_BN2DEC 104 +# define BN_F_BN_BN2HEX 105 +# define BN_F_BN_CTX_GET 116 +# define BN_F_BN_CTX_NEW 106 +# define BN_F_BN_CTX_START 129 +# define BN_F_BN_DIV 107 +# define BN_F_BN_DIV_NO_BRANCH 138 +# define BN_F_BN_DIV_RECP 130 +# define BN_F_BN_EXP 123 +# define BN_F_BN_EXPAND2 108 +# define BN_F_BN_EXPAND_INTERNAL 120 +# define BN_F_BN_GF2M_MOD 131 +# define BN_F_BN_GF2M_MOD_EXP 132 +# define BN_F_BN_GF2M_MOD_MUL 133 +# define BN_F_BN_GF2M_MOD_SOLVE_QUAD 134 +# define BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR 135 +# define BN_F_BN_GF2M_MOD_SQR 136 +# define BN_F_BN_GF2M_MOD_SQRT 137 +# define BN_F_BN_MOD_EXP2_MONT 118 +# define BN_F_BN_MOD_EXP_MONT 109 +# define BN_F_BN_MOD_EXP_MONT_CONSTTIME 124 +# define BN_F_BN_MOD_EXP_MONT_WORD 117 +# define BN_F_BN_MOD_EXP_RECP 125 +# define BN_F_BN_MOD_EXP_SIMPLE 126 +# define BN_F_BN_MOD_INVERSE 110 +# define BN_F_BN_MOD_INVERSE_NO_BRANCH 139 +# define BN_F_BN_MOD_LSHIFT_QUICK 119 +# define BN_F_BN_MOD_MUL_RECIPROCAL 111 +# define BN_F_BN_MOD_SQRT 121 +# define BN_F_BN_MPI2BN 112 +# define BN_F_BN_NEW 113 +# define BN_F_BN_RAND 114 +# define BN_F_BN_RAND_RANGE 122 +# define BN_F_BN_USUB 115 /* Reason codes. */ -#define BN_R_ARG2_LT_ARG3 100 -#define BN_R_BAD_RECIPROCAL 101 -#define BN_R_BIGNUM_TOO_LONG 114 -#define BN_R_CALLED_WITH_EVEN_MODULUS 102 -#define BN_R_DIV_BY_ZERO 103 -#define BN_R_ENCODING_ERROR 104 -#define BN_R_EXPAND_ON_STATIC_BIGNUM_DATA 105 -#define BN_R_INPUT_NOT_REDUCED 110 -#define BN_R_INVALID_LENGTH 106 -#define BN_R_INVALID_RANGE 115 -#define BN_R_NOT_A_SQUARE 111 -#define BN_R_NOT_INITIALIZED 107 -#define BN_R_NO_INVERSE 108 -#define BN_R_NO_SOLUTION 116 -#define BN_R_P_IS_NOT_PRIME 112 -#define BN_R_TOO_MANY_ITERATIONS 113 -#define BN_R_TOO_MANY_TEMPORARY_VARIABLES 109 +# define BN_R_ARG2_LT_ARG3 100 +# define BN_R_BAD_RECIPROCAL 101 +# define BN_R_BIGNUM_TOO_LONG 114 +# define BN_R_CALLED_WITH_EVEN_MODULUS 102 +# define BN_R_DIV_BY_ZERO 103 +# define BN_R_ENCODING_ERROR 104 +# define BN_R_EXPAND_ON_STATIC_BIGNUM_DATA 105 +# define BN_R_INPUT_NOT_REDUCED 110 +# define BN_R_INVALID_LENGTH 106 +# define BN_R_INVALID_RANGE 115 +# define BN_R_NOT_A_SQUARE 111 +# define BN_R_NOT_INITIALIZED 107 +# define BN_R_NO_INVERSE 108 +# define BN_R_NO_SOLUTION 116 +# define BN_R_P_IS_NOT_PRIME 112 +# define BN_R_TOO_MANY_ITERATIONS 113 +# define BN_R_TOO_MANY_TEMPORARY_VARIABLES 109 #ifdef __cplusplus } diff --git a/openssl/crypto/bn/bn_add.c b/openssl/crypto/bn/bn_add.c index 940516370..2f3d11044 100644 --- a/openssl/crypto/bn/bn_add.c +++ b/openssl/crypto/bn/bn_add.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -62,252 +62,252 @@ /* r can == a or b */ int BN_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b) - { - const BIGNUM *tmp; - int a_neg = a->neg, ret; +{ + const BIGNUM *tmp; + int a_neg = a->neg, ret; - bn_check_top(a); - bn_check_top(b); + bn_check_top(a); + bn_check_top(b); - /* a + b a+b - * a + -b a-b - * -a + b b-a - * -a + -b -(a+b) - */ - if (a_neg ^ b->neg) - { - /* only one is negative */ - if (a_neg) - { tmp=a; a=b; b=tmp; } + /*- + * a + b a+b + * a + -b a-b + * -a + b b-a + * -a + -b -(a+b) + */ + if (a_neg ^ b->neg) { + /* only one is negative */ + if (a_neg) { + tmp = a; + a = b; + b = tmp; + } - /* we are now a - b */ + /* we are now a - b */ - if (BN_ucmp(a,b) < 0) - { - if (!BN_usub(r,b,a)) return(0); - r->neg=1; - } - else - { - if (!BN_usub(r,a,b)) return(0); - r->neg=0; - } - return(1); - } + if (BN_ucmp(a, b) < 0) { + if (!BN_usub(r, b, a)) + return (0); + r->neg = 1; + } else { + if (!BN_usub(r, a, b)) + return (0); + r->neg = 0; + } + return (1); + } - ret = BN_uadd(r,a,b); - r->neg = a_neg; - bn_check_top(r); - return ret; - } + ret = BN_uadd(r, a, b); + r->neg = a_neg; + bn_check_top(r); + return ret; +} /* unsigned add of b to a */ int BN_uadd(BIGNUM *r, const BIGNUM *a, const BIGNUM *b) - { - int max,min,dif; - BN_ULONG *ap,*bp,*rp,carry,t1,t2; - const BIGNUM *tmp; +{ + int max, min, dif; + BN_ULONG *ap, *bp, *rp, carry, t1, t2; + const BIGNUM *tmp; - bn_check_top(a); - bn_check_top(b); + bn_check_top(a); + bn_check_top(b); - if (a->top < b->top) - { tmp=a; a=b; b=tmp; } - max = a->top; - min = b->top; - dif = max - min; + if (a->top < b->top) { + tmp = a; + a = b; + b = tmp; + } + max = a->top; + min = b->top; + dif = max - min; - if (bn_wexpand(r,max+1) == NULL) - return 0; + if (bn_wexpand(r, max + 1) == NULL) + return 0; - r->top=max; + r->top = max; + ap = a->d; + bp = b->d; + rp = r->d; - ap=a->d; - bp=b->d; - rp=r->d; + carry = bn_add_words(rp, ap, bp, min); + rp += min; + ap += min; + bp += min; - carry=bn_add_words(rp,ap,bp,min); - rp+=min; - ap+=min; - bp+=min; - - if (carry) - { - while (dif) - { - dif--; - t1 = *(ap++); - t2 = (t1+1) & BN_MASK2; - *(rp++) = t2; - if (t2) - { - carry=0; - break; - } - } - if (carry) - { - /* carry != 0 => dif == 0 */ - *rp = 1; - r->top++; - } - } - if (dif && rp != ap) - while (dif--) - /* copy remaining words if ap != rp */ - *(rp++) = *(ap++); - r->neg = 0; - bn_check_top(r); - return 1; - } + if (carry) { + while (dif) { + dif--; + t1 = *(ap++); + t2 = (t1 + 1) & BN_MASK2; + *(rp++) = t2; + if (t2) { + carry = 0; + break; + } + } + if (carry) { + /* carry != 0 => dif == 0 */ + *rp = 1; + r->top++; + } + } + if (dif && rp != ap) + while (dif--) + /* copy remaining words if ap != rp */ + *(rp++) = *(ap++); + r->neg = 0; + bn_check_top(r); + return 1; +} /* unsigned subtraction of b from a, a must be larger than b. */ int BN_usub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b) - { - int max,min,dif; - register BN_ULONG t1,t2,*ap,*bp,*rp; - int i,carry; +{ + int max, min, dif; + register BN_ULONG t1, t2, *ap, *bp, *rp; + int i, carry; #if defined(IRIX_CC_BUG) && !defined(LINT) - int dummy; + int dummy; #endif - bn_check_top(a); - bn_check_top(b); + bn_check_top(a); + bn_check_top(b); - max = a->top; - min = b->top; - dif = max - min; + max = a->top; + min = b->top; + dif = max - min; - if (dif < 0) /* hmm... should not be happening */ - { - BNerr(BN_F_BN_USUB,BN_R_ARG2_LT_ARG3); - return(0); - } + if (dif < 0) { /* hmm... should not be happening */ + BNerr(BN_F_BN_USUB, BN_R_ARG2_LT_ARG3); + return (0); + } - if (bn_wexpand(r,max) == NULL) return(0); + if (bn_wexpand(r, max) == NULL) + return (0); - ap=a->d; - bp=b->d; - rp=r->d; + ap = a->d; + bp = b->d; + rp = r->d; #if 1 - carry=0; - for (i = min; i != 0; i--) - { - t1= *(ap++); - t2= *(bp++); - if (carry) - { - carry=(t1 <= t2); - t1=(t1-t2-1)&BN_MASK2; - } - else - { - carry=(t1 < t2); - t1=(t1-t2)&BN_MASK2; - } -#if defined(IRIX_CC_BUG) && !defined(LINT) - dummy=t1; -#endif - *(rp++)=t1&BN_MASK2; - } + carry = 0; + for (i = min; i != 0; i--) { + t1 = *(ap++); + t2 = *(bp++); + if (carry) { + carry = (t1 <= t2); + t1 = (t1 - t2 - 1) & BN_MASK2; + } else { + carry = (t1 < t2); + t1 = (t1 - t2) & BN_MASK2; + } +# if defined(IRIX_CC_BUG) && !defined(LINT) + dummy = t1; +# endif + *(rp++) = t1 & BN_MASK2; + } #else - carry=bn_sub_words(rp,ap,bp,min); - ap+=min; - bp+=min; - rp+=min; + carry = bn_sub_words(rp, ap, bp, min); + ap += min; + bp += min; + rp += min; #endif - if (carry) /* subtracted */ - { - if (!dif) - /* error: a < b */ - return 0; - while (dif) - { - dif--; - t1 = *(ap++); - t2 = (t1-1)&BN_MASK2; - *(rp++) = t2; - if (t1) - break; - } - } + if (carry) { /* subtracted */ + if (!dif) + /* error: a < b */ + return 0; + while (dif) { + dif--; + t1 = *(ap++); + t2 = (t1 - 1) & BN_MASK2; + *(rp++) = t2; + if (t1) + break; + } + } #if 0 - memcpy(rp,ap,sizeof(*rp)*(max-i)); + memcpy(rp, ap, sizeof(*rp) * (max - i)); #else - if (rp != ap) - { - for (;;) - { - if (!dif--) break; - rp[0]=ap[0]; - if (!dif--) break; - rp[1]=ap[1]; - if (!dif--) break; - rp[2]=ap[2]; - if (!dif--) break; - rp[3]=ap[3]; - rp+=4; - ap+=4; - } - } + if (rp != ap) { + for (;;) { + if (!dif--) + break; + rp[0] = ap[0]; + if (!dif--) + break; + rp[1] = ap[1]; + if (!dif--) + break; + rp[2] = ap[2]; + if (!dif--) + break; + rp[3] = ap[3]; + rp += 4; + ap += 4; + } + } #endif - r->top=max; - r->neg=0; - bn_correct_top(r); - return(1); - } + r->top = max; + r->neg = 0; + bn_correct_top(r); + return (1); +} int BN_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b) - { - int max; - int add=0,neg=0; - const BIGNUM *tmp; - - bn_check_top(a); - bn_check_top(b); +{ + int max; + int add = 0, neg = 0; + const BIGNUM *tmp; - /* a - b a-b - * a - -b a+b - * -a - b -(a+b) - * -a - -b b-a - */ - if (a->neg) - { - if (b->neg) - { tmp=a; a=b; b=tmp; } - else - { add=1; neg=1; } - } - else - { - if (b->neg) { add=1; neg=0; } - } + bn_check_top(a); + bn_check_top(b); - if (add) - { - if (!BN_uadd(r,a,b)) return(0); - r->neg=neg; - return(1); - } + /*- + * a - b a-b + * a - -b a+b + * -a - b -(a+b) + * -a - -b b-a + */ + if (a->neg) { + if (b->neg) { + tmp = a; + a = b; + b = tmp; + } else { + add = 1; + neg = 1; + } + } else { + if (b->neg) { + add = 1; + neg = 0; + } + } - /* We are actually doing a - b :-) */ + if (add) { + if (!BN_uadd(r, a, b)) + return (0); + r->neg = neg; + return (1); + } - max=(a->top > b->top)?a->top:b->top; - if (bn_wexpand(r,max) == NULL) return(0); - if (BN_ucmp(a,b) < 0) - { - if (!BN_usub(r,b,a)) return(0); - r->neg=1; - } - else - { - if (!BN_usub(r,a,b)) return(0); - r->neg=0; - } - bn_check_top(r); - return(1); - } + /* We are actually doing a - b :-) */ + max = (a->top > b->top) ? a->top : b->top; + if (bn_wexpand(r, max) == NULL) + return (0); + if (BN_ucmp(a, b) < 0) { + if (!BN_usub(r, b, a)) + return (0); + r->neg = 1; + } else { + if (!BN_usub(r, a, b)) + return (0); + r->neg = 0; + } + bn_check_top(r); + return (1); +} diff --git a/openssl/crypto/bn/bn_asm.c b/openssl/crypto/bn/bn_asm.c index c43c91cc0..03a33cffe 100644 --- a/openssl/crypto/bn/bn_asm.c +++ b/openssl/crypto/bn/bn_asm.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -57,7 +57,7 @@ */ #ifndef BN_DEBUG -# undef NDEBUG /* avoid conflicting definitions */ +# undef NDEBUG /* avoid conflicting definitions */ # define NDEBUG #endif @@ -68,769 +68,823 @@ #if defined(BN_LLONG) || defined(BN_UMULT_HIGH) -BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) - { - BN_ULONG c1=0; - - assert(num >= 0); - if (num <= 0) return(c1); - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (num&~3) - { - mul_add(rp[0],ap[0],w,c1); - mul_add(rp[1],ap[1],w,c1); - mul_add(rp[2],ap[2],w,c1); - mul_add(rp[3],ap[3],w,c1); - ap+=4; rp+=4; num-=4; - } -#endif - while (num) - { - mul_add(rp[0],ap[0],w,c1); - ap++; rp++; num--; - } - - return(c1); - } +BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, + BN_ULONG w) +{ + BN_ULONG c1 = 0; + + assert(num >= 0); + if (num <= 0) + return (c1); + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (num & ~3) { + mul_add(rp[0], ap[0], w, c1); + mul_add(rp[1], ap[1], w, c1); + mul_add(rp[2], ap[2], w, c1); + mul_add(rp[3], ap[3], w, c1); + ap += 4; + rp += 4; + num -= 4; + } +# endif + while (num) { + mul_add(rp[0], ap[0], w, c1); + ap++; + rp++; + num--; + } + + return (c1); +} BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) - { - BN_ULONG c1=0; - - assert(num >= 0); - if (num <= 0) return(c1); - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (num&~3) - { - mul(rp[0],ap[0],w,c1); - mul(rp[1],ap[1],w,c1); - mul(rp[2],ap[2],w,c1); - mul(rp[3],ap[3],w,c1); - ap+=4; rp+=4; num-=4; - } -#endif - while (num) - { - mul(rp[0],ap[0],w,c1); - ap++; rp++; num--; - } - return(c1); - } +{ + BN_ULONG c1 = 0; + + assert(num >= 0); + if (num <= 0) + return (c1); + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (num & ~3) { + mul(rp[0], ap[0], w, c1); + mul(rp[1], ap[1], w, c1); + mul(rp[2], ap[2], w, c1); + mul(rp[3], ap[3], w, c1); + ap += 4; + rp += 4; + num -= 4; + } +# endif + while (num) { + mul(rp[0], ap[0], w, c1); + ap++; + rp++; + num--; + } + return (c1); +} void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n) - { - assert(n >= 0); - if (n <= 0) return; - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (n&~3) - { - sqr(r[0],r[1],a[0]); - sqr(r[2],r[3],a[1]); - sqr(r[4],r[5],a[2]); - sqr(r[6],r[7],a[3]); - a+=4; r+=8; n-=4; - } -#endif - while (n) - { - sqr(r[0],r[1],a[0]); - a++; r+=2; n--; - } - } - -#else /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ - -BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) - { - BN_ULONG c=0; - BN_ULONG bl,bh; - - assert(num >= 0); - if (num <= 0) return((BN_ULONG)0); - - bl=LBITS(w); - bh=HBITS(w); - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (num&~3) - { - mul_add(rp[0],ap[0],bl,bh,c); - mul_add(rp[1],ap[1],bl,bh,c); - mul_add(rp[2],ap[2],bl,bh,c); - mul_add(rp[3],ap[3],bl,bh,c); - ap+=4; rp+=4; num-=4; - } -#endif - while (num) - { - mul_add(rp[0],ap[0],bl,bh,c); - ap++; rp++; num--; - } - return(c); - } +{ + assert(n >= 0); + if (n <= 0) + return; + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (n & ~3) { + sqr(r[0], r[1], a[0]); + sqr(r[2], r[3], a[1]); + sqr(r[4], r[5], a[2]); + sqr(r[6], r[7], a[3]); + a += 4; + r += 8; + n -= 4; + } +# endif + while (n) { + sqr(r[0], r[1], a[0]); + a++; + r += 2; + n--; + } +} + +#else /* !(defined(BN_LLONG) || + * defined(BN_UMULT_HIGH)) */ + +BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num, + BN_ULONG w) +{ + BN_ULONG c = 0; + BN_ULONG bl, bh; + + assert(num >= 0); + if (num <= 0) + return ((BN_ULONG)0); + + bl = LBITS(w); + bh = HBITS(w); + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (num & ~3) { + mul_add(rp[0], ap[0], bl, bh, c); + mul_add(rp[1], ap[1], bl, bh, c); + mul_add(rp[2], ap[2], bl, bh, c); + mul_add(rp[3], ap[3], bl, bh, c); + ap += 4; + rp += 4; + num -= 4; + } +# endif + while (num) { + mul_add(rp[0], ap[0], bl, bh, c); + ap++; + rp++; + num--; + } + return (c); +} BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w) - { - BN_ULONG carry=0; - BN_ULONG bl,bh; - - assert(num >= 0); - if (num <= 0) return((BN_ULONG)0); - - bl=LBITS(w); - bh=HBITS(w); - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (num&~3) - { - mul(rp[0],ap[0],bl,bh,carry); - mul(rp[1],ap[1],bl,bh,carry); - mul(rp[2],ap[2],bl,bh,carry); - mul(rp[3],ap[3],bl,bh,carry); - ap+=4; rp+=4; num-=4; - } -#endif - while (num) - { - mul(rp[0],ap[0],bl,bh,carry); - ap++; rp++; num--; - } - return(carry); - } +{ + BN_ULONG carry = 0; + BN_ULONG bl, bh; + + assert(num >= 0); + if (num <= 0) + return ((BN_ULONG)0); + + bl = LBITS(w); + bh = HBITS(w); + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (num & ~3) { + mul(rp[0], ap[0], bl, bh, carry); + mul(rp[1], ap[1], bl, bh, carry); + mul(rp[2], ap[2], bl, bh, carry); + mul(rp[3], ap[3], bl, bh, carry); + ap += 4; + rp += 4; + num -= 4; + } +# endif + while (num) { + mul(rp[0], ap[0], bl, bh, carry); + ap++; + rp++; + num--; + } + return (carry); +} void bn_sqr_words(BN_ULONG *r, const BN_ULONG *a, int n) - { - assert(n >= 0); - if (n <= 0) return; - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (n&~3) - { - sqr64(r[0],r[1],a[0]); - sqr64(r[2],r[3],a[1]); - sqr64(r[4],r[5],a[2]); - sqr64(r[6],r[7],a[3]); - a+=4; r+=8; n-=4; - } -#endif - while (n) - { - sqr64(r[0],r[1],a[0]); - a++; r+=2; n--; - } - } - -#endif /* !(defined(BN_LLONG) || defined(BN_UMULT_HIGH)) */ +{ + assert(n >= 0); + if (n <= 0) + return; + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (n & ~3) { + sqr64(r[0], r[1], a[0]); + sqr64(r[2], r[3], a[1]); + sqr64(r[4], r[5], a[2]); + sqr64(r[6], r[7], a[3]); + a += 4; + r += 8; + n -= 4; + } +# endif + while (n) { + sqr64(r[0], r[1], a[0]); + a++; + r += 2; + n--; + } +} + +#endif /* !(defined(BN_LLONG) || + * defined(BN_UMULT_HIGH)) */ #if defined(BN_LLONG) && defined(BN_DIV2W) BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) - { - return((BN_ULONG)(((((BN_ULLONG)h)<<BN_BITS2)|l)/(BN_ULLONG)d)); - } +{ + return ((BN_ULONG)(((((BN_ULLONG) h) << BN_BITS2) | l) / (BN_ULLONG) d)); +} #else /* Divide h,l by d and return the result. */ /* I need to test this some more :-( */ BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d) - { - BN_ULONG dh,dl,q,ret=0,th,tl,t; - int i,count=2; - - if (d == 0) return(BN_MASK2); - - i=BN_num_bits_word(d); - assert((i == BN_BITS2) || (h <= (BN_ULONG)1<<i)); - - i=BN_BITS2-i; - if (h >= d) h-=d; - - if (i) - { - d<<=i; - h=(h<<i)|(l>>(BN_BITS2-i)); - l<<=i; - } - dh=(d&BN_MASK2h)>>BN_BITS4; - dl=(d&BN_MASK2l); - for (;;) - { - if ((h>>BN_BITS4) == dh) - q=BN_MASK2l; - else - q=h/dh; - - th=q*dh; - tl=dl*q; - for (;;) - { - t=h-th; - if ((t&BN_MASK2h) || - ((tl) <= ( - (t<<BN_BITS4)| - ((l&BN_MASK2h)>>BN_BITS4)))) - break; - q--; - th-=dh; - tl-=dl; - } - t=(tl>>BN_BITS4); - tl=(tl<<BN_BITS4)&BN_MASK2h; - th+=t; - - if (l < tl) th++; - l-=tl; - if (h < th) - { - h+=d; - q--; - } - h-=th; - - if (--count == 0) break; - - ret=q<<BN_BITS4; - h=((h<<BN_BITS4)|(l>>BN_BITS4))&BN_MASK2; - l=(l&BN_MASK2l)<<BN_BITS4; - } - ret|=q; - return(ret); - } -#endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */ +{ + BN_ULONG dh, dl, q, ret = 0, th, tl, t; + int i, count = 2; + + if (d == 0) + return (BN_MASK2); + + i = BN_num_bits_word(d); + assert((i == BN_BITS2) || (h <= (BN_ULONG)1 << i)); + + i = BN_BITS2 - i; + if (h >= d) + h -= d; + + if (i) { + d <<= i; + h = (h << i) | (l >> (BN_BITS2 - i)); + l <<= i; + } + dh = (d & BN_MASK2h) >> BN_BITS4; + dl = (d & BN_MASK2l); + for (;;) { + if ((h >> BN_BITS4) == dh) + q = BN_MASK2l; + else + q = h / dh; + + th = q * dh; + tl = dl * q; + for (;;) { + t = h - th; + if ((t & BN_MASK2h) || + ((tl) <= ((t << BN_BITS4) | ((l & BN_MASK2h) >> BN_BITS4)))) + break; + q--; + th -= dh; + tl -= dl; + } + t = (tl >> BN_BITS4); + tl = (tl << BN_BITS4) & BN_MASK2h; + th += t; + + if (l < tl) + th++; + l -= tl; + if (h < th) { + h += d; + q--; + } + h -= th; + + if (--count == 0) + break; + + ret = q << BN_BITS4; + h = ((h << BN_BITS4) | (l >> BN_BITS4)) & BN_MASK2; + l = (l & BN_MASK2l) << BN_BITS4; + } + ret |= q; + return (ret); +} +#endif /* !defined(BN_LLONG) && defined(BN_DIV2W) */ #ifdef BN_LLONG -BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n) - { - BN_ULLONG ll=0; - - assert(n >= 0); - if (n <= 0) return((BN_ULONG)0); - -#ifndef OPENSSL_SMALL_FOOTPRINT - while (n&~3) - { - ll+=(BN_ULLONG)a[0]+b[0]; - r[0]=(BN_ULONG)ll&BN_MASK2; - ll>>=BN_BITS2; - ll+=(BN_ULLONG)a[1]+b[1]; - r[1]=(BN_ULONG)ll&BN_MASK2; - ll>>=BN_BITS2; - ll+=(BN_ULLONG)a[2]+b[2]; - r[2]=(BN_ULONG)ll&BN_MASK2; - ll>>=BN_BITS2; - ll+=(BN_ULLONG)a[3]+b[3]; - r[3]=(BN_ULONG)ll&BN_MASK2; - ll>>=BN_BITS2; - a+=4; b+=4; r+=4; n-=4; - } -#endif - while (n) - { - ll+=(BN_ULLONG)a[0]+b[0]; - r[0]=(BN_ULONG)ll&BN_MASK2; - ll>>=BN_BITS2; - a++; b++; r++; n--; - } - return((BN_ULONG)ll); - } -#else /* !BN_LLONG */ -BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n) - { - BN_ULONG c,l,t; - - assert(n >= 0); - if (n <= 0) return((BN_ULONG)0); - - c=0; -#ifndef OPENSSL_SMALL_FOOTPRINT - while (n&~3) - { - t=a[0]; - t=(t+c)&BN_MASK2; - c=(t < c); - l=(t+b[0])&BN_MASK2; - c+=(l < t); - r[0]=l; - t=a[1]; - t=(t+c)&BN_MASK2; - c=(t < c); - l=(t+b[1])&BN_MASK2; - c+=(l < t); - r[1]=l; - t=a[2]; - t=(t+c)&BN_MASK2; - c=(t < c); - l=(t+b[2])&BN_MASK2; - c+=(l < t); - r[2]=l; - t=a[3]; - t=(t+c)&BN_MASK2; - c=(t < c); - l=(t+b[3])&BN_MASK2; - c+=(l < t); - r[3]=l; - a+=4; b+=4; r+=4; n-=4; - } -#endif - while(n) - { - t=a[0]; - t=(t+c)&BN_MASK2; - c=(t < c); - l=(t+b[0])&BN_MASK2; - c+=(l < t); - r[0]=l; - a++; b++; r++; n--; - } - return((BN_ULONG)c); - } -#endif /* !BN_LLONG */ - -BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, int n) - { - BN_ULONG t1,t2; - int c=0; - - assert(n >= 0); - if (n <= 0) return((BN_ULONG)0); +BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, + int n) +{ + BN_ULLONG ll = 0; + + assert(n >= 0); + if (n <= 0) + return ((BN_ULONG)0); + +# ifndef OPENSSL_SMALL_FOOTPRINT + while (n & ~3) { + ll += (BN_ULLONG) a[0] + b[0]; + r[0] = (BN_ULONG)ll & BN_MASK2; + ll >>= BN_BITS2; + ll += (BN_ULLONG) a[1] + b[1]; + r[1] = (BN_ULONG)ll & BN_MASK2; + ll >>= BN_BITS2; + ll += (BN_ULLONG) a[2] + b[2]; + r[2] = (BN_ULONG)ll & BN_MASK2; + ll >>= BN_BITS2; + ll += (BN_ULLONG) a[3] + b[3]; + r[3] = (BN_ULONG)ll & BN_MASK2; + ll >>= BN_BITS2; + a += 4; + b += 4; + r += 4; + n -= 4; + } +# endif + while (n) { + ll += (BN_ULLONG) a[0] + b[0]; + r[0] = (BN_ULONG)ll & BN_MASK2; + ll >>= BN_BITS2; + a++; + b++; + r++; + n--; + } + return ((BN_ULONG)ll); +} +#else /* !BN_LLONG */ +BN_ULONG bn_add_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, + int n) +{ + BN_ULONG c, l, t; + + assert(n >= 0); + if (n <= 0) + return ((BN_ULONG)0); + + c = 0; +# ifndef OPENSSL_SMALL_FOOTPRINT + while (n & ~3) { + t = a[0]; + t = (t + c) & BN_MASK2; + c = (t < c); + l = (t + b[0]) & BN_MASK2; + c += (l < t); + r[0] = l; + t = a[1]; + t = (t + c) & BN_MASK2; + c = (t < c); + l = (t + b[1]) & BN_MASK2; + c += (l < t); + r[1] = l; + t = a[2]; + t = (t + c) & BN_MASK2; + c = (t < c); + l = (t + b[2]) & BN_MASK2; + c += (l < t); + r[2] = l; + t = a[3]; + t = (t + c) & BN_MASK2; + c = (t < c); + l = (t + b[3]) & BN_MASK2; + c += (l < t); + r[3] = l; + a += 4; + b += 4; + r += 4; + n -= 4; + } +# endif + while (n) { + t = a[0]; + t = (t + c) & BN_MASK2; + c = (t < c); + l = (t + b[0]) & BN_MASK2; + c += (l < t); + r[0] = l; + a++; + b++; + r++; + n--; + } + return ((BN_ULONG)c); +} +#endif /* !BN_LLONG */ + +BN_ULONG bn_sub_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, + int n) +{ + BN_ULONG t1, t2; + int c = 0; + + assert(n >= 0); + if (n <= 0) + return ((BN_ULONG)0); #ifndef OPENSSL_SMALL_FOOTPRINT - while (n&~3) - { - t1=a[0]; t2=b[0]; - r[0]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - t1=a[1]; t2=b[1]; - r[1]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - t1=a[2]; t2=b[2]; - r[2]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - t1=a[3]; t2=b[3]; - r[3]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - a+=4; b+=4; r+=4; n-=4; - } + while (n & ~3) { + t1 = a[0]; + t2 = b[0]; + r[0] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + t1 = a[1]; + t2 = b[1]; + r[1] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + t1 = a[2]; + t2 = b[2]; + r[2] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + t1 = a[3]; + t2 = b[3]; + r[3] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + a += 4; + b += 4; + r += 4; + n -= 4; + } #endif - while (n) - { - t1=a[0]; t2=b[0]; - r[0]=(t1-t2-c)&BN_MASK2; - if (t1 != t2) c=(t1 < t2); - a++; b++; r++; n--; - } - return(c); - } + while (n) { + t1 = a[0]; + t2 = b[0]; + r[0] = (t1 - t2 - c) & BN_MASK2; + if (t1 != t2) + c = (t1 < t2); + a++; + b++; + r++; + n--; + } + return (c); +} #if defined(BN_MUL_COMBA) && !defined(OPENSSL_SMALL_FOOTPRINT) -#undef bn_mul_comba8 -#undef bn_mul_comba4 -#undef bn_sqr_comba8 -#undef bn_sqr_comba4 +# undef bn_mul_comba8 +# undef bn_mul_comba4 +# undef bn_sqr_comba8 +# undef bn_sqr_comba4 /* mul_add_c(a,b,c0,c1,c2) -- c+=a*b for three word number c=(c2,c1,c0) */ /* mul_add_c2(a,b,c0,c1,c2) -- c+=2*a*b for three word number c=(c2,c1,c0) */ /* sqr_add_c(a,i,c0,c1,c2) -- c+=a[i]^2 for three word number c=(c2,c1,c0) */ -/* sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number c=(c2,c1,c0) */ +/* + * sqr_add_c2(a,i,c0,c1,c2) -- c+=2*a[i]*a[j] for three word number + * c=(c2,c1,c0) + */ -#ifdef BN_LLONG -#define mul_add_c(a,b,c0,c1,c2) \ - t=(BN_ULLONG)a*b; \ - t1=(BN_ULONG)Lw(t); \ - t2=(BN_ULONG)Hw(t); \ - c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ - c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; - -#define mul_add_c2(a,b,c0,c1,c2) \ - t=(BN_ULLONG)a*b; \ - tt=(t+t)&BN_MASK; \ - if (tt < t) c2++; \ - t1=(BN_ULONG)Lw(tt); \ - t2=(BN_ULONG)Hw(tt); \ - c0=(c0+t1)&BN_MASK2; \ - if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \ - c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; - -#define sqr_add_c(a,i,c0,c1,c2) \ - t=(BN_ULLONG)a[i]*a[i]; \ - t1=(BN_ULONG)Lw(t); \ - t2=(BN_ULONG)Hw(t); \ - c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ - c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; - -#define sqr_add_c2(a,i,j,c0,c1,c2) \ - mul_add_c2((a)[i],(a)[j],c0,c1,c2) - -#elif defined(BN_UMULT_LOHI) - -#define mul_add_c(a,b,c0,c1,c2) { \ - BN_ULONG ta=(a),tb=(b); \ - BN_UMULT_LOHI(t1,t2,ta,tb); \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define mul_add_c2(a,b,c0,c1,c2) { \ - BN_ULONG ta=(a),tb=(b),t0; \ - BN_UMULT_LOHI(t0,t1,ta,tb); \ - t2 = t1+t1; c2 += (t2<t1)?1:0; \ - t1 = t0+t0; t2 += (t1<t0)?1:0; \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define sqr_add_c(a,i,c0,c1,c2) { \ - BN_ULONG ta=(a)[i]; \ - BN_UMULT_LOHI(t1,t2,ta,ta); \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define sqr_add_c2(a,i,j,c0,c1,c2) \ - mul_add_c2((a)[i],(a)[j],c0,c1,c2) - -#elif defined(BN_UMULT_HIGH) - -#define mul_add_c(a,b,c0,c1,c2) { \ - BN_ULONG ta=(a),tb=(b); \ - t1 = ta * tb; \ - t2 = BN_UMULT_HIGH(ta,tb); \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define mul_add_c2(a,b,c0,c1,c2) { \ - BN_ULONG ta=(a),tb=(b),t0; \ - t1 = BN_UMULT_HIGH(ta,tb); \ - t0 = ta * tb; \ - t2 = t1+t1; c2 += (t2<t1)?1:0; \ - t1 = t0+t0; t2 += (t1<t0)?1:0; \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define sqr_add_c(a,i,c0,c1,c2) { \ - BN_ULONG ta=(a)[i]; \ - t1 = ta * ta; \ - t2 = BN_UMULT_HIGH(ta,ta); \ - c0 += t1; t2 += (c0<t1)?1:0; \ - c1 += t2; c2 += (c1<t2)?1:0; \ - } - -#define sqr_add_c2(a,i,j,c0,c1,c2) \ - mul_add_c2((a)[i],(a)[j],c0,c1,c2) - -#else /* !BN_LLONG */ -#define mul_add_c(a,b,c0,c1,c2) \ - t1=LBITS(a); t2=HBITS(a); \ - bl=LBITS(b); bh=HBITS(b); \ - mul64(t1,t2,bl,bh); \ - c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ - c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; - -#define mul_add_c2(a,b,c0,c1,c2) \ - t1=LBITS(a); t2=HBITS(a); \ - bl=LBITS(b); bh=HBITS(b); \ - mul64(t1,t2,bl,bh); \ - if (t2 & BN_TBIT) c2++; \ - t2=(t2+t2)&BN_MASK2; \ - if (t1 & BN_TBIT) t2++; \ - t1=(t1+t1)&BN_MASK2; \ - c0=(c0+t1)&BN_MASK2; \ - if ((c0 < t1) && (((++t2)&BN_MASK2) == 0)) c2++; \ - c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; - -#define sqr_add_c(a,i,c0,c1,c2) \ - sqr64(t1,t2,(a)[i]); \ - c0=(c0+t1)&BN_MASK2; if ((c0) < t1) t2++; \ - c1=(c1+t2)&BN_MASK2; if ((c1) < t2) c2++; - -#define sqr_add_c2(a,i,j,c0,c1,c2) \ - mul_add_c2((a)[i],(a)[j],c0,c1,c2) -#endif /* !BN_LLONG */ +# ifdef BN_LLONG +/* + * Keep in mind that additions to multiplication result can not + * overflow, because its high half cannot be all-ones. + */ +# define mul_add_c(a,b,c0,c1,c2) do { \ + BN_ULONG hi; \ + BN_ULLONG t = (BN_ULLONG)(a)*(b); \ + t += c0; /* no carry */ \ + c0 = (BN_ULONG)Lw(t); \ + hi = (BN_ULONG)Hw(t); \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + } while(0) + +# define mul_add_c2(a,b,c0,c1,c2) do { \ + BN_ULONG hi; \ + BN_ULLONG t = (BN_ULLONG)(a)*(b); \ + BN_ULLONG tt = t+c0; /* no carry */ \ + c0 = (BN_ULONG)Lw(tt); \ + hi = (BN_ULONG)Hw(tt); \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + t += c0; /* no carry */ \ + c0 = (BN_ULONG)Lw(t); \ + hi = (BN_ULONG)Hw(t); \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + } while(0) + +# define sqr_add_c(a,i,c0,c1,c2) do { \ + BN_ULONG hi; \ + BN_ULLONG t = (BN_ULLONG)a[i]*a[i]; \ + t += c0; /* no carry */ \ + c0 = (BN_ULONG)Lw(t); \ + hi = (BN_ULONG)Hw(t); \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + } while(0) + +# define sqr_add_c2(a,i,j,c0,c1,c2) \ + mul_add_c2((a)[i],(a)[j],c0,c1,c2) + +# elif defined(BN_UMULT_LOHI) +/* + * Keep in mind that additions to hi can not overflow, because + * the high word of a multiplication result cannot be all-ones. + */ +# define mul_add_c(a,b,c0,c1,c2) do { \ + BN_ULONG ta = (a), tb = (b); \ + BN_ULONG lo, hi; \ + BN_UMULT_LOHI(lo,hi,ta,tb); \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define mul_add_c2(a,b,c0,c1,c2) do { \ + BN_ULONG ta = (a), tb = (b); \ + BN_ULONG lo, hi, tt; \ + BN_UMULT_LOHI(lo,hi,ta,tb); \ + c0 += lo; tt = hi+((c0<lo)?1:0); \ + c1 += tt; c2 += (c1<tt)?1:0; \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define sqr_add_c(a,i,c0,c1,c2) do { \ + BN_ULONG ta = (a)[i]; \ + BN_ULONG lo, hi; \ + BN_UMULT_LOHI(lo,hi,ta,ta); \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define sqr_add_c2(a,i,j,c0,c1,c2) \ + mul_add_c2((a)[i],(a)[j],c0,c1,c2) + +# elif defined(BN_UMULT_HIGH) +/* + * Keep in mind that additions to hi can not overflow, because + * the high word of a multiplication result cannot be all-ones. + */ +# define mul_add_c(a,b,c0,c1,c2) do { \ + BN_ULONG ta = (a), tb = (b); \ + BN_ULONG lo = ta * tb; \ + BN_ULONG hi = BN_UMULT_HIGH(ta,tb); \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define mul_add_c2(a,b,c0,c1,c2) do { \ + BN_ULONG ta = (a), tb = (b), tt; \ + BN_ULONG lo = ta * tb; \ + BN_ULONG hi = BN_UMULT_HIGH(ta,tb); \ + c0 += lo; tt = hi + ((c0<lo)?1:0); \ + c1 += tt; c2 += (c1<tt)?1:0; \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define sqr_add_c(a,i,c0,c1,c2) do { \ + BN_ULONG ta = (a)[i]; \ + BN_ULONG lo = ta * ta; \ + BN_ULONG hi = BN_UMULT_HIGH(ta,ta); \ + c0 += lo; hi += (c0<lo)?1:0; \ + c1 += hi; c2 += (c1<hi)?1:0; \ + } while(0) + +# define sqr_add_c2(a,i,j,c0,c1,c2) \ + mul_add_c2((a)[i],(a)[j],c0,c1,c2) + +# else /* !BN_LLONG */ +/* + * Keep in mind that additions to hi can not overflow, because + * the high word of a multiplication result cannot be all-ones. + */ +# define mul_add_c(a,b,c0,c1,c2) do { \ + BN_ULONG lo = LBITS(a), hi = HBITS(a); \ + BN_ULONG bl = LBITS(b), bh = HBITS(b); \ + mul64(lo,hi,bl,bh); \ + c0 = (c0+lo)&BN_MASK2; if (c0<lo) hi++; \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + } while(0) + +# define mul_add_c2(a,b,c0,c1,c2) do { \ + BN_ULONG tt; \ + BN_ULONG lo = LBITS(a), hi = HBITS(a); \ + BN_ULONG bl = LBITS(b), bh = HBITS(b); \ + mul64(lo,hi,bl,bh); \ + tt = hi; \ + c0 = (c0+lo)&BN_MASK2; if (c0<lo) tt++; \ + c1 = (c1+tt)&BN_MASK2; if (c1<tt) c2++; \ + c0 = (c0+lo)&BN_MASK2; if (c0<lo) hi++; \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + } while(0) + +# define sqr_add_c(a,i,c0,c1,c2) do { \ + BN_ULONG lo, hi; \ + sqr64(lo,hi,(a)[i]); \ + c0 = (c0+lo)&BN_MASK2; if (c0<lo) hi++; \ + c1 = (c1+hi)&BN_MASK2; if (c1<hi) c2++; \ + } while(0) + +# define sqr_add_c2(a,i,j,c0,c1,c2) \ + mul_add_c2((a)[i],(a)[j],c0,c1,c2) +# endif /* !BN_LLONG */ void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) - { -#ifdef BN_LLONG - BN_ULLONG t; -#else - BN_ULONG bl,bh; -#endif - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - mul_add_c(a[0],b[0],c1,c2,c3); - r[0]=c1; - c1=0; - mul_add_c(a[0],b[1],c2,c3,c1); - mul_add_c(a[1],b[0],c2,c3,c1); - r[1]=c2; - c2=0; - mul_add_c(a[2],b[0],c3,c1,c2); - mul_add_c(a[1],b[1],c3,c1,c2); - mul_add_c(a[0],b[2],c3,c1,c2); - r[2]=c3; - c3=0; - mul_add_c(a[0],b[3],c1,c2,c3); - mul_add_c(a[1],b[2],c1,c2,c3); - mul_add_c(a[2],b[1],c1,c2,c3); - mul_add_c(a[3],b[0],c1,c2,c3); - r[3]=c1; - c1=0; - mul_add_c(a[4],b[0],c2,c3,c1); - mul_add_c(a[3],b[1],c2,c3,c1); - mul_add_c(a[2],b[2],c2,c3,c1); - mul_add_c(a[1],b[3],c2,c3,c1); - mul_add_c(a[0],b[4],c2,c3,c1); - r[4]=c2; - c2=0; - mul_add_c(a[0],b[5],c3,c1,c2); - mul_add_c(a[1],b[4],c3,c1,c2); - mul_add_c(a[2],b[3],c3,c1,c2); - mul_add_c(a[3],b[2],c3,c1,c2); - mul_add_c(a[4],b[1],c3,c1,c2); - mul_add_c(a[5],b[0],c3,c1,c2); - r[5]=c3; - c3=0; - mul_add_c(a[6],b[0],c1,c2,c3); - mul_add_c(a[5],b[1],c1,c2,c3); - mul_add_c(a[4],b[2],c1,c2,c3); - mul_add_c(a[3],b[3],c1,c2,c3); - mul_add_c(a[2],b[4],c1,c2,c3); - mul_add_c(a[1],b[5],c1,c2,c3); - mul_add_c(a[0],b[6],c1,c2,c3); - r[6]=c1; - c1=0; - mul_add_c(a[0],b[7],c2,c3,c1); - mul_add_c(a[1],b[6],c2,c3,c1); - mul_add_c(a[2],b[5],c2,c3,c1); - mul_add_c(a[3],b[4],c2,c3,c1); - mul_add_c(a[4],b[3],c2,c3,c1); - mul_add_c(a[5],b[2],c2,c3,c1); - mul_add_c(a[6],b[1],c2,c3,c1); - mul_add_c(a[7],b[0],c2,c3,c1); - r[7]=c2; - c2=0; - mul_add_c(a[7],b[1],c3,c1,c2); - mul_add_c(a[6],b[2],c3,c1,c2); - mul_add_c(a[5],b[3],c3,c1,c2); - mul_add_c(a[4],b[4],c3,c1,c2); - mul_add_c(a[3],b[5],c3,c1,c2); - mul_add_c(a[2],b[6],c3,c1,c2); - mul_add_c(a[1],b[7],c3,c1,c2); - r[8]=c3; - c3=0; - mul_add_c(a[2],b[7],c1,c2,c3); - mul_add_c(a[3],b[6],c1,c2,c3); - mul_add_c(a[4],b[5],c1,c2,c3); - mul_add_c(a[5],b[4],c1,c2,c3); - mul_add_c(a[6],b[3],c1,c2,c3); - mul_add_c(a[7],b[2],c1,c2,c3); - r[9]=c1; - c1=0; - mul_add_c(a[7],b[3],c2,c3,c1); - mul_add_c(a[6],b[4],c2,c3,c1); - mul_add_c(a[5],b[5],c2,c3,c1); - mul_add_c(a[4],b[6],c2,c3,c1); - mul_add_c(a[3],b[7],c2,c3,c1); - r[10]=c2; - c2=0; - mul_add_c(a[4],b[7],c3,c1,c2); - mul_add_c(a[5],b[6],c3,c1,c2); - mul_add_c(a[6],b[5],c3,c1,c2); - mul_add_c(a[7],b[4],c3,c1,c2); - r[11]=c3; - c3=0; - mul_add_c(a[7],b[5],c1,c2,c3); - mul_add_c(a[6],b[6],c1,c2,c3); - mul_add_c(a[5],b[7],c1,c2,c3); - r[12]=c1; - c1=0; - mul_add_c(a[6],b[7],c2,c3,c1); - mul_add_c(a[7],b[6],c2,c3,c1); - r[13]=c2; - c2=0; - mul_add_c(a[7],b[7],c3,c1,c2); - r[14]=c3; - r[15]=c1; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + mul_add_c(a[0], b[0], c1, c2, c3); + r[0] = c1; + c1 = 0; + mul_add_c(a[0], b[1], c2, c3, c1); + mul_add_c(a[1], b[0], c2, c3, c1); + r[1] = c2; + c2 = 0; + mul_add_c(a[2], b[0], c3, c1, c2); + mul_add_c(a[1], b[1], c3, c1, c2); + mul_add_c(a[0], b[2], c3, c1, c2); + r[2] = c3; + c3 = 0; + mul_add_c(a[0], b[3], c1, c2, c3); + mul_add_c(a[1], b[2], c1, c2, c3); + mul_add_c(a[2], b[1], c1, c2, c3); + mul_add_c(a[3], b[0], c1, c2, c3); + r[3] = c1; + c1 = 0; + mul_add_c(a[4], b[0], c2, c3, c1); + mul_add_c(a[3], b[1], c2, c3, c1); + mul_add_c(a[2], b[2], c2, c3, c1); + mul_add_c(a[1], b[3], c2, c3, c1); + mul_add_c(a[0], b[4], c2, c3, c1); + r[4] = c2; + c2 = 0; + mul_add_c(a[0], b[5], c3, c1, c2); + mul_add_c(a[1], b[4], c3, c1, c2); + mul_add_c(a[2], b[3], c3, c1, c2); + mul_add_c(a[3], b[2], c3, c1, c2); + mul_add_c(a[4], b[1], c3, c1, c2); + mul_add_c(a[5], b[0], c3, c1, c2); + r[5] = c3; + c3 = 0; + mul_add_c(a[6], b[0], c1, c2, c3); + mul_add_c(a[5], b[1], c1, c2, c3); + mul_add_c(a[4], b[2], c1, c2, c3); + mul_add_c(a[3], b[3], c1, c2, c3); + mul_add_c(a[2], b[4], c1, c2, c3); + mul_add_c(a[1], b[5], c1, c2, c3); + mul_add_c(a[0], b[6], c1, c2, c3); + r[6] = c1; + c1 = 0; + mul_add_c(a[0], b[7], c2, c3, c1); + mul_add_c(a[1], b[6], c2, c3, c1); + mul_add_c(a[2], b[5], c2, c3, c1); + mul_add_c(a[3], b[4], c2, c3, c1); + mul_add_c(a[4], b[3], c2, c3, c1); + mul_add_c(a[5], b[2], c2, c3, c1); + mul_add_c(a[6], b[1], c2, c3, c1); + mul_add_c(a[7], b[0], c2, c3, c1); + r[7] = c2; + c2 = 0; + mul_add_c(a[7], b[1], c3, c1, c2); + mul_add_c(a[6], b[2], c3, c1, c2); + mul_add_c(a[5], b[3], c3, c1, c2); + mul_add_c(a[4], b[4], c3, c1, c2); + mul_add_c(a[3], b[5], c3, c1, c2); + mul_add_c(a[2], b[6], c3, c1, c2); + mul_add_c(a[1], b[7], c3, c1, c2); + r[8] = c3; + c3 = 0; + mul_add_c(a[2], b[7], c1, c2, c3); + mul_add_c(a[3], b[6], c1, c2, c3); + mul_add_c(a[4], b[5], c1, c2, c3); + mul_add_c(a[5], b[4], c1, c2, c3); + mul_add_c(a[6], b[3], c1, c2, c3); + mul_add_c(a[7], b[2], c1, c2, c3); + r[9] = c1; + c1 = 0; + mul_add_c(a[7], b[3], c2, c3, c1); + mul_add_c(a[6], b[4], c2, c3, c1); + mul_add_c(a[5], b[5], c2, c3, c1); + mul_add_c(a[4], b[6], c2, c3, c1); + mul_add_c(a[3], b[7], c2, c3, c1); + r[10] = c2; + c2 = 0; + mul_add_c(a[4], b[7], c3, c1, c2); + mul_add_c(a[5], b[6], c3, c1, c2); + mul_add_c(a[6], b[5], c3, c1, c2); + mul_add_c(a[7], b[4], c3, c1, c2); + r[11] = c3; + c3 = 0; + mul_add_c(a[7], b[5], c1, c2, c3); + mul_add_c(a[6], b[6], c1, c2, c3); + mul_add_c(a[5], b[7], c1, c2, c3); + r[12] = c1; + c1 = 0; + mul_add_c(a[6], b[7], c2, c3, c1); + mul_add_c(a[7], b[6], c2, c3, c1); + r[13] = c2; + c2 = 0; + mul_add_c(a[7], b[7], c3, c1, c2); + r[14] = c3; + r[15] = c1; +} void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) - { -#ifdef BN_LLONG - BN_ULLONG t; -#else - BN_ULONG bl,bh; -#endif - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - mul_add_c(a[0],b[0],c1,c2,c3); - r[0]=c1; - c1=0; - mul_add_c(a[0],b[1],c2,c3,c1); - mul_add_c(a[1],b[0],c2,c3,c1); - r[1]=c2; - c2=0; - mul_add_c(a[2],b[0],c3,c1,c2); - mul_add_c(a[1],b[1],c3,c1,c2); - mul_add_c(a[0],b[2],c3,c1,c2); - r[2]=c3; - c3=0; - mul_add_c(a[0],b[3],c1,c2,c3); - mul_add_c(a[1],b[2],c1,c2,c3); - mul_add_c(a[2],b[1],c1,c2,c3); - mul_add_c(a[3],b[0],c1,c2,c3); - r[3]=c1; - c1=0; - mul_add_c(a[3],b[1],c2,c3,c1); - mul_add_c(a[2],b[2],c2,c3,c1); - mul_add_c(a[1],b[3],c2,c3,c1); - r[4]=c2; - c2=0; - mul_add_c(a[2],b[3],c3,c1,c2); - mul_add_c(a[3],b[2],c3,c1,c2); - r[5]=c3; - c3=0; - mul_add_c(a[3],b[3],c1,c2,c3); - r[6]=c1; - r[7]=c2; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + mul_add_c(a[0], b[0], c1, c2, c3); + r[0] = c1; + c1 = 0; + mul_add_c(a[0], b[1], c2, c3, c1); + mul_add_c(a[1], b[0], c2, c3, c1); + r[1] = c2; + c2 = 0; + mul_add_c(a[2], b[0], c3, c1, c2); + mul_add_c(a[1], b[1], c3, c1, c2); + mul_add_c(a[0], b[2], c3, c1, c2); + r[2] = c3; + c3 = 0; + mul_add_c(a[0], b[3], c1, c2, c3); + mul_add_c(a[1], b[2], c1, c2, c3); + mul_add_c(a[2], b[1], c1, c2, c3); + mul_add_c(a[3], b[0], c1, c2, c3); + r[3] = c1; + c1 = 0; + mul_add_c(a[3], b[1], c2, c3, c1); + mul_add_c(a[2], b[2], c2, c3, c1); + mul_add_c(a[1], b[3], c2, c3, c1); + r[4] = c2; + c2 = 0; + mul_add_c(a[2], b[3], c3, c1, c2); + mul_add_c(a[3], b[2], c3, c1, c2); + r[5] = c3; + c3 = 0; + mul_add_c(a[3], b[3], c1, c2, c3); + r[6] = c1; + r[7] = c2; +} void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a) - { -#ifdef BN_LLONG - BN_ULLONG t,tt; -#else - BN_ULONG bl,bh; -#endif - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - sqr_add_c(a,0,c1,c2,c3); - r[0]=c1; - c1=0; - sqr_add_c2(a,1,0,c2,c3,c1); - r[1]=c2; - c2=0; - sqr_add_c(a,1,c3,c1,c2); - sqr_add_c2(a,2,0,c3,c1,c2); - r[2]=c3; - c3=0; - sqr_add_c2(a,3,0,c1,c2,c3); - sqr_add_c2(a,2,1,c1,c2,c3); - r[3]=c1; - c1=0; - sqr_add_c(a,2,c2,c3,c1); - sqr_add_c2(a,3,1,c2,c3,c1); - sqr_add_c2(a,4,0,c2,c3,c1); - r[4]=c2; - c2=0; - sqr_add_c2(a,5,0,c3,c1,c2); - sqr_add_c2(a,4,1,c3,c1,c2); - sqr_add_c2(a,3,2,c3,c1,c2); - r[5]=c3; - c3=0; - sqr_add_c(a,3,c1,c2,c3); - sqr_add_c2(a,4,2,c1,c2,c3); - sqr_add_c2(a,5,1,c1,c2,c3); - sqr_add_c2(a,6,0,c1,c2,c3); - r[6]=c1; - c1=0; - sqr_add_c2(a,7,0,c2,c3,c1); - sqr_add_c2(a,6,1,c2,c3,c1); - sqr_add_c2(a,5,2,c2,c3,c1); - sqr_add_c2(a,4,3,c2,c3,c1); - r[7]=c2; - c2=0; - sqr_add_c(a,4,c3,c1,c2); - sqr_add_c2(a,5,3,c3,c1,c2); - sqr_add_c2(a,6,2,c3,c1,c2); - sqr_add_c2(a,7,1,c3,c1,c2); - r[8]=c3; - c3=0; - sqr_add_c2(a,7,2,c1,c2,c3); - sqr_add_c2(a,6,3,c1,c2,c3); - sqr_add_c2(a,5,4,c1,c2,c3); - r[9]=c1; - c1=0; - sqr_add_c(a,5,c2,c3,c1); - sqr_add_c2(a,6,4,c2,c3,c1); - sqr_add_c2(a,7,3,c2,c3,c1); - r[10]=c2; - c2=0; - sqr_add_c2(a,7,4,c3,c1,c2); - sqr_add_c2(a,6,5,c3,c1,c2); - r[11]=c3; - c3=0; - sqr_add_c(a,6,c1,c2,c3); - sqr_add_c2(a,7,5,c1,c2,c3); - r[12]=c1; - c1=0; - sqr_add_c2(a,7,6,c2,c3,c1); - r[13]=c2; - c2=0; - sqr_add_c(a,7,c3,c1,c2); - r[14]=c3; - r[15]=c1; - } +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + sqr_add_c(a, 0, c1, c2, c3); + r[0] = c1; + c1 = 0; + sqr_add_c2(a, 1, 0, c2, c3, c1); + r[1] = c2; + c2 = 0; + sqr_add_c(a, 1, c3, c1, c2); + sqr_add_c2(a, 2, 0, c3, c1, c2); + r[2] = c3; + c3 = 0; + sqr_add_c2(a, 3, 0, c1, c2, c3); + sqr_add_c2(a, 2, 1, c1, c2, c3); + r[3] = c1; + c1 = 0; + sqr_add_c(a, 2, c2, c3, c1); + sqr_add_c2(a, 3, 1, c2, c3, c1); + sqr_add_c2(a, 4, 0, c2, c3, c1); + r[4] = c2; + c2 = 0; + sqr_add_c2(a, 5, 0, c3, c1, c2); + sqr_add_c2(a, 4, 1, c3, c1, c2); + sqr_add_c2(a, 3, 2, c3, c1, c2); + r[5] = c3; + c3 = 0; + sqr_add_c(a, 3, c1, c2, c3); + sqr_add_c2(a, 4, 2, c1, c2, c3); + sqr_add_c2(a, 5, 1, c1, c2, c3); + sqr_add_c2(a, 6, 0, c1, c2, c3); + r[6] = c1; + c1 = 0; + sqr_add_c2(a, 7, 0, c2, c3, c1); + sqr_add_c2(a, 6, 1, c2, c3, c1); + sqr_add_c2(a, 5, 2, c2, c3, c1); + sqr_add_c2(a, 4, 3, c2, c3, c1); + r[7] = c2; + c2 = 0; + sqr_add_c(a, 4, c3, c1, c2); + sqr_add_c2(a, 5, 3, c3, c1, c2); + sqr_add_c2(a, 6, 2, c3, c1, c2); + sqr_add_c2(a, 7, 1, c3, c1, c2); + r[8] = c3; + c3 = 0; + sqr_add_c2(a, 7, 2, c1, c2, c3); + sqr_add_c2(a, 6, 3, c1, c2, c3); + sqr_add_c2(a, 5, 4, c1, c2, c3); + r[9] = c1; + c1 = 0; + sqr_add_c(a, 5, c2, c3, c1); + sqr_add_c2(a, 6, 4, c2, c3, c1); + sqr_add_c2(a, 7, 3, c2, c3, c1); + r[10] = c2; + c2 = 0; + sqr_add_c2(a, 7, 4, c3, c1, c2); + sqr_add_c2(a, 6, 5, c3, c1, c2); + r[11] = c3; + c3 = 0; + sqr_add_c(a, 6, c1, c2, c3); + sqr_add_c2(a, 7, 5, c1, c2, c3); + r[12] = c1; + c1 = 0; + sqr_add_c2(a, 7, 6, c2, c3, c1); + r[13] = c2; + c2 = 0; + sqr_add_c(a, 7, c3, c1, c2); + r[14] = c3; + r[15] = c1; +} void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a) - { -#ifdef BN_LLONG - BN_ULLONG t,tt; -#else - BN_ULONG bl,bh; -#endif - BN_ULONG t1,t2; - BN_ULONG c1,c2,c3; - - c1=0; - c2=0; - c3=0; - sqr_add_c(a,0,c1,c2,c3); - r[0]=c1; - c1=0; - sqr_add_c2(a,1,0,c2,c3,c1); - r[1]=c2; - c2=0; - sqr_add_c(a,1,c3,c1,c2); - sqr_add_c2(a,2,0,c3,c1,c2); - r[2]=c3; - c3=0; - sqr_add_c2(a,3,0,c1,c2,c3); - sqr_add_c2(a,2,1,c1,c2,c3); - r[3]=c1; - c1=0; - sqr_add_c(a,2,c2,c3,c1); - sqr_add_c2(a,3,1,c2,c3,c1); - r[4]=c2; - c2=0; - sqr_add_c2(a,3,2,c3,c1,c2); - r[5]=c3; - c3=0; - sqr_add_c(a,3,c1,c2,c3); - r[6]=c1; - r[7]=c2; - } - -#ifdef OPENSSL_NO_ASM -#ifdef OPENSSL_BN_ASM_MONT -#include <alloca.h> +{ + BN_ULONG c1, c2, c3; + + c1 = 0; + c2 = 0; + c3 = 0; + sqr_add_c(a, 0, c1, c2, c3); + r[0] = c1; + c1 = 0; + sqr_add_c2(a, 1, 0, c2, c3, c1); + r[1] = c2; + c2 = 0; + sqr_add_c(a, 1, c3, c1, c2); + sqr_add_c2(a, 2, 0, c3, c1, c2); + r[2] = c3; + c3 = 0; + sqr_add_c2(a, 3, 0, c1, c2, c3); + sqr_add_c2(a, 2, 1, c1, c2, c3); + r[3] = c1; + c1 = 0; + sqr_add_c(a, 2, c2, c3, c1); + sqr_add_c2(a, 3, 1, c2, c3, c1); + r[4] = c2; + c2 = 0; + sqr_add_c2(a, 3, 2, c3, c1, c2); + r[5] = c3; + c3 = 0; + sqr_add_c(a, 3, c1, c2, c3); + r[6] = c1; + r[7] = c2; +} + +# ifdef OPENSSL_NO_ASM +# ifdef OPENSSL_BN_ASM_MONT +# include <alloca.h> /* * This is essentially reference implementation, which may or may not * result in performance improvement. E.g. on IA-32 this routine was @@ -844,187 +898,196 @@ void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a) * versions. Assembler vs. assembler improvement coefficients can * [and are known to] differ and are to be documented elsewhere. */ -int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0p, int num) - { - BN_ULONG c0,c1,ml,*tp,n0; -#ifdef mul64 - BN_ULONG mh; -#endif - volatile BN_ULONG *vp; - int i=0,j; - -#if 0 /* template for platform-specific implementation */ - if (ap==bp) return bn_sqr_mont(rp,ap,np,n0p,num); -#endif - vp = tp = alloca((num+2)*sizeof(BN_ULONG)); - - n0 = *n0p; - - c0 = 0; - ml = bp[0]; -#ifdef mul64 - mh = HBITS(ml); - ml = LBITS(ml); - for (j=0;j<num;++j) - mul(tp[j],ap[j],ml,mh,c0); -#else - for (j=0;j<num;++j) - mul(tp[j],ap[j],ml,c0); -#endif - - tp[num] = c0; - tp[num+1] = 0; - goto enter; - - for(i=0;i<num;i++) - { - c0 = 0; - ml = bp[i]; -#ifdef mul64 - mh = HBITS(ml); - ml = LBITS(ml); - for (j=0;j<num;++j) - mul_add(tp[j],ap[j],ml,mh,c0); -#else - for (j=0;j<num;++j) - mul_add(tp[j],ap[j],ml,c0); -#endif - c1 = (tp[num] + c0)&BN_MASK2; - tp[num] = c1; - tp[num+1] = (c1<c0?1:0); - enter: - c1 = tp[0]; - ml = (c1*n0)&BN_MASK2; - c0 = 0; -#ifdef mul64 - mh = HBITS(ml); - ml = LBITS(ml); - mul_add(c1,np[0],ml,mh,c0); -#else - mul_add(c1,ml,np[0],c0); -#endif - for(j=1;j<num;j++) - { - c1 = tp[j]; -#ifdef mul64 - mul_add(c1,np[j],ml,mh,c0); -#else - mul_add(c1,ml,np[j],c0); -#endif - tp[j-1] = c1&BN_MASK2; - } - c1 = (tp[num] + c0)&BN_MASK2; - tp[num-1] = c1; - tp[num] = tp[num+1] + (c1<c0?1:0); - } - - if (tp[num]!=0 || tp[num-1]>=np[num-1]) - { - c0 = bn_sub_words(rp,tp,np,num); - if (tp[num]!=0 || c0==0) - { - for(i=0;i<num+2;i++) vp[i] = 0; - return 1; - } - } - for(i=0;i<num;i++) rp[i] = tp[i], vp[i] = 0; - vp[num] = 0; - vp[num+1] = 0; - return 1; - } -#else +int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + const BN_ULONG *np, const BN_ULONG *n0p, int num) +{ + BN_ULONG c0, c1, ml, *tp, n0; +# ifdef mul64 + BN_ULONG mh; +# endif + volatile BN_ULONG *vp; + int i = 0, j; + +# if 0 /* template for platform-specific + * implementation */ + if (ap == bp) + return bn_sqr_mont(rp, ap, np, n0p, num); +# endif + vp = tp = alloca((num + 2) * sizeof(BN_ULONG)); + + n0 = *n0p; + + c0 = 0; + ml = bp[0]; +# ifdef mul64 + mh = HBITS(ml); + ml = LBITS(ml); + for (j = 0; j < num; ++j) + mul(tp[j], ap[j], ml, mh, c0); +# else + for (j = 0; j < num; ++j) + mul(tp[j], ap[j], ml, c0); +# endif + + tp[num] = c0; + tp[num + 1] = 0; + goto enter; + + for (i = 0; i < num; i++) { + c0 = 0; + ml = bp[i]; +# ifdef mul64 + mh = HBITS(ml); + ml = LBITS(ml); + for (j = 0; j < num; ++j) + mul_add(tp[j], ap[j], ml, mh, c0); +# else + for (j = 0; j < num; ++j) + mul_add(tp[j], ap[j], ml, c0); +# endif + c1 = (tp[num] + c0) & BN_MASK2; + tp[num] = c1; + tp[num + 1] = (c1 < c0 ? 1 : 0); + enter: + c1 = tp[0]; + ml = (c1 * n0) & BN_MASK2; + c0 = 0; +# ifdef mul64 + mh = HBITS(ml); + ml = LBITS(ml); + mul_add(c1, np[0], ml, mh, c0); +# else + mul_add(c1, ml, np[0], c0); +# endif + for (j = 1; j < num; j++) { + c1 = tp[j]; +# ifdef mul64 + mul_add(c1, np[j], ml, mh, c0); +# else + mul_add(c1, ml, np[j], c0); +# endif + tp[j - 1] = c1 & BN_MASK2; + } + c1 = (tp[num] + c0) & BN_MASK2; + tp[num - 1] = c1; + tp[num] = tp[num + 1] + (c1 < c0 ? 1 : 0); + } + + if (tp[num] != 0 || tp[num - 1] >= np[num - 1]) { + c0 = bn_sub_words(rp, tp, np, num); + if (tp[num] != 0 || c0 == 0) { + for (i = 0; i < num + 2; i++) + vp[i] = 0; + return 1; + } + } + for (i = 0; i < num; i++) + rp[i] = tp[i], vp[i] = 0; + vp[num] = 0; + vp[num + 1] = 0; + return 1; +} +# else /* * Return value of 0 indicates that multiplication/convolution was not * performed to signal the caller to fall down to alternative/original * code-path. */ -int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num) -{ return 0; } -#endif /* OPENSSL_BN_ASM_MONT */ -#endif +int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + const BN_ULONG *np, const BN_ULONG *n0, int num) +{ + return 0; +} +# endif /* OPENSSL_BN_ASM_MONT */ +# endif -#else /* !BN_MUL_COMBA */ +#else /* !BN_MUL_COMBA */ /* hmm... is it faster just to do a multiply? */ -#undef bn_sqr_comba4 +# undef bn_sqr_comba4 void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a) - { - BN_ULONG t[8]; - bn_sqr_normal(r,a,4,t); - } +{ + BN_ULONG t[8]; + bn_sqr_normal(r, a, 4, t); +} -#undef bn_sqr_comba8 +# undef bn_sqr_comba8 void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a) - { - BN_ULONG t[16]; - bn_sqr_normal(r,a,8,t); - } +{ + BN_ULONG t[16]; + bn_sqr_normal(r, a, 8, t); +} void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) - { - r[4]=bn_mul_words( &(r[0]),a,4,b[0]); - r[5]=bn_mul_add_words(&(r[1]),a,4,b[1]); - r[6]=bn_mul_add_words(&(r[2]),a,4,b[2]); - r[7]=bn_mul_add_words(&(r[3]),a,4,b[3]); - } +{ + r[4] = bn_mul_words(&(r[0]), a, 4, b[0]); + r[5] = bn_mul_add_words(&(r[1]), a, 4, b[1]); + r[6] = bn_mul_add_words(&(r[2]), a, 4, b[2]); + r[7] = bn_mul_add_words(&(r[3]), a, 4, b[3]); +} void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b) - { - r[ 8]=bn_mul_words( &(r[0]),a,8,b[0]); - r[ 9]=bn_mul_add_words(&(r[1]),a,8,b[1]); - r[10]=bn_mul_add_words(&(r[2]),a,8,b[2]); - r[11]=bn_mul_add_words(&(r[3]),a,8,b[3]); - r[12]=bn_mul_add_words(&(r[4]),a,8,b[4]); - r[13]=bn_mul_add_words(&(r[5]),a,8,b[5]); - r[14]=bn_mul_add_words(&(r[6]),a,8,b[6]); - r[15]=bn_mul_add_words(&(r[7]),a,8,b[7]); - } - -#ifdef OPENSSL_NO_ASM -#ifdef OPENSSL_BN_ASM_MONT -#include <alloca.h> -int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0p, int num) - { - BN_ULONG c0,c1,*tp,n0=*n0p; - volatile BN_ULONG *vp; - int i=0,j; - - vp = tp = alloca((num+2)*sizeof(BN_ULONG)); - - for(i=0;i<=num;i++) tp[i]=0; - - for(i=0;i<num;i++) - { - c0 = bn_mul_add_words(tp,ap,num,bp[i]); - c1 = (tp[num] + c0)&BN_MASK2; - tp[num] = c1; - tp[num+1] = (c1<c0?1:0); - - c0 = bn_mul_add_words(tp,np,num,tp[0]*n0); - c1 = (tp[num] + c0)&BN_MASK2; - tp[num] = c1; - tp[num+1] += (c1<c0?1:0); - for(j=0;j<=num;j++) tp[j]=tp[j+1]; - } - - if (tp[num]!=0 || tp[num-1]>=np[num-1]) - { - c0 = bn_sub_words(rp,tp,np,num); - if (tp[num]!=0 || c0==0) - { - for(i=0;i<num+2;i++) vp[i] = 0; - return 1; - } - } - for(i=0;i<num;i++) rp[i] = tp[i], vp[i] = 0; - vp[num] = 0; - vp[num+1] = 0; - return 1; - } -#else -int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num) -{ return 0; } -#endif /* OPENSSL_BN_ASM_MONT */ -#endif - -#endif /* !BN_MUL_COMBA */ +{ + r[8] = bn_mul_words(&(r[0]), a, 8, b[0]); + r[9] = bn_mul_add_words(&(r[1]), a, 8, b[1]); + r[10] = bn_mul_add_words(&(r[2]), a, 8, b[2]); + r[11] = bn_mul_add_words(&(r[3]), a, 8, b[3]); + r[12] = bn_mul_add_words(&(r[4]), a, 8, b[4]); + r[13] = bn_mul_add_words(&(r[5]), a, 8, b[5]); + r[14] = bn_mul_add_words(&(r[6]), a, 8, b[6]); + r[15] = bn_mul_add_words(&(r[7]), a, 8, b[7]); +} + +# ifdef OPENSSL_NO_ASM +# ifdef OPENSSL_BN_ASM_MONT +# include <alloca.h> +int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + const BN_ULONG *np, const BN_ULONG *n0p, int num) +{ + BN_ULONG c0, c1, *tp, n0 = *n0p; + volatile BN_ULONG *vp; + int i = 0, j; + + vp = tp = alloca((num + 2) * sizeof(BN_ULONG)); + + for (i = 0; i <= num; i++) + tp[i] = 0; + + for (i = 0; i < num; i++) { + c0 = bn_mul_add_words(tp, ap, num, bp[i]); + c1 = (tp[num] + c0) & BN_MASK2; + tp[num] = c1; + tp[num + 1] = (c1 < c0 ? 1 : 0); + + c0 = bn_mul_add_words(tp, np, num, tp[0] * n0); + c1 = (tp[num] + c0) & BN_MASK2; + tp[num] = c1; + tp[num + 1] += (c1 < c0 ? 1 : 0); + for (j = 0; j <= num; j++) + tp[j] = tp[j + 1]; + } + + if (tp[num] != 0 || tp[num - 1] >= np[num - 1]) { + c0 = bn_sub_words(rp, tp, np, num); + if (tp[num] != 0 || c0 == 0) { + for (i = 0; i < num + 2; i++) + vp[i] = 0; + return 1; + } + } + for (i = 0; i < num; i++) + rp[i] = tp[i], vp[i] = 0; + vp[num] = 0; + vp[num + 1] = 0; + return 1; +} +# else +int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + const BN_ULONG *np, const BN_ULONG *n0, int num) +{ + return 0; +} +# endif /* OPENSSL_BN_ASM_MONT */ +# endif + +#endif /* !BN_MUL_COMBA */ diff --git a/openssl/crypto/bn/bn_blind.c b/openssl/crypto/bn/bn_blind.c index 9ed8bc2b4..d448daa3c 100644 --- a/openssl/crypto/bn/bn_blind.c +++ b/openssl/crypto/bn/bn_blind.c @@ -7,7 +7,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -58,21 +58,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -87,10 +87,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -102,7 +102,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -113,273 +113,273 @@ #include "cryptlib.h" #include "bn_lcl.h" -#define BN_BLINDING_COUNTER 32 +#define BN_BLINDING_COUNTER 32 -struct bn_blinding_st - { - BIGNUM *A; - BIGNUM *Ai; - BIGNUM *e; - BIGNUM *mod; /* just a reference */ +struct bn_blinding_st { + BIGNUM *A; + BIGNUM *Ai; + BIGNUM *e; + BIGNUM *mod; /* just a reference */ #ifndef OPENSSL_NO_DEPRECATED - unsigned long thread_id; /* added in OpenSSL 0.9.6j and 0.9.7b; - * used only by crypto/rsa/rsa_eay.c, rsa_lib.c */ + unsigned long thread_id; /* added in OpenSSL 0.9.6j and 0.9.7b; used + * only by crypto/rsa/rsa_eay.c, rsa_lib.c */ #endif - CRYPTO_THREADID tid; - int counter; - unsigned long flags; - BN_MONT_CTX *m_ctx; - int (*bn_mod_exp)(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, - BN_MONT_CTX *m_ctx); - }; + CRYPTO_THREADID tid; + int counter; + unsigned long flags; + BN_MONT_CTX *m_ctx; + int (*bn_mod_exp) (BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx); +}; BN_BLINDING *BN_BLINDING_new(const BIGNUM *A, const BIGNUM *Ai, BIGNUM *mod) - { - BN_BLINDING *ret=NULL; - - bn_check_top(mod); - - if ((ret=(BN_BLINDING *)OPENSSL_malloc(sizeof(BN_BLINDING))) == NULL) - { - BNerr(BN_F_BN_BLINDING_NEW,ERR_R_MALLOC_FAILURE); - return(NULL); - } - memset(ret,0,sizeof(BN_BLINDING)); - if (A != NULL) - { - if ((ret->A = BN_dup(A)) == NULL) goto err; - } - if (Ai != NULL) - { - if ((ret->Ai = BN_dup(Ai)) == NULL) goto err; - } - - /* save a copy of mod in the BN_BLINDING structure */ - if ((ret->mod = BN_dup(mod)) == NULL) goto err; - if (BN_get_flags(mod, BN_FLG_CONSTTIME) != 0) - BN_set_flags(ret->mod, BN_FLG_CONSTTIME); - - /* Set the counter to the special value -1 - * to indicate that this is never-used fresh blinding - * that does not need updating before first use. */ - ret->counter = -1; - CRYPTO_THREADID_current(&ret->tid); - return(ret); -err: - if (ret != NULL) BN_BLINDING_free(ret); - return(NULL); - } +{ + BN_BLINDING *ret = NULL; + + bn_check_top(mod); + + if ((ret = (BN_BLINDING *)OPENSSL_malloc(sizeof(BN_BLINDING))) == NULL) { + BNerr(BN_F_BN_BLINDING_NEW, ERR_R_MALLOC_FAILURE); + return (NULL); + } + memset(ret, 0, sizeof(BN_BLINDING)); + if (A != NULL) { + if ((ret->A = BN_dup(A)) == NULL) + goto err; + } + if (Ai != NULL) { + if ((ret->Ai = BN_dup(Ai)) == NULL) + goto err; + } + + /* save a copy of mod in the BN_BLINDING structure */ + if ((ret->mod = BN_dup(mod)) == NULL) + goto err; + if (BN_get_flags(mod, BN_FLG_CONSTTIME) != 0) + BN_set_flags(ret->mod, BN_FLG_CONSTTIME); + + /* + * Set the counter to the special value -1 to indicate that this is + * never-used fresh blinding that does not need updating before first + * use. + */ + ret->counter = -1; + CRYPTO_THREADID_current(&ret->tid); + return (ret); + err: + if (ret != NULL) + BN_BLINDING_free(ret); + return (NULL); +} void BN_BLINDING_free(BN_BLINDING *r) - { - if(r == NULL) - return; - - if (r->A != NULL) BN_free(r->A ); - if (r->Ai != NULL) BN_free(r->Ai); - if (r->e != NULL) BN_free(r->e ); - if (r->mod != NULL) BN_free(r->mod); - OPENSSL_free(r); - } +{ + if (r == NULL) + return; + + if (r->A != NULL) + BN_free(r->A); + if (r->Ai != NULL) + BN_free(r->Ai); + if (r->e != NULL) + BN_free(r->e); + if (r->mod != NULL) + BN_free(r->mod); + OPENSSL_free(r); +} int BN_BLINDING_update(BN_BLINDING *b, BN_CTX *ctx) - { - int ret=0; - - if ((b->A == NULL) || (b->Ai == NULL)) - { - BNerr(BN_F_BN_BLINDING_UPDATE,BN_R_NOT_INITIALIZED); - goto err; - } - - if (b->counter == -1) - b->counter = 0; - - if (++b->counter == BN_BLINDING_COUNTER && b->e != NULL && - !(b->flags & BN_BLINDING_NO_RECREATE)) - { - /* re-create blinding parameters */ - if (!BN_BLINDING_create_param(b, NULL, NULL, ctx, NULL, NULL)) - goto err; - } - else if (!(b->flags & BN_BLINDING_NO_UPDATE)) - { - if (!BN_mod_mul(b->A,b->A,b->A,b->mod,ctx)) goto err; - if (!BN_mod_mul(b->Ai,b->Ai,b->Ai,b->mod,ctx)) goto err; - } - - ret=1; -err: - if (b->counter == BN_BLINDING_COUNTER) - b->counter = 0; - return(ret); - } +{ + int ret = 0; + + if ((b->A == NULL) || (b->Ai == NULL)) { + BNerr(BN_F_BN_BLINDING_UPDATE, BN_R_NOT_INITIALIZED); + goto err; + } + + if (b->counter == -1) + b->counter = 0; + + if (++b->counter == BN_BLINDING_COUNTER && b->e != NULL && + !(b->flags & BN_BLINDING_NO_RECREATE)) { + /* re-create blinding parameters */ + if (!BN_BLINDING_create_param(b, NULL, NULL, ctx, NULL, NULL)) + goto err; + } else if (!(b->flags & BN_BLINDING_NO_UPDATE)) { + if (!BN_mod_mul(b->A, b->A, b->A, b->mod, ctx)) + goto err; + if (!BN_mod_mul(b->Ai, b->Ai, b->Ai, b->mod, ctx)) + goto err; + } + + ret = 1; + err: + if (b->counter == BN_BLINDING_COUNTER) + b->counter = 0; + return (ret); +} int BN_BLINDING_convert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx) - { - return BN_BLINDING_convert_ex(n, NULL, b, ctx); - } +{ + return BN_BLINDING_convert_ex(n, NULL, b, ctx); +} int BN_BLINDING_convert_ex(BIGNUM *n, BIGNUM *r, BN_BLINDING *b, BN_CTX *ctx) - { - int ret = 1; +{ + int ret = 1; - bn_check_top(n); + bn_check_top(n); - if ((b->A == NULL) || (b->Ai == NULL)) - { - BNerr(BN_F_BN_BLINDING_CONVERT_EX,BN_R_NOT_INITIALIZED); - return(0); - } + if ((b->A == NULL) || (b->Ai == NULL)) { + BNerr(BN_F_BN_BLINDING_CONVERT_EX, BN_R_NOT_INITIALIZED); + return (0); + } - if (b->counter == -1) - /* Fresh blinding, doesn't need updating. */ - b->counter = 0; - else if (!BN_BLINDING_update(b,ctx)) - return(0); + if (b->counter == -1) + /* Fresh blinding, doesn't need updating. */ + b->counter = 0; + else if (!BN_BLINDING_update(b, ctx)) + return (0); - if (r != NULL) - { - if (!BN_copy(r, b->Ai)) ret=0; - } + if (r != NULL) { + if (!BN_copy(r, b->Ai)) + ret = 0; + } - if (!BN_mod_mul(n,n,b->A,b->mod,ctx)) ret=0; - - return ret; - } + if (!BN_mod_mul(n, n, b->A, b->mod, ctx)) + ret = 0; + + return ret; +} int BN_BLINDING_invert(BIGNUM *n, BN_BLINDING *b, BN_CTX *ctx) - { - return BN_BLINDING_invert_ex(n, NULL, b, ctx); - } - -int BN_BLINDING_invert_ex(BIGNUM *n, const BIGNUM *r, BN_BLINDING *b, BN_CTX *ctx) - { - int ret; - - bn_check_top(n); - - if (r != NULL) - ret = BN_mod_mul(n, n, r, b->mod, ctx); - else - { - if (b->Ai == NULL) - { - BNerr(BN_F_BN_BLINDING_INVERT_EX,BN_R_NOT_INITIALIZED); - return(0); - } - ret = BN_mod_mul(n, n, b->Ai, b->mod, ctx); - } - - bn_check_top(n); - return(ret); - } +{ + return BN_BLINDING_invert_ex(n, NULL, b, ctx); +} + +int BN_BLINDING_invert_ex(BIGNUM *n, const BIGNUM *r, BN_BLINDING *b, + BN_CTX *ctx) +{ + int ret; + + bn_check_top(n); + + if (r != NULL) + ret = BN_mod_mul(n, n, r, b->mod, ctx); + else { + if (b->Ai == NULL) { + BNerr(BN_F_BN_BLINDING_INVERT_EX, BN_R_NOT_INITIALIZED); + return (0); + } + ret = BN_mod_mul(n, n, b->Ai, b->mod, ctx); + } + + bn_check_top(n); + return (ret); +} #ifndef OPENSSL_NO_DEPRECATED unsigned long BN_BLINDING_get_thread_id(const BN_BLINDING *b) - { - return b->thread_id; - } +{ + return b->thread_id; +} void BN_BLINDING_set_thread_id(BN_BLINDING *b, unsigned long n) - { - b->thread_id = n; - } +{ + b->thread_id = n; +} #endif CRYPTO_THREADID *BN_BLINDING_thread_id(BN_BLINDING *b) - { - return &b->tid; - } +{ + return &b->tid; +} unsigned long BN_BLINDING_get_flags(const BN_BLINDING *b) - { - return b->flags; - } +{ + return b->flags; +} void BN_BLINDING_set_flags(BN_BLINDING *b, unsigned long flags) - { - b->flags = flags; - } +{ + b->flags = flags; +} BN_BLINDING *BN_BLINDING_create_param(BN_BLINDING *b, - const BIGNUM *e, BIGNUM *m, BN_CTX *ctx, - int (*bn_mod_exp)(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *m_ctx), - BN_MONT_CTX *m_ctx) + const BIGNUM *e, BIGNUM *m, BN_CTX *ctx, + int (*bn_mod_exp) (BIGNUM *r, + const BIGNUM *a, + const BIGNUM *p, + const BIGNUM *m, + BN_CTX *ctx, + BN_MONT_CTX *m_ctx), + BN_MONT_CTX *m_ctx) { - int retry_counter = 32; - BN_BLINDING *ret = NULL; - - if (b == NULL) - ret = BN_BLINDING_new(NULL, NULL, m); - else - ret = b; - - if (ret == NULL) - goto err; - - if (ret->A == NULL && (ret->A = BN_new()) == NULL) - goto err; - if (ret->Ai == NULL && (ret->Ai = BN_new()) == NULL) - goto err; - - if (e != NULL) - { - if (ret->e != NULL) - BN_free(ret->e); - ret->e = BN_dup(e); - } - if (ret->e == NULL) - goto err; - - if (bn_mod_exp != NULL) - ret->bn_mod_exp = bn_mod_exp; - if (m_ctx != NULL) - ret->m_ctx = m_ctx; - - do { - if (!BN_rand_range(ret->A, ret->mod)) goto err; - if (BN_mod_inverse(ret->Ai, ret->A, ret->mod, ctx) == NULL) - { - /* this should almost never happen for good RSA keys */ - unsigned long error = ERR_peek_last_error(); - if (ERR_GET_REASON(error) == BN_R_NO_INVERSE) - { - if (retry_counter-- == 0) - { - BNerr(BN_F_BN_BLINDING_CREATE_PARAM, - BN_R_TOO_MANY_ITERATIONS); - goto err; - } - ERR_clear_error(); - } - else - goto err; - } - else - break; - } while (1); - - if (ret->bn_mod_exp != NULL && ret->m_ctx != NULL) - { - if (!ret->bn_mod_exp(ret->A, ret->A, ret->e, ret->mod, ctx, ret->m_ctx)) - goto err; - } - else - { - if (!BN_mod_exp(ret->A, ret->A, ret->e, ret->mod, ctx)) - goto err; - } - - return ret; -err: - if (b == NULL && ret != NULL) - { - BN_BLINDING_free(ret); - ret = NULL; - } - - return ret; + int retry_counter = 32; + BN_BLINDING *ret = NULL; + + if (b == NULL) + ret = BN_BLINDING_new(NULL, NULL, m); + else + ret = b; + + if (ret == NULL) + goto err; + + if (ret->A == NULL && (ret->A = BN_new()) == NULL) + goto err; + if (ret->Ai == NULL && (ret->Ai = BN_new()) == NULL) + goto err; + + if (e != NULL) { + if (ret->e != NULL) + BN_free(ret->e); + ret->e = BN_dup(e); + } + if (ret->e == NULL) + goto err; + + if (bn_mod_exp != NULL) + ret->bn_mod_exp = bn_mod_exp; + if (m_ctx != NULL) + ret->m_ctx = m_ctx; + + do { + if (!BN_rand_range(ret->A, ret->mod)) + goto err; + if (BN_mod_inverse(ret->Ai, ret->A, ret->mod, ctx) == NULL) { + /* + * this should almost never happen for good RSA keys + */ + unsigned long error = ERR_peek_last_error(); + if (ERR_GET_REASON(error) == BN_R_NO_INVERSE) { + if (retry_counter-- == 0) { + BNerr(BN_F_BN_BLINDING_CREATE_PARAM, + BN_R_TOO_MANY_ITERATIONS); + goto err; + } + ERR_clear_error(); + } else + goto err; + } else + break; + } while (1); + + if (ret->bn_mod_exp != NULL && ret->m_ctx != NULL) { + if (!ret->bn_mod_exp + (ret->A, ret->A, ret->e, ret->mod, ctx, ret->m_ctx)) + goto err; + } else { + if (!BN_mod_exp(ret->A, ret->A, ret->e, ret->mod, ctx)) + goto err; + } + + return ret; + err: + if (b == NULL && ret != NULL) { + BN_BLINDING_free(ret); + ret = NULL; + } + + return ret; } diff --git a/openssl/crypto/bn/bn_const.c b/openssl/crypto/bn/bn_const.c index eb60a25b3..12c3208c2 100644 --- a/openssl/crypto/bn/bn_const.c +++ b/openssl/crypto/bn/bn_const.c @@ -3,7 +3,8 @@ #include "bn.h" -/* "First Oakley Default Group" from RFC2409, section 6.1. +/*- + * "First Oakley Default Group" from RFC2409, section 6.1. * * The prime is: 2^768 - 2 ^704 - 1 + 2^64 * { [2^638 pi] + 149686 } * @@ -12,21 +13,26 @@ */ BIGNUM *get_rfc2409_prime_768(BIGNUM *bn) - { - static const unsigned char RFC2409_PRIME_768[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x3A,0x36,0x20,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC2409_PRIME_768,sizeof(RFC2409_PRIME_768),bn); - } +{ + static const unsigned char RFC2409_PRIME_768[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x3A, 0x36, 0x20, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC2409_PRIME_768, sizeof(RFC2409_PRIME_768), bn); +} -/* "Second Oakley Default Group" from RFC2409, section 6.2. +/*- + * "Second Oakley Default Group" from RFC2409, section 6.2. * * The prime is: 2^1024 - 2^960 - 1 + 2^64 * { [2^894 pi] + 129093 }. * @@ -35,24 +41,30 @@ BIGNUM *get_rfc2409_prime_768(BIGNUM *bn) */ BIGNUM *get_rfc2409_prime_1024(BIGNUM *bn) - { - static const unsigned char RFC2409_PRIME_1024[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE6,0x53,0x81, - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC2409_PRIME_1024,sizeof(RFC2409_PRIME_1024),bn); - } +{ + static const unsigned char RFC2409_PRIME_1024[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE6, 0x53, 0x81, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC2409_PRIME_1024, sizeof(RFC2409_PRIME_1024), bn); +} -/* "1536-bit MODP Group" from RFC3526, Section 2. +/*- + * "1536-bit MODP Group" from RFC3526, Section 2. * * The prime is: 2^1536 - 2^1472 - 1 + 2^64 * { [2^1406 pi] + 741804 } * @@ -61,29 +73,38 @@ BIGNUM *get_rfc2409_prime_1024(BIGNUM *bn) */ BIGNUM *get_rfc3526_prime_1536(BIGNUM *bn) - { - static const unsigned char RFC3526_PRIME_1536[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE4,0x5B,0x3D, - 0xC2,0x00,0x7C,0xB8,0xA1,0x63,0xBF,0x05,0x98,0xDA,0x48,0x36, - 0x1C,0x55,0xD3,0x9A,0x69,0x16,0x3F,0xA8,0xFD,0x24,0xCF,0x5F, - 0x83,0x65,0x5D,0x23,0xDC,0xA3,0xAD,0x96,0x1C,0x62,0xF3,0x56, - 0x20,0x85,0x52,0xBB,0x9E,0xD5,0x29,0x07,0x70,0x96,0x96,0x6D, - 0x67,0x0C,0x35,0x4E,0x4A,0xBC,0x98,0x04,0xF1,0x74,0x6C,0x08, - 0xCA,0x23,0x73,0x27,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC3526_PRIME_1536,sizeof(RFC3526_PRIME_1536),bn); - } +{ + static const unsigned char RFC3526_PRIME_1536[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE4, 0x5B, 0x3D, + 0xC2, 0x00, 0x7C, 0xB8, 0xA1, 0x63, 0xBF, 0x05, + 0x98, 0xDA, 0x48, 0x36, 0x1C, 0x55, 0xD3, 0x9A, + 0x69, 0x16, 0x3F, 0xA8, 0xFD, 0x24, 0xCF, 0x5F, + 0x83, 0x65, 0x5D, 0x23, 0xDC, 0xA3, 0xAD, 0x96, + 0x1C, 0x62, 0xF3, 0x56, 0x20, 0x85, 0x52, 0xBB, + 0x9E, 0xD5, 0x29, 0x07, 0x70, 0x96, 0x96, 0x6D, + 0x67, 0x0C, 0x35, 0x4E, 0x4A, 0xBC, 0x98, 0x04, + 0xF1, 0x74, 0x6C, 0x08, 0xCA, 0x23, 0x73, 0x27, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC3526_PRIME_1536, sizeof(RFC3526_PRIME_1536), bn); +} -/* "2048-bit MODP Group" from RFC3526, Section 3. +/*- + * "2048-bit MODP Group" from RFC3526, Section 3. * * The prime is: 2^2048 - 2^1984 - 1 + 2^64 * { [2^1918 pi] + 124476 } * @@ -91,35 +112,46 @@ BIGNUM *get_rfc3526_prime_1536(BIGNUM *bn) */ BIGNUM *get_rfc3526_prime_2048(BIGNUM *bn) - { - static const unsigned char RFC3526_PRIME_2048[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE4,0x5B,0x3D, - 0xC2,0x00,0x7C,0xB8,0xA1,0x63,0xBF,0x05,0x98,0xDA,0x48,0x36, - 0x1C,0x55,0xD3,0x9A,0x69,0x16,0x3F,0xA8,0xFD,0x24,0xCF,0x5F, - 0x83,0x65,0x5D,0x23,0xDC,0xA3,0xAD,0x96,0x1C,0x62,0xF3,0x56, - 0x20,0x85,0x52,0xBB,0x9E,0xD5,0x29,0x07,0x70,0x96,0x96,0x6D, - 0x67,0x0C,0x35,0x4E,0x4A,0xBC,0x98,0x04,0xF1,0x74,0x6C,0x08, - 0xCA,0x18,0x21,0x7C,0x32,0x90,0x5E,0x46,0x2E,0x36,0xCE,0x3B, - 0xE3,0x9E,0x77,0x2C,0x18,0x0E,0x86,0x03,0x9B,0x27,0x83,0xA2, - 0xEC,0x07,0xA2,0x8F,0xB5,0xC5,0x5D,0xF0,0x6F,0x4C,0x52,0xC9, - 0xDE,0x2B,0xCB,0xF6,0x95,0x58,0x17,0x18,0x39,0x95,0x49,0x7C, - 0xEA,0x95,0x6A,0xE5,0x15,0xD2,0x26,0x18,0x98,0xFA,0x05,0x10, - 0x15,0x72,0x8E,0x5A,0x8A,0xAC,0xAA,0x68,0xFF,0xFF,0xFF,0xFF, - 0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC3526_PRIME_2048,sizeof(RFC3526_PRIME_2048),bn); - } +{ + static const unsigned char RFC3526_PRIME_2048[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE4, 0x5B, 0x3D, + 0xC2, 0x00, 0x7C, 0xB8, 0xA1, 0x63, 0xBF, 0x05, + 0x98, 0xDA, 0x48, 0x36, 0x1C, 0x55, 0xD3, 0x9A, + 0x69, 0x16, 0x3F, 0xA8, 0xFD, 0x24, 0xCF, 0x5F, + 0x83, 0x65, 0x5D, 0x23, 0xDC, 0xA3, 0xAD, 0x96, + 0x1C, 0x62, 0xF3, 0x56, 0x20, 0x85, 0x52, 0xBB, + 0x9E, 0xD5, 0x29, 0x07, 0x70, 0x96, 0x96, 0x6D, + 0x67, 0x0C, 0x35, 0x4E, 0x4A, 0xBC, 0x98, 0x04, + 0xF1, 0x74, 0x6C, 0x08, 0xCA, 0x18, 0x21, 0x7C, + 0x32, 0x90, 0x5E, 0x46, 0x2E, 0x36, 0xCE, 0x3B, + 0xE3, 0x9E, 0x77, 0x2C, 0x18, 0x0E, 0x86, 0x03, + 0x9B, 0x27, 0x83, 0xA2, 0xEC, 0x07, 0xA2, 0x8F, + 0xB5, 0xC5, 0x5D, 0xF0, 0x6F, 0x4C, 0x52, 0xC9, + 0xDE, 0x2B, 0xCB, 0xF6, 0x95, 0x58, 0x17, 0x18, + 0x39, 0x95, 0x49, 0x7C, 0xEA, 0x95, 0x6A, 0xE5, + 0x15, 0xD2, 0x26, 0x18, 0x98, 0xFA, 0x05, 0x10, + 0x15, 0x72, 0x8E, 0x5A, 0x8A, 0xAC, 0xAA, 0x68, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC3526_PRIME_2048, sizeof(RFC3526_PRIME_2048), bn); +} -/* "3072-bit MODP Group" from RFC3526, Section 4. +/*- + * "3072-bit MODP Group" from RFC3526, Section 4. * * The prime is: 2^3072 - 2^3008 - 1 + 2^64 * { [2^2942 pi] + 1690314 } * @@ -127,45 +159,62 @@ BIGNUM *get_rfc3526_prime_2048(BIGNUM *bn) */ BIGNUM *get_rfc3526_prime_3072(BIGNUM *bn) - { - static const unsigned char RFC3526_PRIME_3072[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE4,0x5B,0x3D, - 0xC2,0x00,0x7C,0xB8,0xA1,0x63,0xBF,0x05,0x98,0xDA,0x48,0x36, - 0x1C,0x55,0xD3,0x9A,0x69,0x16,0x3F,0xA8,0xFD,0x24,0xCF,0x5F, - 0x83,0x65,0x5D,0x23,0xDC,0xA3,0xAD,0x96,0x1C,0x62,0xF3,0x56, - 0x20,0x85,0x52,0xBB,0x9E,0xD5,0x29,0x07,0x70,0x96,0x96,0x6D, - 0x67,0x0C,0x35,0x4E,0x4A,0xBC,0x98,0x04,0xF1,0x74,0x6C,0x08, - 0xCA,0x18,0x21,0x7C,0x32,0x90,0x5E,0x46,0x2E,0x36,0xCE,0x3B, - 0xE3,0x9E,0x77,0x2C,0x18,0x0E,0x86,0x03,0x9B,0x27,0x83,0xA2, - 0xEC,0x07,0xA2,0x8F,0xB5,0xC5,0x5D,0xF0,0x6F,0x4C,0x52,0xC9, - 0xDE,0x2B,0xCB,0xF6,0x95,0x58,0x17,0x18,0x39,0x95,0x49,0x7C, - 0xEA,0x95,0x6A,0xE5,0x15,0xD2,0x26,0x18,0x98,0xFA,0x05,0x10, - 0x15,0x72,0x8E,0x5A,0x8A,0xAA,0xC4,0x2D,0xAD,0x33,0x17,0x0D, - 0x04,0x50,0x7A,0x33,0xA8,0x55,0x21,0xAB,0xDF,0x1C,0xBA,0x64, - 0xEC,0xFB,0x85,0x04,0x58,0xDB,0xEF,0x0A,0x8A,0xEA,0x71,0x57, - 0x5D,0x06,0x0C,0x7D,0xB3,0x97,0x0F,0x85,0xA6,0xE1,0xE4,0xC7, - 0xAB,0xF5,0xAE,0x8C,0xDB,0x09,0x33,0xD7,0x1E,0x8C,0x94,0xE0, - 0x4A,0x25,0x61,0x9D,0xCE,0xE3,0xD2,0x26,0x1A,0xD2,0xEE,0x6B, - 0xF1,0x2F,0xFA,0x06,0xD9,0x8A,0x08,0x64,0xD8,0x76,0x02,0x73, - 0x3E,0xC8,0x6A,0x64,0x52,0x1F,0x2B,0x18,0x17,0x7B,0x20,0x0C, - 0xBB,0xE1,0x17,0x57,0x7A,0x61,0x5D,0x6C,0x77,0x09,0x88,0xC0, - 0xBA,0xD9,0x46,0xE2,0x08,0xE2,0x4F,0xA0,0x74,0xE5,0xAB,0x31, - 0x43,0xDB,0x5B,0xFC,0xE0,0xFD,0x10,0x8E,0x4B,0x82,0xD1,0x20, - 0xA9,0x3A,0xD2,0xCA,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC3526_PRIME_3072,sizeof(RFC3526_PRIME_3072),bn); - } +{ + static const unsigned char RFC3526_PRIME_3072[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE4, 0x5B, 0x3D, + 0xC2, 0x00, 0x7C, 0xB8, 0xA1, 0x63, 0xBF, 0x05, + 0x98, 0xDA, 0x48, 0x36, 0x1C, 0x55, 0xD3, 0x9A, + 0x69, 0x16, 0x3F, 0xA8, 0xFD, 0x24, 0xCF, 0x5F, + 0x83, 0x65, 0x5D, 0x23, 0xDC, 0xA3, 0xAD, 0x96, + 0x1C, 0x62, 0xF3, 0x56, 0x20, 0x85, 0x52, 0xBB, + 0x9E, 0xD5, 0x29, 0x07, 0x70, 0x96, 0x96, 0x6D, + 0x67, 0x0C, 0x35, 0x4E, 0x4A, 0xBC, 0x98, 0x04, + 0xF1, 0x74, 0x6C, 0x08, 0xCA, 0x18, 0x21, 0x7C, + 0x32, 0x90, 0x5E, 0x46, 0x2E, 0x36, 0xCE, 0x3B, + 0xE3, 0x9E, 0x77, 0x2C, 0x18, 0x0E, 0x86, 0x03, + 0x9B, 0x27, 0x83, 0xA2, 0xEC, 0x07, 0xA2, 0x8F, + 0xB5, 0xC5, 0x5D, 0xF0, 0x6F, 0x4C, 0x52, 0xC9, + 0xDE, 0x2B, 0xCB, 0xF6, 0x95, 0x58, 0x17, 0x18, + 0x39, 0x95, 0x49, 0x7C, 0xEA, 0x95, 0x6A, 0xE5, + 0x15, 0xD2, 0x26, 0x18, 0x98, 0xFA, 0x05, 0x10, + 0x15, 0x72, 0x8E, 0x5A, 0x8A, 0xAA, 0xC4, 0x2D, + 0xAD, 0x33, 0x17, 0x0D, 0x04, 0x50, 0x7A, 0x33, + 0xA8, 0x55, 0x21, 0xAB, 0xDF, 0x1C, 0xBA, 0x64, + 0xEC, 0xFB, 0x85, 0x04, 0x58, 0xDB, 0xEF, 0x0A, + 0x8A, 0xEA, 0x71, 0x57, 0x5D, 0x06, 0x0C, 0x7D, + 0xB3, 0x97, 0x0F, 0x85, 0xA6, 0xE1, 0xE4, 0xC7, + 0xAB, 0xF5, 0xAE, 0x8C, 0xDB, 0x09, 0x33, 0xD7, + 0x1E, 0x8C, 0x94, 0xE0, 0x4A, 0x25, 0x61, 0x9D, + 0xCE, 0xE3, 0xD2, 0x26, 0x1A, 0xD2, 0xEE, 0x6B, + 0xF1, 0x2F, 0xFA, 0x06, 0xD9, 0x8A, 0x08, 0x64, + 0xD8, 0x76, 0x02, 0x73, 0x3E, 0xC8, 0x6A, 0x64, + 0x52, 0x1F, 0x2B, 0x18, 0x17, 0x7B, 0x20, 0x0C, + 0xBB, 0xE1, 0x17, 0x57, 0x7A, 0x61, 0x5D, 0x6C, + 0x77, 0x09, 0x88, 0xC0, 0xBA, 0xD9, 0x46, 0xE2, + 0x08, 0xE2, 0x4F, 0xA0, 0x74, 0xE5, 0xAB, 0x31, + 0x43, 0xDB, 0x5B, 0xFC, 0xE0, 0xFD, 0x10, 0x8E, + 0x4B, 0x82, 0xD1, 0x20, 0xA9, 0x3A, 0xD2, 0xCA, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC3526_PRIME_3072, sizeof(RFC3526_PRIME_3072), bn); +} -/* "4096-bit MODP Group" from RFC3526, Section 5. +/*- + * "4096-bit MODP Group" from RFC3526, Section 5. * * The prime is: 2^4096 - 2^4032 - 1 + 2^64 * { [2^3966 pi] + 240904 } * @@ -173,56 +222,78 @@ BIGNUM *get_rfc3526_prime_3072(BIGNUM *bn) */ BIGNUM *get_rfc3526_prime_4096(BIGNUM *bn) - { - static const unsigned char RFC3526_PRIME_4096[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE4,0x5B,0x3D, - 0xC2,0x00,0x7C,0xB8,0xA1,0x63,0xBF,0x05,0x98,0xDA,0x48,0x36, - 0x1C,0x55,0xD3,0x9A,0x69,0x16,0x3F,0xA8,0xFD,0x24,0xCF,0x5F, - 0x83,0x65,0x5D,0x23,0xDC,0xA3,0xAD,0x96,0x1C,0x62,0xF3,0x56, - 0x20,0x85,0x52,0xBB,0x9E,0xD5,0x29,0x07,0x70,0x96,0x96,0x6D, - 0x67,0x0C,0x35,0x4E,0x4A,0xBC,0x98,0x04,0xF1,0x74,0x6C,0x08, - 0xCA,0x18,0x21,0x7C,0x32,0x90,0x5E,0x46,0x2E,0x36,0xCE,0x3B, - 0xE3,0x9E,0x77,0x2C,0x18,0x0E,0x86,0x03,0x9B,0x27,0x83,0xA2, - 0xEC,0x07,0xA2,0x8F,0xB5,0xC5,0x5D,0xF0,0x6F,0x4C,0x52,0xC9, - 0xDE,0x2B,0xCB,0xF6,0x95,0x58,0x17,0x18,0x39,0x95,0x49,0x7C, - 0xEA,0x95,0x6A,0xE5,0x15,0xD2,0x26,0x18,0x98,0xFA,0x05,0x10, - 0x15,0x72,0x8E,0x5A,0x8A,0xAA,0xC4,0x2D,0xAD,0x33,0x17,0x0D, - 0x04,0x50,0x7A,0x33,0xA8,0x55,0x21,0xAB,0xDF,0x1C,0xBA,0x64, - 0xEC,0xFB,0x85,0x04,0x58,0xDB,0xEF,0x0A,0x8A,0xEA,0x71,0x57, - 0x5D,0x06,0x0C,0x7D,0xB3,0x97,0x0F,0x85,0xA6,0xE1,0xE4,0xC7, - 0xAB,0xF5,0xAE,0x8C,0xDB,0x09,0x33,0xD7,0x1E,0x8C,0x94,0xE0, - 0x4A,0x25,0x61,0x9D,0xCE,0xE3,0xD2,0x26,0x1A,0xD2,0xEE,0x6B, - 0xF1,0x2F,0xFA,0x06,0xD9,0x8A,0x08,0x64,0xD8,0x76,0x02,0x73, - 0x3E,0xC8,0x6A,0x64,0x52,0x1F,0x2B,0x18,0x17,0x7B,0x20,0x0C, - 0xBB,0xE1,0x17,0x57,0x7A,0x61,0x5D,0x6C,0x77,0x09,0x88,0xC0, - 0xBA,0xD9,0x46,0xE2,0x08,0xE2,0x4F,0xA0,0x74,0xE5,0xAB,0x31, - 0x43,0xDB,0x5B,0xFC,0xE0,0xFD,0x10,0x8E,0x4B,0x82,0xD1,0x20, - 0xA9,0x21,0x08,0x01,0x1A,0x72,0x3C,0x12,0xA7,0x87,0xE6,0xD7, - 0x88,0x71,0x9A,0x10,0xBD,0xBA,0x5B,0x26,0x99,0xC3,0x27,0x18, - 0x6A,0xF4,0xE2,0x3C,0x1A,0x94,0x68,0x34,0xB6,0x15,0x0B,0xDA, - 0x25,0x83,0xE9,0xCA,0x2A,0xD4,0x4C,0xE8,0xDB,0xBB,0xC2,0xDB, - 0x04,0xDE,0x8E,0xF9,0x2E,0x8E,0xFC,0x14,0x1F,0xBE,0xCA,0xA6, - 0x28,0x7C,0x59,0x47,0x4E,0x6B,0xC0,0x5D,0x99,0xB2,0x96,0x4F, - 0xA0,0x90,0xC3,0xA2,0x23,0x3B,0xA1,0x86,0x51,0x5B,0xE7,0xED, - 0x1F,0x61,0x29,0x70,0xCE,0xE2,0xD7,0xAF,0xB8,0x1B,0xDD,0x76, - 0x21,0x70,0x48,0x1C,0xD0,0x06,0x91,0x27,0xD5,0xB0,0x5A,0xA9, - 0x93,0xB4,0xEA,0x98,0x8D,0x8F,0xDD,0xC1,0x86,0xFF,0xB7,0xDC, - 0x90,0xA6,0xC0,0x8F,0x4D,0xF4,0x35,0xC9,0x34,0x06,0x31,0x99, - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC3526_PRIME_4096,sizeof(RFC3526_PRIME_4096),bn); - } +{ + static const unsigned char RFC3526_PRIME_4096[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE4, 0x5B, 0x3D, + 0xC2, 0x00, 0x7C, 0xB8, 0xA1, 0x63, 0xBF, 0x05, + 0x98, 0xDA, 0x48, 0x36, 0x1C, 0x55, 0xD3, 0x9A, + 0x69, 0x16, 0x3F, 0xA8, 0xFD, 0x24, 0xCF, 0x5F, + 0x83, 0x65, 0x5D, 0x23, 0xDC, 0xA3, 0xAD, 0x96, + 0x1C, 0x62, 0xF3, 0x56, 0x20, 0x85, 0x52, 0xBB, + 0x9E, 0xD5, 0x29, 0x07, 0x70, 0x96, 0x96, 0x6D, + 0x67, 0x0C, 0x35, 0x4E, 0x4A, 0xBC, 0x98, 0x04, + 0xF1, 0x74, 0x6C, 0x08, 0xCA, 0x18, 0x21, 0x7C, + 0x32, 0x90, 0x5E, 0x46, 0x2E, 0x36, 0xCE, 0x3B, + 0xE3, 0x9E, 0x77, 0x2C, 0x18, 0x0E, 0x86, 0x03, + 0x9B, 0x27, 0x83, 0xA2, 0xEC, 0x07, 0xA2, 0x8F, + 0xB5, 0xC5, 0x5D, 0xF0, 0x6F, 0x4C, 0x52, 0xC9, + 0xDE, 0x2B, 0xCB, 0xF6, 0x95, 0x58, 0x17, 0x18, + 0x39, 0x95, 0x49, 0x7C, 0xEA, 0x95, 0x6A, 0xE5, + 0x15, 0xD2, 0x26, 0x18, 0x98, 0xFA, 0x05, 0x10, + 0x15, 0x72, 0x8E, 0x5A, 0x8A, 0xAA, 0xC4, 0x2D, + 0xAD, 0x33, 0x17, 0x0D, 0x04, 0x50, 0x7A, 0x33, + 0xA8, 0x55, 0x21, 0xAB, 0xDF, 0x1C, 0xBA, 0x64, + 0xEC, 0xFB, 0x85, 0x04, 0x58, 0xDB, 0xEF, 0x0A, + 0x8A, 0xEA, 0x71, 0x57, 0x5D, 0x06, 0x0C, 0x7D, + 0xB3, 0x97, 0x0F, 0x85, 0xA6, 0xE1, 0xE4, 0xC7, + 0xAB, 0xF5, 0xAE, 0x8C, 0xDB, 0x09, 0x33, 0xD7, + 0x1E, 0x8C, 0x94, 0xE0, 0x4A, 0x25, 0x61, 0x9D, + 0xCE, 0xE3, 0xD2, 0x26, 0x1A, 0xD2, 0xEE, 0x6B, + 0xF1, 0x2F, 0xFA, 0x06, 0xD9, 0x8A, 0x08, 0x64, + 0xD8, 0x76, 0x02, 0x73, 0x3E, 0xC8, 0x6A, 0x64, + 0x52, 0x1F, 0x2B, 0x18, 0x17, 0x7B, 0x20, 0x0C, + 0xBB, 0xE1, 0x17, 0x57, 0x7A, 0x61, 0x5D, 0x6C, + 0x77, 0x09, 0x88, 0xC0, 0xBA, 0xD9, 0x46, 0xE2, + 0x08, 0xE2, 0x4F, 0xA0, 0x74, 0xE5, 0xAB, 0x31, + 0x43, 0xDB, 0x5B, 0xFC, 0xE0, 0xFD, 0x10, 0x8E, + 0x4B, 0x82, 0xD1, 0x20, 0xA9, 0x21, 0x08, 0x01, + 0x1A, 0x72, 0x3C, 0x12, 0xA7, 0x87, 0xE6, 0xD7, + 0x88, 0x71, 0x9A, 0x10, 0xBD, 0xBA, 0x5B, 0x26, + 0x99, 0xC3, 0x27, 0x18, 0x6A, 0xF4, 0xE2, 0x3C, + 0x1A, 0x94, 0x68, 0x34, 0xB6, 0x15, 0x0B, 0xDA, + 0x25, 0x83, 0xE9, 0xCA, 0x2A, 0xD4, 0x4C, 0xE8, + 0xDB, 0xBB, 0xC2, 0xDB, 0x04, 0xDE, 0x8E, 0xF9, + 0x2E, 0x8E, 0xFC, 0x14, 0x1F, 0xBE, 0xCA, 0xA6, + 0x28, 0x7C, 0x59, 0x47, 0x4E, 0x6B, 0xC0, 0x5D, + 0x99, 0xB2, 0x96, 0x4F, 0xA0, 0x90, 0xC3, 0xA2, + 0x23, 0x3B, 0xA1, 0x86, 0x51, 0x5B, 0xE7, 0xED, + 0x1F, 0x61, 0x29, 0x70, 0xCE, 0xE2, 0xD7, 0xAF, + 0xB8, 0x1B, 0xDD, 0x76, 0x21, 0x70, 0x48, 0x1C, + 0xD0, 0x06, 0x91, 0x27, 0xD5, 0xB0, 0x5A, 0xA9, + 0x93, 0xB4, 0xEA, 0x98, 0x8D, 0x8F, 0xDD, 0xC1, + 0x86, 0xFF, 0xB7, 0xDC, 0x90, 0xA6, 0xC0, 0x8F, + 0x4D, 0xF4, 0x35, 0xC9, 0x34, 0x06, 0x31, 0x99, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC3526_PRIME_4096, sizeof(RFC3526_PRIME_4096), bn); +} -/* "6144-bit MODP Group" from RFC3526, Section 6. +/*- + * "6144-bit MODP Group" from RFC3526, Section 6. * * The prime is: 2^6144 - 2^6080 - 1 + 2^64 * { [2^6014 pi] + 929484 } * @@ -230,77 +301,110 @@ BIGNUM *get_rfc3526_prime_4096(BIGNUM *bn) */ BIGNUM *get_rfc3526_prime_6144(BIGNUM *bn) - { - static const unsigned char RFC3526_PRIME_6144[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE4,0x5B,0x3D, - 0xC2,0x00,0x7C,0xB8,0xA1,0x63,0xBF,0x05,0x98,0xDA,0x48,0x36, - 0x1C,0x55,0xD3,0x9A,0x69,0x16,0x3F,0xA8,0xFD,0x24,0xCF,0x5F, - 0x83,0x65,0x5D,0x23,0xDC,0xA3,0xAD,0x96,0x1C,0x62,0xF3,0x56, - 0x20,0x85,0x52,0xBB,0x9E,0xD5,0x29,0x07,0x70,0x96,0x96,0x6D, - 0x67,0x0C,0x35,0x4E,0x4A,0xBC,0x98,0x04,0xF1,0x74,0x6C,0x08, - 0xCA,0x18,0x21,0x7C,0x32,0x90,0x5E,0x46,0x2E,0x36,0xCE,0x3B, - 0xE3,0x9E,0x77,0x2C,0x18,0x0E,0x86,0x03,0x9B,0x27,0x83,0xA2, - 0xEC,0x07,0xA2,0x8F,0xB5,0xC5,0x5D,0xF0,0x6F,0x4C,0x52,0xC9, - 0xDE,0x2B,0xCB,0xF6,0x95,0x58,0x17,0x18,0x39,0x95,0x49,0x7C, - 0xEA,0x95,0x6A,0xE5,0x15,0xD2,0x26,0x18,0x98,0xFA,0x05,0x10, - 0x15,0x72,0x8E,0x5A,0x8A,0xAA,0xC4,0x2D,0xAD,0x33,0x17,0x0D, - 0x04,0x50,0x7A,0x33,0xA8,0x55,0x21,0xAB,0xDF,0x1C,0xBA,0x64, - 0xEC,0xFB,0x85,0x04,0x58,0xDB,0xEF,0x0A,0x8A,0xEA,0x71,0x57, - 0x5D,0x06,0x0C,0x7D,0xB3,0x97,0x0F,0x85,0xA6,0xE1,0xE4,0xC7, - 0xAB,0xF5,0xAE,0x8C,0xDB,0x09,0x33,0xD7,0x1E,0x8C,0x94,0xE0, - 0x4A,0x25,0x61,0x9D,0xCE,0xE3,0xD2,0x26,0x1A,0xD2,0xEE,0x6B, - 0xF1,0x2F,0xFA,0x06,0xD9,0x8A,0x08,0x64,0xD8,0x76,0x02,0x73, - 0x3E,0xC8,0x6A,0x64,0x52,0x1F,0x2B,0x18,0x17,0x7B,0x20,0x0C, - 0xBB,0xE1,0x17,0x57,0x7A,0x61,0x5D,0x6C,0x77,0x09,0x88,0xC0, - 0xBA,0xD9,0x46,0xE2,0x08,0xE2,0x4F,0xA0,0x74,0xE5,0xAB,0x31, - 0x43,0xDB,0x5B,0xFC,0xE0,0xFD,0x10,0x8E,0x4B,0x82,0xD1,0x20, - 0xA9,0x21,0x08,0x01,0x1A,0x72,0x3C,0x12,0xA7,0x87,0xE6,0xD7, - 0x88,0x71,0x9A,0x10,0xBD,0xBA,0x5B,0x26,0x99,0xC3,0x27,0x18, - 0x6A,0xF4,0xE2,0x3C,0x1A,0x94,0x68,0x34,0xB6,0x15,0x0B,0xDA, - 0x25,0x83,0xE9,0xCA,0x2A,0xD4,0x4C,0xE8,0xDB,0xBB,0xC2,0xDB, - 0x04,0xDE,0x8E,0xF9,0x2E,0x8E,0xFC,0x14,0x1F,0xBE,0xCA,0xA6, - 0x28,0x7C,0x59,0x47,0x4E,0x6B,0xC0,0x5D,0x99,0xB2,0x96,0x4F, - 0xA0,0x90,0xC3,0xA2,0x23,0x3B,0xA1,0x86,0x51,0x5B,0xE7,0xED, - 0x1F,0x61,0x29,0x70,0xCE,0xE2,0xD7,0xAF,0xB8,0x1B,0xDD,0x76, - 0x21,0x70,0x48,0x1C,0xD0,0x06,0x91,0x27,0xD5,0xB0,0x5A,0xA9, - 0x93,0xB4,0xEA,0x98,0x8D,0x8F,0xDD,0xC1,0x86,0xFF,0xB7,0xDC, - 0x90,0xA6,0xC0,0x8F,0x4D,0xF4,0x35,0xC9,0x34,0x02,0x84,0x92, - 0x36,0xC3,0xFA,0xB4,0xD2,0x7C,0x70,0x26,0xC1,0xD4,0xDC,0xB2, - 0x60,0x26,0x46,0xDE,0xC9,0x75,0x1E,0x76,0x3D,0xBA,0x37,0xBD, - 0xF8,0xFF,0x94,0x06,0xAD,0x9E,0x53,0x0E,0xE5,0xDB,0x38,0x2F, - 0x41,0x30,0x01,0xAE,0xB0,0x6A,0x53,0xED,0x90,0x27,0xD8,0x31, - 0x17,0x97,0x27,0xB0,0x86,0x5A,0x89,0x18,0xDA,0x3E,0xDB,0xEB, - 0xCF,0x9B,0x14,0xED,0x44,0xCE,0x6C,0xBA,0xCE,0xD4,0xBB,0x1B, - 0xDB,0x7F,0x14,0x47,0xE6,0xCC,0x25,0x4B,0x33,0x20,0x51,0x51, - 0x2B,0xD7,0xAF,0x42,0x6F,0xB8,0xF4,0x01,0x37,0x8C,0xD2,0xBF, - 0x59,0x83,0xCA,0x01,0xC6,0x4B,0x92,0xEC,0xF0,0x32,0xEA,0x15, - 0xD1,0x72,0x1D,0x03,0xF4,0x82,0xD7,0xCE,0x6E,0x74,0xFE,0xF6, - 0xD5,0x5E,0x70,0x2F,0x46,0x98,0x0C,0x82,0xB5,0xA8,0x40,0x31, - 0x90,0x0B,0x1C,0x9E,0x59,0xE7,0xC9,0x7F,0xBE,0xC7,0xE8,0xF3, - 0x23,0xA9,0x7A,0x7E,0x36,0xCC,0x88,0xBE,0x0F,0x1D,0x45,0xB7, - 0xFF,0x58,0x5A,0xC5,0x4B,0xD4,0x07,0xB2,0x2B,0x41,0x54,0xAA, - 0xCC,0x8F,0x6D,0x7E,0xBF,0x48,0xE1,0xD8,0x14,0xCC,0x5E,0xD2, - 0x0F,0x80,0x37,0xE0,0xA7,0x97,0x15,0xEE,0xF2,0x9B,0xE3,0x28, - 0x06,0xA1,0xD5,0x8B,0xB7,0xC5,0xDA,0x76,0xF5,0x50,0xAA,0x3D, - 0x8A,0x1F,0xBF,0xF0,0xEB,0x19,0xCC,0xB1,0xA3,0x13,0xD5,0x5C, - 0xDA,0x56,0xC9,0xEC,0x2E,0xF2,0x96,0x32,0x38,0x7F,0xE8,0xD7, - 0x6E,0x3C,0x04,0x68,0x04,0x3E,0x8F,0x66,0x3F,0x48,0x60,0xEE, - 0x12,0xBF,0x2D,0x5B,0x0B,0x74,0x74,0xD6,0xE6,0x94,0xF9,0x1E, - 0x6D,0xCC,0x40,0x24,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC3526_PRIME_6144,sizeof(RFC3526_PRIME_6144),bn); - } +{ + static const unsigned char RFC3526_PRIME_6144[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE4, 0x5B, 0x3D, + 0xC2, 0x00, 0x7C, 0xB8, 0xA1, 0x63, 0xBF, 0x05, + 0x98, 0xDA, 0x48, 0x36, 0x1C, 0x55, 0xD3, 0x9A, + 0x69, 0x16, 0x3F, 0xA8, 0xFD, 0x24, 0xCF, 0x5F, + 0x83, 0x65, 0x5D, 0x23, 0xDC, 0xA3, 0xAD, 0x96, + 0x1C, 0x62, 0xF3, 0x56, 0x20, 0x85, 0x52, 0xBB, + 0x9E, 0xD5, 0x29, 0x07, 0x70, 0x96, 0x96, 0x6D, + 0x67, 0x0C, 0x35, 0x4E, 0x4A, 0xBC, 0x98, 0x04, + 0xF1, 0x74, 0x6C, 0x08, 0xCA, 0x18, 0x21, 0x7C, + 0x32, 0x90, 0x5E, 0x46, 0x2E, 0x36, 0xCE, 0x3B, + 0xE3, 0x9E, 0x77, 0x2C, 0x18, 0x0E, 0x86, 0x03, + 0x9B, 0x27, 0x83, 0xA2, 0xEC, 0x07, 0xA2, 0x8F, + 0xB5, 0xC5, 0x5D, 0xF0, 0x6F, 0x4C, 0x52, 0xC9, + 0xDE, 0x2B, 0xCB, 0xF6, 0x95, 0x58, 0x17, 0x18, + 0x39, 0x95, 0x49, 0x7C, 0xEA, 0x95, 0x6A, 0xE5, + 0x15, 0xD2, 0x26, 0x18, 0x98, 0xFA, 0x05, 0x10, + 0x15, 0x72, 0x8E, 0x5A, 0x8A, 0xAA, 0xC4, 0x2D, + 0xAD, 0x33, 0x17, 0x0D, 0x04, 0x50, 0x7A, 0x33, + 0xA8, 0x55, 0x21, 0xAB, 0xDF, 0x1C, 0xBA, 0x64, + 0xEC, 0xFB, 0x85, 0x04, 0x58, 0xDB, 0xEF, 0x0A, + 0x8A, 0xEA, 0x71, 0x57, 0x5D, 0x06, 0x0C, 0x7D, + 0xB3, 0x97, 0x0F, 0x85, 0xA6, 0xE1, 0xE4, 0xC7, + 0xAB, 0xF5, 0xAE, 0x8C, 0xDB, 0x09, 0x33, 0xD7, + 0x1E, 0x8C, 0x94, 0xE0, 0x4A, 0x25, 0x61, 0x9D, + 0xCE, 0xE3, 0xD2, 0x26, 0x1A, 0xD2, 0xEE, 0x6B, + 0xF1, 0x2F, 0xFA, 0x06, 0xD9, 0x8A, 0x08, 0x64, + 0xD8, 0x76, 0x02, 0x73, 0x3E, 0xC8, 0x6A, 0x64, + 0x52, 0x1F, 0x2B, 0x18, 0x17, 0x7B, 0x20, 0x0C, + 0xBB, 0xE1, 0x17, 0x57, 0x7A, 0x61, 0x5D, 0x6C, + 0x77, 0x09, 0x88, 0xC0, 0xBA, 0xD9, 0x46, 0xE2, + 0x08, 0xE2, 0x4F, 0xA0, 0x74, 0xE5, 0xAB, 0x31, + 0x43, 0xDB, 0x5B, 0xFC, 0xE0, 0xFD, 0x10, 0x8E, + 0x4B, 0x82, 0xD1, 0x20, 0xA9, 0x21, 0x08, 0x01, + 0x1A, 0x72, 0x3C, 0x12, 0xA7, 0x87, 0xE6, 0xD7, + 0x88, 0x71, 0x9A, 0x10, 0xBD, 0xBA, 0x5B, 0x26, + 0x99, 0xC3, 0x27, 0x18, 0x6A, 0xF4, 0xE2, 0x3C, + 0x1A, 0x94, 0x68, 0x34, 0xB6, 0x15, 0x0B, 0xDA, + 0x25, 0x83, 0xE9, 0xCA, 0x2A, 0xD4, 0x4C, 0xE8, + 0xDB, 0xBB, 0xC2, 0xDB, 0x04, 0xDE, 0x8E, 0xF9, + 0x2E, 0x8E, 0xFC, 0x14, 0x1F, 0xBE, 0xCA, 0xA6, + 0x28, 0x7C, 0x59, 0x47, 0x4E, 0x6B, 0xC0, 0x5D, + 0x99, 0xB2, 0x96, 0x4F, 0xA0, 0x90, 0xC3, 0xA2, + 0x23, 0x3B, 0xA1, 0x86, 0x51, 0x5B, 0xE7, 0xED, + 0x1F, 0x61, 0x29, 0x70, 0xCE, 0xE2, 0xD7, 0xAF, + 0xB8, 0x1B, 0xDD, 0x76, 0x21, 0x70, 0x48, 0x1C, + 0xD0, 0x06, 0x91, 0x27, 0xD5, 0xB0, 0x5A, 0xA9, + 0x93, 0xB4, 0xEA, 0x98, 0x8D, 0x8F, 0xDD, 0xC1, + 0x86, 0xFF, 0xB7, 0xDC, 0x90, 0xA6, 0xC0, 0x8F, + 0x4D, 0xF4, 0x35, 0xC9, 0x34, 0x02, 0x84, 0x92, + 0x36, 0xC3, 0xFA, 0xB4, 0xD2, 0x7C, 0x70, 0x26, + 0xC1, 0xD4, 0xDC, 0xB2, 0x60, 0x26, 0x46, 0xDE, + 0xC9, 0x75, 0x1E, 0x76, 0x3D, 0xBA, 0x37, 0xBD, + 0xF8, 0xFF, 0x94, 0x06, 0xAD, 0x9E, 0x53, 0x0E, + 0xE5, 0xDB, 0x38, 0x2F, 0x41, 0x30, 0x01, 0xAE, + 0xB0, 0x6A, 0x53, 0xED, 0x90, 0x27, 0xD8, 0x31, + 0x17, 0x97, 0x27, 0xB0, 0x86, 0x5A, 0x89, 0x18, + 0xDA, 0x3E, 0xDB, 0xEB, 0xCF, 0x9B, 0x14, 0xED, + 0x44, 0xCE, 0x6C, 0xBA, 0xCE, 0xD4, 0xBB, 0x1B, + 0xDB, 0x7F, 0x14, 0x47, 0xE6, 0xCC, 0x25, 0x4B, + 0x33, 0x20, 0x51, 0x51, 0x2B, 0xD7, 0xAF, 0x42, + 0x6F, 0xB8, 0xF4, 0x01, 0x37, 0x8C, 0xD2, 0xBF, + 0x59, 0x83, 0xCA, 0x01, 0xC6, 0x4B, 0x92, 0xEC, + 0xF0, 0x32, 0xEA, 0x15, 0xD1, 0x72, 0x1D, 0x03, + 0xF4, 0x82, 0xD7, 0xCE, 0x6E, 0x74, 0xFE, 0xF6, + 0xD5, 0x5E, 0x70, 0x2F, 0x46, 0x98, 0x0C, 0x82, + 0xB5, 0xA8, 0x40, 0x31, 0x90, 0x0B, 0x1C, 0x9E, + 0x59, 0xE7, 0xC9, 0x7F, 0xBE, 0xC7, 0xE8, 0xF3, + 0x23, 0xA9, 0x7A, 0x7E, 0x36, 0xCC, 0x88, 0xBE, + 0x0F, 0x1D, 0x45, 0xB7, 0xFF, 0x58, 0x5A, 0xC5, + 0x4B, 0xD4, 0x07, 0xB2, 0x2B, 0x41, 0x54, 0xAA, + 0xCC, 0x8F, 0x6D, 0x7E, 0xBF, 0x48, 0xE1, 0xD8, + 0x14, 0xCC, 0x5E, 0xD2, 0x0F, 0x80, 0x37, 0xE0, + 0xA7, 0x97, 0x15, 0xEE, 0xF2, 0x9B, 0xE3, 0x28, + 0x06, 0xA1, 0xD5, 0x8B, 0xB7, 0xC5, 0xDA, 0x76, + 0xF5, 0x50, 0xAA, 0x3D, 0x8A, 0x1F, 0xBF, 0xF0, + 0xEB, 0x19, 0xCC, 0xB1, 0xA3, 0x13, 0xD5, 0x5C, + 0xDA, 0x56, 0xC9, 0xEC, 0x2E, 0xF2, 0x96, 0x32, + 0x38, 0x7F, 0xE8, 0xD7, 0x6E, 0x3C, 0x04, 0x68, + 0x04, 0x3E, 0x8F, 0x66, 0x3F, 0x48, 0x60, 0xEE, + 0x12, 0xBF, 0x2D, 0x5B, 0x0B, 0x74, 0x74, 0xD6, + 0xE6, 0x94, 0xF9, 0x1E, 0x6D, 0xCC, 0x40, 0x24, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC3526_PRIME_6144, sizeof(RFC3526_PRIME_6144), bn); +} -/* "8192-bit MODP Group" from RFC3526, Section 7. +/*- + * "8192-bit MODP Group" from RFC3526, Section 7. * * The prime is: 2^8192 - 2^8128 - 1 + 2^64 * { [2^8062 pi] + 4743158 } * @@ -308,95 +412,136 @@ BIGNUM *get_rfc3526_prime_6144(BIGNUM *bn) */ BIGNUM *get_rfc3526_prime_8192(BIGNUM *bn) - { - static const unsigned char RFC3526_PRIME_8192[]={ - 0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xFF,0xC9,0x0F,0xDA,0xA2, - 0x21,0x68,0xC2,0x34,0xC4,0xC6,0x62,0x8B,0x80,0xDC,0x1C,0xD1, - 0x29,0x02,0x4E,0x08,0x8A,0x67,0xCC,0x74,0x02,0x0B,0xBE,0xA6, - 0x3B,0x13,0x9B,0x22,0x51,0x4A,0x08,0x79,0x8E,0x34,0x04,0xDD, - 0xEF,0x95,0x19,0xB3,0xCD,0x3A,0x43,0x1B,0x30,0x2B,0x0A,0x6D, - 0xF2,0x5F,0x14,0x37,0x4F,0xE1,0x35,0x6D,0x6D,0x51,0xC2,0x45, - 0xE4,0x85,0xB5,0x76,0x62,0x5E,0x7E,0xC6,0xF4,0x4C,0x42,0xE9, - 0xA6,0x37,0xED,0x6B,0x0B,0xFF,0x5C,0xB6,0xF4,0x06,0xB7,0xED, - 0xEE,0x38,0x6B,0xFB,0x5A,0x89,0x9F,0xA5,0xAE,0x9F,0x24,0x11, - 0x7C,0x4B,0x1F,0xE6,0x49,0x28,0x66,0x51,0xEC,0xE4,0x5B,0x3D, - 0xC2,0x00,0x7C,0xB8,0xA1,0x63,0xBF,0x05,0x98,0xDA,0x48,0x36, - 0x1C,0x55,0xD3,0x9A,0x69,0x16,0x3F,0xA8,0xFD,0x24,0xCF,0x5F, - 0x83,0x65,0x5D,0x23,0xDC,0xA3,0xAD,0x96,0x1C,0x62,0xF3,0x56, - 0x20,0x85,0x52,0xBB,0x9E,0xD5,0x29,0x07,0x70,0x96,0x96,0x6D, - 0x67,0x0C,0x35,0x4E,0x4A,0xBC,0x98,0x04,0xF1,0x74,0x6C,0x08, - 0xCA,0x18,0x21,0x7C,0x32,0x90,0x5E,0x46,0x2E,0x36,0xCE,0x3B, - 0xE3,0x9E,0x77,0x2C,0x18,0x0E,0x86,0x03,0x9B,0x27,0x83,0xA2, - 0xEC,0x07,0xA2,0x8F,0xB5,0xC5,0x5D,0xF0,0x6F,0x4C,0x52,0xC9, - 0xDE,0x2B,0xCB,0xF6,0x95,0x58,0x17,0x18,0x39,0x95,0x49,0x7C, - 0xEA,0x95,0x6A,0xE5,0x15,0xD2,0x26,0x18,0x98,0xFA,0x05,0x10, - 0x15,0x72,0x8E,0x5A,0x8A,0xAA,0xC4,0x2D,0xAD,0x33,0x17,0x0D, - 0x04,0x50,0x7A,0x33,0xA8,0x55,0x21,0xAB,0xDF,0x1C,0xBA,0x64, - 0xEC,0xFB,0x85,0x04,0x58,0xDB,0xEF,0x0A,0x8A,0xEA,0x71,0x57, - 0x5D,0x06,0x0C,0x7D,0xB3,0x97,0x0F,0x85,0xA6,0xE1,0xE4,0xC7, - 0xAB,0xF5,0xAE,0x8C,0xDB,0x09,0x33,0xD7,0x1E,0x8C,0x94,0xE0, - 0x4A,0x25,0x61,0x9D,0xCE,0xE3,0xD2,0x26,0x1A,0xD2,0xEE,0x6B, - 0xF1,0x2F,0xFA,0x06,0xD9,0x8A,0x08,0x64,0xD8,0x76,0x02,0x73, - 0x3E,0xC8,0x6A,0x64,0x52,0x1F,0x2B,0x18,0x17,0x7B,0x20,0x0C, - 0xBB,0xE1,0x17,0x57,0x7A,0x61,0x5D,0x6C,0x77,0x09,0x88,0xC0, - 0xBA,0xD9,0x46,0xE2,0x08,0xE2,0x4F,0xA0,0x74,0xE5,0xAB,0x31, - 0x43,0xDB,0x5B,0xFC,0xE0,0xFD,0x10,0x8E,0x4B,0x82,0xD1,0x20, - 0xA9,0x21,0x08,0x01,0x1A,0x72,0x3C,0x12,0xA7,0x87,0xE6,0xD7, - 0x88,0x71,0x9A,0x10,0xBD,0xBA,0x5B,0x26,0x99,0xC3,0x27,0x18, - 0x6A,0xF4,0xE2,0x3C,0x1A,0x94,0x68,0x34,0xB6,0x15,0x0B,0xDA, - 0x25,0x83,0xE9,0xCA,0x2A,0xD4,0x4C,0xE8,0xDB,0xBB,0xC2,0xDB, - 0x04,0xDE,0x8E,0xF9,0x2E,0x8E,0xFC,0x14,0x1F,0xBE,0xCA,0xA6, - 0x28,0x7C,0x59,0x47,0x4E,0x6B,0xC0,0x5D,0x99,0xB2,0x96,0x4F, - 0xA0,0x90,0xC3,0xA2,0x23,0x3B,0xA1,0x86,0x51,0x5B,0xE7,0xED, - 0x1F,0x61,0x29,0x70,0xCE,0xE2,0xD7,0xAF,0xB8,0x1B,0xDD,0x76, - 0x21,0x70,0x48,0x1C,0xD0,0x06,0x91,0x27,0xD5,0xB0,0x5A,0xA9, - 0x93,0xB4,0xEA,0x98,0x8D,0x8F,0xDD,0xC1,0x86,0xFF,0xB7,0xDC, - 0x90,0xA6,0xC0,0x8F,0x4D,0xF4,0x35,0xC9,0x34,0x02,0x84,0x92, - 0x36,0xC3,0xFA,0xB4,0xD2,0x7C,0x70,0x26,0xC1,0xD4,0xDC,0xB2, - 0x60,0x26,0x46,0xDE,0xC9,0x75,0x1E,0x76,0x3D,0xBA,0x37,0xBD, - 0xF8,0xFF,0x94,0x06,0xAD,0x9E,0x53,0x0E,0xE5,0xDB,0x38,0x2F, - 0x41,0x30,0x01,0xAE,0xB0,0x6A,0x53,0xED,0x90,0x27,0xD8,0x31, - 0x17,0x97,0x27,0xB0,0x86,0x5A,0x89,0x18,0xDA,0x3E,0xDB,0xEB, - 0xCF,0x9B,0x14,0xED,0x44,0xCE,0x6C,0xBA,0xCE,0xD4,0xBB,0x1B, - 0xDB,0x7F,0x14,0x47,0xE6,0xCC,0x25,0x4B,0x33,0x20,0x51,0x51, - 0x2B,0xD7,0xAF,0x42,0x6F,0xB8,0xF4,0x01,0x37,0x8C,0xD2,0xBF, - 0x59,0x83,0xCA,0x01,0xC6,0x4B,0x92,0xEC,0xF0,0x32,0xEA,0x15, - 0xD1,0x72,0x1D,0x03,0xF4,0x82,0xD7,0xCE,0x6E,0x74,0xFE,0xF6, - 0xD5,0x5E,0x70,0x2F,0x46,0x98,0x0C,0x82,0xB5,0xA8,0x40,0x31, - 0x90,0x0B,0x1C,0x9E,0x59,0xE7,0xC9,0x7F,0xBE,0xC7,0xE8,0xF3, - 0x23,0xA9,0x7A,0x7E,0x36,0xCC,0x88,0xBE,0x0F,0x1D,0x45,0xB7, - 0xFF,0x58,0x5A,0xC5,0x4B,0xD4,0x07,0xB2,0x2B,0x41,0x54,0xAA, - 0xCC,0x8F,0x6D,0x7E,0xBF,0x48,0xE1,0xD8,0x14,0xCC,0x5E,0xD2, - 0x0F,0x80,0x37,0xE0,0xA7,0x97,0x15,0xEE,0xF2,0x9B,0xE3,0x28, - 0x06,0xA1,0xD5,0x8B,0xB7,0xC5,0xDA,0x76,0xF5,0x50,0xAA,0x3D, - 0x8A,0x1F,0xBF,0xF0,0xEB,0x19,0xCC,0xB1,0xA3,0x13,0xD5,0x5C, - 0xDA,0x56,0xC9,0xEC,0x2E,0xF2,0x96,0x32,0x38,0x7F,0xE8,0xD7, - 0x6E,0x3C,0x04,0x68,0x04,0x3E,0x8F,0x66,0x3F,0x48,0x60,0xEE, - 0x12,0xBF,0x2D,0x5B,0x0B,0x74,0x74,0xD6,0xE6,0x94,0xF9,0x1E, - 0x6D,0xBE,0x11,0x59,0x74,0xA3,0x92,0x6F,0x12,0xFE,0xE5,0xE4, - 0x38,0x77,0x7C,0xB6,0xA9,0x32,0xDF,0x8C,0xD8,0xBE,0xC4,0xD0, - 0x73,0xB9,0x31,0xBA,0x3B,0xC8,0x32,0xB6,0x8D,0x9D,0xD3,0x00, - 0x74,0x1F,0xA7,0xBF,0x8A,0xFC,0x47,0xED,0x25,0x76,0xF6,0x93, - 0x6B,0xA4,0x24,0x66,0x3A,0xAB,0x63,0x9C,0x5A,0xE4,0xF5,0x68, - 0x34,0x23,0xB4,0x74,0x2B,0xF1,0xC9,0x78,0x23,0x8F,0x16,0xCB, - 0xE3,0x9D,0x65,0x2D,0xE3,0xFD,0xB8,0xBE,0xFC,0x84,0x8A,0xD9, - 0x22,0x22,0x2E,0x04,0xA4,0x03,0x7C,0x07,0x13,0xEB,0x57,0xA8, - 0x1A,0x23,0xF0,0xC7,0x34,0x73,0xFC,0x64,0x6C,0xEA,0x30,0x6B, - 0x4B,0xCB,0xC8,0x86,0x2F,0x83,0x85,0xDD,0xFA,0x9D,0x4B,0x7F, - 0xA2,0xC0,0x87,0xE8,0x79,0x68,0x33,0x03,0xED,0x5B,0xDD,0x3A, - 0x06,0x2B,0x3C,0xF5,0xB3,0xA2,0x78,0xA6,0x6D,0x2A,0x13,0xF8, - 0x3F,0x44,0xF8,0x2D,0xDF,0x31,0x0E,0xE0,0x74,0xAB,0x6A,0x36, - 0x45,0x97,0xE8,0x99,0xA0,0x25,0x5D,0xC1,0x64,0xF3,0x1C,0xC5, - 0x08,0x46,0x85,0x1D,0xF9,0xAB,0x48,0x19,0x5D,0xED,0x7E,0xA1, - 0xB1,0xD5,0x10,0xBD,0x7E,0xE7,0x4D,0x73,0xFA,0xF3,0x6B,0xC3, - 0x1E,0xCF,0xA2,0x68,0x35,0x90,0x46,0xF4,0xEB,0x87,0x9F,0x92, - 0x40,0x09,0x43,0x8B,0x48,0x1C,0x6C,0xD7,0x88,0x9A,0x00,0x2E, - 0xD5,0xEE,0x38,0x2B,0xC9,0x19,0x0D,0xA6,0xFC,0x02,0x6E,0x47, - 0x95,0x58,0xE4,0x47,0x56,0x77,0xE9,0xAA,0x9E,0x30,0x50,0xE2, - 0x76,0x56,0x94,0xDF,0xC8,0x1F,0x56,0xE8,0x80,0xB9,0x6E,0x71, - 0x60,0xC9,0x80,0xDD,0x98,0xED,0xD3,0xDF,0xFF,0xFF,0xFF,0xFF, - 0xFF,0xFF,0xFF,0xFF, - }; - return BN_bin2bn(RFC3526_PRIME_8192,sizeof(RFC3526_PRIME_8192),bn); - } - +{ + static const unsigned char RFC3526_PRIME_8192[] = { + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + 0xC9, 0x0F, 0xDA, 0xA2, 0x21, 0x68, 0xC2, 0x34, + 0xC4, 0xC6, 0x62, 0x8B, 0x80, 0xDC, 0x1C, 0xD1, + 0x29, 0x02, 0x4E, 0x08, 0x8A, 0x67, 0xCC, 0x74, + 0x02, 0x0B, 0xBE, 0xA6, 0x3B, 0x13, 0x9B, 0x22, + 0x51, 0x4A, 0x08, 0x79, 0x8E, 0x34, 0x04, 0xDD, + 0xEF, 0x95, 0x19, 0xB3, 0xCD, 0x3A, 0x43, 0x1B, + 0x30, 0x2B, 0x0A, 0x6D, 0xF2, 0x5F, 0x14, 0x37, + 0x4F, 0xE1, 0x35, 0x6D, 0x6D, 0x51, 0xC2, 0x45, + 0xE4, 0x85, 0xB5, 0x76, 0x62, 0x5E, 0x7E, 0xC6, + 0xF4, 0x4C, 0x42, 0xE9, 0xA6, 0x37, 0xED, 0x6B, + 0x0B, 0xFF, 0x5C, 0xB6, 0xF4, 0x06, 0xB7, 0xED, + 0xEE, 0x38, 0x6B, 0xFB, 0x5A, 0x89, 0x9F, 0xA5, + 0xAE, 0x9F, 0x24, 0x11, 0x7C, 0x4B, 0x1F, 0xE6, + 0x49, 0x28, 0x66, 0x51, 0xEC, 0xE4, 0x5B, 0x3D, + 0xC2, 0x00, 0x7C, 0xB8, 0xA1, 0x63, 0xBF, 0x05, + 0x98, 0xDA, 0x48, 0x36, 0x1C, 0x55, 0xD3, 0x9A, + 0x69, 0x16, 0x3F, 0xA8, 0xFD, 0x24, 0xCF, 0x5F, + 0x83, 0x65, 0x5D, 0x23, 0xDC, 0xA3, 0xAD, 0x96, + 0x1C, 0x62, 0xF3, 0x56, 0x20, 0x85, 0x52, 0xBB, + 0x9E, 0xD5, 0x29, 0x07, 0x70, 0x96, 0x96, 0x6D, + 0x67, 0x0C, 0x35, 0x4E, 0x4A, 0xBC, 0x98, 0x04, + 0xF1, 0x74, 0x6C, 0x08, 0xCA, 0x18, 0x21, 0x7C, + 0x32, 0x90, 0x5E, 0x46, 0x2E, 0x36, 0xCE, 0x3B, + 0xE3, 0x9E, 0x77, 0x2C, 0x18, 0x0E, 0x86, 0x03, + 0x9B, 0x27, 0x83, 0xA2, 0xEC, 0x07, 0xA2, 0x8F, + 0xB5, 0xC5, 0x5D, 0xF0, 0x6F, 0x4C, 0x52, 0xC9, + 0xDE, 0x2B, 0xCB, 0xF6, 0x95, 0x58, 0x17, 0x18, + 0x39, 0x95, 0x49, 0x7C, 0xEA, 0x95, 0x6A, 0xE5, + 0x15, 0xD2, 0x26, 0x18, 0x98, 0xFA, 0x05, 0x10, + 0x15, 0x72, 0x8E, 0x5A, 0x8A, 0xAA, 0xC4, 0x2D, + 0xAD, 0x33, 0x17, 0x0D, 0x04, 0x50, 0x7A, 0x33, + 0xA8, 0x55, 0x21, 0xAB, 0xDF, 0x1C, 0xBA, 0x64, + 0xEC, 0xFB, 0x85, 0x04, 0x58, 0xDB, 0xEF, 0x0A, + 0x8A, 0xEA, 0x71, 0x57, 0x5D, 0x06, 0x0C, 0x7D, + 0xB3, 0x97, 0x0F, 0x85, 0xA6, 0xE1, 0xE4, 0xC7, + 0xAB, 0xF5, 0xAE, 0x8C, 0xDB, 0x09, 0x33, 0xD7, + 0x1E, 0x8C, 0x94, 0xE0, 0x4A, 0x25, 0x61, 0x9D, + 0xCE, 0xE3, 0xD2, 0x26, 0x1A, 0xD2, 0xEE, 0x6B, + 0xF1, 0x2F, 0xFA, 0x06, 0xD9, 0x8A, 0x08, 0x64, + 0xD8, 0x76, 0x02, 0x73, 0x3E, 0xC8, 0x6A, 0x64, + 0x52, 0x1F, 0x2B, 0x18, 0x17, 0x7B, 0x20, 0x0C, + 0xBB, 0xE1, 0x17, 0x57, 0x7A, 0x61, 0x5D, 0x6C, + 0x77, 0x09, 0x88, 0xC0, 0xBA, 0xD9, 0x46, 0xE2, + 0x08, 0xE2, 0x4F, 0xA0, 0x74, 0xE5, 0xAB, 0x31, + 0x43, 0xDB, 0x5B, 0xFC, 0xE0, 0xFD, 0x10, 0x8E, + 0x4B, 0x82, 0xD1, 0x20, 0xA9, 0x21, 0x08, 0x01, + 0x1A, 0x72, 0x3C, 0x12, 0xA7, 0x87, 0xE6, 0xD7, + 0x88, 0x71, 0x9A, 0x10, 0xBD, 0xBA, 0x5B, 0x26, + 0x99, 0xC3, 0x27, 0x18, 0x6A, 0xF4, 0xE2, 0x3C, + 0x1A, 0x94, 0x68, 0x34, 0xB6, 0x15, 0x0B, 0xDA, + 0x25, 0x83, 0xE9, 0xCA, 0x2A, 0xD4, 0x4C, 0xE8, + 0xDB, 0xBB, 0xC2, 0xDB, 0x04, 0xDE, 0x8E, 0xF9, + 0x2E, 0x8E, 0xFC, 0x14, 0x1F, 0xBE, 0xCA, 0xA6, + 0x28, 0x7C, 0x59, 0x47, 0x4E, 0x6B, 0xC0, 0x5D, + 0x99, 0xB2, 0x96, 0x4F, 0xA0, 0x90, 0xC3, 0xA2, + 0x23, 0x3B, 0xA1, 0x86, 0x51, 0x5B, 0xE7, 0xED, + 0x1F, 0x61, 0x29, 0x70, 0xCE, 0xE2, 0xD7, 0xAF, + 0xB8, 0x1B, 0xDD, 0x76, 0x21, 0x70, 0x48, 0x1C, + 0xD0, 0x06, 0x91, 0x27, 0xD5, 0xB0, 0x5A, 0xA9, + 0x93, 0xB4, 0xEA, 0x98, 0x8D, 0x8F, 0xDD, 0xC1, + 0x86, 0xFF, 0xB7, 0xDC, 0x90, 0xA6, 0xC0, 0x8F, + 0x4D, 0xF4, 0x35, 0xC9, 0x34, 0x02, 0x84, 0x92, + 0x36, 0xC3, 0xFA, 0xB4, 0xD2, 0x7C, 0x70, 0x26, + 0xC1, 0xD4, 0xDC, 0xB2, 0x60, 0x26, 0x46, 0xDE, + 0xC9, 0x75, 0x1E, 0x76, 0x3D, 0xBA, 0x37, 0xBD, + 0xF8, 0xFF, 0x94, 0x06, 0xAD, 0x9E, 0x53, 0x0E, + 0xE5, 0xDB, 0x38, 0x2F, 0x41, 0x30, 0x01, 0xAE, + 0xB0, 0x6A, 0x53, 0xED, 0x90, 0x27, 0xD8, 0x31, + 0x17, 0x97, 0x27, 0xB0, 0x86, 0x5A, 0x89, 0x18, + 0xDA, 0x3E, 0xDB, 0xEB, 0xCF, 0x9B, 0x14, 0xED, + 0x44, 0xCE, 0x6C, 0xBA, 0xCE, 0xD4, 0xBB, 0x1B, + 0xDB, 0x7F, 0x14, 0x47, 0xE6, 0xCC, 0x25, 0x4B, + 0x33, 0x20, 0x51, 0x51, 0x2B, 0xD7, 0xAF, 0x42, + 0x6F, 0xB8, 0xF4, 0x01, 0x37, 0x8C, 0xD2, 0xBF, + 0x59, 0x83, 0xCA, 0x01, 0xC6, 0x4B, 0x92, 0xEC, + 0xF0, 0x32, 0xEA, 0x15, 0xD1, 0x72, 0x1D, 0x03, + 0xF4, 0x82, 0xD7, 0xCE, 0x6E, 0x74, 0xFE, 0xF6, + 0xD5, 0x5E, 0x70, 0x2F, 0x46, 0x98, 0x0C, 0x82, + 0xB5, 0xA8, 0x40, 0x31, 0x90, 0x0B, 0x1C, 0x9E, + 0x59, 0xE7, 0xC9, 0x7F, 0xBE, 0xC7, 0xE8, 0xF3, + 0x23, 0xA9, 0x7A, 0x7E, 0x36, 0xCC, 0x88, 0xBE, + 0x0F, 0x1D, 0x45, 0xB7, 0xFF, 0x58, 0x5A, 0xC5, + 0x4B, 0xD4, 0x07, 0xB2, 0x2B, 0x41, 0x54, 0xAA, + 0xCC, 0x8F, 0x6D, 0x7E, 0xBF, 0x48, 0xE1, 0xD8, + 0x14, 0xCC, 0x5E, 0xD2, 0x0F, 0x80, 0x37, 0xE0, + 0xA7, 0x97, 0x15, 0xEE, 0xF2, 0x9B, 0xE3, 0x28, + 0x06, 0xA1, 0xD5, 0x8B, 0xB7, 0xC5, 0xDA, 0x76, + 0xF5, 0x50, 0xAA, 0x3D, 0x8A, 0x1F, 0xBF, 0xF0, + 0xEB, 0x19, 0xCC, 0xB1, 0xA3, 0x13, 0xD5, 0x5C, + 0xDA, 0x56, 0xC9, 0xEC, 0x2E, 0xF2, 0x96, 0x32, + 0x38, 0x7F, 0xE8, 0xD7, 0x6E, 0x3C, 0x04, 0x68, + 0x04, 0x3E, 0x8F, 0x66, 0x3F, 0x48, 0x60, 0xEE, + 0x12, 0xBF, 0x2D, 0x5B, 0x0B, 0x74, 0x74, 0xD6, + 0xE6, 0x94, 0xF9, 0x1E, 0x6D, 0xBE, 0x11, 0x59, + 0x74, 0xA3, 0x92, 0x6F, 0x12, 0xFE, 0xE5, 0xE4, + 0x38, 0x77, 0x7C, 0xB6, 0xA9, 0x32, 0xDF, 0x8C, + 0xD8, 0xBE, 0xC4, 0xD0, 0x73, 0xB9, 0x31, 0xBA, + 0x3B, 0xC8, 0x32, 0xB6, 0x8D, 0x9D, 0xD3, 0x00, + 0x74, 0x1F, 0xA7, 0xBF, 0x8A, 0xFC, 0x47, 0xED, + 0x25, 0x76, 0xF6, 0x93, 0x6B, 0xA4, 0x24, 0x66, + 0x3A, 0xAB, 0x63, 0x9C, 0x5A, 0xE4, 0xF5, 0x68, + 0x34, 0x23, 0xB4, 0x74, 0x2B, 0xF1, 0xC9, 0x78, + 0x23, 0x8F, 0x16, 0xCB, 0xE3, 0x9D, 0x65, 0x2D, + 0xE3, 0xFD, 0xB8, 0xBE, 0xFC, 0x84, 0x8A, 0xD9, + 0x22, 0x22, 0x2E, 0x04, 0xA4, 0x03, 0x7C, 0x07, + 0x13, 0xEB, 0x57, 0xA8, 0x1A, 0x23, 0xF0, 0xC7, + 0x34, 0x73, 0xFC, 0x64, 0x6C, 0xEA, 0x30, 0x6B, + 0x4B, 0xCB, 0xC8, 0x86, 0x2F, 0x83, 0x85, 0xDD, + 0xFA, 0x9D, 0x4B, 0x7F, 0xA2, 0xC0, 0x87, 0xE8, + 0x79, 0x68, 0x33, 0x03, 0xED, 0x5B, 0xDD, 0x3A, + 0x06, 0x2B, 0x3C, 0xF5, 0xB3, 0xA2, 0x78, 0xA6, + 0x6D, 0x2A, 0x13, 0xF8, 0x3F, 0x44, 0xF8, 0x2D, + 0xDF, 0x31, 0x0E, 0xE0, 0x74, 0xAB, 0x6A, 0x36, + 0x45, 0x97, 0xE8, 0x99, 0xA0, 0x25, 0x5D, 0xC1, + 0x64, 0xF3, 0x1C, 0xC5, 0x08, 0x46, 0x85, 0x1D, + 0xF9, 0xAB, 0x48, 0x19, 0x5D, 0xED, 0x7E, 0xA1, + 0xB1, 0xD5, 0x10, 0xBD, 0x7E, 0xE7, 0x4D, 0x73, + 0xFA, 0xF3, 0x6B, 0xC3, 0x1E, 0xCF, 0xA2, 0x68, + 0x35, 0x90, 0x46, 0xF4, 0xEB, 0x87, 0x9F, 0x92, + 0x40, 0x09, 0x43, 0x8B, 0x48, 0x1C, 0x6C, 0xD7, + 0x88, 0x9A, 0x00, 0x2E, 0xD5, 0xEE, 0x38, 0x2B, + 0xC9, 0x19, 0x0D, 0xA6, 0xFC, 0x02, 0x6E, 0x47, + 0x95, 0x58, 0xE4, 0x47, 0x56, 0x77, 0xE9, 0xAA, + 0x9E, 0x30, 0x50, 0xE2, 0x76, 0x56, 0x94, 0xDF, + 0xC8, 0x1F, 0x56, 0xE8, 0x80, 0xB9, 0x6E, 0x71, + 0x60, 0xC9, 0x80, 0xDD, 0x98, 0xED, 0xD3, 0xDF, + 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, + }; + return BN_bin2bn(RFC3526_PRIME_8192, sizeof(RFC3526_PRIME_8192), bn); +} diff --git a/openssl/crypto/bn/bn_ctx.c b/openssl/crypto/bn/bn_ctx.c index 3f2256f67..526c6a046 100644 --- a/openssl/crypto/bn/bn_ctx.c +++ b/openssl/crypto/bn/bn_ctx.c @@ -8,7 +8,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -55,9 +55,9 @@ */ #if !defined(BN_CTX_DEBUG) && !defined(BN_DEBUG) -#ifndef NDEBUG -#define NDEBUG -#endif +# ifndef NDEBUG +# define NDEBUG +# endif #endif #include <stdio.h> @@ -66,7 +66,8 @@ #include "cryptlib.h" #include "bn_lcl.h" -/* TODO list +/*- + * TODO list * * 1. Check a bunch of "(words+1)" type hacks in various bignum functions and * check they can be safely removed. @@ -79,376 +80,369 @@ */ /* How many bignums are in each "pool item"; */ -#define BN_CTX_POOL_SIZE 16 +#define BN_CTX_POOL_SIZE 16 /* The stack frame info is resizing, set a first-time expansion size; */ -#define BN_CTX_START_FRAMES 32 +#define BN_CTX_START_FRAMES 32 /***********/ /* BN_POOL */ /***********/ /* A bundle of bignums that can be linked with other bundles */ -typedef struct bignum_pool_item - { - /* The bignum values */ - BIGNUM vals[BN_CTX_POOL_SIZE]; - /* Linked-list admin */ - struct bignum_pool_item *prev, *next; - } BN_POOL_ITEM; +typedef struct bignum_pool_item { + /* The bignum values */ + BIGNUM vals[BN_CTX_POOL_SIZE]; + /* Linked-list admin */ + struct bignum_pool_item *prev, *next; +} BN_POOL_ITEM; /* A linked-list of bignums grouped in bundles */ -typedef struct bignum_pool - { - /* Linked-list admin */ - BN_POOL_ITEM *head, *current, *tail; - /* Stack depth and allocation size */ - unsigned used, size; - } BN_POOL; -static void BN_POOL_init(BN_POOL *); -static void BN_POOL_finish(BN_POOL *); +typedef struct bignum_pool { + /* Linked-list admin */ + BN_POOL_ITEM *head, *current, *tail; + /* Stack depth and allocation size */ + unsigned used, size; +} BN_POOL; +static void BN_POOL_init(BN_POOL *); +static void BN_POOL_finish(BN_POOL *); #ifndef OPENSSL_NO_DEPRECATED -static void BN_POOL_reset(BN_POOL *); +static void BN_POOL_reset(BN_POOL *); #endif -static BIGNUM * BN_POOL_get(BN_POOL *); -static void BN_POOL_release(BN_POOL *, unsigned int); +static BIGNUM *BN_POOL_get(BN_POOL *); +static void BN_POOL_release(BN_POOL *, unsigned int); /************/ /* BN_STACK */ /************/ /* A wrapper to manage the "stack frames" */ -typedef struct bignum_ctx_stack - { - /* Array of indexes into the bignum stack */ - unsigned int *indexes; - /* Number of stack frames, and the size of the allocated array */ - unsigned int depth, size; - } BN_STACK; -static void BN_STACK_init(BN_STACK *); -static void BN_STACK_finish(BN_STACK *); +typedef struct bignum_ctx_stack { + /* Array of indexes into the bignum stack */ + unsigned int *indexes; + /* Number of stack frames, and the size of the allocated array */ + unsigned int depth, size; +} BN_STACK; +static void BN_STACK_init(BN_STACK *); +static void BN_STACK_finish(BN_STACK *); #ifndef OPENSSL_NO_DEPRECATED -static void BN_STACK_reset(BN_STACK *); +static void BN_STACK_reset(BN_STACK *); #endif -static int BN_STACK_push(BN_STACK *, unsigned int); -static unsigned int BN_STACK_pop(BN_STACK *); +static int BN_STACK_push(BN_STACK *, unsigned int); +static unsigned int BN_STACK_pop(BN_STACK *); /**********/ /* BN_CTX */ /**********/ /* The opaque BN_CTX type */ -struct bignum_ctx - { - /* The bignum bundles */ - BN_POOL pool; - /* The "stack frames", if you will */ - BN_STACK stack; - /* The number of bignums currently assigned */ - unsigned int used; - /* Depth of stack overflow */ - int err_stack; - /* Block "gets" until an "end" (compatibility behaviour) */ - int too_many; - }; +struct bignum_ctx { + /* The bignum bundles */ + BN_POOL pool; + /* The "stack frames", if you will */ + BN_STACK stack; + /* The number of bignums currently assigned */ + unsigned int used; + /* Depth of stack overflow */ + int err_stack; + /* Block "gets" until an "end" (compatibility behaviour) */ + int too_many; +}; /* Enable this to find BN_CTX bugs */ #ifdef BN_CTX_DEBUG static const char *ctxdbg_cur = NULL; static void ctxdbg(BN_CTX *ctx) - { - unsigned int bnidx = 0, fpidx = 0; - BN_POOL_ITEM *item = ctx->pool.head; - BN_STACK *stack = &ctx->stack; - fprintf(stderr,"(%08x): ", (unsigned int)ctx); - while(bnidx < ctx->used) - { - fprintf(stderr,"%03x ", item->vals[bnidx++ % BN_CTX_POOL_SIZE].dmax); - if(!(bnidx % BN_CTX_POOL_SIZE)) - item = item->next; - } - fprintf(stderr,"\n"); - bnidx = 0; - fprintf(stderr," : "); - while(fpidx < stack->depth) - { - while(bnidx++ < stack->indexes[fpidx]) - fprintf(stderr," "); - fprintf(stderr,"^^^ "); - bnidx++; - fpidx++; - } - fprintf(stderr,"\n"); - } -#define CTXDBG_ENTRY(str, ctx) do { \ - ctxdbg_cur = (str); \ - fprintf(stderr,"Starting %s\n", ctxdbg_cur); \ - ctxdbg(ctx); \ - } while(0) -#define CTXDBG_EXIT(ctx) do { \ - fprintf(stderr,"Ending %s\n", ctxdbg_cur); \ - ctxdbg(ctx); \ - } while(0) -#define CTXDBG_RET(ctx,ret) +{ + unsigned int bnidx = 0, fpidx = 0; + BN_POOL_ITEM *item = ctx->pool.head; + BN_STACK *stack = &ctx->stack; + fprintf(stderr, "(%16p): ", ctx); + while (bnidx < ctx->used) { + fprintf(stderr, "%03x ", item->vals[bnidx++ % BN_CTX_POOL_SIZE].dmax); + if (!(bnidx % BN_CTX_POOL_SIZE)) + item = item->next; + } + fprintf(stderr, "\n"); + bnidx = 0; + fprintf(stderr, " : "); + while (fpidx < stack->depth) { + while (bnidx++ < stack->indexes[fpidx]) + fprintf(stderr, " "); + fprintf(stderr, "^^^ "); + bnidx++; + fpidx++; + } + fprintf(stderr, "\n"); +} + +# define CTXDBG_ENTRY(str, ctx) do { \ + ctxdbg_cur = (str); \ + fprintf(stderr,"Starting %s\n", ctxdbg_cur); \ + ctxdbg(ctx); \ + } while(0) +# define CTXDBG_EXIT(ctx) do { \ + fprintf(stderr,"Ending %s\n", ctxdbg_cur); \ + ctxdbg(ctx); \ + } while(0) +# define CTXDBG_RET(ctx,ret) #else -#define CTXDBG_ENTRY(str, ctx) -#define CTXDBG_EXIT(ctx) -#define CTXDBG_RET(ctx,ret) +# define CTXDBG_ENTRY(str, ctx) +# define CTXDBG_EXIT(ctx) +# define CTXDBG_RET(ctx,ret) #endif -/* This function is an evil legacy and should not be used. This implementation - * is WYSIWYG, though I've done my best. */ +/* + * This function is an evil legacy and should not be used. This + * implementation is WYSIWYG, though I've done my best. + */ #ifndef OPENSSL_NO_DEPRECATED void BN_CTX_init(BN_CTX *ctx) - { - /* Assume the caller obtained the context via BN_CTX_new() and so is - * trying to reset it for use. Nothing else makes sense, least of all - * binary compatibility from a time when they could declare a static - * variable. */ - BN_POOL_reset(&ctx->pool); - BN_STACK_reset(&ctx->stack); - ctx->used = 0; - ctx->err_stack = 0; - ctx->too_many = 0; - } +{ + /* + * Assume the caller obtained the context via BN_CTX_new() and so is + * trying to reset it for use. Nothing else makes sense, least of all + * binary compatibility from a time when they could declare a static + * variable. + */ + BN_POOL_reset(&ctx->pool); + BN_STACK_reset(&ctx->stack); + ctx->used = 0; + ctx->err_stack = 0; + ctx->too_many = 0; +} #endif BN_CTX *BN_CTX_new(void) - { - BN_CTX *ret = OPENSSL_malloc(sizeof(BN_CTX)); - if(!ret) - { - BNerr(BN_F_BN_CTX_NEW,ERR_R_MALLOC_FAILURE); - return NULL; - } - /* Initialise the structure */ - BN_POOL_init(&ret->pool); - BN_STACK_init(&ret->stack); - ret->used = 0; - ret->err_stack = 0; - ret->too_many = 0; - return ret; - } +{ + BN_CTX *ret = OPENSSL_malloc(sizeof(BN_CTX)); + if (!ret) { + BNerr(BN_F_BN_CTX_NEW, ERR_R_MALLOC_FAILURE); + return NULL; + } + /* Initialise the structure */ + BN_POOL_init(&ret->pool); + BN_STACK_init(&ret->stack); + ret->used = 0; + ret->err_stack = 0; + ret->too_many = 0; + return ret; +} void BN_CTX_free(BN_CTX *ctx) - { - if (ctx == NULL) - return; +{ + if (ctx == NULL) + return; #ifdef BN_CTX_DEBUG - { - BN_POOL_ITEM *pool = ctx->pool.head; - fprintf(stderr,"BN_CTX_free, stack-size=%d, pool-bignums=%d\n", - ctx->stack.size, ctx->pool.size); - fprintf(stderr,"dmaxs: "); - while(pool) { - unsigned loop = 0; - while(loop < BN_CTX_POOL_SIZE) - fprintf(stderr,"%02x ", pool->vals[loop++].dmax); - pool = pool->next; - } - fprintf(stderr,"\n"); - } + { + BN_POOL_ITEM *pool = ctx->pool.head; + fprintf(stderr, "BN_CTX_free, stack-size=%d, pool-bignums=%d\n", + ctx->stack.size, ctx->pool.size); + fprintf(stderr, "dmaxs: "); + while (pool) { + unsigned loop = 0; + while (loop < BN_CTX_POOL_SIZE) + fprintf(stderr, "%02x ", pool->vals[loop++].dmax); + pool = pool->next; + } + fprintf(stderr, "\n"); + } #endif - BN_STACK_finish(&ctx->stack); - BN_POOL_finish(&ctx->pool); - OPENSSL_free(ctx); - } + BN_STACK_finish(&ctx->stack); + BN_POOL_finish(&ctx->pool); + OPENSSL_free(ctx); +} void BN_CTX_start(BN_CTX *ctx) - { - CTXDBG_ENTRY("BN_CTX_start", ctx); - /* If we're already overflowing ... */ - if(ctx->err_stack || ctx->too_many) - ctx->err_stack++; - /* (Try to) get a new frame pointer */ - else if(!BN_STACK_push(&ctx->stack, ctx->used)) - { - BNerr(BN_F_BN_CTX_START,BN_R_TOO_MANY_TEMPORARY_VARIABLES); - ctx->err_stack++; - } - CTXDBG_EXIT(ctx); - } +{ + CTXDBG_ENTRY("BN_CTX_start", ctx); + /* If we're already overflowing ... */ + if (ctx->err_stack || ctx->too_many) + ctx->err_stack++; + /* (Try to) get a new frame pointer */ + else if (!BN_STACK_push(&ctx->stack, ctx->used)) { + BNerr(BN_F_BN_CTX_START, BN_R_TOO_MANY_TEMPORARY_VARIABLES); + ctx->err_stack++; + } + CTXDBG_EXIT(ctx); +} void BN_CTX_end(BN_CTX *ctx) - { - CTXDBG_ENTRY("BN_CTX_end", ctx); - if(ctx->err_stack) - ctx->err_stack--; - else - { - unsigned int fp = BN_STACK_pop(&ctx->stack); - /* Does this stack frame have anything to release? */ - if(fp < ctx->used) - BN_POOL_release(&ctx->pool, ctx->used - fp); - ctx->used = fp; - /* Unjam "too_many" in case "get" had failed */ - ctx->too_many = 0; - } - CTXDBG_EXIT(ctx); - } +{ + CTXDBG_ENTRY("BN_CTX_end", ctx); + if (ctx->err_stack) + ctx->err_stack--; + else { + unsigned int fp = BN_STACK_pop(&ctx->stack); + /* Does this stack frame have anything to release? */ + if (fp < ctx->used) + BN_POOL_release(&ctx->pool, ctx->used - fp); + ctx->used = fp; + /* Unjam "too_many" in case "get" had failed */ + ctx->too_many = 0; + } + CTXDBG_EXIT(ctx); +} BIGNUM *BN_CTX_get(BN_CTX *ctx) - { - BIGNUM *ret; - CTXDBG_ENTRY("BN_CTX_get", ctx); - if(ctx->err_stack || ctx->too_many) return NULL; - if((ret = BN_POOL_get(&ctx->pool)) == NULL) - { - /* Setting too_many prevents repeated "get" attempts from - * cluttering the error stack. */ - ctx->too_many = 1; - BNerr(BN_F_BN_CTX_GET,BN_R_TOO_MANY_TEMPORARY_VARIABLES); - return NULL; - } - /* OK, make sure the returned bignum is "zero" */ - BN_zero(ret); - ctx->used++; - CTXDBG_RET(ctx, ret); - return ret; - } +{ + BIGNUM *ret; + CTXDBG_ENTRY("BN_CTX_get", ctx); + if (ctx->err_stack || ctx->too_many) + return NULL; + if ((ret = BN_POOL_get(&ctx->pool)) == NULL) { + /* + * Setting too_many prevents repeated "get" attempts from cluttering + * the error stack. + */ + ctx->too_many = 1; + BNerr(BN_F_BN_CTX_GET, BN_R_TOO_MANY_TEMPORARY_VARIABLES); + return NULL; + } + /* OK, make sure the returned bignum is "zero" */ + BN_zero(ret); + ctx->used++; + CTXDBG_RET(ctx, ret); + return ret; +} /************/ /* BN_STACK */ /************/ static void BN_STACK_init(BN_STACK *st) - { - st->indexes = NULL; - st->depth = st->size = 0; - } +{ + st->indexes = NULL; + st->depth = st->size = 0; +} static void BN_STACK_finish(BN_STACK *st) - { - if(st->size) OPENSSL_free(st->indexes); - } +{ + if (st->size) + OPENSSL_free(st->indexes); +} #ifndef OPENSSL_NO_DEPRECATED static void BN_STACK_reset(BN_STACK *st) - { - st->depth = 0; - } +{ + st->depth = 0; +} #endif static int BN_STACK_push(BN_STACK *st, unsigned int idx) - { - if(st->depth == st->size) - /* Need to expand */ - { - unsigned int newsize = (st->size ? - (st->size * 3 / 2) : BN_CTX_START_FRAMES); - unsigned int *newitems = OPENSSL_malloc(newsize * - sizeof(unsigned int)); - if(!newitems) return 0; - if(st->depth) - memcpy(newitems, st->indexes, st->depth * - sizeof(unsigned int)); - if(st->size) OPENSSL_free(st->indexes); - st->indexes = newitems; - st->size = newsize; - } - st->indexes[(st->depth)++] = idx; - return 1; - } +{ + if (st->depth == st->size) + /* Need to expand */ + { + unsigned int newsize = (st->size ? + (st->size * 3 / 2) : BN_CTX_START_FRAMES); + unsigned int *newitems = OPENSSL_malloc(newsize * + sizeof(unsigned int)); + if (!newitems) + return 0; + if (st->depth) + memcpy(newitems, st->indexes, st->depth * sizeof(unsigned int)); + if (st->size) + OPENSSL_free(st->indexes); + st->indexes = newitems; + st->size = newsize; + } + st->indexes[(st->depth)++] = idx; + return 1; +} static unsigned int BN_STACK_pop(BN_STACK *st) - { - return st->indexes[--(st->depth)]; - } +{ + return st->indexes[--(st->depth)]; +} /***********/ /* BN_POOL */ /***********/ static void BN_POOL_init(BN_POOL *p) - { - p->head = p->current = p->tail = NULL; - p->used = p->size = 0; - } +{ + p->head = p->current = p->tail = NULL; + p->used = p->size = 0; +} static void BN_POOL_finish(BN_POOL *p) - { - while(p->head) - { - unsigned int loop = 0; - BIGNUM *bn = p->head->vals; - while(loop++ < BN_CTX_POOL_SIZE) - { - if(bn->d) BN_clear_free(bn); - bn++; - } - p->current = p->head->next; - OPENSSL_free(p->head); - p->head = p->current; - } - } +{ + while (p->head) { + unsigned int loop = 0; + BIGNUM *bn = p->head->vals; + while (loop++ < BN_CTX_POOL_SIZE) { + if (bn->d) + BN_clear_free(bn); + bn++; + } + p->current = p->head->next; + OPENSSL_free(p->head); + p->head = p->current; + } +} #ifndef OPENSSL_NO_DEPRECATED static void BN_POOL_reset(BN_POOL *p) - { - BN_POOL_ITEM *item = p->head; - while(item) - { - unsigned int loop = 0; - BIGNUM *bn = item->vals; - while(loop++ < BN_CTX_POOL_SIZE) - { - if(bn->d) BN_clear(bn); - bn++; - } - item = item->next; - } - p->current = p->head; - p->used = 0; - } +{ + BN_POOL_ITEM *item = p->head; + while (item) { + unsigned int loop = 0; + BIGNUM *bn = item->vals; + while (loop++ < BN_CTX_POOL_SIZE) { + if (bn->d) + BN_clear(bn); + bn++; + } + item = item->next; + } + p->current = p->head; + p->used = 0; +} #endif static BIGNUM *BN_POOL_get(BN_POOL *p) - { - if(p->used == p->size) - { - BIGNUM *bn; - unsigned int loop = 0; - BN_POOL_ITEM *item = OPENSSL_malloc(sizeof(BN_POOL_ITEM)); - if(!item) return NULL; - /* Initialise the structure */ - bn = item->vals; - while(loop++ < BN_CTX_POOL_SIZE) - BN_init(bn++); - item->prev = p->tail; - item->next = NULL; - /* Link it in */ - if(!p->head) - p->head = p->current = p->tail = item; - else - { - p->tail->next = item; - p->tail = item; - p->current = item; - } - p->size += BN_CTX_POOL_SIZE; - p->used++; - /* Return the first bignum from the new pool */ - return item->vals; - } - if(!p->used) - p->current = p->head; - else if((p->used % BN_CTX_POOL_SIZE) == 0) - p->current = p->current->next; - return p->current->vals + ((p->used++) % BN_CTX_POOL_SIZE); - } +{ + if (p->used == p->size) { + BIGNUM *bn; + unsigned int loop = 0; + BN_POOL_ITEM *item = OPENSSL_malloc(sizeof(BN_POOL_ITEM)); + if (!item) + return NULL; + /* Initialise the structure */ + bn = item->vals; + while (loop++ < BN_CTX_POOL_SIZE) + BN_init(bn++); + item->prev = p->tail; + item->next = NULL; + /* Link it in */ + if (!p->head) + p->head = p->current = p->tail = item; + else { + p->tail->next = item; + p->tail = item; + p->current = item; + } + p->size += BN_CTX_POOL_SIZE; + p->used++; + /* Return the first bignum from the new pool */ + return item->vals; + } + if (!p->used) + p->current = p->head; + else if ((p->used % BN_CTX_POOL_SIZE) == 0) + p->current = p->current->next; + return p->current->vals + ((p->used++) % BN_CTX_POOL_SIZE); +} static void BN_POOL_release(BN_POOL *p, unsigned int num) - { - unsigned int offset = (p->used - 1) % BN_CTX_POOL_SIZE; - p->used -= num; - while(num--) - { - bn_check_top(p->current->vals + offset); - if(!offset) - { - offset = BN_CTX_POOL_SIZE - 1; - p->current = p->current->prev; - } - else - offset--; - } - } - +{ + unsigned int offset = (p->used - 1) % BN_CTX_POOL_SIZE; + p->used -= num; + while (num--) { + bn_check_top(p->current->vals + offset); + if (!offset) { + offset = BN_CTX_POOL_SIZE - 1; + p->current = p->current->prev; + } else + offset--; + } +} diff --git a/openssl/crypto/bn/bn_depr.c b/openssl/crypto/bn/bn_depr.c index 27535e4fc..34895f598 100644 --- a/openssl/crypto/bn/bn_depr.c +++ b/openssl/crypto/bn/bn_depr.c @@ -7,7 +7,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -53,8 +53,10 @@ * */ -/* Support for deprecated functions goes here - static linkage will only slurp - * this code if applications are using them directly. */ +/* + * Support for deprecated functions goes here - static linkage will only + * slurp this code if applications are using them directly. + */ #include <stdio.h> #include <time.h> @@ -62,51 +64,52 @@ #include "bn_lcl.h" #include <openssl/rand.h> -static void *dummy=&dummy; +static void *dummy = &dummy; #ifndef OPENSSL_NO_DEPRECATED BIGNUM *BN_generate_prime(BIGNUM *ret, int bits, int safe, - const BIGNUM *add, const BIGNUM *rem, - void (*callback)(int,int,void *), void *cb_arg) - { - BN_GENCB cb; - BIGNUM *rnd=NULL; - int found = 0; + const BIGNUM *add, const BIGNUM *rem, + void (*callback) (int, int, void *), void *cb_arg) +{ + BN_GENCB cb; + BIGNUM *rnd = NULL; + int found = 0; - BN_GENCB_set_old(&cb, callback, cb_arg); + BN_GENCB_set_old(&cb, callback, cb_arg); - if (ret == NULL) - { - if ((rnd=BN_new()) == NULL) goto err; - } - else - rnd=ret; - if(!BN_generate_prime_ex(rnd, bits, safe, add, rem, &cb)) - goto err; + if (ret == NULL) { + if ((rnd = BN_new()) == NULL) + goto err; + } else + rnd = ret; + if (!BN_generate_prime_ex(rnd, bits, safe, add, rem, &cb)) + goto err; - /* we have a prime :-) */ - found = 1; -err: - if (!found && (ret == NULL) && (rnd != NULL)) BN_free(rnd); - return(found ? rnd : NULL); - } + /* we have a prime :-) */ + found = 1; + err: + if (!found && (ret == NULL) && (rnd != NULL)) + BN_free(rnd); + return (found ? rnd : NULL); +} -int BN_is_prime(const BIGNUM *a, int checks, void (*callback)(int,int,void *), - BN_CTX *ctx_passed, void *cb_arg) - { - BN_GENCB cb; - BN_GENCB_set_old(&cb, callback, cb_arg); - return BN_is_prime_ex(a, checks, ctx_passed, &cb); - } +int BN_is_prime(const BIGNUM *a, int checks, + void (*callback) (int, int, void *), BN_CTX *ctx_passed, + void *cb_arg) +{ + BN_GENCB cb; + BN_GENCB_set_old(&cb, callback, cb_arg); + return BN_is_prime_ex(a, checks, ctx_passed, &cb); +} int BN_is_prime_fasttest(const BIGNUM *a, int checks, - void (*callback)(int,int,void *), - BN_CTX *ctx_passed, void *cb_arg, - int do_trial_division) - { - BN_GENCB cb; - BN_GENCB_set_old(&cb, callback, cb_arg); - return BN_is_prime_fasttest_ex(a, checks, ctx_passed, - do_trial_division, &cb); - } + void (*callback) (int, int, void *), + BN_CTX *ctx_passed, void *cb_arg, + int do_trial_division) +{ + BN_GENCB cb; + BN_GENCB_set_old(&cb, callback, cb_arg); + return BN_is_prime_fasttest_ex(a, checks, ctx_passed, + do_trial_division, &cb); +} #endif diff --git a/openssl/crypto/bn/bn_div.c b/openssl/crypto/bn/bn_div.c index 7b2403185..72e6ce3f7 100644 --- a/openssl/crypto/bn/bn_div.c +++ b/openssl/crypto/bn/bn_div.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -61,77 +61,86 @@ #include "cryptlib.h" #include "bn_lcl.h" - /* The old slow way */ #if 0 int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, - BN_CTX *ctx) - { - int i,nm,nd; - int ret = 0; - BIGNUM *D; - - bn_check_top(m); - bn_check_top(d); - if (BN_is_zero(d)) - { - BNerr(BN_F_BN_DIV,BN_R_DIV_BY_ZERO); - return(0); - } - - if (BN_ucmp(m,d) < 0) - { - if (rem != NULL) - { if (BN_copy(rem,m) == NULL) return(0); } - if (dv != NULL) BN_zero(dv); - return(1); - } - - BN_CTX_start(ctx); - D = BN_CTX_get(ctx); - if (dv == NULL) dv = BN_CTX_get(ctx); - if (rem == NULL) rem = BN_CTX_get(ctx); - if (D == NULL || dv == NULL || rem == NULL) - goto end; - - nd=BN_num_bits(d); - nm=BN_num_bits(m); - if (BN_copy(D,d) == NULL) goto end; - if (BN_copy(rem,m) == NULL) goto end; - - /* The next 2 are needed so we can do a dv->d[0]|=1 later - * since BN_lshift1 will only work once there is a value :-) */ - BN_zero(dv); - if(bn_wexpand(dv,1) == NULL) goto end; - dv->top=1; - - if (!BN_lshift(D,D,nm-nd)) goto end; - for (i=nm-nd; i>=0; i--) - { - if (!BN_lshift1(dv,dv)) goto end; - if (BN_ucmp(rem,D) >= 0) - { - dv->d[0]|=1; - if (!BN_usub(rem,rem,D)) goto end; - } + BN_CTX *ctx) +{ + int i, nm, nd; + int ret = 0; + BIGNUM *D; + + bn_check_top(m); + bn_check_top(d); + if (BN_is_zero(d)) { + BNerr(BN_F_BN_DIV, BN_R_DIV_BY_ZERO); + return (0); + } + + if (BN_ucmp(m, d) < 0) { + if (rem != NULL) { + if (BN_copy(rem, m) == NULL) + return (0); + } + if (dv != NULL) + BN_zero(dv); + return (1); + } + + BN_CTX_start(ctx); + D = BN_CTX_get(ctx); + if (dv == NULL) + dv = BN_CTX_get(ctx); + if (rem == NULL) + rem = BN_CTX_get(ctx); + if (D == NULL || dv == NULL || rem == NULL) + goto end; + + nd = BN_num_bits(d); + nm = BN_num_bits(m); + if (BN_copy(D, d) == NULL) + goto end; + if (BN_copy(rem, m) == NULL) + goto end; + + /* + * The next 2 are needed so we can do a dv->d[0]|=1 later since + * BN_lshift1 will only work once there is a value :-) + */ + BN_zero(dv); + if (bn_wexpand(dv, 1) == NULL) + goto end; + dv->top = 1; + + if (!BN_lshift(D, D, nm - nd)) + goto end; + for (i = nm - nd; i >= 0; i--) { + if (!BN_lshift1(dv, dv)) + goto end; + if (BN_ucmp(rem, D) >= 0) { + dv->d[0] |= 1; + if (!BN_usub(rem, rem, D)) + goto end; + } /* CAN IMPROVE (and have now :=) */ - if (!BN_rshift1(D,D)) goto end; - } - rem->neg=BN_is_zero(rem)?0:m->neg; - dv->neg=m->neg^d->neg; - ret = 1; + if (!BN_rshift1(D, D)) + goto end; + } + rem->neg = BN_is_zero(rem) ? 0 : m->neg; + dv->neg = m->neg ^ d->neg; + ret = 1; end: - BN_CTX_end(ctx); - return(ret); - } + BN_CTX_end(ctx); + return (ret); +} #else -#if !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) \ +# if !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) \ && !defined(PEDANTIC) && !defined(BN_DIV3W) -# if defined(__GNUC__) && __GNUC__>=2 -# if defined(__i386) || defined (__i386__) - /* +# if defined(__GNUC__) && __GNUC__>=2 +# if defined(__i386) || defined (__i386__) + /*- * There were two reasons for implementing this template: * - GNU C generates a call to a function (__udivdi3 to be exact) * in reply to ((((BN_ULLONG)n0)<<BN_BITS2)|n1)/d0 (I fail to @@ -139,39 +148,39 @@ int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, * - divl doesn't only calculate quotient, but also leaves * remainder in %edx which we can definitely use here:-) * - * <appro@fy.chalmers.se> + * <appro@fy.chalmers.se> */ -#undef bn_div_words -# define bn_div_words(n0,n1,d0) \ - ({ asm volatile ( \ - "divl %4" \ - : "=a"(q), "=d"(rem) \ - : "a"(n1), "d"(n0), "g"(d0) \ - : "cc"); \ - q; \ - }) -# define REMAINDER_IS_ALREADY_CALCULATED -# elif defined(__x86_64) && defined(SIXTY_FOUR_BIT_LONG) +# undef bn_div_words +# define bn_div_words(n0,n1,d0) \ + ({ asm volatile ( \ + "divl %4" \ + : "=a"(q), "=d"(rem) \ + : "a"(n1), "d"(n0), "g"(d0) \ + : "cc"); \ + q; \ + }) +# define REMAINDER_IS_ALREADY_CALCULATED +# elif defined(__x86_64) && defined(SIXTY_FOUR_BIT_LONG) /* * Same story here, but it's 128-bit by 64-bit division. Wow! - * <appro@fy.chalmers.se> + * <appro@fy.chalmers.se> */ -# undef bn_div_words -# define bn_div_words(n0,n1,d0) \ - ({ asm volatile ( \ - "divq %4" \ - : "=a"(q), "=d"(rem) \ - : "a"(n1), "d"(n0), "g"(d0) \ - : "cc"); \ - q; \ - }) -# define REMAINDER_IS_ALREADY_CALCULATED -# endif /* __<cpu> */ -# endif /* __GNUC__ */ -#endif /* OPENSSL_NO_ASM */ - - -/* BN_div computes dv := num / divisor, rounding towards +# undef bn_div_words +# define bn_div_words(n0,n1,d0) \ + ({ asm volatile ( \ + "divq %4" \ + : "=a"(q), "=d"(rem) \ + : "a"(n1), "d"(n0), "g"(d0) \ + : "cc"); \ + q; \ + }) +# define REMAINDER_IS_ALREADY_CALCULATED +# endif /* __<cpu> */ +# endif /* __GNUC__ */ +# endif /* OPENSSL_NO_ASM */ + +/*- + * BN_div computes dv := num / divisor, rounding towards * zero, and sets up rm such that dv*divisor + rm = num holds. * Thus: * dv->neg == num->neg ^ divisor->neg (unless the result is zero) @@ -179,270 +188,290 @@ int BN_div(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, * If 'dv' or 'rm' is NULL, the respective value is not returned. */ int BN_div(BIGNUM *dv, BIGNUM *rm, const BIGNUM *num, const BIGNUM *divisor, - BN_CTX *ctx) - { - int norm_shift,i,loop; - BIGNUM *tmp,wnum,*snum,*sdiv,*res; - BN_ULONG *resp,*wnump; - BN_ULONG d0,d1; - int num_n,div_n; - int no_branch=0; - - /* Invalid zero-padding would have particularly bad consequences - * in the case of 'num', so don't just rely on bn_check_top() for this one - * (bn_check_top() works only for BN_DEBUG builds) */ - if (num->top > 0 && num->d[num->top - 1] == 0) - { - BNerr(BN_F_BN_DIV,BN_R_NOT_INITIALIZED); - return 0; - } - - bn_check_top(num); - - if ((BN_get_flags(num, BN_FLG_CONSTTIME) != 0) || (BN_get_flags(divisor, BN_FLG_CONSTTIME) != 0)) - { - no_branch=1; - } - - bn_check_top(dv); - bn_check_top(rm); - /* bn_check_top(num); */ /* 'num' has been checked already */ - bn_check_top(divisor); - - if (BN_is_zero(divisor)) - { - BNerr(BN_F_BN_DIV,BN_R_DIV_BY_ZERO); - return(0); - } - - if (!no_branch && BN_ucmp(num,divisor) < 0) - { - if (rm != NULL) - { if (BN_copy(rm,num) == NULL) return(0); } - if (dv != NULL) BN_zero(dv); - return(1); - } - - BN_CTX_start(ctx); - tmp=BN_CTX_get(ctx); - snum=BN_CTX_get(ctx); - sdiv=BN_CTX_get(ctx); - if (dv == NULL) - res=BN_CTX_get(ctx); - else res=dv; - if (sdiv == NULL || res == NULL || tmp == NULL || snum == NULL) - goto err; - - /* First we normalise the numbers */ - norm_shift=BN_BITS2-((BN_num_bits(divisor))%BN_BITS2); - if (!(BN_lshift(sdiv,divisor,norm_shift))) goto err; - sdiv->neg=0; - norm_shift+=BN_BITS2; - if (!(BN_lshift(snum,num,norm_shift))) goto err; - snum->neg=0; - - if (no_branch) - { - /* Since we don't know whether snum is larger than sdiv, - * we pad snum with enough zeroes without changing its - * value. - */ - if (snum->top <= sdiv->top+1) - { - if (bn_wexpand(snum, sdiv->top + 2) == NULL) goto err; - for (i = snum->top; i < sdiv->top + 2; i++) snum->d[i] = 0; - snum->top = sdiv->top + 2; - } - else - { - if (bn_wexpand(snum, snum->top + 1) == NULL) goto err; - snum->d[snum->top] = 0; - snum->top ++; - } - } - - div_n=sdiv->top; - num_n=snum->top; - loop=num_n-div_n; - /* Lets setup a 'window' into snum - * This is the part that corresponds to the current - * 'area' being divided */ - wnum.neg = 0; - wnum.d = &(snum->d[loop]); - wnum.top = div_n; - /* only needed when BN_ucmp messes up the values between top and max */ - wnum.dmax = snum->dmax - loop; /* so we don't step out of bounds */ - - /* Get the top 2 words of sdiv */ - /* div_n=sdiv->top; */ - d0=sdiv->d[div_n-1]; - d1=(div_n == 1)?0:sdiv->d[div_n-2]; - - /* pointer to the 'top' of snum */ - wnump= &(snum->d[num_n-1]); - - /* Setup to 'res' */ - res->neg= (num->neg^divisor->neg); - if (!bn_wexpand(res,(loop+1))) goto err; - res->top=loop-no_branch; - resp= &(res->d[loop-1]); - - /* space for temp */ - if (!bn_wexpand(tmp,(div_n+1))) goto err; - - if (!no_branch) - { - if (BN_ucmp(&wnum,sdiv) >= 0) - { - /* If BN_DEBUG_RAND is defined BN_ucmp changes (via - * bn_pollute) the const bignum arguments => - * clean the values between top and max again */ - bn_clear_top2max(&wnum); - bn_sub_words(wnum.d, wnum.d, sdiv->d, div_n); - *resp=1; - } - else - res->top--; - } - - /* if res->top == 0 then clear the neg value otherwise decrease - * the resp pointer */ - if (res->top == 0) - res->neg = 0; - else - resp--; - - for (i=0; i<loop-1; i++, wnump--, resp--) - { - BN_ULONG q,l0; - /* the first part of the loop uses the top two words of - * snum and sdiv to calculate a BN_ULONG q such that - * | wnum - sdiv * q | < sdiv */ -#if defined(BN_DIV3W) && !defined(OPENSSL_NO_ASM) - BN_ULONG bn_div_3_words(BN_ULONG*,BN_ULONG,BN_ULONG); - q=bn_div_3_words(wnump,d1,d0); -#else - BN_ULONG n0,n1,rem=0; - - n0=wnump[0]; - n1=wnump[-1]; - if (n0 == d0) - q=BN_MASK2; - else /* n0 < d0 */ - { -#ifdef BN_LLONG - BN_ULLONG t2; - -#if defined(BN_LLONG) && defined(BN_DIV2W) && !defined(bn_div_words) - q=(BN_ULONG)(((((BN_ULLONG)n0)<<BN_BITS2)|n1)/d0); -#else - q=bn_div_words(n0,n1,d0); -#ifdef BN_DEBUG_LEVITTE - fprintf(stderr,"DEBUG: bn_div_words(0x%08X,0x%08X,0x%08\ -X) -> 0x%08X\n", - n0, n1, d0, q); -#endif -#endif - -#ifndef REMAINDER_IS_ALREADY_CALCULATED - /* - * rem doesn't have to be BN_ULLONG. The least we - * know it's less that d0, isn't it? - */ - rem=(n1-q*d0)&BN_MASK2; -#endif - t2=(BN_ULLONG)d1*q; - - for (;;) - { - if (t2 <= ((((BN_ULLONG)rem)<<BN_BITS2)|wnump[-2])) - break; - q--; - rem += d0; - if (rem < d0) break; /* don't let rem overflow */ - t2 -= d1; - } -#else /* !BN_LLONG */ - BN_ULONG t2l,t2h; - - q=bn_div_words(n0,n1,d0); -#ifdef BN_DEBUG_LEVITTE - fprintf(stderr,"DEBUG: bn_div_words(0x%08X,0x%08X,0x%08\ -X) -> 0x%08X\n", - n0, n1, d0, q); -#endif -#ifndef REMAINDER_IS_ALREADY_CALCULATED - rem=(n1-q*d0)&BN_MASK2; -#endif - -#if defined(BN_UMULT_LOHI) - BN_UMULT_LOHI(t2l,t2h,d1,q); -#elif defined(BN_UMULT_HIGH) - t2l = d1 * q; - t2h = BN_UMULT_HIGH(d1,q); -#else - { - BN_ULONG ql, qh; - t2l=LBITS(d1); t2h=HBITS(d1); - ql =LBITS(q); qh =HBITS(q); - mul64(t2l,t2h,ql,qh); /* t2=(BN_ULLONG)d1*q; */ - } -#endif - - for (;;) - { - if ((t2h < rem) || - ((t2h == rem) && (t2l <= wnump[-2]))) - break; - q--; - rem += d0; - if (rem < d0) break; /* don't let rem overflow */ - if (t2l < d1) t2h--; t2l -= d1; - } -#endif /* !BN_LLONG */ - } -#endif /* !BN_DIV3W */ - - l0=bn_mul_words(tmp->d,sdiv->d,div_n,q); - tmp->d[div_n]=l0; - wnum.d--; - /* ingore top values of the bignums just sub the two - * BN_ULONG arrays with bn_sub_words */ - if (bn_sub_words(wnum.d, wnum.d, tmp->d, div_n+1)) - { - /* Note: As we have considered only the leading - * two BN_ULONGs in the calculation of q, sdiv * q - * might be greater than wnum (but then (q-1) * sdiv - * is less or equal than wnum) - */ - q--; - if (bn_add_words(wnum.d, wnum.d, sdiv->d, div_n)) - /* we can't have an overflow here (assuming - * that q != 0, but if q == 0 then tmp is - * zero anyway) */ - (*wnump)++; - } - /* store part of the result */ - *resp = q; - } - bn_correct_top(snum); - if (rm != NULL) - { - /* Keep a copy of the neg flag in num because if rm==num - * BN_rshift() will overwrite it. - */ - int neg = num->neg; - BN_rshift(rm,snum,norm_shift); - if (!BN_is_zero(rm)) - rm->neg = neg; - bn_check_top(rm); - } - if (no_branch) bn_correct_top(res); - BN_CTX_end(ctx); - return(1); -err: - bn_check_top(rm); - BN_CTX_end(ctx); - return(0); - } + BN_CTX *ctx) +{ + int norm_shift, i, loop; + BIGNUM *tmp, wnum, *snum, *sdiv, *res; + BN_ULONG *resp, *wnump; + BN_ULONG d0, d1; + int num_n, div_n; + int no_branch = 0; + + /* + * Invalid zero-padding would have particularly bad consequences so don't + * just rely on bn_check_top() here (bn_check_top() works only for + * BN_DEBUG builds) + */ + if ((num->top > 0 && num->d[num->top - 1] == 0) || + (divisor->top > 0 && divisor->d[divisor->top - 1] == 0)) { + BNerr(BN_F_BN_DIV, BN_R_NOT_INITIALIZED); + return 0; + } + + bn_check_top(num); + bn_check_top(divisor); + + if ((BN_get_flags(num, BN_FLG_CONSTTIME) != 0) + || (BN_get_flags(divisor, BN_FLG_CONSTTIME) != 0)) { + no_branch = 1; + } + + bn_check_top(dv); + bn_check_top(rm); + /*- bn_check_top(num); *//* + * 'num' has been checked already + */ + /*- bn_check_top(divisor); *//* + * 'divisor' has been checked already + */ + + if (BN_is_zero(divisor)) { + BNerr(BN_F_BN_DIV, BN_R_DIV_BY_ZERO); + return (0); + } + + if (!no_branch && BN_ucmp(num, divisor) < 0) { + if (rm != NULL) { + if (BN_copy(rm, num) == NULL) + return (0); + } + if (dv != NULL) + BN_zero(dv); + return (1); + } + + BN_CTX_start(ctx); + tmp = BN_CTX_get(ctx); + snum = BN_CTX_get(ctx); + sdiv = BN_CTX_get(ctx); + if (dv == NULL) + res = BN_CTX_get(ctx); + else + res = dv; + if (sdiv == NULL || res == NULL || tmp == NULL || snum == NULL) + goto err; + + /* First we normalise the numbers */ + norm_shift = BN_BITS2 - ((BN_num_bits(divisor)) % BN_BITS2); + if (!(BN_lshift(sdiv, divisor, norm_shift))) + goto err; + sdiv->neg = 0; + norm_shift += BN_BITS2; + if (!(BN_lshift(snum, num, norm_shift))) + goto err; + snum->neg = 0; + + if (no_branch) { + /* + * Since we don't know whether snum is larger than sdiv, we pad snum + * with enough zeroes without changing its value. + */ + if (snum->top <= sdiv->top + 1) { + if (bn_wexpand(snum, sdiv->top + 2) == NULL) + goto err; + for (i = snum->top; i < sdiv->top + 2; i++) + snum->d[i] = 0; + snum->top = sdiv->top + 2; + } else { + if (bn_wexpand(snum, snum->top + 1) == NULL) + goto err; + snum->d[snum->top] = 0; + snum->top++; + } + } + + div_n = sdiv->top; + num_n = snum->top; + loop = num_n - div_n; + /* + * Lets setup a 'window' into snum This is the part that corresponds to + * the current 'area' being divided + */ + wnum.neg = 0; + wnum.d = &(snum->d[loop]); + wnum.top = div_n; + /* + * only needed when BN_ucmp messes up the values between top and max + */ + wnum.dmax = snum->dmax - loop; /* so we don't step out of bounds */ + + /* Get the top 2 words of sdiv */ + /* div_n=sdiv->top; */ + d0 = sdiv->d[div_n - 1]; + d1 = (div_n == 1) ? 0 : sdiv->d[div_n - 2]; + + /* pointer to the 'top' of snum */ + wnump = &(snum->d[num_n - 1]); + + /* Setup to 'res' */ + res->neg = (num->neg ^ divisor->neg); + if (!bn_wexpand(res, (loop + 1))) + goto err; + res->top = loop - no_branch; + resp = &(res->d[loop - 1]); + + /* space for temp */ + if (!bn_wexpand(tmp, (div_n + 1))) + goto err; + + if (!no_branch) { + if (BN_ucmp(&wnum, sdiv) >= 0) { + /* + * If BN_DEBUG_RAND is defined BN_ucmp changes (via bn_pollute) + * the const bignum arguments => clean the values between top and + * max again + */ + bn_clear_top2max(&wnum); + bn_sub_words(wnum.d, wnum.d, sdiv->d, div_n); + *resp = 1; + } else + res->top--; + } + + /* + * if res->top == 0 then clear the neg value otherwise decrease the resp + * pointer + */ + if (res->top == 0) + res->neg = 0; + else + resp--; + + for (i = 0; i < loop - 1; i++, wnump--, resp--) { + BN_ULONG q, l0; + /* + * the first part of the loop uses the top two words of snum and sdiv + * to calculate a BN_ULONG q such that | wnum - sdiv * q | < sdiv + */ +# if defined(BN_DIV3W) && !defined(OPENSSL_NO_ASM) + BN_ULONG bn_div_3_words(BN_ULONG *, BN_ULONG, BN_ULONG); + q = bn_div_3_words(wnump, d1, d0); +# else + BN_ULONG n0, n1, rem = 0; + + n0 = wnump[0]; + n1 = wnump[-1]; + if (n0 == d0) + q = BN_MASK2; + else { /* n0 < d0 */ + +# ifdef BN_LLONG + BN_ULLONG t2; + +# if defined(BN_LLONG) && defined(BN_DIV2W) && !defined(bn_div_words) + q = (BN_ULONG)(((((BN_ULLONG) n0) << BN_BITS2) | n1) / d0); +# else + q = bn_div_words(n0, n1, d0); +# ifdef BN_DEBUG_LEVITTE + fprintf(stderr, "DEBUG: bn_div_words(0x%08X,0x%08X,0x%08\ +X) -> 0x%08X\n", n0, n1, d0, q); +# endif +# endif + +# ifndef REMAINDER_IS_ALREADY_CALCULATED + /* + * rem doesn't have to be BN_ULLONG. The least we + * know it's less that d0, isn't it? + */ + rem = (n1 - q * d0) & BN_MASK2; +# endif + t2 = (BN_ULLONG) d1 *q; + + for (;;) { + if (t2 <= ((((BN_ULLONG) rem) << BN_BITS2) | wnump[-2])) + break; + q--; + rem += d0; + if (rem < d0) + break; /* don't let rem overflow */ + t2 -= d1; + } +# else /* !BN_LLONG */ + BN_ULONG t2l, t2h; + + q = bn_div_words(n0, n1, d0); +# ifdef BN_DEBUG_LEVITTE + fprintf(stderr, "DEBUG: bn_div_words(0x%08X,0x%08X,0x%08\ +X) -> 0x%08X\n", n0, n1, d0, q); +# endif +# ifndef REMAINDER_IS_ALREADY_CALCULATED + rem = (n1 - q * d0) & BN_MASK2; +# endif + +# if defined(BN_UMULT_LOHI) + BN_UMULT_LOHI(t2l, t2h, d1, q); +# elif defined(BN_UMULT_HIGH) + t2l = d1 * q; + t2h = BN_UMULT_HIGH(d1, q); +# else + { + BN_ULONG ql, qh; + t2l = LBITS(d1); + t2h = HBITS(d1); + ql = LBITS(q); + qh = HBITS(q); + mul64(t2l, t2h, ql, qh); /* t2=(BN_ULLONG)d1*q; */ + } +# endif + + for (;;) { + if ((t2h < rem) || ((t2h == rem) && (t2l <= wnump[-2]))) + break; + q--; + rem += d0; + if (rem < d0) + break; /* don't let rem overflow */ + if (t2l < d1) + t2h--; + t2l -= d1; + } +# endif /* !BN_LLONG */ + } +# endif /* !BN_DIV3W */ + + l0 = bn_mul_words(tmp->d, sdiv->d, div_n, q); + tmp->d[div_n] = l0; + wnum.d--; + /* + * ingore top values of the bignums just sub the two BN_ULONG arrays + * with bn_sub_words + */ + if (bn_sub_words(wnum.d, wnum.d, tmp->d, div_n + 1)) { + /* + * Note: As we have considered only the leading two BN_ULONGs in + * the calculation of q, sdiv * q might be greater than wnum (but + * then (q-1) * sdiv is less or equal than wnum) + */ + q--; + if (bn_add_words(wnum.d, wnum.d, sdiv->d, div_n)) + /* + * we can't have an overflow here (assuming that q != 0, but + * if q == 0 then tmp is zero anyway) + */ + (*wnump)++; + } + /* store part of the result */ + *resp = q; + } + bn_correct_top(snum); + if (rm != NULL) { + /* + * Keep a copy of the neg flag in num because if rm==num BN_rshift() + * will overwrite it. + */ + int neg = num->neg; + BN_rshift(rm, snum, norm_shift); + if (!BN_is_zero(rm)) + rm->neg = neg; + bn_check_top(rm); + } + if (no_branch) + bn_correct_top(res); + BN_CTX_end(ctx); + return (1); + err: + bn_check_top(rm); + BN_CTX_end(ctx); + return (0); +} #endif diff --git a/openssl/crypto/bn/bn_err.c b/openssl/crypto/bn/bn_err.c index cfe2eb94a..faa7e226b 100644 --- a/openssl/crypto/bn/bn_err.c +++ b/openssl/crypto/bn/bn_err.c @@ -7,7 +7,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -53,7 +53,8 @@ * */ -/* NOTE: this file was auto generated by the mkerr.pl script: any changes +/* + * NOTE: this file was auto generated by the mkerr.pl script: any changes * made to it will be overwritten when the script next updates this file, * only reason strings will be preserved. */ @@ -65,86 +66,85 @@ /* BEGIN ERROR CODES */ #ifndef OPENSSL_NO_ERR -#define ERR_FUNC(func) ERR_PACK(ERR_LIB_BN,func,0) -#define ERR_REASON(reason) ERR_PACK(ERR_LIB_BN,0,reason) +# define ERR_FUNC(func) ERR_PACK(ERR_LIB_BN,func,0) +# define ERR_REASON(reason) ERR_PACK(ERR_LIB_BN,0,reason) -static ERR_STRING_DATA BN_str_functs[]= - { -{ERR_FUNC(BN_F_BNRAND), "BNRAND"}, -{ERR_FUNC(BN_F_BN_BLINDING_CONVERT_EX), "BN_BLINDING_convert_ex"}, -{ERR_FUNC(BN_F_BN_BLINDING_CREATE_PARAM), "BN_BLINDING_create_param"}, -{ERR_FUNC(BN_F_BN_BLINDING_INVERT_EX), "BN_BLINDING_invert_ex"}, -{ERR_FUNC(BN_F_BN_BLINDING_NEW), "BN_BLINDING_new"}, -{ERR_FUNC(BN_F_BN_BLINDING_UPDATE), "BN_BLINDING_update"}, -{ERR_FUNC(BN_F_BN_BN2DEC), "BN_bn2dec"}, -{ERR_FUNC(BN_F_BN_BN2HEX), "BN_bn2hex"}, -{ERR_FUNC(BN_F_BN_CTX_GET), "BN_CTX_get"}, -{ERR_FUNC(BN_F_BN_CTX_NEW), "BN_CTX_new"}, -{ERR_FUNC(BN_F_BN_CTX_START), "BN_CTX_start"}, -{ERR_FUNC(BN_F_BN_DIV), "BN_div"}, -{ERR_FUNC(BN_F_BN_DIV_NO_BRANCH), "BN_div_no_branch"}, -{ERR_FUNC(BN_F_BN_DIV_RECP), "BN_div_recp"}, -{ERR_FUNC(BN_F_BN_EXP), "BN_exp"}, -{ERR_FUNC(BN_F_BN_EXPAND2), "bn_expand2"}, -{ERR_FUNC(BN_F_BN_EXPAND_INTERNAL), "BN_EXPAND_INTERNAL"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD), "BN_GF2m_mod"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD_EXP), "BN_GF2m_mod_exp"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD_MUL), "BN_GF2m_mod_mul"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD_SOLVE_QUAD), "BN_GF2m_mod_solve_quad"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR), "BN_GF2m_mod_solve_quad_arr"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD_SQR), "BN_GF2m_mod_sqr"}, -{ERR_FUNC(BN_F_BN_GF2M_MOD_SQRT), "BN_GF2m_mod_sqrt"}, -{ERR_FUNC(BN_F_BN_MOD_EXP2_MONT), "BN_mod_exp2_mont"}, -{ERR_FUNC(BN_F_BN_MOD_EXP_MONT), "BN_mod_exp_mont"}, -{ERR_FUNC(BN_F_BN_MOD_EXP_MONT_CONSTTIME), "BN_mod_exp_mont_consttime"}, -{ERR_FUNC(BN_F_BN_MOD_EXP_MONT_WORD), "BN_mod_exp_mont_word"}, -{ERR_FUNC(BN_F_BN_MOD_EXP_RECP), "BN_mod_exp_recp"}, -{ERR_FUNC(BN_F_BN_MOD_EXP_SIMPLE), "BN_mod_exp_simple"}, -{ERR_FUNC(BN_F_BN_MOD_INVERSE), "BN_mod_inverse"}, -{ERR_FUNC(BN_F_BN_MOD_INVERSE_NO_BRANCH), "BN_mod_inverse_no_branch"}, -{ERR_FUNC(BN_F_BN_MOD_LSHIFT_QUICK), "BN_mod_lshift_quick"}, -{ERR_FUNC(BN_F_BN_MOD_MUL_RECIPROCAL), "BN_mod_mul_reciprocal"}, -{ERR_FUNC(BN_F_BN_MOD_SQRT), "BN_mod_sqrt"}, -{ERR_FUNC(BN_F_BN_MPI2BN), "BN_mpi2bn"}, -{ERR_FUNC(BN_F_BN_NEW), "BN_new"}, -{ERR_FUNC(BN_F_BN_RAND), "BN_rand"}, -{ERR_FUNC(BN_F_BN_RAND_RANGE), "BN_rand_range"}, -{ERR_FUNC(BN_F_BN_USUB), "BN_usub"}, -{0,NULL} - }; +static ERR_STRING_DATA BN_str_functs[] = { + {ERR_FUNC(BN_F_BNRAND), "BNRAND"}, + {ERR_FUNC(BN_F_BN_BLINDING_CONVERT_EX), "BN_BLINDING_convert_ex"}, + {ERR_FUNC(BN_F_BN_BLINDING_CREATE_PARAM), "BN_BLINDING_create_param"}, + {ERR_FUNC(BN_F_BN_BLINDING_INVERT_EX), "BN_BLINDING_invert_ex"}, + {ERR_FUNC(BN_F_BN_BLINDING_NEW), "BN_BLINDING_new"}, + {ERR_FUNC(BN_F_BN_BLINDING_UPDATE), "BN_BLINDING_update"}, + {ERR_FUNC(BN_F_BN_BN2DEC), "BN_bn2dec"}, + {ERR_FUNC(BN_F_BN_BN2HEX), "BN_bn2hex"}, + {ERR_FUNC(BN_F_BN_CTX_GET), "BN_CTX_get"}, + {ERR_FUNC(BN_F_BN_CTX_NEW), "BN_CTX_new"}, + {ERR_FUNC(BN_F_BN_CTX_START), "BN_CTX_start"}, + {ERR_FUNC(BN_F_BN_DIV), "BN_div"}, + {ERR_FUNC(BN_F_BN_DIV_NO_BRANCH), "BN_div_no_branch"}, + {ERR_FUNC(BN_F_BN_DIV_RECP), "BN_div_recp"}, + {ERR_FUNC(BN_F_BN_EXP), "BN_exp"}, + {ERR_FUNC(BN_F_BN_EXPAND2), "bn_expand2"}, + {ERR_FUNC(BN_F_BN_EXPAND_INTERNAL), "BN_EXPAND_INTERNAL"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD), "BN_GF2m_mod"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD_EXP), "BN_GF2m_mod_exp"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD_MUL), "BN_GF2m_mod_mul"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD_SOLVE_QUAD), "BN_GF2m_mod_solve_quad"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR), "BN_GF2m_mod_solve_quad_arr"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD_SQR), "BN_GF2m_mod_sqr"}, + {ERR_FUNC(BN_F_BN_GF2M_MOD_SQRT), "BN_GF2m_mod_sqrt"}, + {ERR_FUNC(BN_F_BN_MOD_EXP2_MONT), "BN_mod_exp2_mont"}, + {ERR_FUNC(BN_F_BN_MOD_EXP_MONT), "BN_mod_exp_mont"}, + {ERR_FUNC(BN_F_BN_MOD_EXP_MONT_CONSTTIME), "BN_mod_exp_mont_consttime"}, + {ERR_FUNC(BN_F_BN_MOD_EXP_MONT_WORD), "BN_mod_exp_mont_word"}, + {ERR_FUNC(BN_F_BN_MOD_EXP_RECP), "BN_mod_exp_recp"}, + {ERR_FUNC(BN_F_BN_MOD_EXP_SIMPLE), "BN_mod_exp_simple"}, + {ERR_FUNC(BN_F_BN_MOD_INVERSE), "BN_mod_inverse"}, + {ERR_FUNC(BN_F_BN_MOD_INVERSE_NO_BRANCH), "BN_mod_inverse_no_branch"}, + {ERR_FUNC(BN_F_BN_MOD_LSHIFT_QUICK), "BN_mod_lshift_quick"}, + {ERR_FUNC(BN_F_BN_MOD_MUL_RECIPROCAL), "BN_mod_mul_reciprocal"}, + {ERR_FUNC(BN_F_BN_MOD_SQRT), "BN_mod_sqrt"}, + {ERR_FUNC(BN_F_BN_MPI2BN), "BN_mpi2bn"}, + {ERR_FUNC(BN_F_BN_NEW), "BN_new"}, + {ERR_FUNC(BN_F_BN_RAND), "BN_rand"}, + {ERR_FUNC(BN_F_BN_RAND_RANGE), "BN_rand_range"}, + {ERR_FUNC(BN_F_BN_USUB), "BN_usub"}, + {0, NULL} +}; -static ERR_STRING_DATA BN_str_reasons[]= - { -{ERR_REASON(BN_R_ARG2_LT_ARG3) ,"arg2 lt arg3"}, -{ERR_REASON(BN_R_BAD_RECIPROCAL) ,"bad reciprocal"}, -{ERR_REASON(BN_R_BIGNUM_TOO_LONG) ,"bignum too long"}, -{ERR_REASON(BN_R_CALLED_WITH_EVEN_MODULUS),"called with even modulus"}, -{ERR_REASON(BN_R_DIV_BY_ZERO) ,"div by zero"}, -{ERR_REASON(BN_R_ENCODING_ERROR) ,"encoding error"}, -{ERR_REASON(BN_R_EXPAND_ON_STATIC_BIGNUM_DATA),"expand on static bignum data"}, -{ERR_REASON(BN_R_INPUT_NOT_REDUCED) ,"input not reduced"}, -{ERR_REASON(BN_R_INVALID_LENGTH) ,"invalid length"}, -{ERR_REASON(BN_R_INVALID_RANGE) ,"invalid range"}, -{ERR_REASON(BN_R_NOT_A_SQUARE) ,"not a square"}, -{ERR_REASON(BN_R_NOT_INITIALIZED) ,"not initialized"}, -{ERR_REASON(BN_R_NO_INVERSE) ,"no inverse"}, -{ERR_REASON(BN_R_NO_SOLUTION) ,"no solution"}, -{ERR_REASON(BN_R_P_IS_NOT_PRIME) ,"p is not prime"}, -{ERR_REASON(BN_R_TOO_MANY_ITERATIONS) ,"too many iterations"}, -{ERR_REASON(BN_R_TOO_MANY_TEMPORARY_VARIABLES),"too many temporary variables"}, -{0,NULL} - }; +static ERR_STRING_DATA BN_str_reasons[] = { + {ERR_REASON(BN_R_ARG2_LT_ARG3), "arg2 lt arg3"}, + {ERR_REASON(BN_R_BAD_RECIPROCAL), "bad reciprocal"}, + {ERR_REASON(BN_R_BIGNUM_TOO_LONG), "bignum too long"}, + {ERR_REASON(BN_R_CALLED_WITH_EVEN_MODULUS), "called with even modulus"}, + {ERR_REASON(BN_R_DIV_BY_ZERO), "div by zero"}, + {ERR_REASON(BN_R_ENCODING_ERROR), "encoding error"}, + {ERR_REASON(BN_R_EXPAND_ON_STATIC_BIGNUM_DATA), + "expand on static bignum data"}, + {ERR_REASON(BN_R_INPUT_NOT_REDUCED), "input not reduced"}, + {ERR_REASON(BN_R_INVALID_LENGTH), "invalid length"}, + {ERR_REASON(BN_R_INVALID_RANGE), "invalid range"}, + {ERR_REASON(BN_R_NOT_A_SQUARE), "not a square"}, + {ERR_REASON(BN_R_NOT_INITIALIZED), "not initialized"}, + {ERR_REASON(BN_R_NO_INVERSE), "no inverse"}, + {ERR_REASON(BN_R_NO_SOLUTION), "no solution"}, + {ERR_REASON(BN_R_P_IS_NOT_PRIME), "p is not prime"}, + {ERR_REASON(BN_R_TOO_MANY_ITERATIONS), "too many iterations"}, + {ERR_REASON(BN_R_TOO_MANY_TEMPORARY_VARIABLES), + "too many temporary variables"}, + {0, NULL} +}; #endif void ERR_load_BN_strings(void) - { +{ #ifndef OPENSSL_NO_ERR - if (ERR_func_error_string(BN_str_functs[0].error) == NULL) - { - ERR_load_strings(0,BN_str_functs); - ERR_load_strings(0,BN_str_reasons); - } + if (ERR_func_error_string(BN_str_functs[0].error) == NULL) { + ERR_load_strings(0, BN_str_functs); + ERR_load_strings(0, BN_str_reasons); + } #endif - } +} diff --git a/openssl/crypto/bn/bn_exp.c b/openssl/crypto/bn/bn_exp.c index 611fa3262..28a9fd53b 100644 --- a/openssl/crypto/bn/bn_exp.c +++ b/openssl/crypto/bn/bn_exp.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -109,7 +109,6 @@ * */ - #include "cryptlib.h" #include "bn_lcl.h" @@ -123,982 +122,1279 @@ # ifndef alloca # define alloca(s) __builtin_alloca((s)) # endif +#elif defined(__sun) +# include <alloca.h> +#endif + +#undef RSAZ_ENABLED +#if defined(OPENSSL_BN_ASM_MONT) && \ + (defined(__x86_64) || defined(__x86_64__) || \ + defined(_M_AMD64) || defined(_M_X64)) +# include "rsaz_exp.h" +# define RSAZ_ENABLED +#endif + +#undef SPARC_T4_MONT +#if defined(OPENSSL_BN_ASM_MONT) && (defined(__sparc__) || defined(__sparc)) +# include "sparc_arch.h" +extern unsigned int OPENSSL_sparcv9cap_P[]; +# define SPARC_T4_MONT #endif /* maximum precomputation table size for *variable* sliding windows */ -#define TABLE_SIZE 32 +#define TABLE_SIZE 32 /* this one works - simple but works */ int BN_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) - { - int i,bits,ret=0; - BIGNUM *v,*rr; - - if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) - { - /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ - BNerr(BN_F_BN_EXP,ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); - return -1; - } - - BN_CTX_start(ctx); - if ((r == a) || (r == p)) - rr = BN_CTX_get(ctx); - else - rr = r; - v = BN_CTX_get(ctx); - if (rr == NULL || v == NULL) goto err; - - if (BN_copy(v,a) == NULL) goto err; - bits=BN_num_bits(p); - - if (BN_is_odd(p)) - { if (BN_copy(rr,a) == NULL) goto err; } - else { if (!BN_one(rr)) goto err; } - - for (i=1; i<bits; i++) - { - if (!BN_sqr(v,v,ctx)) goto err; - if (BN_is_bit_set(p,i)) - { - if (!BN_mul(rr,rr,v,ctx)) goto err; - } - } - ret=1; -err: - if (r != rr) BN_copy(r,rr); - BN_CTX_end(ctx); - bn_check_top(r); - return(ret); - } - +{ + int i, bits, ret = 0; + BIGNUM *v, *rr; + + if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) { + /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ + BNerr(BN_F_BN_EXP, ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); + return -1; + } + + BN_CTX_start(ctx); + if ((r == a) || (r == p)) + rr = BN_CTX_get(ctx); + else + rr = r; + v = BN_CTX_get(ctx); + if (rr == NULL || v == NULL) + goto err; + + if (BN_copy(v, a) == NULL) + goto err; + bits = BN_num_bits(p); + + if (BN_is_odd(p)) { + if (BN_copy(rr, a) == NULL) + goto err; + } else { + if (!BN_one(rr)) + goto err; + } + + for (i = 1; i < bits; i++) { + if (!BN_sqr(v, v, ctx)) + goto err; + if (BN_is_bit_set(p, i)) { + if (!BN_mul(rr, rr, v, ctx)) + goto err; + } + } + ret = 1; + err: + if (r != rr) + BN_copy(r, rr); + BN_CTX_end(ctx); + bn_check_top(r); + return (ret); +} int BN_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, const BIGNUM *m, - BN_CTX *ctx) - { - int ret; - - bn_check_top(a); - bn_check_top(p); - bn_check_top(m); - - /* For even modulus m = 2^k*m_odd, it might make sense to compute - * a^p mod m_odd and a^p mod 2^k separately (with Montgomery - * exponentiation for the odd part), using appropriate exponent - * reductions, and combine the results using the CRT. - * - * For now, we use Montgomery only if the modulus is odd; otherwise, - * exponentiation using the reciprocal-based quick remaindering - * algorithm is used. - * - * (Timing obtained with expspeed.c [computations a^p mod m - * where a, p, m are of the same length: 256, 512, 1024, 2048, - * 4096, 8192 bits], compared to the running time of the - * standard algorithm: - * - * BN_mod_exp_mont 33 .. 40 % [AMD K6-2, Linux, debug configuration] - * 55 .. 77 % [UltraSparc processor, but - * debug-solaris-sparcv8-gcc conf.] - * - * BN_mod_exp_recp 50 .. 70 % [AMD K6-2, Linux, debug configuration] - * 62 .. 118 % [UltraSparc, debug-solaris-sparcv8-gcc] - * - * On the Sparc, BN_mod_exp_recp was faster than BN_mod_exp_mont - * at 2048 and more bits, but at 512 and 1024 bits, it was - * slower even than the standard algorithm! - * - * "Real" timings [linux-elf, solaris-sparcv9-gcc configurations] - * should be obtained when the new Montgomery reduction code - * has been integrated into OpenSSL.) - */ + BN_CTX *ctx) +{ + int ret; + + bn_check_top(a); + bn_check_top(p); + bn_check_top(m); + + /*- + * For even modulus m = 2^k*m_odd, it might make sense to compute + * a^p mod m_odd and a^p mod 2^k separately (with Montgomery + * exponentiation for the odd part), using appropriate exponent + * reductions, and combine the results using the CRT. + * + * For now, we use Montgomery only if the modulus is odd; otherwise, + * exponentiation using the reciprocal-based quick remaindering + * algorithm is used. + * + * (Timing obtained with expspeed.c [computations a^p mod m + * where a, p, m are of the same length: 256, 512, 1024, 2048, + * 4096, 8192 bits], compared to the running time of the + * standard algorithm: + * + * BN_mod_exp_mont 33 .. 40 % [AMD K6-2, Linux, debug configuration] + * 55 .. 77 % [UltraSparc processor, but + * debug-solaris-sparcv8-gcc conf.] + * + * BN_mod_exp_recp 50 .. 70 % [AMD K6-2, Linux, debug configuration] + * 62 .. 118 % [UltraSparc, debug-solaris-sparcv8-gcc] + * + * On the Sparc, BN_mod_exp_recp was faster than BN_mod_exp_mont + * at 2048 and more bits, but at 512 and 1024 bits, it was + * slower even than the standard algorithm! + * + * "Real" timings [linux-elf, solaris-sparcv9-gcc configurations] + * should be obtained when the new Montgomery reduction code + * has been integrated into OpenSSL.) + */ #define MONT_MUL_MOD #define MONT_EXP_WORD #define RECP_MUL_MOD #ifdef MONT_MUL_MOD - /* I have finally been able to take out this pre-condition of - * the top bit being set. It was caused by an error in BN_div - * with negatives. There was also another problem when for a^b%m - * a >= m. eay 07-May-97 */ -/* if ((m->d[m->top-1]&BN_TBIT) && BN_is_odd(m)) */ - - if (BN_is_odd(m)) - { -# ifdef MONT_EXP_WORD - if (a->top == 1 && !a->neg && (BN_get_flags(p, BN_FLG_CONSTTIME) == 0)) - { - BN_ULONG A = a->d[0]; - ret=BN_mod_exp_mont_word(r,A,p,m,ctx,NULL); - } - else -# endif - ret=BN_mod_exp_mont(r,a,p,m,ctx,NULL); - } - else + /* + * I have finally been able to take out this pre-condition of the top bit + * being set. It was caused by an error in BN_div with negatives. There + * was also another problem when for a^b%m a >= m. eay 07-May-97 + */ + /* if ((m->d[m->top-1]&BN_TBIT) && BN_is_odd(m)) */ + + if (BN_is_odd(m)) { +# ifdef MONT_EXP_WORD + if (a->top == 1 && !a->neg + && (BN_get_flags(p, BN_FLG_CONSTTIME) == 0)) { + BN_ULONG A = a->d[0]; + ret = BN_mod_exp_mont_word(r, A, p, m, ctx, NULL); + } else +# endif + ret = BN_mod_exp_mont(r, a, p, m, ctx, NULL); + } else #endif #ifdef RECP_MUL_MOD - { ret=BN_mod_exp_recp(r,a,p,m,ctx); } + { + ret = BN_mod_exp_recp(r, a, p, m, ctx); + } #else - { ret=BN_mod_exp_simple(r,a,p,m,ctx); } + { + ret = BN_mod_exp_simple(r, a, p, m, ctx); + } #endif - bn_check_top(r); - return(ret); - } - + bn_check_top(r); + return (ret); +} int BN_mod_exp_recp(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx) - { - int i,j,bits,ret=0,wstart,wend,window,wvalue; - int start=1; - BIGNUM *aa; - /* Table of variables obtained from 'ctx' */ - BIGNUM *val[TABLE_SIZE]; - BN_RECP_CTX recp; - - if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) - { - /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ - BNerr(BN_F_BN_MOD_EXP_RECP,ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); - return -1; - } - - bits=BN_num_bits(p); - - if (bits == 0) - { - ret = BN_one(r); - return ret; - } - - BN_CTX_start(ctx); - aa = BN_CTX_get(ctx); - val[0] = BN_CTX_get(ctx); - if(!aa || !val[0]) goto err; - - BN_RECP_CTX_init(&recp); - if (m->neg) - { - /* ignore sign of 'm' */ - if (!BN_copy(aa, m)) goto err; - aa->neg = 0; - if (BN_RECP_CTX_set(&recp,aa,ctx) <= 0) goto err; - } - else - { - if (BN_RECP_CTX_set(&recp,m,ctx) <= 0) goto err; - } - - if (!BN_nnmod(val[0],a,m,ctx)) goto err; /* 1 */ - if (BN_is_zero(val[0])) - { - BN_zero(r); - ret = 1; - goto err; - } - - window = BN_window_bits_for_exponent_size(bits); - if (window > 1) - { - if (!BN_mod_mul_reciprocal(aa,val[0],val[0],&recp,ctx)) - goto err; /* 2 */ - j=1<<(window-1); - for (i=1; i<j; i++) - { - if(((val[i] = BN_CTX_get(ctx)) == NULL) || - !BN_mod_mul_reciprocal(val[i],val[i-1], - aa,&recp,ctx)) - goto err; - } - } - - start=1; /* This is used to avoid multiplication etc - * when there is only the value '1' in the - * buffer. */ - wvalue=0; /* The 'value' of the window */ - wstart=bits-1; /* The top bit of the window */ - wend=0; /* The bottom bit of the window */ - - if (!BN_one(r)) goto err; - - for (;;) - { - if (BN_is_bit_set(p,wstart) == 0) - { - if (!start) - if (!BN_mod_mul_reciprocal(r,r,r,&recp,ctx)) - goto err; - if (wstart == 0) break; - wstart--; - continue; - } - /* We now have wstart on a 'set' bit, we now need to work out - * how bit a window to do. To do this we need to scan - * forward until the last set bit before the end of the - * window */ - j=wstart; - wvalue=1; - wend=0; - for (i=1; i<window; i++) - { - if (wstart-i < 0) break; - if (BN_is_bit_set(p,wstart-i)) - { - wvalue<<=(i-wend); - wvalue|=1; - wend=i; - } - } - - /* wend is the size of the current window */ - j=wend+1; - /* add the 'bytes above' */ - if (!start) - for (i=0; i<j; i++) - { - if (!BN_mod_mul_reciprocal(r,r,r,&recp,ctx)) - goto err; - } - - /* wvalue will be an odd number < 2^window */ - if (!BN_mod_mul_reciprocal(r,r,val[wvalue>>1],&recp,ctx)) - goto err; - - /* move the 'window' down further */ - wstart-=wend+1; - wvalue=0; - start=0; - if (wstart < 0) break; - } - ret=1; -err: - BN_CTX_end(ctx); - BN_RECP_CTX_free(&recp); - bn_check_top(r); - return(ret); - } - + const BIGNUM *m, BN_CTX *ctx) +{ + int i, j, bits, ret = 0, wstart, wend, window, wvalue; + int start = 1; + BIGNUM *aa; + /* Table of variables obtained from 'ctx' */ + BIGNUM *val[TABLE_SIZE]; + BN_RECP_CTX recp; + + if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) { + /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ + BNerr(BN_F_BN_MOD_EXP_RECP, ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); + return -1; + } + + bits = BN_num_bits(p); + + if (bits == 0) { + ret = BN_one(r); + return ret; + } + + BN_CTX_start(ctx); + aa = BN_CTX_get(ctx); + val[0] = BN_CTX_get(ctx); + if (!aa || !val[0]) + goto err; + + BN_RECP_CTX_init(&recp); + if (m->neg) { + /* ignore sign of 'm' */ + if (!BN_copy(aa, m)) + goto err; + aa->neg = 0; + if (BN_RECP_CTX_set(&recp, aa, ctx) <= 0) + goto err; + } else { + if (BN_RECP_CTX_set(&recp, m, ctx) <= 0) + goto err; + } + + if (!BN_nnmod(val[0], a, m, ctx)) + goto err; /* 1 */ + if (BN_is_zero(val[0])) { + BN_zero(r); + ret = 1; + goto err; + } + + window = BN_window_bits_for_exponent_size(bits); + if (window > 1) { + if (!BN_mod_mul_reciprocal(aa, val[0], val[0], &recp, ctx)) + goto err; /* 2 */ + j = 1 << (window - 1); + for (i = 1; i < j; i++) { + if (((val[i] = BN_CTX_get(ctx)) == NULL) || + !BN_mod_mul_reciprocal(val[i], val[i - 1], aa, &recp, ctx)) + goto err; + } + } + + start = 1; /* This is used to avoid multiplication etc + * when there is only the value '1' in the + * buffer. */ + wvalue = 0; /* The 'value' of the window */ + wstart = bits - 1; /* The top bit of the window */ + wend = 0; /* The bottom bit of the window */ + + if (!BN_one(r)) + goto err; + + for (;;) { + if (BN_is_bit_set(p, wstart) == 0) { + if (!start) + if (!BN_mod_mul_reciprocal(r, r, r, &recp, ctx)) + goto err; + if (wstart == 0) + break; + wstart--; + continue; + } + /* + * We now have wstart on a 'set' bit, we now need to work out how bit + * a window to do. To do this we need to scan forward until the last + * set bit before the end of the window + */ + j = wstart; + wvalue = 1; + wend = 0; + for (i = 1; i < window; i++) { + if (wstart - i < 0) + break; + if (BN_is_bit_set(p, wstart - i)) { + wvalue <<= (i - wend); + wvalue |= 1; + wend = i; + } + } + + /* wend is the size of the current window */ + j = wend + 1; + /* add the 'bytes above' */ + if (!start) + for (i = 0; i < j; i++) { + if (!BN_mod_mul_reciprocal(r, r, r, &recp, ctx)) + goto err; + } + + /* wvalue will be an odd number < 2^window */ + if (!BN_mod_mul_reciprocal(r, r, val[wvalue >> 1], &recp, ctx)) + goto err; + + /* move the 'window' down further */ + wstart -= wend + 1; + wvalue = 0; + start = 0; + if (wstart < 0) + break; + } + ret = 1; + err: + BN_CTX_end(ctx); + BN_RECP_CTX_free(&recp); + bn_check_top(r); + return (ret); +} int BN_mod_exp_mont(BIGNUM *rr, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont) - { - int i,j,bits,ret=0,wstart,wend,window,wvalue; - int start=1; - BIGNUM *d,*r; - const BIGNUM *aa; - /* Table of variables obtained from 'ctx' */ - BIGNUM *val[TABLE_SIZE]; - BN_MONT_CTX *mont=NULL; - - if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) - { - return BN_mod_exp_mont_consttime(rr, a, p, m, ctx, in_mont); - } - - bn_check_top(a); - bn_check_top(p); - bn_check_top(m); - - if (!BN_is_odd(m)) - { - BNerr(BN_F_BN_MOD_EXP_MONT,BN_R_CALLED_WITH_EVEN_MODULUS); - return(0); - } - bits=BN_num_bits(p); - if (bits == 0) - { - ret = BN_one(rr); - return ret; - } - - BN_CTX_start(ctx); - d = BN_CTX_get(ctx); - r = BN_CTX_get(ctx); - val[0] = BN_CTX_get(ctx); - if (!d || !r || !val[0]) goto err; - - /* If this is not done, things will break in the montgomery - * part */ - - if (in_mont != NULL) - mont=in_mont; - else - { - if ((mont=BN_MONT_CTX_new()) == NULL) goto err; - if (!BN_MONT_CTX_set(mont,m,ctx)) goto err; - } - - if (a->neg || BN_ucmp(a,m) >= 0) - { - if (!BN_nnmod(val[0],a,m,ctx)) - goto err; - aa= val[0]; - } - else - aa=a; - if (BN_is_zero(aa)) - { - BN_zero(rr); - ret = 1; - goto err; - } - if (!BN_to_montgomery(val[0],aa,mont,ctx)) goto err; /* 1 */ - - window = BN_window_bits_for_exponent_size(bits); - if (window > 1) - { - if (!BN_mod_mul_montgomery(d,val[0],val[0],mont,ctx)) goto err; /* 2 */ - j=1<<(window-1); - for (i=1; i<j; i++) - { - if(((val[i] = BN_CTX_get(ctx)) == NULL) || - !BN_mod_mul_montgomery(val[i],val[i-1], - d,mont,ctx)) - goto err; - } - } - - start=1; /* This is used to avoid multiplication etc - * when there is only the value '1' in the - * buffer. */ - wvalue=0; /* The 'value' of the window */ - wstart=bits-1; /* The top bit of the window */ - wend=0; /* The bottom bit of the window */ - - if (!BN_to_montgomery(r,BN_value_one(),mont,ctx)) goto err; - for (;;) - { - if (BN_is_bit_set(p,wstart) == 0) - { - if (!start) - { - if (!BN_mod_mul_montgomery(r,r,r,mont,ctx)) - goto err; - } - if (wstart == 0) break; - wstart--; - continue; - } - /* We now have wstart on a 'set' bit, we now need to work out - * how bit a window to do. To do this we need to scan - * forward until the last set bit before the end of the - * window */ - j=wstart; - wvalue=1; - wend=0; - for (i=1; i<window; i++) - { - if (wstart-i < 0) break; - if (BN_is_bit_set(p,wstart-i)) - { - wvalue<<=(i-wend); - wvalue|=1; - wend=i; - } - } - - /* wend is the size of the current window */ - j=wend+1; - /* add the 'bytes above' */ - if (!start) - for (i=0; i<j; i++) - { - if (!BN_mod_mul_montgomery(r,r,r,mont,ctx)) - goto err; - } - - /* wvalue will be an odd number < 2^window */ - if (!BN_mod_mul_montgomery(r,r,val[wvalue>>1],mont,ctx)) - goto err; - - /* move the 'window' down further */ - wstart-=wend+1; - wvalue=0; - start=0; - if (wstart < 0) break; - } - if (!BN_from_montgomery(rr,r,mont,ctx)) goto err; - ret=1; -err: - if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont); - BN_CTX_end(ctx); - bn_check_top(rr); - return(ret); - } - - -/* BN_mod_exp_mont_consttime() stores the precomputed powers in a specific layout - * so that accessing any of these table values shows the same access pattern as far - * as cache lines are concerned. The following functions are used to transfer a BIGNUM - * from/to that table. */ - -static int MOD_EXP_CTIME_COPY_TO_PREBUF(const BIGNUM *b, int top, unsigned char *buf, int idx, int width) - { - size_t i, j; - - if (top > b->top) - top = b->top; /* this works because 'buf' is explicitly zeroed */ - for (i = 0, j=idx; i < top * sizeof b->d[0]; i++, j+=width) - { - buf[j] = ((unsigned char*)b->d)[i]; - } - - return 1; - } - -static int MOD_EXP_CTIME_COPY_FROM_PREBUF(BIGNUM *b, int top, unsigned char *buf, int idx, int width) - { - size_t i, j; - - if (bn_wexpand(b, top) == NULL) - return 0; - - for (i=0, j=idx; i < top * sizeof b->d[0]; i++, j+=width) - { - ((unsigned char*)b->d)[i] = buf[j]; - } - - b->top = top; - bn_correct_top(b); - return 1; - } - -/* Given a pointer value, compute the next address that is a cache line multiple. */ -#define MOD_EXP_CTIME_ALIGN(x_) \ - ((unsigned char*)(x_) + (MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH - (((size_t)(x_)) & (MOD_EXP_CTIME_MIN_CACHE_LINE_MASK)))) + const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont) +{ + int i, j, bits, ret = 0, wstart, wend, window, wvalue; + int start = 1; + BIGNUM *d, *r; + const BIGNUM *aa; + /* Table of variables obtained from 'ctx' */ + BIGNUM *val[TABLE_SIZE]; + BN_MONT_CTX *mont = NULL; + + if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) { + return BN_mod_exp_mont_consttime(rr, a, p, m, ctx, in_mont); + } + + bn_check_top(a); + bn_check_top(p); + bn_check_top(m); + + if (!BN_is_odd(m)) { + BNerr(BN_F_BN_MOD_EXP_MONT, BN_R_CALLED_WITH_EVEN_MODULUS); + return (0); + } + bits = BN_num_bits(p); + if (bits == 0) { + ret = BN_one(rr); + return ret; + } + + BN_CTX_start(ctx); + d = BN_CTX_get(ctx); + r = BN_CTX_get(ctx); + val[0] = BN_CTX_get(ctx); + if (!d || !r || !val[0]) + goto err; + + /* + * If this is not done, things will break in the montgomery part + */ + + if (in_mont != NULL) + mont = in_mont; + else { + if ((mont = BN_MONT_CTX_new()) == NULL) + goto err; + if (!BN_MONT_CTX_set(mont, m, ctx)) + goto err; + } + + if (a->neg || BN_ucmp(a, m) >= 0) { + if (!BN_nnmod(val[0], a, m, ctx)) + goto err; + aa = val[0]; + } else + aa = a; + if (BN_is_zero(aa)) { + BN_zero(rr); + ret = 1; + goto err; + } + if (!BN_to_montgomery(val[0], aa, mont, ctx)) + goto err; /* 1 */ + + window = BN_window_bits_for_exponent_size(bits); + if (window > 1) { + if (!BN_mod_mul_montgomery(d, val[0], val[0], mont, ctx)) + goto err; /* 2 */ + j = 1 << (window - 1); + for (i = 1; i < j; i++) { + if (((val[i] = BN_CTX_get(ctx)) == NULL) || + !BN_mod_mul_montgomery(val[i], val[i - 1], d, mont, ctx)) + goto err; + } + } + + start = 1; /* This is used to avoid multiplication etc + * when there is only the value '1' in the + * buffer. */ + wvalue = 0; /* The 'value' of the window */ + wstart = bits - 1; /* The top bit of the window */ + wend = 0; /* The bottom bit of the window */ + +#if 1 /* by Shay Gueron's suggestion */ + j = m->top; /* borrow j */ + if (m->d[j - 1] & (((BN_ULONG)1) << (BN_BITS2 - 1))) { + if (bn_wexpand(r, j) == NULL) + goto err; + /* 2^(top*BN_BITS2) - m */ + r->d[0] = (0 - m->d[0]) & BN_MASK2; + for (i = 1; i < j; i++) + r->d[i] = (~m->d[i]) & BN_MASK2; + r->top = j; + /* + * Upper words will be zero if the corresponding words of 'm' were + * 0xfff[...], so decrement r->top accordingly. + */ + bn_correct_top(r); + } else +#endif + if (!BN_to_montgomery(r, BN_value_one(), mont, ctx)) + goto err; + for (;;) { + if (BN_is_bit_set(p, wstart) == 0) { + if (!start) { + if (!BN_mod_mul_montgomery(r, r, r, mont, ctx)) + goto err; + } + if (wstart == 0) + break; + wstart--; + continue; + } + /* + * We now have wstart on a 'set' bit, we now need to work out how bit + * a window to do. To do this we need to scan forward until the last + * set bit before the end of the window + */ + j = wstart; + wvalue = 1; + wend = 0; + for (i = 1; i < window; i++) { + if (wstart - i < 0) + break; + if (BN_is_bit_set(p, wstart - i)) { + wvalue <<= (i - wend); + wvalue |= 1; + wend = i; + } + } + + /* wend is the size of the current window */ + j = wend + 1; + /* add the 'bytes above' */ + if (!start) + for (i = 0; i < j; i++) { + if (!BN_mod_mul_montgomery(r, r, r, mont, ctx)) + goto err; + } + + /* wvalue will be an odd number < 2^window */ + if (!BN_mod_mul_montgomery(r, r, val[wvalue >> 1], mont, ctx)) + goto err; + + /* move the 'window' down further */ + wstart -= wend + 1; + wvalue = 0; + start = 0; + if (wstart < 0) + break; + } +#if defined(SPARC_T4_MONT) + if (OPENSSL_sparcv9cap_P[0] & (SPARCV9_VIS3 | SPARCV9_PREFER_FPU)) { + j = mont->N.top; /* borrow j */ + val[0]->d[0] = 1; /* borrow val[0] */ + for (i = 1; i < j; i++) + val[0]->d[i] = 0; + val[0]->top = j; + if (!BN_mod_mul_montgomery(rr, r, val[0], mont, ctx)) + goto err; + } else +#endif + if (!BN_from_montgomery(rr, r, mont, ctx)) + goto err; + ret = 1; + err: + if ((in_mont == NULL) && (mont != NULL)) + BN_MONT_CTX_free(mont); + BN_CTX_end(ctx); + bn_check_top(rr); + return (ret); +} + +#if defined(SPARC_T4_MONT) +static BN_ULONG bn_get_bits(const BIGNUM *a, int bitpos) +{ + BN_ULONG ret = 0; + int wordpos; + + wordpos = bitpos / BN_BITS2; + bitpos %= BN_BITS2; + if (wordpos >= 0 && wordpos < a->top) { + ret = a->d[wordpos] & BN_MASK2; + if (bitpos) { + ret >>= bitpos; + if (++wordpos < a->top) + ret |= a->d[wordpos] << (BN_BITS2 - bitpos); + } + } + + return ret & BN_MASK2; +} +#endif -/* This variant of BN_mod_exp_mont() uses fixed windows and the special - * precomputation memory layout to limit data-dependency to a minimum - * to protect secret exponents (cf. the hyper-threading timing attacks - * pointed out by Colin Percival, - * http://www.daemonology.net/hyperthreading-considered-harmful/) +/* + * BN_mod_exp_mont_consttime() stores the precomputed powers in a specific + * layout so that accessing any of these table values shows the same access + * pattern as far as cache lines are concerned. The following functions are + * used to transfer a BIGNUM from/to that table. + */ + +static int MOD_EXP_CTIME_COPY_TO_PREBUF(const BIGNUM *b, int top, + unsigned char *buf, int idx, + int width) +{ + size_t i, j; + + if (top > b->top) + top = b->top; /* this works because 'buf' is explicitly + * zeroed */ + for (i = 0, j = idx; i < top * sizeof b->d[0]; i++, j += width) { + buf[j] = ((unsigned char *)b->d)[i]; + } + + return 1; +} + +static int MOD_EXP_CTIME_COPY_FROM_PREBUF(BIGNUM *b, int top, + unsigned char *buf, int idx, + int width) +{ + size_t i, j; + + if (bn_wexpand(b, top) == NULL) + return 0; + + for (i = 0, j = idx; i < top * sizeof b->d[0]; i++, j += width) { + ((unsigned char *)b->d)[i] = buf[j]; + } + + b->top = top; + bn_correct_top(b); + return 1; +} + +/* + * Given a pointer value, compute the next address that is a cache line + * multiple. + */ +#define MOD_EXP_CTIME_ALIGN(x_) \ + ((unsigned char*)(x_) + (MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH - (((size_t)(x_)) & (MOD_EXP_CTIME_MIN_CACHE_LINE_MASK)))) + +/* + * This variant of BN_mod_exp_mont() uses fixed windows and the special + * precomputation memory layout to limit data-dependency to a minimum to + * protect secret exponents (cf. the hyper-threading timing attacks pointed + * out by Colin Percival, + * http://www.daemong-consideredperthreading-considered-harmful/) */ int BN_mod_exp_mont_consttime(BIGNUM *rr, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont) - { - int i,bits,ret=0,window,wvalue; - int top; - BN_MONT_CTX *mont=NULL; - - int numPowers; - unsigned char *powerbufFree=NULL; - int powerbufLen = 0; - unsigned char *powerbuf=NULL; - BIGNUM tmp, am; - - bn_check_top(a); - bn_check_top(p); - bn_check_top(m); - - top = m->top; - - if (!(m->d[0] & 1)) - { - BNerr(BN_F_BN_MOD_EXP_MONT_CONSTTIME,BN_R_CALLED_WITH_EVEN_MODULUS); - return(0); - } - bits=BN_num_bits(p); - if (bits == 0) - { - ret = BN_one(rr); - return ret; - } - - BN_CTX_start(ctx); - - /* Allocate a montgomery context if it was not supplied by the caller. - * If this is not done, things will break in the montgomery part. - */ - if (in_mont != NULL) - mont=in_mont; - else - { - if ((mont=BN_MONT_CTX_new()) == NULL) goto err; - if (!BN_MONT_CTX_set(mont,m,ctx)) goto err; - } - - /* Get the window size to use with size of p. */ - window = BN_window_bits_for_ctime_exponent_size(bits); -#if defined(OPENSSL_BN_ASM_MONT5) - if (window==6 && bits<=1024) window=5; /* ~5% improvement of 2048-bit RSA sign */ + const BIGNUM *m, BN_CTX *ctx, + BN_MONT_CTX *in_mont) +{ + int i, bits, ret = 0, window, wvalue; + int top; + BN_MONT_CTX *mont = NULL; + + int numPowers; + unsigned char *powerbufFree = NULL; + int powerbufLen = 0; + unsigned char *powerbuf = NULL; + BIGNUM tmp, am; +#if defined(SPARC_T4_MONT) + unsigned int t4 = 0; #endif - /* Allocate a buffer large enough to hold all of the pre-computed - * powers of am, am itself and tmp. - */ - numPowers = 1 << window; - powerbufLen = sizeof(m->d[0])*(top*numPowers + - ((2*top)>numPowers?(2*top):numPowers)); -#ifdef alloca - if (powerbufLen < 3072) - powerbufFree = alloca(powerbufLen+MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH); - else + bn_check_top(a); + bn_check_top(p); + bn_check_top(m); + + top = m->top; + + if (!(m->d[0] & 1)) { + BNerr(BN_F_BN_MOD_EXP_MONT_CONSTTIME, BN_R_CALLED_WITH_EVEN_MODULUS); + return (0); + } + bits = BN_num_bits(p); + if (bits == 0) { + ret = BN_one(rr); + return ret; + } + + BN_CTX_start(ctx); + + /* + * Allocate a montgomery context if it was not supplied by the caller. If + * this is not done, things will break in the montgomery part. + */ + if (in_mont != NULL) + mont = in_mont; + else { + if ((mont = BN_MONT_CTX_new()) == NULL) + goto err; + if (!BN_MONT_CTX_set(mont, m, ctx)) + goto err; + } + +#ifdef RSAZ_ENABLED + /* + * If the size of the operands allow it, perform the optimized + * RSAZ exponentiation. For further information see + * crypto/bn/rsaz_exp.c and accompanying assembly modules. + */ + if ((16 == a->top) && (16 == p->top) && (BN_num_bits(m) == 1024) + && rsaz_avx2_eligible()) { + if (NULL == bn_wexpand(rr, 16)) + goto err; + RSAZ_1024_mod_exp_avx2(rr->d, a->d, p->d, m->d, mont->RR.d, + mont->n0[0]); + rr->top = 16; + rr->neg = 0; + bn_correct_top(rr); + ret = 1; + goto err; + } else if ((8 == a->top) && (8 == p->top) && (BN_num_bits(m) == 512)) { + if (NULL == bn_wexpand(rr, 8)) + goto err; + RSAZ_512_mod_exp(rr->d, a->d, p->d, m->d, mont->n0[0], mont->RR.d); + rr->top = 8; + rr->neg = 0; + bn_correct_top(rr); + ret = 1; + goto err; + } #endif - if ((powerbufFree=(unsigned char*)OPENSSL_malloc(powerbufLen+MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH)) == NULL) - goto err; - - powerbuf = MOD_EXP_CTIME_ALIGN(powerbufFree); - memset(powerbuf, 0, powerbufLen); + /* Get the window size to use with size of p. */ + window = BN_window_bits_for_ctime_exponent_size(bits); +#if defined(SPARC_T4_MONT) + if (window >= 5 && (top & 15) == 0 && top <= 64 && + (OPENSSL_sparcv9cap_P[1] & (CFR_MONTMUL | CFR_MONTSQR)) == + (CFR_MONTMUL | CFR_MONTSQR) && (t4 = OPENSSL_sparcv9cap_P[0])) + window = 5; + else +#endif +#if defined(OPENSSL_BN_ASM_MONT5) + if (window >= 5) { + window = 5; /* ~5% improvement for RSA2048 sign, and even + * for RSA4096 */ + if ((top & 7) == 0) + powerbufLen += 2 * top * sizeof(m->d[0]); + } +#endif + (void)0; + + /* + * Allocate a buffer large enough to hold all of the pre-computed powers + * of am, am itself and tmp. + */ + numPowers = 1 << window; + powerbufLen += sizeof(m->d[0]) * (top * numPowers + + ((2 * top) > + numPowers ? (2 * top) : numPowers)); #ifdef alloca - if (powerbufLen < 3072) - powerbufFree = NULL; + if (powerbufLen < 3072) + powerbufFree = + alloca(powerbufLen + MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH); + else #endif + if ((powerbufFree = + (unsigned char *)OPENSSL_malloc(powerbufLen + + MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH)) + == NULL) + goto err; - /* lay down tmp and am right after powers table */ - tmp.d = (BN_ULONG *)(powerbuf + sizeof(m->d[0])*top*numPowers); - am.d = tmp.d + top; - tmp.top = am.top = 0; - tmp.dmax = am.dmax = top; - tmp.neg = am.neg = 0; - tmp.flags = am.flags = BN_FLG_STATIC_DATA; - - /* prepare a^0 in Montgomery domain */ -#if 1 - if (!BN_to_montgomery(&tmp,BN_value_one(),mont,ctx)) goto err; -#else - tmp.d[0] = (0-m->d[0])&BN_MASK2; /* 2^(top*BN_BITS2) - m */ - for (i=1;i<top;i++) - tmp.d[i] = (~m->d[i])&BN_MASK2; - tmp.top = top; -#endif + powerbuf = MOD_EXP_CTIME_ALIGN(powerbufFree); + memset(powerbuf, 0, powerbufLen); - /* prepare a^1 in Montgomery domain */ - if (a->neg || BN_ucmp(a,m) >= 0) - { - if (!BN_mod(&am,a,m,ctx)) goto err; - if (!BN_to_montgomery(&am,&am,mont,ctx)) goto err; - } - else if (!BN_to_montgomery(&am,a,mont,ctx)) goto err; +#ifdef alloca + if (powerbufLen < 3072) + powerbufFree = NULL; +#endif + /* lay down tmp and am right after powers table */ + tmp.d = (BN_ULONG *)(powerbuf + sizeof(m->d[0]) * top * numPowers); + am.d = tmp.d + top; + tmp.top = am.top = 0; + tmp.dmax = am.dmax = top; + tmp.neg = am.neg = 0; + tmp.flags = am.flags = BN_FLG_STATIC_DATA; + + /* prepare a^0 in Montgomery domain */ +#if 1 /* by Shay Gueron's suggestion */ + if (m->d[top - 1] & (((BN_ULONG)1) << (BN_BITS2 - 1))) { + /* 2^(top*BN_BITS2) - m */ + tmp.d[0] = (0 - m->d[0]) & BN_MASK2; + for (i = 1; i < top; i++) + tmp.d[i] = (~m->d[i]) & BN_MASK2; + tmp.top = top; + } else +#endif + if (!BN_to_montgomery(&tmp, BN_value_one(), mont, ctx)) + goto err; + + /* prepare a^1 in Montgomery domain */ + if (a->neg || BN_ucmp(a, m) >= 0) { + if (!BN_mod(&am, a, m, ctx)) + goto err; + if (!BN_to_montgomery(&am, &am, mont, ctx)) + goto err; + } else if (!BN_to_montgomery(&am, a, mont, ctx)) + goto err; + +#if defined(SPARC_T4_MONT) + if (t4) { + typedef int (*bn_pwr5_mont_f) (BN_ULONG *tp, const BN_ULONG *np, + const BN_ULONG *n0, const void *table, + int power, int bits); + int bn_pwr5_mont_t4_8(BN_ULONG *tp, const BN_ULONG *np, + const BN_ULONG *n0, const void *table, + int power, int bits); + int bn_pwr5_mont_t4_16(BN_ULONG *tp, const BN_ULONG *np, + const BN_ULONG *n0, const void *table, + int power, int bits); + int bn_pwr5_mont_t4_24(BN_ULONG *tp, const BN_ULONG *np, + const BN_ULONG *n0, const void *table, + int power, int bits); + int bn_pwr5_mont_t4_32(BN_ULONG *tp, const BN_ULONG *np, + const BN_ULONG *n0, const void *table, + int power, int bits); + static const bn_pwr5_mont_f pwr5_funcs[4] = { + bn_pwr5_mont_t4_8, bn_pwr5_mont_t4_16, + bn_pwr5_mont_t4_24, bn_pwr5_mont_t4_32 + }; + bn_pwr5_mont_f pwr5_worker = pwr5_funcs[top / 16 - 1]; + + typedef int (*bn_mul_mont_f) (BN_ULONG *rp, const BN_ULONG *ap, + const void *bp, const BN_ULONG *np, + const BN_ULONG *n0); + int bn_mul_mont_t4_8(BN_ULONG *rp, const BN_ULONG *ap, const void *bp, + const BN_ULONG *np, const BN_ULONG *n0); + int bn_mul_mont_t4_16(BN_ULONG *rp, const BN_ULONG *ap, + const void *bp, const BN_ULONG *np, + const BN_ULONG *n0); + int bn_mul_mont_t4_24(BN_ULONG *rp, const BN_ULONG *ap, + const void *bp, const BN_ULONG *np, + const BN_ULONG *n0); + int bn_mul_mont_t4_32(BN_ULONG *rp, const BN_ULONG *ap, + const void *bp, const BN_ULONG *np, + const BN_ULONG *n0); + static const bn_mul_mont_f mul_funcs[4] = { + bn_mul_mont_t4_8, bn_mul_mont_t4_16, + bn_mul_mont_t4_24, bn_mul_mont_t4_32 + }; + bn_mul_mont_f mul_worker = mul_funcs[top / 16 - 1]; + + void bn_mul_mont_vis3(BN_ULONG *rp, const BN_ULONG *ap, + const void *bp, const BN_ULONG *np, + const BN_ULONG *n0, int num); + void bn_mul_mont_t4(BN_ULONG *rp, const BN_ULONG *ap, + const void *bp, const BN_ULONG *np, + const BN_ULONG *n0, int num); + void bn_mul_mont_gather5_t4(BN_ULONG *rp, const BN_ULONG *ap, + const void *table, const BN_ULONG *np, + const BN_ULONG *n0, int num, int power); + void bn_flip_n_scatter5_t4(const BN_ULONG *inp, size_t num, + void *table, size_t power); + void bn_gather5_t4(BN_ULONG *out, size_t num, + void *table, size_t power); + void bn_flip_t4(BN_ULONG *dst, BN_ULONG *src, size_t num); + + BN_ULONG *np = mont->N.d, *n0 = mont->n0; + int stride = 5 * (6 - (top / 16 - 1)); /* multiple of 5, but less + * than 32 */ + + /* + * BN_to_montgomery can contaminate words above .top [in + * BN_DEBUG[_DEBUG] build]... + */ + for (i = am.top; i < top; i++) + am.d[i] = 0; + for (i = tmp.top; i < top; i++) + tmp.d[i] = 0; + + bn_flip_n_scatter5_t4(tmp.d, top, powerbuf, 0); + bn_flip_n_scatter5_t4(am.d, top, powerbuf, 1); + if (!(*mul_worker) (tmp.d, am.d, am.d, np, n0) && + !(*mul_worker) (tmp.d, am.d, am.d, np, n0)) + bn_mul_mont_vis3(tmp.d, am.d, am.d, np, n0, top); + bn_flip_n_scatter5_t4(tmp.d, top, powerbuf, 2); + + for (i = 3; i < 32; i++) { + /* Calculate a^i = a^(i-1) * a */ + if (!(*mul_worker) (tmp.d, tmp.d, am.d, np, n0) && + !(*mul_worker) (tmp.d, tmp.d, am.d, np, n0)) + bn_mul_mont_vis3(tmp.d, tmp.d, am.d, np, n0, top); + bn_flip_n_scatter5_t4(tmp.d, top, powerbuf, i); + } + + /* switch to 64-bit domain */ + np = alloca(top * sizeof(BN_ULONG)); + top /= 2; + bn_flip_t4(np, mont->N.d, top); + + bits--; + for (wvalue = 0, i = bits % 5; i >= 0; i--, bits--) + wvalue = (wvalue << 1) + BN_is_bit_set(p, bits); + bn_gather5_t4(tmp.d, top, powerbuf, wvalue); + + /* + * Scan the exponent one window at a time starting from the most + * significant bits. + */ + while (bits >= 0) { + if (bits < stride) + stride = bits + 1; + bits -= stride; + wvalue = bn_get_bits(p, bits + 1); + + if ((*pwr5_worker) (tmp.d, np, n0, powerbuf, wvalue, stride)) + continue; + /* retry once and fall back */ + if ((*pwr5_worker) (tmp.d, np, n0, powerbuf, wvalue, stride)) + continue; + + bits += stride - 5; + wvalue >>= stride - 5; + wvalue &= 31; + bn_mul_mont_t4(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont_t4(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont_t4(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont_t4(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont_t4(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont_gather5_t4(tmp.d, tmp.d, powerbuf, np, n0, top, + wvalue); + } + + bn_flip_t4(tmp.d, tmp.d, top); + top *= 2; + /* back to 32-bit domain */ + tmp.top = top; + bn_correct_top(&tmp); + OPENSSL_cleanse(np, top * sizeof(BN_ULONG)); + } else +#endif #if defined(OPENSSL_BN_ASM_MONT5) - /* This optimization uses ideas from http://eprint.iacr.org/2011/239, - * specifically optimization of cache-timing attack countermeasures - * and pre-computation optimization. */ - - /* Dedicated window==4 case improves 512-bit RSA sign by ~15%, but as - * 512-bit RSA is hardly relevant, we omit it to spare size... */ - if (window==5 && top>1) - { - void bn_mul_mont_gather5(BN_ULONG *rp,const BN_ULONG *ap, - const void *table,const BN_ULONG *np, - const BN_ULONG *n0,int num,int power); - void bn_scatter5(const BN_ULONG *inp,size_t num, - void *table,size_t power); - void bn_gather5(BN_ULONG *out,size_t num, - void *table,size_t power); - - BN_ULONG *np=mont->N.d, *n0=mont->n0; - - /* BN_to_montgomery can contaminate words above .top - * [in BN_DEBUG[_DEBUG] build]... */ - for (i=am.top; i<top; i++) am.d[i]=0; - for (i=tmp.top; i<top; i++) tmp.d[i]=0; - - bn_scatter5(tmp.d,top,powerbuf,0); - bn_scatter5(am.d,am.top,powerbuf,1); - bn_mul_mont(tmp.d,am.d,am.d,np,n0,top); - bn_scatter5(tmp.d,top,powerbuf,2); - -#if 0 - for (i=3; i<32; i++) - { - /* Calculate a^i = a^(i-1) * a */ - bn_mul_mont_gather5(tmp.d,am.d,powerbuf,np,n0,top,i-1); - bn_scatter5(tmp.d,top,powerbuf,i); - } -#else - /* same as above, but uses squaring for 1/2 of operations */ - for (i=4; i<32; i*=2) - { - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_scatter5(tmp.d,top,powerbuf,i); - } - for (i=3; i<8; i+=2) - { - int j; - bn_mul_mont_gather5(tmp.d,am.d,powerbuf,np,n0,top,i-1); - bn_scatter5(tmp.d,top,powerbuf,i); - for (j=2*i; j<32; j*=2) - { - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_scatter5(tmp.d,top,powerbuf,j); - } - } - for (; i<16; i+=2) - { - bn_mul_mont_gather5(tmp.d,am.d,powerbuf,np,n0,top,i-1); - bn_scatter5(tmp.d,top,powerbuf,i); - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_scatter5(tmp.d,top,powerbuf,2*i); - } - for (; i<32; i+=2) - { - bn_mul_mont_gather5(tmp.d,am.d,powerbuf,np,n0,top,i-1); - bn_scatter5(tmp.d,top,powerbuf,i); - } + if (window == 5 && top > 1) { + /* + * This optimization uses ideas from http://eprint.iacr.org/2011/239, + * specifically optimization of cache-timing attack countermeasures + * and pre-computation optimization. + */ + + /* + * Dedicated window==4 case improves 512-bit RSA sign by ~15%, but as + * 512-bit RSA is hardly relevant, we omit it to spare size... + */ + void bn_mul_mont_gather5(BN_ULONG *rp, const BN_ULONG *ap, + const void *table, const BN_ULONG *np, + const BN_ULONG *n0, int num, int power); + void bn_scatter5(const BN_ULONG *inp, size_t num, + void *table, size_t power); + void bn_gather5(BN_ULONG *out, size_t num, void *table, size_t power); + void bn_power5(BN_ULONG *rp, const BN_ULONG *ap, + const void *table, const BN_ULONG *np, + const BN_ULONG *n0, int num, int power); + int bn_get_bits5(const BN_ULONG *ap, int off); + int bn_from_montgomery(BN_ULONG *rp, const BN_ULONG *ap, + const BN_ULONG *not_used, const BN_ULONG *np, + const BN_ULONG *n0, int num); + + BN_ULONG *np = mont->N.d, *n0 = mont->n0, *np2; + + /* + * BN_to_montgomery can contaminate words above .top [in + * BN_DEBUG[_DEBUG] build]... + */ + for (i = am.top; i < top; i++) + am.d[i] = 0; + for (i = tmp.top; i < top; i++) + tmp.d[i] = 0; + + if (top & 7) + np2 = np; + else + for (np2 = am.d + top, i = 0; i < top; i++) + np2[2 * i] = np[i]; + + bn_scatter5(tmp.d, top, powerbuf, 0); + bn_scatter5(am.d, am.top, powerbuf, 1); + bn_mul_mont(tmp.d, am.d, am.d, np, n0, top); + bn_scatter5(tmp.d, top, powerbuf, 2); + +# if 0 + for (i = 3; i < 32; i++) { + /* Calculate a^i = a^(i-1) * a */ + bn_mul_mont_gather5(tmp.d, am.d, powerbuf, np2, n0, top, i - 1); + bn_scatter5(tmp.d, top, powerbuf, i); + } +# else + /* same as above, but uses squaring for 1/2 of operations */ + for (i = 4; i < 32; i *= 2) { + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_scatter5(tmp.d, top, powerbuf, i); + } + for (i = 3; i < 8; i += 2) { + int j; + bn_mul_mont_gather5(tmp.d, am.d, powerbuf, np2, n0, top, i - 1); + bn_scatter5(tmp.d, top, powerbuf, i); + for (j = 2 * i; j < 32; j *= 2) { + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_scatter5(tmp.d, top, powerbuf, j); + } + } + for (; i < 16; i += 2) { + bn_mul_mont_gather5(tmp.d, am.d, powerbuf, np2, n0, top, i - 1); + bn_scatter5(tmp.d, top, powerbuf, i); + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_scatter5(tmp.d, top, powerbuf, 2 * i); + } + for (; i < 32; i += 2) { + bn_mul_mont_gather5(tmp.d, am.d, powerbuf, np2, n0, top, i - 1); + bn_scatter5(tmp.d, top, powerbuf, i); + } +# endif + bits--; + for (wvalue = 0, i = bits % 5; i >= 0; i--, bits--) + wvalue = (wvalue << 1) + BN_is_bit_set(p, bits); + bn_gather5(tmp.d, top, powerbuf, wvalue); + + /* + * Scan the exponent one window at a time starting from the most + * significant bits. + */ + if (top & 7) + while (bits >= 0) { + for (wvalue = 0, i = 0; i < 5; i++, bits--) + wvalue = (wvalue << 1) + BN_is_bit_set(p, bits); + + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont(tmp.d, tmp.d, tmp.d, np, n0, top); + bn_mul_mont_gather5(tmp.d, tmp.d, powerbuf, np, n0, top, + wvalue); + } else { + while (bits >= 0) { + wvalue = bn_get_bits5(p->d, bits - 4); + bits -= 5; + bn_power5(tmp.d, tmp.d, powerbuf, np2, n0, top, wvalue); + } + } + + ret = bn_from_montgomery(tmp.d, tmp.d, NULL, np2, n0, top); + tmp.top = top; + bn_correct_top(&tmp); + if (ret) { + if (!BN_copy(rr, &tmp)) + ret = 0; + goto err; /* non-zero ret means it's not error */ + } + } else #endif - bits--; - for (wvalue=0, i=bits%5; i>=0; i--,bits--) - wvalue = (wvalue<<1)+BN_is_bit_set(p,bits); - bn_gather5(tmp.d,top,powerbuf,wvalue); - - /* Scan the exponent one window at a time starting from the most - * significant bits. - */ - while (bits >= 0) - { - for (wvalue=0, i=0; i<5; i++,bits--) - wvalue = (wvalue<<1)+BN_is_bit_set(p,bits); - - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_mul_mont(tmp.d,tmp.d,tmp.d,np,n0,top); - bn_mul_mont_gather5(tmp.d,tmp.d,powerbuf,np,n0,top,wvalue); - } - - tmp.top=top; - bn_correct_top(&tmp); - } - else + { + if (!MOD_EXP_CTIME_COPY_TO_PREBUF(&tmp, top, powerbuf, 0, numPowers)) + goto err; + if (!MOD_EXP_CTIME_COPY_TO_PREBUF(&am, top, powerbuf, 1, numPowers)) + goto err; + + /* + * If the window size is greater than 1, then calculate + * val[i=2..2^winsize-1]. Powers are computed as a*a^(i-1) (even + * powers could instead be computed as (a^(i/2))^2 to use the slight + * performance advantage of sqr over mul). + */ + if (window > 1) { + if (!BN_mod_mul_montgomery(&tmp, &am, &am, mont, ctx)) + goto err; + if (!MOD_EXP_CTIME_COPY_TO_PREBUF + (&tmp, top, powerbuf, 2, numPowers)) + goto err; + for (i = 3; i < numPowers; i++) { + /* Calculate a^i = a^(i-1) * a */ + if (!BN_mod_mul_montgomery(&tmp, &am, &tmp, mont, ctx)) + goto err; + if (!MOD_EXP_CTIME_COPY_TO_PREBUF + (&tmp, top, powerbuf, i, numPowers)) + goto err; + } + } + + bits--; + for (wvalue = 0, i = bits % window; i >= 0; i--, bits--) + wvalue = (wvalue << 1) + BN_is_bit_set(p, bits); + if (!MOD_EXP_CTIME_COPY_FROM_PREBUF + (&tmp, top, powerbuf, wvalue, numPowers)) + goto err; + + /* + * Scan the exponent one window at a time starting from the most + * significant bits. + */ + while (bits >= 0) { + wvalue = 0; /* The 'value' of the window */ + + /* Scan the window, squaring the result as we go */ + for (i = 0; i < window; i++, bits--) { + if (!BN_mod_mul_montgomery(&tmp, &tmp, &tmp, mont, ctx)) + goto err; + wvalue = (wvalue << 1) + BN_is_bit_set(p, bits); + } + + /* + * Fetch the appropriate pre-computed value from the pre-buf + */ + if (!MOD_EXP_CTIME_COPY_FROM_PREBUF + (&am, top, powerbuf, wvalue, numPowers)) + goto err; + + /* Multiply the result into the intermediate result */ + if (!BN_mod_mul_montgomery(&tmp, &tmp, &am, mont, ctx)) + goto err; + } + } + + /* Convert the final result from montgomery to standard format */ +#if defined(SPARC_T4_MONT) + if (OPENSSL_sparcv9cap_P[0] & (SPARCV9_VIS3 | SPARCV9_PREFER_FPU)) { + am.d[0] = 1; /* borrow am */ + for (i = 1; i < top; i++) + am.d[i] = 0; + if (!BN_mod_mul_montgomery(rr, &tmp, &am, mont, ctx)) + goto err; + } else #endif - { - if (!MOD_EXP_CTIME_COPY_TO_PREBUF(&tmp, top, powerbuf, 0, numPowers)) goto err; - if (!MOD_EXP_CTIME_COPY_TO_PREBUF(&am, top, powerbuf, 1, numPowers)) goto err; - - /* If the window size is greater than 1, then calculate - * val[i=2..2^winsize-1]. Powers are computed as a*a^(i-1) - * (even powers could instead be computed as (a^(i/2))^2 - * to use the slight performance advantage of sqr over mul). - */ - if (window > 1) - { - if (!BN_mod_mul_montgomery(&tmp,&am,&am,mont,ctx)) goto err; - if (!MOD_EXP_CTIME_COPY_TO_PREBUF(&tmp, top, powerbuf, 2, numPowers)) goto err; - for (i=3; i<numPowers; i++) - { - /* Calculate a^i = a^(i-1) * a */ - if (!BN_mod_mul_montgomery(&tmp,&am,&tmp,mont,ctx)) - goto err; - if (!MOD_EXP_CTIME_COPY_TO_PREBUF(&tmp, top, powerbuf, i, numPowers)) goto err; - } - } - - bits--; - for (wvalue=0, i=bits%window; i>=0; i--,bits--) - wvalue = (wvalue<<1)+BN_is_bit_set(p,bits); - if (!MOD_EXP_CTIME_COPY_FROM_PREBUF(&tmp,top,powerbuf,wvalue,numPowers)) goto err; - - /* Scan the exponent one window at a time starting from the most - * significant bits. - */ - while (bits >= 0) - { - wvalue=0; /* The 'value' of the window */ - - /* Scan the window, squaring the result as we go */ - for (i=0; i<window; i++,bits--) - { - if (!BN_mod_mul_montgomery(&tmp,&tmp,&tmp,mont,ctx)) goto err; - wvalue = (wvalue<<1)+BN_is_bit_set(p,bits); - } - - /* Fetch the appropriate pre-computed value from the pre-buf */ - if (!MOD_EXP_CTIME_COPY_FROM_PREBUF(&am, top, powerbuf, wvalue, numPowers)) goto err; - - /* Multiply the result into the intermediate result */ - if (!BN_mod_mul_montgomery(&tmp,&tmp,&am,mont,ctx)) goto err; - } - } - - /* Convert the final result from montgomery to standard format */ - if (!BN_from_montgomery(rr,&tmp,mont,ctx)) goto err; - ret=1; -err: - if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont); - if (powerbuf!=NULL) - { - OPENSSL_cleanse(powerbuf,powerbufLen); - if (powerbufFree) OPENSSL_free(powerbufFree); - } - BN_CTX_end(ctx); - return(ret); - } + if (!BN_from_montgomery(rr, &tmp, mont, ctx)) + goto err; + ret = 1; + err: + if ((in_mont == NULL) && (mont != NULL)) + BN_MONT_CTX_free(mont); + if (powerbuf != NULL) { + OPENSSL_cleanse(powerbuf, powerbufLen); + if (powerbufFree) + OPENSSL_free(powerbufFree); + } + BN_CTX_end(ctx); + return (ret); +} int BN_mod_exp_mont_word(BIGNUM *rr, BN_ULONG a, const BIGNUM *p, const BIGNUM *m, BN_CTX *ctx, BN_MONT_CTX *in_mont) - { - BN_MONT_CTX *mont = NULL; - int b, bits, ret=0; - int r_is_one; - BN_ULONG w, next_w; - BIGNUM *d, *r, *t; - BIGNUM *swap_tmp; +{ + BN_MONT_CTX *mont = NULL; + int b, bits, ret = 0; + int r_is_one; + BN_ULONG w, next_w; + BIGNUM *d, *r, *t; + BIGNUM *swap_tmp; #define BN_MOD_MUL_WORD(r, w, m) \ - (BN_mul_word(r, (w)) && \ - (/* BN_ucmp(r, (m)) < 0 ? 1 :*/ \ - (BN_mod(t, r, m, ctx) && (swap_tmp = r, r = t, t = swap_tmp, 1)))) - /* BN_MOD_MUL_WORD is only used with 'w' large, - * so the BN_ucmp test is probably more overhead - * than always using BN_mod (which uses BN_copy if - * a similar test returns true). */ - /* We can use BN_mod and do not need BN_nnmod because our - * accumulator is never negative (the result of BN_mod does - * not depend on the sign of the modulus). - */ + (BN_mul_word(r, (w)) && \ + (/* BN_ucmp(r, (m)) < 0 ? 1 :*/ \ + (BN_mod(t, r, m, ctx) && (swap_tmp = r, r = t, t = swap_tmp, 1)))) + /* + * BN_MOD_MUL_WORD is only used with 'w' large, so the BN_ucmp test is + * probably more overhead than always using BN_mod (which uses BN_copy if + * a similar test returns true). + */ + /* + * We can use BN_mod and do not need BN_nnmod because our accumulator is + * never negative (the result of BN_mod does not depend on the sign of + * the modulus). + */ #define BN_TO_MONTGOMERY_WORD(r, w, mont) \ - (BN_set_word(r, (w)) && BN_to_montgomery(r, r, (mont), ctx)) - - if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) - { - /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ - BNerr(BN_F_BN_MOD_EXP_MONT_WORD,ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); - return -1; - } - - bn_check_top(p); - bn_check_top(m); - - if (!BN_is_odd(m)) - { - BNerr(BN_F_BN_MOD_EXP_MONT_WORD,BN_R_CALLED_WITH_EVEN_MODULUS); - return(0); - } - if (m->top == 1) - a %= m->d[0]; /* make sure that 'a' is reduced */ - - bits = BN_num_bits(p); - if (bits == 0) - { - /* x**0 mod 1 is still zero. */ - if (BN_is_one(m)) - { - ret = 1; - BN_zero(rr); - } - else - ret = BN_one(rr); - return ret; - } - if (a == 0) - { - BN_zero(rr); - ret = 1; - return ret; - } - - BN_CTX_start(ctx); - d = BN_CTX_get(ctx); - r = BN_CTX_get(ctx); - t = BN_CTX_get(ctx); - if (d == NULL || r == NULL || t == NULL) goto err; - - if (in_mont != NULL) - mont=in_mont; - else - { - if ((mont = BN_MONT_CTX_new()) == NULL) goto err; - if (!BN_MONT_CTX_set(mont, m, ctx)) goto err; - } - - r_is_one = 1; /* except for Montgomery factor */ - - /* bits-1 >= 0 */ - - /* The result is accumulated in the product r*w. */ - w = a; /* bit 'bits-1' of 'p' is always set */ - for (b = bits-2; b >= 0; b--) - { - /* First, square r*w. */ - next_w = w*w; - if ((next_w/w) != w) /* overflow */ - { - if (r_is_one) - { - if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) goto err; - r_is_one = 0; - } - else - { - if (!BN_MOD_MUL_WORD(r, w, m)) goto err; - } - next_w = 1; - } - w = next_w; - if (!r_is_one) - { - if (!BN_mod_mul_montgomery(r, r, r, mont, ctx)) goto err; - } - - /* Second, multiply r*w by 'a' if exponent bit is set. */ - if (BN_is_bit_set(p, b)) - { - next_w = w*a; - if ((next_w/a) != w) /* overflow */ - { - if (r_is_one) - { - if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) goto err; - r_is_one = 0; - } - else - { - if (!BN_MOD_MUL_WORD(r, w, m)) goto err; - } - next_w = a; - } - w = next_w; - } - } - - /* Finally, set r:=r*w. */ - if (w != 1) - { - if (r_is_one) - { - if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) goto err; - r_is_one = 0; - } - else - { - if (!BN_MOD_MUL_WORD(r, w, m)) goto err; - } - } - - if (r_is_one) /* can happen only if a == 1*/ - { - if (!BN_one(rr)) goto err; - } - else - { - if (!BN_from_montgomery(rr, r, mont, ctx)) goto err; - } - ret = 1; -err: - if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont); - BN_CTX_end(ctx); - bn_check_top(rr); - return(ret); - } - + (BN_set_word(r, (w)) && BN_to_montgomery(r, r, (mont), ctx)) + + if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) { + /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ + BNerr(BN_F_BN_MOD_EXP_MONT_WORD, ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); + return -1; + } + + bn_check_top(p); + bn_check_top(m); + + if (!BN_is_odd(m)) { + BNerr(BN_F_BN_MOD_EXP_MONT_WORD, BN_R_CALLED_WITH_EVEN_MODULUS); + return (0); + } + if (m->top == 1) + a %= m->d[0]; /* make sure that 'a' is reduced */ + + bits = BN_num_bits(p); + if (bits == 0) { + /* x**0 mod 1 is still zero. */ + if (BN_is_one(m)) { + ret = 1; + BN_zero(rr); + } else + ret = BN_one(rr); + return ret; + } + if (a == 0) { + BN_zero(rr); + ret = 1; + return ret; + } + + BN_CTX_start(ctx); + d = BN_CTX_get(ctx); + r = BN_CTX_get(ctx); + t = BN_CTX_get(ctx); + if (d == NULL || r == NULL || t == NULL) + goto err; + + if (in_mont != NULL) + mont = in_mont; + else { + if ((mont = BN_MONT_CTX_new()) == NULL) + goto err; + if (!BN_MONT_CTX_set(mont, m, ctx)) + goto err; + } + + r_is_one = 1; /* except for Montgomery factor */ + + /* bits-1 >= 0 */ + + /* The result is accumulated in the product r*w. */ + w = a; /* bit 'bits-1' of 'p' is always set */ + for (b = bits - 2; b >= 0; b--) { + /* First, square r*w. */ + next_w = w * w; + if ((next_w / w) != w) { /* overflow */ + if (r_is_one) { + if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) + goto err; + r_is_one = 0; + } else { + if (!BN_MOD_MUL_WORD(r, w, m)) + goto err; + } + next_w = 1; + } + w = next_w; + if (!r_is_one) { + if (!BN_mod_mul_montgomery(r, r, r, mont, ctx)) + goto err; + } + + /* Second, multiply r*w by 'a' if exponent bit is set. */ + if (BN_is_bit_set(p, b)) { + next_w = w * a; + if ((next_w / a) != w) { /* overflow */ + if (r_is_one) { + if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) + goto err; + r_is_one = 0; + } else { + if (!BN_MOD_MUL_WORD(r, w, m)) + goto err; + } + next_w = a; + } + w = next_w; + } + } + + /* Finally, set r:=r*w. */ + if (w != 1) { + if (r_is_one) { + if (!BN_TO_MONTGOMERY_WORD(r, w, mont)) + goto err; + r_is_one = 0; + } else { + if (!BN_MOD_MUL_WORD(r, w, m)) + goto err; + } + } + + if (r_is_one) { /* can happen only if a == 1 */ + if (!BN_one(rr)) + goto err; + } else { + if (!BN_from_montgomery(rr, r, mont, ctx)) + goto err; + } + ret = 1; + err: + if ((in_mont == NULL) && (mont != NULL)) + BN_MONT_CTX_free(mont); + BN_CTX_end(ctx); + bn_check_top(rr); + return (ret); +} /* The old fallback, simple version :-) */ int BN_mod_exp_simple(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, - const BIGNUM *m, BN_CTX *ctx) - { - int i,j,bits,ret=0,wstart,wend,window,wvalue; - int start=1; - BIGNUM *d; - /* Table of variables obtained from 'ctx' */ - BIGNUM *val[TABLE_SIZE]; - - if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) - { - /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ - BNerr(BN_F_BN_MOD_EXP_SIMPLE,ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); - return -1; - } - - bits=BN_num_bits(p); - - if (bits == 0) - { - ret = BN_one(r); - return ret; - } - - BN_CTX_start(ctx); - d = BN_CTX_get(ctx); - val[0] = BN_CTX_get(ctx); - if(!d || !val[0]) goto err; - - if (!BN_nnmod(val[0],a,m,ctx)) goto err; /* 1 */ - if (BN_is_zero(val[0])) - { - BN_zero(r); - ret = 1; - goto err; - } - - window = BN_window_bits_for_exponent_size(bits); - if (window > 1) - { - if (!BN_mod_mul(d,val[0],val[0],m,ctx)) - goto err; /* 2 */ - j=1<<(window-1); - for (i=1; i<j; i++) - { - if(((val[i] = BN_CTX_get(ctx)) == NULL) || - !BN_mod_mul(val[i],val[i-1],d,m,ctx)) - goto err; - } - } - - start=1; /* This is used to avoid multiplication etc - * when there is only the value '1' in the - * buffer. */ - wvalue=0; /* The 'value' of the window */ - wstart=bits-1; /* The top bit of the window */ - wend=0; /* The bottom bit of the window */ - - if (!BN_one(r)) goto err; - - for (;;) - { - if (BN_is_bit_set(p,wstart) == 0) - { - if (!start) - if (!BN_mod_mul(r,r,r,m,ctx)) - goto err; - if (wstart == 0) break; - wstart--; - continue; - } - /* We now have wstart on a 'set' bit, we now need to work out - * how bit a window to do. To do this we need to scan - * forward until the last set bit before the end of the - * window */ - j=wstart; - wvalue=1; - wend=0; - for (i=1; i<window; i++) - { - if (wstart-i < 0) break; - if (BN_is_bit_set(p,wstart-i)) - { - wvalue<<=(i-wend); - wvalue|=1; - wend=i; - } - } - - /* wend is the size of the current window */ - j=wend+1; - /* add the 'bytes above' */ - if (!start) - for (i=0; i<j; i++) - { - if (!BN_mod_mul(r,r,r,m,ctx)) - goto err; - } - - /* wvalue will be an odd number < 2^window */ - if (!BN_mod_mul(r,r,val[wvalue>>1],m,ctx)) - goto err; - - /* move the 'window' down further */ - wstart-=wend+1; - wvalue=0; - start=0; - if (wstart < 0) break; - } - ret=1; -err: - BN_CTX_end(ctx); - bn_check_top(r); - return(ret); - } + const BIGNUM *m, BN_CTX *ctx) +{ + int i, j, bits, ret = 0, wstart, wend, window, wvalue; + int start = 1; + BIGNUM *d; + /* Table of variables obtained from 'ctx' */ + BIGNUM *val[TABLE_SIZE]; + + if (BN_get_flags(p, BN_FLG_CONSTTIME) != 0) { + /* BN_FLG_CONSTTIME only supported by BN_mod_exp_mont() */ + BNerr(BN_F_BN_MOD_EXP_SIMPLE, ERR_R_SHOULD_NOT_HAVE_BEEN_CALLED); + return -1; + } + + bits = BN_num_bits(p); + + if (bits == 0) { + ret = BN_one(r); + return ret; + } + + BN_CTX_start(ctx); + d = BN_CTX_get(ctx); + val[0] = BN_CTX_get(ctx); + if (!d || !val[0]) + goto err; + + if (!BN_nnmod(val[0], a, m, ctx)) + goto err; /* 1 */ + if (BN_is_zero(val[0])) { + BN_zero(r); + ret = 1; + goto err; + } + + window = BN_window_bits_for_exponent_size(bits); + if (window > 1) { + if (!BN_mod_mul(d, val[0], val[0], m, ctx)) + goto err; /* 2 */ + j = 1 << (window - 1); + for (i = 1; i < j; i++) { + if (((val[i] = BN_CTX_get(ctx)) == NULL) || + !BN_mod_mul(val[i], val[i - 1], d, m, ctx)) + goto err; + } + } + + start = 1; /* This is used to avoid multiplication etc + * when there is only the value '1' in the + * buffer. */ + wvalue = 0; /* The 'value' of the window */ + wstart = bits - 1; /* The top bit of the window */ + wend = 0; /* The bottom bit of the window */ + + if (!BN_one(r)) + goto err; + + for (;;) { + if (BN_is_bit_set(p, wstart) == 0) { + if (!start) + if (!BN_mod_mul(r, r, r, m, ctx)) + goto err; + if (wstart == 0) + break; + wstart--; + continue; + } + /* + * We now have wstart on a 'set' bit, we now need to work out how bit + * a window to do. To do this we need to scan forward until the last + * set bit before the end of the window + */ + j = wstart; + wvalue = 1; + wend = 0; + for (i = 1; i < window; i++) { + if (wstart - i < 0) + break; + if (BN_is_bit_set(p, wstart - i)) { + wvalue <<= (i - wend); + wvalue |= 1; + wend = i; + } + } + + /* wend is the size of the current window */ + j = wend + 1; + /* add the 'bytes above' */ + if (!start) + for (i = 0; i < j; i++) { + if (!BN_mod_mul(r, r, r, m, ctx)) + goto err; + } + + /* wvalue will be an odd number < 2^window */ + if (!BN_mod_mul(r, r, val[wvalue >> 1], m, ctx)) + goto err; + + /* move the 'window' down further */ + wstart -= wend + 1; + wvalue = 0; + start = 0; + if (wstart < 0) + break; + } + ret = 1; + err: + BN_CTX_end(ctx); + bn_check_top(r); + return (ret); +} diff --git a/openssl/crypto/bn/bn_exp2.c b/openssl/crypto/bn/bn_exp2.c index bd0c34b91..43fd2044c 100644 --- a/openssl/crypto/bn/bn_exp2.c +++ b/openssl/crypto/bn/bn_exp2.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -113,200 +113,191 @@ #include "cryptlib.h" #include "bn_lcl.h" -#define TABLE_SIZE 32 +#define TABLE_SIZE 32 int BN_mod_exp2_mont(BIGNUM *rr, const BIGNUM *a1, const BIGNUM *p1, - const BIGNUM *a2, const BIGNUM *p2, const BIGNUM *m, - BN_CTX *ctx, BN_MONT_CTX *in_mont) - { - int i,j,bits,b,bits1,bits2,ret=0,wpos1,wpos2,window1,window2,wvalue1,wvalue2; - int r_is_one=1; - BIGNUM *d,*r; - const BIGNUM *a_mod_m; - /* Tables of variables obtained from 'ctx' */ - BIGNUM *val1[TABLE_SIZE], *val2[TABLE_SIZE]; - BN_MONT_CTX *mont=NULL; + const BIGNUM *a2, const BIGNUM *p2, const BIGNUM *m, + BN_CTX *ctx, BN_MONT_CTX *in_mont) +{ + int i, j, bits, b, bits1, bits2, ret = + 0, wpos1, wpos2, window1, window2, wvalue1, wvalue2; + int r_is_one = 1; + BIGNUM *d, *r; + const BIGNUM *a_mod_m; + /* Tables of variables obtained from 'ctx' */ + BIGNUM *val1[TABLE_SIZE], *val2[TABLE_SIZE]; + BN_MONT_CTX *mont = NULL; + + bn_check_top(a1); + bn_check_top(p1); + bn_check_top(a2); + bn_check_top(p2); + bn_check_top(m); + + if (!(m->d[0] & 1)) { + BNerr(BN_F_BN_MOD_EXP2_MONT, BN_R_CALLED_WITH_EVEN_MODULUS); + return (0); + } + bits1 = BN_num_bits(p1); + bits2 = BN_num_bits(p2); + if ((bits1 == 0) && (bits2 == 0)) { + ret = BN_one(rr); + return ret; + } - bn_check_top(a1); - bn_check_top(p1); - bn_check_top(a2); - bn_check_top(p2); - bn_check_top(m); + bits = (bits1 > bits2) ? bits1 : bits2; - if (!(m->d[0] & 1)) - { - BNerr(BN_F_BN_MOD_EXP2_MONT,BN_R_CALLED_WITH_EVEN_MODULUS); - return(0); - } - bits1=BN_num_bits(p1); - bits2=BN_num_bits(p2); - if ((bits1 == 0) && (bits2 == 0)) - { - ret = BN_one(rr); - return ret; - } - - bits=(bits1 > bits2)?bits1:bits2; + BN_CTX_start(ctx); + d = BN_CTX_get(ctx); + r = BN_CTX_get(ctx); + val1[0] = BN_CTX_get(ctx); + val2[0] = BN_CTX_get(ctx); + if (!d || !r || !val1[0] || !val2[0]) + goto err; - BN_CTX_start(ctx); - d = BN_CTX_get(ctx); - r = BN_CTX_get(ctx); - val1[0] = BN_CTX_get(ctx); - val2[0] = BN_CTX_get(ctx); - if(!d || !r || !val1[0] || !val2[0]) goto err; + if (in_mont != NULL) + mont = in_mont; + else { + if ((mont = BN_MONT_CTX_new()) == NULL) + goto err; + if (!BN_MONT_CTX_set(mont, m, ctx)) + goto err; + } - if (in_mont != NULL) - mont=in_mont; - else - { - if ((mont=BN_MONT_CTX_new()) == NULL) goto err; - if (!BN_MONT_CTX_set(mont,m,ctx)) goto err; - } + window1 = BN_window_bits_for_exponent_size(bits1); + window2 = BN_window_bits_for_exponent_size(bits2); - window1 = BN_window_bits_for_exponent_size(bits1); - window2 = BN_window_bits_for_exponent_size(bits2); + /* + * Build table for a1: val1[i] := a1^(2*i + 1) mod m for i = 0 .. 2^(window1-1) + */ + if (a1->neg || BN_ucmp(a1, m) >= 0) { + if (!BN_mod(val1[0], a1, m, ctx)) + goto err; + a_mod_m = val1[0]; + } else + a_mod_m = a1; + if (BN_is_zero(a_mod_m)) { + BN_zero(rr); + ret = 1; + goto err; + } - /* - * Build table for a1: val1[i] := a1^(2*i + 1) mod m for i = 0 .. 2^(window1-1) - */ - if (a1->neg || BN_ucmp(a1,m) >= 0) - { - if (!BN_mod(val1[0],a1,m,ctx)) - goto err; - a_mod_m = val1[0]; - } - else - a_mod_m = a1; - if (BN_is_zero(a_mod_m)) - { - BN_zero(rr); - ret = 1; - goto err; - } + if (!BN_to_montgomery(val1[0], a_mod_m, mont, ctx)) + goto err; + if (window1 > 1) { + if (!BN_mod_mul_montgomery(d, val1[0], val1[0], mont, ctx)) + goto err; - if (!BN_to_montgomery(val1[0],a_mod_m,mont,ctx)) goto err; - if (window1 > 1) - { - if (!BN_mod_mul_montgomery(d,val1[0],val1[0],mont,ctx)) goto err; + j = 1 << (window1 - 1); + for (i = 1; i < j; i++) { + if (((val1[i] = BN_CTX_get(ctx)) == NULL) || + !BN_mod_mul_montgomery(val1[i], val1[i - 1], d, mont, ctx)) + goto err; + } + } - j=1<<(window1-1); - for (i=1; i<j; i++) - { - if(((val1[i] = BN_CTX_get(ctx)) == NULL) || - !BN_mod_mul_montgomery(val1[i],val1[i-1], - d,mont,ctx)) - goto err; - } - } + /* + * Build table for a2: val2[i] := a2^(2*i + 1) mod m for i = 0 .. 2^(window2-1) + */ + if (a2->neg || BN_ucmp(a2, m) >= 0) { + if (!BN_mod(val2[0], a2, m, ctx)) + goto err; + a_mod_m = val2[0]; + } else + a_mod_m = a2; + if (BN_is_zero(a_mod_m)) { + BN_zero(rr); + ret = 1; + goto err; + } + if (!BN_to_montgomery(val2[0], a_mod_m, mont, ctx)) + goto err; + if (window2 > 1) { + if (!BN_mod_mul_montgomery(d, val2[0], val2[0], mont, ctx)) + goto err; + j = 1 << (window2 - 1); + for (i = 1; i < j; i++) { + if (((val2[i] = BN_CTX_get(ctx)) == NULL) || + !BN_mod_mul_montgomery(val2[i], val2[i - 1], d, mont, ctx)) + goto err; + } + } - /* - * Build table for a2: val2[i] := a2^(2*i + 1) mod m for i = 0 .. 2^(window2-1) - */ - if (a2->neg || BN_ucmp(a2,m) >= 0) - { - if (!BN_mod(val2[0],a2,m,ctx)) - goto err; - a_mod_m = val2[0]; - } - else - a_mod_m = a2; - if (BN_is_zero(a_mod_m)) - { - BN_zero(rr); - ret = 1; - goto err; - } - if (!BN_to_montgomery(val2[0],a_mod_m,mont,ctx)) goto err; - if (window2 > 1) - { - if (!BN_mod_mul_montgomery(d,val2[0],val2[0],mont,ctx)) goto err; + /* Now compute the power product, using independent windows. */ + r_is_one = 1; + wvalue1 = 0; /* The 'value' of the first window */ + wvalue2 = 0; /* The 'value' of the second window */ + wpos1 = 0; /* If wvalue1 > 0, the bottom bit of the + * first window */ + wpos2 = 0; /* If wvalue2 > 0, the bottom bit of the + * second window */ - j=1<<(window2-1); - for (i=1; i<j; i++) - { - if(((val2[i] = BN_CTX_get(ctx)) == NULL) || - !BN_mod_mul_montgomery(val2[i],val2[i-1], - d,mont,ctx)) - goto err; - } - } + if (!BN_to_montgomery(r, BN_value_one(), mont, ctx)) + goto err; + for (b = bits - 1; b >= 0; b--) { + if (!r_is_one) { + if (!BN_mod_mul_montgomery(r, r, r, mont, ctx)) + goto err; + } + if (!wvalue1) + if (BN_is_bit_set(p1, b)) { + /* + * consider bits b-window1+1 .. b for this window + */ + i = b - window1 + 1; + while (!BN_is_bit_set(p1, i)) /* works for i<0 */ + i++; + wpos1 = i; + wvalue1 = 1; + for (i = b - 1; i >= wpos1; i--) { + wvalue1 <<= 1; + if (BN_is_bit_set(p1, i)) + wvalue1++; + } + } - /* Now compute the power product, using independent windows. */ - r_is_one=1; - wvalue1=0; /* The 'value' of the first window */ - wvalue2=0; /* The 'value' of the second window */ - wpos1=0; /* If wvalue1 > 0, the bottom bit of the first window */ - wpos2=0; /* If wvalue2 > 0, the bottom bit of the second window */ + if (!wvalue2) + if (BN_is_bit_set(p2, b)) { + /* + * consider bits b-window2+1 .. b for this window + */ + i = b - window2 + 1; + while (!BN_is_bit_set(p2, i)) + i++; + wpos2 = i; + wvalue2 = 1; + for (i = b - 1; i >= wpos2; i--) { + wvalue2 <<= 1; + if (BN_is_bit_set(p2, i)) + wvalue2++; + } + } - if (!BN_to_montgomery(r,BN_value_one(),mont,ctx)) goto err; - for (b=bits-1; b>=0; b--) - { - if (!r_is_one) - { - if (!BN_mod_mul_montgomery(r,r,r,mont,ctx)) - goto err; - } - - if (!wvalue1) - if (BN_is_bit_set(p1, b)) - { - /* consider bits b-window1+1 .. b for this window */ - i = b-window1+1; - while (!BN_is_bit_set(p1, i)) /* works for i<0 */ - i++; - wpos1 = i; - wvalue1 = 1; - for (i = b-1; i >= wpos1; i--) - { - wvalue1 <<= 1; - if (BN_is_bit_set(p1, i)) - wvalue1++; - } - } - - if (!wvalue2) - if (BN_is_bit_set(p2, b)) - { - /* consider bits b-window2+1 .. b for this window */ - i = b-window2+1; - while (!BN_is_bit_set(p2, i)) - i++; - wpos2 = i; - wvalue2 = 1; - for (i = b-1; i >= wpos2; i--) - { - wvalue2 <<= 1; - if (BN_is_bit_set(p2, i)) - wvalue2++; - } - } + if (wvalue1 && b == wpos1) { + /* wvalue1 is odd and < 2^window1 */ + if (!BN_mod_mul_montgomery(r, r, val1[wvalue1 >> 1], mont, ctx)) + goto err; + wvalue1 = 0; + r_is_one = 0; + } - if (wvalue1 && b == wpos1) - { - /* wvalue1 is odd and < 2^window1 */ - if (!BN_mod_mul_montgomery(r,r,val1[wvalue1>>1],mont,ctx)) - goto err; - wvalue1 = 0; - r_is_one = 0; - } - - if (wvalue2 && b == wpos2) - { - /* wvalue2 is odd and < 2^window2 */ - if (!BN_mod_mul_montgomery(r,r,val2[wvalue2>>1],mont,ctx)) - goto err; - wvalue2 = 0; - r_is_one = 0; - } - } - if (!BN_from_montgomery(rr,r,mont,ctx)) - goto err; - ret=1; -err: - if ((in_mont == NULL) && (mont != NULL)) BN_MONT_CTX_free(mont); - BN_CTX_end(ctx); - bn_check_top(rr); - return(ret); - } + if (wvalue2 && b == wpos2) { + /* wvalue2 is odd and < 2^window2 */ + if (!BN_mod_mul_montgomery(r, r, val2[wvalue2 >> 1], mont, ctx)) + goto err; + wvalue2 = 0; + r_is_one = 0; + } + } + if (!BN_from_montgomery(rr, r, mont, ctx)) + goto err; + ret = 1; + err: + if ((in_mont == NULL) && (mont != NULL)) + BN_MONT_CTX_free(mont); + BN_CTX_end(ctx); + bn_check_top(rr); + return (ret); +} diff --git a/openssl/crypto/bn/bn_gcd.c b/openssl/crypto/bn/bn_gcd.c index a808f5317..97c55ab72 100644 --- a/openssl/crypto/bn/bn_gcd.c +++ b/openssl/crypto/bn/bn_gcd.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -115,541 +115,586 @@ static BIGNUM *euclid(BIGNUM *a, BIGNUM *b); int BN_gcd(BIGNUM *r, const BIGNUM *in_a, const BIGNUM *in_b, BN_CTX *ctx) - { - BIGNUM *a,*b,*t; - int ret=0; - - bn_check_top(in_a); - bn_check_top(in_b); - - BN_CTX_start(ctx); - a = BN_CTX_get(ctx); - b = BN_CTX_get(ctx); - if (a == NULL || b == NULL) goto err; - - if (BN_copy(a,in_a) == NULL) goto err; - if (BN_copy(b,in_b) == NULL) goto err; - a->neg = 0; - b->neg = 0; - - if (BN_cmp(a,b) < 0) { t=a; a=b; b=t; } - t=euclid(a,b); - if (t == NULL) goto err; - - if (BN_copy(r,t) == NULL) goto err; - ret=1; -err: - BN_CTX_end(ctx); - bn_check_top(r); - return(ret); - } +{ + BIGNUM *a, *b, *t; + int ret = 0; + + bn_check_top(in_a); + bn_check_top(in_b); + + BN_CTX_start(ctx); + a = BN_CTX_get(ctx); + b = BN_CTX_get(ctx); + if (a == NULL || b == NULL) + goto err; + + if (BN_copy(a, in_a) == NULL) + goto err; + if (BN_copy(b, in_b) == NULL) + goto err; + a->neg = 0; + b->neg = 0; + + if (BN_cmp(a, b) < 0) { + t = a; + a = b; + b = t; + } + t = euclid(a, b); + if (t == NULL) + goto err; + + if (BN_copy(r, t) == NULL) + goto err; + ret = 1; + err: + BN_CTX_end(ctx); + bn_check_top(r); + return (ret); +} static BIGNUM *euclid(BIGNUM *a, BIGNUM *b) - { - BIGNUM *t; - int shifts=0; - - bn_check_top(a); - bn_check_top(b); - - /* 0 <= b <= a */ - while (!BN_is_zero(b)) - { - /* 0 < b <= a */ - - if (BN_is_odd(a)) - { - if (BN_is_odd(b)) - { - if (!BN_sub(a,a,b)) goto err; - if (!BN_rshift1(a,a)) goto err; - if (BN_cmp(a,b) < 0) - { t=a; a=b; b=t; } - } - else /* a odd - b even */ - { - if (!BN_rshift1(b,b)) goto err; - if (BN_cmp(a,b) < 0) - { t=a; a=b; b=t; } - } - } - else /* a is even */ - { - if (BN_is_odd(b)) - { - if (!BN_rshift1(a,a)) goto err; - if (BN_cmp(a,b) < 0) - { t=a; a=b; b=t; } - } - else /* a even - b even */ - { - if (!BN_rshift1(a,a)) goto err; - if (!BN_rshift1(b,b)) goto err; - shifts++; - } - } - /* 0 <= b <= a */ - } - - if (shifts) - { - if (!BN_lshift(a,a,shifts)) goto err; - } - bn_check_top(a); - return(a); -err: - return(NULL); - } - +{ + BIGNUM *t; + int shifts = 0; + + bn_check_top(a); + bn_check_top(b); + + /* 0 <= b <= a */ + while (!BN_is_zero(b)) { + /* 0 < b <= a */ + + if (BN_is_odd(a)) { + if (BN_is_odd(b)) { + if (!BN_sub(a, a, b)) + goto err; + if (!BN_rshift1(a, a)) + goto err; + if (BN_cmp(a, b) < 0) { + t = a; + a = b; + b = t; + } + } else { /* a odd - b even */ + + if (!BN_rshift1(b, b)) + goto err; + if (BN_cmp(a, b) < 0) { + t = a; + a = b; + b = t; + } + } + } else { /* a is even */ + + if (BN_is_odd(b)) { + if (!BN_rshift1(a, a)) + goto err; + if (BN_cmp(a, b) < 0) { + t = a; + a = b; + b = t; + } + } else { /* a even - b even */ + + if (!BN_rshift1(a, a)) + goto err; + if (!BN_rshift1(b, b)) + goto err; + shifts++; + } + } + /* 0 <= b <= a */ + } + + if (shifts) { + if (!BN_lshift(a, a, shifts)) + goto err; + } + bn_check_top(a); + return (a); + err: + return (NULL); +} /* solves ax == 1 (mod n) */ static BIGNUM *BN_mod_inverse_no_branch(BIGNUM *in, - const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx); + const BIGNUM *a, const BIGNUM *n, + BN_CTX *ctx); BIGNUM *BN_mod_inverse(BIGNUM *in, - const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx) - { - BIGNUM *A,*B,*X,*Y,*M,*D,*T,*R=NULL; - BIGNUM *ret=NULL; - int sign; - - if ((BN_get_flags(a, BN_FLG_CONSTTIME) != 0) || (BN_get_flags(n, BN_FLG_CONSTTIME) != 0)) - { - return BN_mod_inverse_no_branch(in, a, n, ctx); - } - - bn_check_top(a); - bn_check_top(n); - - BN_CTX_start(ctx); - A = BN_CTX_get(ctx); - B = BN_CTX_get(ctx); - X = BN_CTX_get(ctx); - D = BN_CTX_get(ctx); - M = BN_CTX_get(ctx); - Y = BN_CTX_get(ctx); - T = BN_CTX_get(ctx); - if (T == NULL) goto err; - - if (in == NULL) - R=BN_new(); - else - R=in; - if (R == NULL) goto err; - - BN_one(X); - BN_zero(Y); - if (BN_copy(B,a) == NULL) goto err; - if (BN_copy(A,n) == NULL) goto err; - A->neg = 0; - if (B->neg || (BN_ucmp(B, A) >= 0)) - { - if (!BN_nnmod(B, B, A, ctx)) goto err; - } - sign = -1; - /* From B = a mod |n|, A = |n| it follows that - * - * 0 <= B < A, - * -sign*X*a == B (mod |n|), - * sign*Y*a == A (mod |n|). - */ - - if (BN_is_odd(n) && (BN_num_bits(n) <= (BN_BITS <= 32 ? 450 : 2048))) - { - /* Binary inversion algorithm; requires odd modulus. - * This is faster than the general algorithm if the modulus - * is sufficiently small (about 400 .. 500 bits on 32-bit - * sytems, but much more on 64-bit systems) */ - int shift; - - while (!BN_is_zero(B)) - { - /* - * 0 < B < |n|, - * 0 < A <= |n|, - * (1) -sign*X*a == B (mod |n|), - * (2) sign*Y*a == A (mod |n|) - */ - - /* Now divide B by the maximum possible power of two in the integers, - * and divide X by the same value mod |n|. - * When we're done, (1) still holds. */ - shift = 0; - while (!BN_is_bit_set(B, shift)) /* note that 0 < B */ - { - shift++; - - if (BN_is_odd(X)) - { - if (!BN_uadd(X, X, n)) goto err; - } - /* now X is even, so we can easily divide it by two */ - if (!BN_rshift1(X, X)) goto err; - } - if (shift > 0) - { - if (!BN_rshift(B, B, shift)) goto err; - } - - - /* Same for A and Y. Afterwards, (2) still holds. */ - shift = 0; - while (!BN_is_bit_set(A, shift)) /* note that 0 < A */ - { - shift++; - - if (BN_is_odd(Y)) - { - if (!BN_uadd(Y, Y, n)) goto err; - } - /* now Y is even */ - if (!BN_rshift1(Y, Y)) goto err; - } - if (shift > 0) - { - if (!BN_rshift(A, A, shift)) goto err; - } - - - /* We still have (1) and (2). - * Both A and B are odd. - * The following computations ensure that - * - * 0 <= B < |n|, - * 0 < A < |n|, - * (1) -sign*X*a == B (mod |n|), - * (2) sign*Y*a == A (mod |n|), - * - * and that either A or B is even in the next iteration. - */ - if (BN_ucmp(B, A) >= 0) - { - /* -sign*(X + Y)*a == B - A (mod |n|) */ - if (!BN_uadd(X, X, Y)) goto err; - /* NB: we could use BN_mod_add_quick(X, X, Y, n), but that - * actually makes the algorithm slower */ - if (!BN_usub(B, B, A)) goto err; - } - else - { - /* sign*(X + Y)*a == A - B (mod |n|) */ - if (!BN_uadd(Y, Y, X)) goto err; - /* as above, BN_mod_add_quick(Y, Y, X, n) would slow things down */ - if (!BN_usub(A, A, B)) goto err; - } - } - } - else - { - /* general inversion algorithm */ - - while (!BN_is_zero(B)) - { - BIGNUM *tmp; - - /* - * 0 < B < A, - * (*) -sign*X*a == B (mod |n|), - * sign*Y*a == A (mod |n|) - */ - - /* (D, M) := (A/B, A%B) ... */ - if (BN_num_bits(A) == BN_num_bits(B)) - { - if (!BN_one(D)) goto err; - if (!BN_sub(M,A,B)) goto err; - } - else if (BN_num_bits(A) == BN_num_bits(B) + 1) - { - /* A/B is 1, 2, or 3 */ - if (!BN_lshift1(T,B)) goto err; - if (BN_ucmp(A,T) < 0) - { - /* A < 2*B, so D=1 */ - if (!BN_one(D)) goto err; - if (!BN_sub(M,A,B)) goto err; - } - else - { - /* A >= 2*B, so D=2 or D=3 */ - if (!BN_sub(M,A,T)) goto err; - if (!BN_add(D,T,B)) goto err; /* use D (:= 3*B) as temp */ - if (BN_ucmp(A,D) < 0) - { - /* A < 3*B, so D=2 */ - if (!BN_set_word(D,2)) goto err; - /* M (= A - 2*B) already has the correct value */ - } - else - { - /* only D=3 remains */ - if (!BN_set_word(D,3)) goto err; - /* currently M = A - 2*B, but we need M = A - 3*B */ - if (!BN_sub(M,M,B)) goto err; - } - } - } - else - { - if (!BN_div(D,M,A,B,ctx)) goto err; - } - - /* Now - * A = D*B + M; - * thus we have - * (**) sign*Y*a == D*B + M (mod |n|). - */ - - tmp=A; /* keep the BIGNUM object, the value does not matter */ - - /* (A, B) := (B, A mod B) ... */ - A=B; - B=M; - /* ... so we have 0 <= B < A again */ - - /* Since the former M is now B and the former B is now A, - * (**) translates into - * sign*Y*a == D*A + B (mod |n|), - * i.e. - * sign*Y*a - D*A == B (mod |n|). - * Similarly, (*) translates into - * -sign*X*a == A (mod |n|). - * - * Thus, - * sign*Y*a + D*sign*X*a == B (mod |n|), - * i.e. - * sign*(Y + D*X)*a == B (mod |n|). - * - * So if we set (X, Y, sign) := (Y + D*X, X, -sign), we arrive back at - * -sign*X*a == B (mod |n|), - * sign*Y*a == A (mod |n|). - * Note that X and Y stay non-negative all the time. - */ - - /* most of the time D is very small, so we can optimize tmp := D*X+Y */ - if (BN_is_one(D)) - { - if (!BN_add(tmp,X,Y)) goto err; - } - else - { - if (BN_is_word(D,2)) - { - if (!BN_lshift1(tmp,X)) goto err; - } - else if (BN_is_word(D,4)) - { - if (!BN_lshift(tmp,X,2)) goto err; - } - else if (D->top == 1) - { - if (!BN_copy(tmp,X)) goto err; - if (!BN_mul_word(tmp,D->d[0])) goto err; - } - else - { - if (!BN_mul(tmp,D,X,ctx)) goto err; - } - if (!BN_add(tmp,tmp,Y)) goto err; - } - - M=Y; /* keep the BIGNUM object, the value does not matter */ - Y=X; - X=tmp; - sign = -sign; - } - } - - /* - * The while loop (Euclid's algorithm) ends when - * A == gcd(a,n); - * we have - * sign*Y*a == A (mod |n|), - * where Y is non-negative. - */ - - if (sign < 0) - { - if (!BN_sub(Y,n,Y)) goto err; - } - /* Now Y*a == A (mod |n|). */ - - - if (BN_is_one(A)) - { - /* Y*a == 1 (mod |n|) */ - if (!Y->neg && BN_ucmp(Y,n) < 0) - { - if (!BN_copy(R,Y)) goto err; - } - else - { - if (!BN_nnmod(R,Y,n,ctx)) goto err; - } - } - else - { - BNerr(BN_F_BN_MOD_INVERSE,BN_R_NO_INVERSE); - goto err; - } - ret=R; -err: - if ((ret == NULL) && (in == NULL)) BN_free(R); - BN_CTX_end(ctx); - bn_check_top(ret); - return(ret); - } - - -/* BN_mod_inverse_no_branch is a special version of BN_mod_inverse. - * It does not contain branches that may leak sensitive information. + const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx) +{ + BIGNUM *A, *B, *X, *Y, *M, *D, *T, *R = NULL; + BIGNUM *ret = NULL; + int sign; + + if ((BN_get_flags(a, BN_FLG_CONSTTIME) != 0) + || (BN_get_flags(n, BN_FLG_CONSTTIME) != 0)) { + return BN_mod_inverse_no_branch(in, a, n, ctx); + } + + bn_check_top(a); + bn_check_top(n); + + BN_CTX_start(ctx); + A = BN_CTX_get(ctx); + B = BN_CTX_get(ctx); + X = BN_CTX_get(ctx); + D = BN_CTX_get(ctx); + M = BN_CTX_get(ctx); + Y = BN_CTX_get(ctx); + T = BN_CTX_get(ctx); + if (T == NULL) + goto err; + + if (in == NULL) + R = BN_new(); + else + R = in; + if (R == NULL) + goto err; + + BN_one(X); + BN_zero(Y); + if (BN_copy(B, a) == NULL) + goto err; + if (BN_copy(A, n) == NULL) + goto err; + A->neg = 0; + if (B->neg || (BN_ucmp(B, A) >= 0)) { + if (!BN_nnmod(B, B, A, ctx)) + goto err; + } + sign = -1; + /*- + * From B = a mod |n|, A = |n| it follows that + * + * 0 <= B < A, + * -sign*X*a == B (mod |n|), + * sign*Y*a == A (mod |n|). + */ + + if (BN_is_odd(n) && (BN_num_bits(n) <= (BN_BITS <= 32 ? 450 : 2048))) { + /* + * Binary inversion algorithm; requires odd modulus. This is faster + * than the general algorithm if the modulus is sufficiently small + * (about 400 .. 500 bits on 32-bit sytems, but much more on 64-bit + * systems) + */ + int shift; + + while (!BN_is_zero(B)) { + /*- + * 0 < B < |n|, + * 0 < A <= |n|, + * (1) -sign*X*a == B (mod |n|), + * (2) sign*Y*a == A (mod |n|) + */ + + /* + * Now divide B by the maximum possible power of two in the + * integers, and divide X by the same value mod |n|. When we're + * done, (1) still holds. + */ + shift = 0; + while (!BN_is_bit_set(B, shift)) { /* note that 0 < B */ + shift++; + + if (BN_is_odd(X)) { + if (!BN_uadd(X, X, n)) + goto err; + } + /* + * now X is even, so we can easily divide it by two + */ + if (!BN_rshift1(X, X)) + goto err; + } + if (shift > 0) { + if (!BN_rshift(B, B, shift)) + goto err; + } + + /* + * Same for A and Y. Afterwards, (2) still holds. + */ + shift = 0; + while (!BN_is_bit_set(A, shift)) { /* note that 0 < A */ + shift++; + + if (BN_is_odd(Y)) { + if (!BN_uadd(Y, Y, n)) + goto err; + } + /* now Y is even */ + if (!BN_rshift1(Y, Y)) + goto err; + } + if (shift > 0) { + if (!BN_rshift(A, A, shift)) + goto err; + } + + /*- + * We still have (1) and (2). + * Both A and B are odd. + * The following computations ensure that + * + * 0 <= B < |n|, + * 0 < A < |n|, + * (1) -sign*X*a == B (mod |n|), + * (2) sign*Y*a == A (mod |n|), + * + * and that either A or B is even in the next iteration. + */ + if (BN_ucmp(B, A) >= 0) { + /* -sign*(X + Y)*a == B - A (mod |n|) */ + if (!BN_uadd(X, X, Y)) + goto err; + /* + * NB: we could use BN_mod_add_quick(X, X, Y, n), but that + * actually makes the algorithm slower + */ + if (!BN_usub(B, B, A)) + goto err; + } else { + /* sign*(X + Y)*a == A - B (mod |n|) */ + if (!BN_uadd(Y, Y, X)) + goto err; + /* + * as above, BN_mod_add_quick(Y, Y, X, n) would slow things + * down + */ + if (!BN_usub(A, A, B)) + goto err; + } + } + } else { + /* general inversion algorithm */ + + while (!BN_is_zero(B)) { + BIGNUM *tmp; + + /*- + * 0 < B < A, + * (*) -sign*X*a == B (mod |n|), + * sign*Y*a == A (mod |n|) + */ + + /* (D, M) := (A/B, A%B) ... */ + if (BN_num_bits(A) == BN_num_bits(B)) { + if (!BN_one(D)) + goto err; + if (!BN_sub(M, A, B)) + goto err; + } else if (BN_num_bits(A) == BN_num_bits(B) + 1) { + /* A/B is 1, 2, or 3 */ + if (!BN_lshift1(T, B)) + goto err; + if (BN_ucmp(A, T) < 0) { + /* A < 2*B, so D=1 */ + if (!BN_one(D)) + goto err; + if (!BN_sub(M, A, B)) + goto err; + } else { + /* A >= 2*B, so D=2 or D=3 */ + if (!BN_sub(M, A, T)) + goto err; + if (!BN_add(D, T, B)) + goto err; /* use D (:= 3*B) as temp */ + if (BN_ucmp(A, D) < 0) { + /* A < 3*B, so D=2 */ + if (!BN_set_word(D, 2)) + goto err; + /* + * M (= A - 2*B) already has the correct value + */ + } else { + /* only D=3 remains */ + if (!BN_set_word(D, 3)) + goto err; + /* + * currently M = A - 2*B, but we need M = A - 3*B + */ + if (!BN_sub(M, M, B)) + goto err; + } + } + } else { + if (!BN_div(D, M, A, B, ctx)) + goto err; + } + + /*- + * Now + * A = D*B + M; + * thus we have + * (**) sign*Y*a == D*B + M (mod |n|). + */ + + tmp = A; /* keep the BIGNUM object, the value does not + * matter */ + + /* (A, B) := (B, A mod B) ... */ + A = B; + B = M; + /* ... so we have 0 <= B < A again */ + + /*- + * Since the former M is now B and the former B is now A, + * (**) translates into + * sign*Y*a == D*A + B (mod |n|), + * i.e. + * sign*Y*a - D*A == B (mod |n|). + * Similarly, (*) translates into + * -sign*X*a == A (mod |n|). + * + * Thus, + * sign*Y*a + D*sign*X*a == B (mod |n|), + * i.e. + * sign*(Y + D*X)*a == B (mod |n|). + * + * So if we set (X, Y, sign) := (Y + D*X, X, -sign), we arrive back at + * -sign*X*a == B (mod |n|), + * sign*Y*a == A (mod |n|). + * Note that X and Y stay non-negative all the time. + */ + + /* + * most of the time D is very small, so we can optimize tmp := + * D*X+Y + */ + if (BN_is_one(D)) { + if (!BN_add(tmp, X, Y)) + goto err; + } else { + if (BN_is_word(D, 2)) { + if (!BN_lshift1(tmp, X)) + goto err; + } else if (BN_is_word(D, 4)) { + if (!BN_lshift(tmp, X, 2)) + goto err; + } else if (D->top == 1) { + if (!BN_copy(tmp, X)) + goto err; + if (!BN_mul_word(tmp, D->d[0])) + goto err; + } else { + if (!BN_mul(tmp, D, X, ctx)) + goto err; + } + if (!BN_add(tmp, tmp, Y)) + goto err; + } + + M = Y; /* keep the BIGNUM object, the value does not + * matter */ + Y = X; + X = tmp; + sign = -sign; + } + } + + /*- + * The while loop (Euclid's algorithm) ends when + * A == gcd(a,n); + * we have + * sign*Y*a == A (mod |n|), + * where Y is non-negative. + */ + + if (sign < 0) { + if (!BN_sub(Y, n, Y)) + goto err; + } + /* Now Y*a == A (mod |n|). */ + + if (BN_is_one(A)) { + /* Y*a == 1 (mod |n|) */ + if (!Y->neg && BN_ucmp(Y, n) < 0) { + if (!BN_copy(R, Y)) + goto err; + } else { + if (!BN_nnmod(R, Y, n, ctx)) + goto err; + } + } else { + BNerr(BN_F_BN_MOD_INVERSE, BN_R_NO_INVERSE); + goto err; + } + ret = R; + err: + if ((ret == NULL) && (in == NULL)) + BN_free(R); + BN_CTX_end(ctx); + bn_check_top(ret); + return (ret); +} + +/* + * BN_mod_inverse_no_branch is a special version of BN_mod_inverse. It does + * not contain branches that may leak sensitive information. */ static BIGNUM *BN_mod_inverse_no_branch(BIGNUM *in, - const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx) - { - BIGNUM *A,*B,*X,*Y,*M,*D,*T,*R=NULL; - BIGNUM local_A, local_B; - BIGNUM *pA, *pB; - BIGNUM *ret=NULL; - int sign; - - bn_check_top(a); - bn_check_top(n); - - BN_CTX_start(ctx); - A = BN_CTX_get(ctx); - B = BN_CTX_get(ctx); - X = BN_CTX_get(ctx); - D = BN_CTX_get(ctx); - M = BN_CTX_get(ctx); - Y = BN_CTX_get(ctx); - T = BN_CTX_get(ctx); - if (T == NULL) goto err; - - if (in == NULL) - R=BN_new(); - else - R=in; - if (R == NULL) goto err; - - BN_one(X); - BN_zero(Y); - if (BN_copy(B,a) == NULL) goto err; - if (BN_copy(A,n) == NULL) goto err; - A->neg = 0; - - if (B->neg || (BN_ucmp(B, A) >= 0)) - { - /* Turn BN_FLG_CONSTTIME flag on, so that when BN_div is invoked, - * BN_div_no_branch will be called eventually. - */ - pB = &local_B; - BN_with_flags(pB, B, BN_FLG_CONSTTIME); - if (!BN_nnmod(B, pB, A, ctx)) goto err; - } - sign = -1; - /* From B = a mod |n|, A = |n| it follows that - * - * 0 <= B < A, - * -sign*X*a == B (mod |n|), - * sign*Y*a == A (mod |n|). - */ - - while (!BN_is_zero(B)) - { - BIGNUM *tmp; - - /* - * 0 < B < A, - * (*) -sign*X*a == B (mod |n|), - * sign*Y*a == A (mod |n|) - */ - - /* Turn BN_FLG_CONSTTIME flag on, so that when BN_div is invoked, - * BN_div_no_branch will be called eventually. - */ - pA = &local_A; - BN_with_flags(pA, A, BN_FLG_CONSTTIME); - - /* (D, M) := (A/B, A%B) ... */ - if (!BN_div(D,M,pA,B,ctx)) goto err; - - /* Now - * A = D*B + M; - * thus we have - * (**) sign*Y*a == D*B + M (mod |n|). - */ - - tmp=A; /* keep the BIGNUM object, the value does not matter */ - - /* (A, B) := (B, A mod B) ... */ - A=B; - B=M; - /* ... so we have 0 <= B < A again */ - - /* Since the former M is now B and the former B is now A, - * (**) translates into - * sign*Y*a == D*A + B (mod |n|), - * i.e. - * sign*Y*a - D*A == B (mod |n|). - * Similarly, (*) translates into - * -sign*X*a == A (mod |n|). - * - * Thus, - * sign*Y*a + D*sign*X*a == B (mod |n|), - * i.e. - * sign*(Y + D*X)*a == B (mod |n|). - * - * So if we set (X, Y, sign) := (Y + D*X, X, -sign), we arrive back at - * -sign*X*a == B (mod |n|), - * sign*Y*a == A (mod |n|). - * Note that X and Y stay non-negative all the time. - */ - - if (!BN_mul(tmp,D,X,ctx)) goto err; - if (!BN_add(tmp,tmp,Y)) goto err; - - M=Y; /* keep the BIGNUM object, the value does not matter */ - Y=X; - X=tmp; - sign = -sign; - } - - /* - * The while loop (Euclid's algorithm) ends when - * A == gcd(a,n); - * we have - * sign*Y*a == A (mod |n|), - * where Y is non-negative. - */ - - if (sign < 0) - { - if (!BN_sub(Y,n,Y)) goto err; - } - /* Now Y*a == A (mod |n|). */ - - if (BN_is_one(A)) - { - /* Y*a == 1 (mod |n|) */ - if (!Y->neg && BN_ucmp(Y,n) < 0) - { - if (!BN_copy(R,Y)) goto err; - } - else - { - if (!BN_nnmod(R,Y,n,ctx)) goto err; - } - } - else - { - BNerr(BN_F_BN_MOD_INVERSE_NO_BRANCH,BN_R_NO_INVERSE); - goto err; - } - ret=R; -err: - if ((ret == NULL) && (in == NULL)) BN_free(R); - BN_CTX_end(ctx); - bn_check_top(ret); - return(ret); - } + const BIGNUM *a, const BIGNUM *n, + BN_CTX *ctx) +{ + BIGNUM *A, *B, *X, *Y, *M, *D, *T, *R = NULL; + BIGNUM local_A, local_B; + BIGNUM *pA, *pB; + BIGNUM *ret = NULL; + int sign; + + bn_check_top(a); + bn_check_top(n); + + BN_CTX_start(ctx); + A = BN_CTX_get(ctx); + B = BN_CTX_get(ctx); + X = BN_CTX_get(ctx); + D = BN_CTX_get(ctx); + M = BN_CTX_get(ctx); + Y = BN_CTX_get(ctx); + T = BN_CTX_get(ctx); + if (T == NULL) + goto err; + + if (in == NULL) + R = BN_new(); + else + R = in; + if (R == NULL) + goto err; + + BN_one(X); + BN_zero(Y); + if (BN_copy(B, a) == NULL) + goto err; + if (BN_copy(A, n) == NULL) + goto err; + A->neg = 0; + + if (B->neg || (BN_ucmp(B, A) >= 0)) { + /* + * Turn BN_FLG_CONSTTIME flag on, so that when BN_div is invoked, + * BN_div_no_branch will be called eventually. + */ + pB = &local_B; + BN_with_flags(pB, B, BN_FLG_CONSTTIME); + if (!BN_nnmod(B, pB, A, ctx)) + goto err; + } + sign = -1; + /*- + * From B = a mod |n|, A = |n| it follows that + * + * 0 <= B < A, + * -sign*X*a == B (mod |n|), + * sign*Y*a == A (mod |n|). + */ + + while (!BN_is_zero(B)) { + BIGNUM *tmp; + + /*- + * 0 < B < A, + * (*) -sign*X*a == B (mod |n|), + * sign*Y*a == A (mod |n|) + */ + + /* + * Turn BN_FLG_CONSTTIME flag on, so that when BN_div is invoked, + * BN_div_no_branch will be called eventually. + */ + pA = &local_A; + BN_with_flags(pA, A, BN_FLG_CONSTTIME); + + /* (D, M) := (A/B, A%B) ... */ + if (!BN_div(D, M, pA, B, ctx)) + goto err; + + /*- + * Now + * A = D*B + M; + * thus we have + * (**) sign*Y*a == D*B + M (mod |n|). + */ + + tmp = A; /* keep the BIGNUM object, the value does not + * matter */ + + /* (A, B) := (B, A mod B) ... */ + A = B; + B = M; + /* ... so we have 0 <= B < A again */ + + /*- + * Since the former M is now B and the former B is now A, + * (**) translates into + * sign*Y*a == D*A + B (mod |n|), + * i.e. + * sign*Y*a - D*A == B (mod |n|). + * Similarly, (*) translates into + * -sign*X*a == A (mod |n|). + * + * Thus, + * sign*Y*a + D*sign*X*a == B (mod |n|), + * i.e. + * sign*(Y + D*X)*a == B (mod |n|). + * + * So if we set (X, Y, sign) := (Y + D*X, X, -sign), we arrive back at + * -sign*X*a == B (mod |n|), + * sign*Y*a == A (mod |n|). + * Note that X and Y stay non-negative all the time. + */ + + if (!BN_mul(tmp, D, X, ctx)) + goto err; + if (!BN_add(tmp, tmp, Y)) + goto err; + + M = Y; /* keep the BIGNUM object, the value does not + * matter */ + Y = X; + X = tmp; + sign = -sign; + } + + /*- + * The while loop (Euclid's algorithm) ends when + * A == gcd(a,n); + * we have + * sign*Y*a == A (mod |n|), + * where Y is non-negative. + */ + + if (sign < 0) { + if (!BN_sub(Y, n, Y)) + goto err; + } + /* Now Y*a == A (mod |n|). */ + + if (BN_is_one(A)) { + /* Y*a == 1 (mod |n|) */ + if (!Y->neg && BN_ucmp(Y, n) < 0) { + if (!BN_copy(R, Y)) + goto err; + } else { + if (!BN_nnmod(R, Y, n, ctx)) + goto err; + } + } else { + BNerr(BN_F_BN_MOD_INVERSE_NO_BRANCH, BN_R_NO_INVERSE); + goto err; + } + ret = R; + err: + if ((ret == NULL) && (in == NULL)) + BN_free(R); + BN_CTX_end(ctx); + bn_check_top(ret); + return (ret); +} diff --git a/openssl/crypto/bn/bn_gf2m.c b/openssl/crypto/bn/bn_gf2m.c index 8a4dc20ad..aeee49a01 100644 --- a/openssl/crypto/bn/bn_gf2m.c +++ b/openssl/crypto/bn/bn_gf2m.c @@ -27,12 +27,13 @@ * */ -/* NOTE: This file is licensed pursuant to the OpenSSL license below - * and may be modified; but after modifications, the above covenant - * may no longer apply! In such cases, the corresponding paragraph - * ["In addition, Sun covenants ... causes the infringement."] and - * this note can be edited out; but please keep the Sun copyright - * notice and attribution. */ +/* + * NOTE: This file is licensed pursuant to the OpenSSL license below and may + * be modified; but after modifications, the above covenant may no longer + * apply! In such cases, the corresponding paragraph ["In addition, Sun + * covenants ... causes the infringement."] and this note can be edited out; + * but please keep the Sun copyright notice and attribution. + */ /* ==================================================================== * Copyright (c) 1998-2002 The OpenSSL Project. All rights reserved. @@ -42,7 +43,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -96,1018 +97,1197 @@ #ifndef OPENSSL_NO_EC2M -/* Maximum number of iterations before BN_GF2m_mod_solve_quad_arr should fail. */ -#define MAX_ITERATIONS 50 +/* + * Maximum number of iterations before BN_GF2m_mod_solve_quad_arr should + * fail. + */ +# define MAX_ITERATIONS 50 + +static const BN_ULONG SQR_tb[16] = { 0, 1, 4, 5, 16, 17, 20, 21, + 64, 65, 68, 69, 80, 81, 84, 85 +}; -static const BN_ULONG SQR_tb[16] = - { 0, 1, 4, 5, 16, 17, 20, 21, - 64, 65, 68, 69, 80, 81, 84, 85 }; /* Platform-specific macros to accelerate squaring. */ -#if defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG) -#define SQR1(w) \ +# if defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG) +# define SQR1(w) \ SQR_tb[(w) >> 60 & 0xF] << 56 | SQR_tb[(w) >> 56 & 0xF] << 48 | \ SQR_tb[(w) >> 52 & 0xF] << 40 | SQR_tb[(w) >> 48 & 0xF] << 32 | \ SQR_tb[(w) >> 44 & 0xF] << 24 | SQR_tb[(w) >> 40 & 0xF] << 16 | \ SQR_tb[(w) >> 36 & 0xF] << 8 | SQR_tb[(w) >> 32 & 0xF] -#define SQR0(w) \ +# define SQR0(w) \ SQR_tb[(w) >> 28 & 0xF] << 56 | SQR_tb[(w) >> 24 & 0xF] << 48 | \ SQR_tb[(w) >> 20 & 0xF] << 40 | SQR_tb[(w) >> 16 & 0xF] << 32 | \ SQR_tb[(w) >> 12 & 0xF] << 24 | SQR_tb[(w) >> 8 & 0xF] << 16 | \ SQR_tb[(w) >> 4 & 0xF] << 8 | SQR_tb[(w) & 0xF] -#endif -#ifdef THIRTY_TWO_BIT -#define SQR1(w) \ +# endif +# ifdef THIRTY_TWO_BIT +# define SQR1(w) \ SQR_tb[(w) >> 28 & 0xF] << 24 | SQR_tb[(w) >> 24 & 0xF] << 16 | \ SQR_tb[(w) >> 20 & 0xF] << 8 | SQR_tb[(w) >> 16 & 0xF] -#define SQR0(w) \ +# define SQR0(w) \ SQR_tb[(w) >> 12 & 0xF] << 24 | SQR_tb[(w) >> 8 & 0xF] << 16 | \ SQR_tb[(w) >> 4 & 0xF] << 8 | SQR_tb[(w) & 0xF] -#endif +# endif -#if !defined(OPENSSL_BN_ASM_GF2m) -/* Product of two polynomials a, b each with degree < BN_BITS2 - 1, - * result is a polynomial r with degree < 2 * BN_BITS - 1 - * The caller MUST ensure that the variables have the right amount - * of space allocated. +# if !defined(OPENSSL_BN_ASM_GF2m) +/* + * Product of two polynomials a, b each with degree < BN_BITS2 - 1, result is + * a polynomial r with degree < 2 * BN_BITS - 1 The caller MUST ensure that + * the variables have the right amount of space allocated. */ -#ifdef THIRTY_TWO_BIT -static void bn_GF2m_mul_1x1(BN_ULONG *r1, BN_ULONG *r0, const BN_ULONG a, const BN_ULONG b) - { - register BN_ULONG h, l, s; - BN_ULONG tab[8], top2b = a >> 30; - register BN_ULONG a1, a2, a4; - - a1 = a & (0x3FFFFFFF); a2 = a1 << 1; a4 = a2 << 1; - - tab[0] = 0; tab[1] = a1; tab[2] = a2; tab[3] = a1^a2; - tab[4] = a4; tab[5] = a1^a4; tab[6] = a2^a4; tab[7] = a1^a2^a4; - - s = tab[b & 0x7]; l = s; - s = tab[b >> 3 & 0x7]; l ^= s << 3; h = s >> 29; - s = tab[b >> 6 & 0x7]; l ^= s << 6; h ^= s >> 26; - s = tab[b >> 9 & 0x7]; l ^= s << 9; h ^= s >> 23; - s = tab[b >> 12 & 0x7]; l ^= s << 12; h ^= s >> 20; - s = tab[b >> 15 & 0x7]; l ^= s << 15; h ^= s >> 17; - s = tab[b >> 18 & 0x7]; l ^= s << 18; h ^= s >> 14; - s = tab[b >> 21 & 0x7]; l ^= s << 21; h ^= s >> 11; - s = tab[b >> 24 & 0x7]; l ^= s << 24; h ^= s >> 8; - s = tab[b >> 27 & 0x7]; l ^= s << 27; h ^= s >> 5; - s = tab[b >> 30 ]; l ^= s << 30; h ^= s >> 2; - - /* compensate for the top two bits of a */ - - if (top2b & 01) { l ^= b << 30; h ^= b >> 2; } - if (top2b & 02) { l ^= b << 31; h ^= b >> 1; } - - *r1 = h; *r0 = l; - } -#endif -#if defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG) -static void bn_GF2m_mul_1x1(BN_ULONG *r1, BN_ULONG *r0, const BN_ULONG a, const BN_ULONG b) - { - register BN_ULONG h, l, s; - BN_ULONG tab[16], top3b = a >> 61; - register BN_ULONG a1, a2, a4, a8; - - a1 = a & (0x1FFFFFFFFFFFFFFFULL); a2 = a1 << 1; a4 = a2 << 1; a8 = a4 << 1; - - tab[ 0] = 0; tab[ 1] = a1; tab[ 2] = a2; tab[ 3] = a1^a2; - tab[ 4] = a4; tab[ 5] = a1^a4; tab[ 6] = a2^a4; tab[ 7] = a1^a2^a4; - tab[ 8] = a8; tab[ 9] = a1^a8; tab[10] = a2^a8; tab[11] = a1^a2^a8; - tab[12] = a4^a8; tab[13] = a1^a4^a8; tab[14] = a2^a4^a8; tab[15] = a1^a2^a4^a8; - - s = tab[b & 0xF]; l = s; - s = tab[b >> 4 & 0xF]; l ^= s << 4; h = s >> 60; - s = tab[b >> 8 & 0xF]; l ^= s << 8; h ^= s >> 56; - s = tab[b >> 12 & 0xF]; l ^= s << 12; h ^= s >> 52; - s = tab[b >> 16 & 0xF]; l ^= s << 16; h ^= s >> 48; - s = tab[b >> 20 & 0xF]; l ^= s << 20; h ^= s >> 44; - s = tab[b >> 24 & 0xF]; l ^= s << 24; h ^= s >> 40; - s = tab[b >> 28 & 0xF]; l ^= s << 28; h ^= s >> 36; - s = tab[b >> 32 & 0xF]; l ^= s << 32; h ^= s >> 32; - s = tab[b >> 36 & 0xF]; l ^= s << 36; h ^= s >> 28; - s = tab[b >> 40 & 0xF]; l ^= s << 40; h ^= s >> 24; - s = tab[b >> 44 & 0xF]; l ^= s << 44; h ^= s >> 20; - s = tab[b >> 48 & 0xF]; l ^= s << 48; h ^= s >> 16; - s = tab[b >> 52 & 0xF]; l ^= s << 52; h ^= s >> 12; - s = tab[b >> 56 & 0xF]; l ^= s << 56; h ^= s >> 8; - s = tab[b >> 60 ]; l ^= s << 60; h ^= s >> 4; - - /* compensate for the top three bits of a */ - - if (top3b & 01) { l ^= b << 61; h ^= b >> 3; } - if (top3b & 02) { l ^= b << 62; h ^= b >> 2; } - if (top3b & 04) { l ^= b << 63; h ^= b >> 1; } - - *r1 = h; *r0 = l; - } -#endif - -/* Product of two polynomials a, b each with degree < 2 * BN_BITS2 - 1, - * result is a polynomial r with degree < 4 * BN_BITS2 - 1 - * The caller MUST ensure that the variables have the right amount - * of space allocated. +# ifdef THIRTY_TWO_BIT +static void bn_GF2m_mul_1x1(BN_ULONG *r1, BN_ULONG *r0, const BN_ULONG a, + const BN_ULONG b) +{ + register BN_ULONG h, l, s; + BN_ULONG tab[8], top2b = a >> 30; + register BN_ULONG a1, a2, a4; + + a1 = a & (0x3FFFFFFF); + a2 = a1 << 1; + a4 = a2 << 1; + + tab[0] = 0; + tab[1] = a1; + tab[2] = a2; + tab[3] = a1 ^ a2; + tab[4] = a4; + tab[5] = a1 ^ a4; + tab[6] = a2 ^ a4; + tab[7] = a1 ^ a2 ^ a4; + + s = tab[b & 0x7]; + l = s; + s = tab[b >> 3 & 0x7]; + l ^= s << 3; + h = s >> 29; + s = tab[b >> 6 & 0x7]; + l ^= s << 6; + h ^= s >> 26; + s = tab[b >> 9 & 0x7]; + l ^= s << 9; + h ^= s >> 23; + s = tab[b >> 12 & 0x7]; + l ^= s << 12; + h ^= s >> 20; + s = tab[b >> 15 & 0x7]; + l ^= s << 15; + h ^= s >> 17; + s = tab[b >> 18 & 0x7]; + l ^= s << 18; + h ^= s >> 14; + s = tab[b >> 21 & 0x7]; + l ^= s << 21; + h ^= s >> 11; + s = tab[b >> 24 & 0x7]; + l ^= s << 24; + h ^= s >> 8; + s = tab[b >> 27 & 0x7]; + l ^= s << 27; + h ^= s >> 5; + s = tab[b >> 30]; + l ^= s << 30; + h ^= s >> 2; + + /* compensate for the top two bits of a */ + + if (top2b & 01) { + l ^= b << 30; + h ^= b >> 2; + } + if (top2b & 02) { + l ^= b << 31; + h ^= b >> 1; + } + + *r1 = h; + *r0 = l; +} +# endif +# if defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG) +static void bn_GF2m_mul_1x1(BN_ULONG *r1, BN_ULONG *r0, const BN_ULONG a, + const BN_ULONG b) +{ + register BN_ULONG h, l, s; + BN_ULONG tab[16], top3b = a >> 61; + register BN_ULONG a1, a2, a4, a8; + + a1 = a & (0x1FFFFFFFFFFFFFFFULL); + a2 = a1 << 1; + a4 = a2 << 1; + a8 = a4 << 1; + + tab[0] = 0; + tab[1] = a1; + tab[2] = a2; + tab[3] = a1 ^ a2; + tab[4] = a4; + tab[5] = a1 ^ a4; + tab[6] = a2 ^ a4; + tab[7] = a1 ^ a2 ^ a4; + tab[8] = a8; + tab[9] = a1 ^ a8; + tab[10] = a2 ^ a8; + tab[11] = a1 ^ a2 ^ a8; + tab[12] = a4 ^ a8; + tab[13] = a1 ^ a4 ^ a8; + tab[14] = a2 ^ a4 ^ a8; + tab[15] = a1 ^ a2 ^ a4 ^ a8; + + s = tab[b & 0xF]; + l = s; + s = tab[b >> 4 & 0xF]; + l ^= s << 4; + h = s >> 60; + s = tab[b >> 8 & 0xF]; + l ^= s << 8; + h ^= s >> 56; + s = tab[b >> 12 & 0xF]; + l ^= s << 12; + h ^= s >> 52; + s = tab[b >> 16 & 0xF]; + l ^= s << 16; + h ^= s >> 48; + s = tab[b >> 20 & 0xF]; + l ^= s << 20; + h ^= s >> 44; + s = tab[b >> 24 & 0xF]; + l ^= s << 24; + h ^= s >> 40; + s = tab[b >> 28 & 0xF]; + l ^= s << 28; + h ^= s >> 36; + s = tab[b >> 32 & 0xF]; + l ^= s << 32; + h ^= s >> 32; + s = tab[b >> 36 & 0xF]; + l ^= s << 36; + h ^= s >> 28; + s = tab[b >> 40 & 0xF]; + l ^= s << 40; + h ^= s >> 24; + s = tab[b >> 44 & 0xF]; + l ^= s << 44; + h ^= s >> 20; + s = tab[b >> 48 & 0xF]; + l ^= s << 48; + h ^= s >> 16; + s = tab[b >> 52 & 0xF]; + l ^= s << 52; + h ^= s >> 12; + s = tab[b >> 56 & 0xF]; + l ^= s << 56; + h ^= s >> 8; + s = tab[b >> 60]; + l ^= s << 60; + h ^= s >> 4; + + /* compensate for the top three bits of a */ + + if (top3b & 01) { + l ^= b << 61; + h ^= b >> 3; + } + if (top3b & 02) { + l ^= b << 62; + h ^= b >> 2; + } + if (top3b & 04) { + l ^= b << 63; + h ^= b >> 1; + } + + *r1 = h; + *r0 = l; +} +# endif + +/* + * Product of two polynomials a, b each with degree < 2 * BN_BITS2 - 1, + * result is a polynomial r with degree < 4 * BN_BITS2 - 1 The caller MUST + * ensure that the variables have the right amount of space allocated. */ -static void bn_GF2m_mul_2x2(BN_ULONG *r, const BN_ULONG a1, const BN_ULONG a0, const BN_ULONG b1, const BN_ULONG b0) - { - BN_ULONG m1, m0; - /* r[3] = h1, r[2] = h0; r[1] = l1; r[0] = l0 */ - bn_GF2m_mul_1x1(r+3, r+2, a1, b1); - bn_GF2m_mul_1x1(r+1, r, a0, b0); - bn_GF2m_mul_1x1(&m1, &m0, a0 ^ a1, b0 ^ b1); - /* Correction on m1 ^= l1 ^ h1; m0 ^= l0 ^ h0; */ - r[2] ^= m1 ^ r[1] ^ r[3]; /* h0 ^= m1 ^ l1 ^ h1; */ - r[1] = r[3] ^ r[2] ^ r[0] ^ m1 ^ m0; /* l1 ^= l0 ^ h0 ^ m0; */ - } -#else -void bn_GF2m_mul_2x2(BN_ULONG *r, BN_ULONG a1, BN_ULONG a0, BN_ULONG b1, BN_ULONG b0); -#endif - -/* Add polynomials a and b and store result in r; r could be a or b, a and b +static void bn_GF2m_mul_2x2(BN_ULONG *r, const BN_ULONG a1, const BN_ULONG a0, + const BN_ULONG b1, const BN_ULONG b0) +{ + BN_ULONG m1, m0; + /* r[3] = h1, r[2] = h0; r[1] = l1; r[0] = l0 */ + bn_GF2m_mul_1x1(r + 3, r + 2, a1, b1); + bn_GF2m_mul_1x1(r + 1, r, a0, b0); + bn_GF2m_mul_1x1(&m1, &m0, a0 ^ a1, b0 ^ b1); + /* Correction on m1 ^= l1 ^ h1; m0 ^= l0 ^ h0; */ + r[2] ^= m1 ^ r[1] ^ r[3]; /* h0 ^= m1 ^ l1 ^ h1; */ + r[1] = r[3] ^ r[2] ^ r[0] ^ m1 ^ m0; /* l1 ^= l0 ^ h0 ^ m0; */ +} +# else +void bn_GF2m_mul_2x2(BN_ULONG *r, BN_ULONG a1, BN_ULONG a0, BN_ULONG b1, + BN_ULONG b0); +# endif + +/* + * Add polynomials a and b and store result in r; r could be a or b, a and b * could be equal; r is the bitwise XOR of a and b. */ -int BN_GF2m_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b) - { - int i; - const BIGNUM *at, *bt; - - bn_check_top(a); - bn_check_top(b); - - if (a->top < b->top) { at = b; bt = a; } - else { at = a; bt = b; } - - if(bn_wexpand(r, at->top) == NULL) - return 0; - - for (i = 0; i < bt->top; i++) - { - r->d[i] = at->d[i] ^ bt->d[i]; - } - for (; i < at->top; i++) - { - r->d[i] = at->d[i]; - } - - r->top = at->top; - bn_correct_top(r); - - return 1; - } - - -/* Some functions allow for representation of the irreducible polynomials +int BN_GF2m_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b) +{ + int i; + const BIGNUM *at, *bt; + + bn_check_top(a); + bn_check_top(b); + + if (a->top < b->top) { + at = b; + bt = a; + } else { + at = a; + bt = b; + } + + if (bn_wexpand(r, at->top) == NULL) + return 0; + + for (i = 0; i < bt->top; i++) { + r->d[i] = at->d[i] ^ bt->d[i]; + } + for (; i < at->top; i++) { + r->d[i] = at->d[i]; + } + + r->top = at->top; + bn_correct_top(r); + + return 1; +} + +/*- + * Some functions allow for representation of the irreducible polynomials * as an int[], say p. The irreducible f(t) is then of the form: * t^p[0] + t^p[1] + ... + t^p[k] * where m = p[0] > p[1] > ... > p[k] = 0. */ - /* Performs modular reduction of a and store result in r. r could be a. */ int BN_GF2m_mod_arr(BIGNUM *r, const BIGNUM *a, const int p[]) - { - int j, k; - int n, dN, d0, d1; - BN_ULONG zz, *z; - - bn_check_top(a); - - if (!p[0]) - { - /* reduction mod 1 => return 0 */ - BN_zero(r); - return 1; - } - - /* Since the algorithm does reduction in the r value, if a != r, copy - * the contents of a into r so we can do reduction in r. - */ - if (a != r) - { - if (!bn_wexpand(r, a->top)) return 0; - for (j = 0; j < a->top; j++) - { - r->d[j] = a->d[j]; - } - r->top = a->top; - } - z = r->d; - - /* start reduction */ - dN = p[0] / BN_BITS2; - for (j = r->top - 1; j > dN;) - { - zz = z[j]; - if (z[j] == 0) { j--; continue; } - z[j] = 0; - - for (k = 1; p[k] != 0; k++) - { - /* reducing component t^p[k] */ - n = p[0] - p[k]; - d0 = n % BN_BITS2; d1 = BN_BITS2 - d0; - n /= BN_BITS2; - z[j-n] ^= (zz>>d0); - if (d0) z[j-n-1] ^= (zz<<d1); - } - - /* reducing component t^0 */ - n = dN; - d0 = p[0] % BN_BITS2; - d1 = BN_BITS2 - d0; - z[j-n] ^= (zz >> d0); - if (d0) z[j-n-1] ^= (zz << d1); - } - - /* final round of reduction */ - while (j == dN) - { - - d0 = p[0] % BN_BITS2; - zz = z[dN] >> d0; - if (zz == 0) break; - d1 = BN_BITS2 - d0; - - /* clear up the top d1 bits */ - if (d0) - z[dN] = (z[dN] << d1) >> d1; - else - z[dN] = 0; - z[0] ^= zz; /* reduction t^0 component */ - - for (k = 1; p[k] != 0; k++) - { - BN_ULONG tmp_ulong; - - /* reducing component t^p[k]*/ - n = p[k] / BN_BITS2; - d0 = p[k] % BN_BITS2; - d1 = BN_BITS2 - d0; - z[n] ^= (zz << d0); - tmp_ulong = zz >> d1; - if (d0 && tmp_ulong) - z[n+1] ^= tmp_ulong; - } - - - } - - bn_correct_top(r); - return 1; - } - -/* Performs modular reduction of a by p and store result in r. r could be a. - * +{ + int j, k; + int n, dN, d0, d1; + BN_ULONG zz, *z; + + bn_check_top(a); + + if (!p[0]) { + /* reduction mod 1 => return 0 */ + BN_zero(r); + return 1; + } + + /* + * Since the algorithm does reduction in the r value, if a != r, copy the + * contents of a into r so we can do reduction in r. + */ + if (a != r) { + if (!bn_wexpand(r, a->top)) + return 0; + for (j = 0; j < a->top; j++) { + r->d[j] = a->d[j]; + } + r->top = a->top; + } + z = r->d; + + /* start reduction */ + dN = p[0] / BN_BITS2; + for (j = r->top - 1; j > dN;) { + zz = z[j]; + if (z[j] == 0) { + j--; + continue; + } + z[j] = 0; + + for (k = 1; p[k] != 0; k++) { + /* reducing component t^p[k] */ + n = p[0] - p[k]; + d0 = n % BN_BITS2; + d1 = BN_BITS2 - d0; + n /= BN_BITS2; + z[j - n] ^= (zz >> d0); + if (d0) + z[j - n - 1] ^= (zz << d1); + } + + /* reducing component t^0 */ + n = dN; + d0 = p[0] % BN_BITS2; + d1 = BN_BITS2 - d0; + z[j - n] ^= (zz >> d0); + if (d0) + z[j - n - 1] ^= (zz << d1); + } + + /* final round of reduction */ + while (j == dN) { + + d0 = p[0] % BN_BITS2; + zz = z[dN] >> d0; + if (zz == 0) + break; + d1 = BN_BITS2 - d0; + + /* clear up the top d1 bits */ + if (d0) + z[dN] = (z[dN] << d1) >> d1; + else + z[dN] = 0; + z[0] ^= zz; /* reduction t^0 component */ + + for (k = 1; p[k] != 0; k++) { + BN_ULONG tmp_ulong; + + /* reducing component t^p[k] */ + n = p[k] / BN_BITS2; + d0 = p[k] % BN_BITS2; + d1 = BN_BITS2 - d0; + z[n] ^= (zz << d0); + tmp_ulong = zz >> d1; + if (d0 && tmp_ulong) + z[n + 1] ^= tmp_ulong; + } + + } + + bn_correct_top(r); + return 1; +} + +/* + * Performs modular reduction of a by p and store result in r. r could be a. * This function calls down to the BN_GF2m_mod_arr implementation; this wrapper - * function is only provided for convenience; for best performance, use the + * function is only provided for convenience; for best performance, use the * BN_GF2m_mod_arr function. */ -int BN_GF2m_mod(BIGNUM *r, const BIGNUM *a, const BIGNUM *p) - { - int ret = 0; - int arr[6]; - bn_check_top(a); - bn_check_top(p); - ret = BN_GF2m_poly2arr(p, arr, sizeof(arr)/sizeof(arr[0])); - if (!ret || ret > (int)(sizeof(arr)/sizeof(arr[0]))) - { - BNerr(BN_F_BN_GF2M_MOD,BN_R_INVALID_LENGTH); - return 0; - } - ret = BN_GF2m_mod_arr(r, a, arr); - bn_check_top(r); - return ret; - } - - -/* Compute the product of two polynomials a and b, reduce modulo p, and store +int BN_GF2m_mod(BIGNUM *r, const BIGNUM *a, const BIGNUM *p) +{ + int ret = 0; + int arr[6]; + bn_check_top(a); + bn_check_top(p); + ret = BN_GF2m_poly2arr(p, arr, sizeof(arr) / sizeof(arr[0])); + if (!ret || ret > (int)(sizeof(arr) / sizeof(arr[0]))) { + BNerr(BN_F_BN_GF2M_MOD, BN_R_INVALID_LENGTH); + return 0; + } + ret = BN_GF2m_mod_arr(r, a, arr); + bn_check_top(r); + return ret; +} + +/* + * Compute the product of two polynomials a and b, reduce modulo p, and store * the result in r. r could be a or b; a could be b. */ -int BN_GF2m_mod_mul_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const int p[], BN_CTX *ctx) - { - int zlen, i, j, k, ret = 0; - BIGNUM *s; - BN_ULONG x1, x0, y1, y0, zz[4]; - - bn_check_top(a); - bn_check_top(b); - - if (a == b) - { - return BN_GF2m_mod_sqr_arr(r, a, p, ctx); - } - - BN_CTX_start(ctx); - if ((s = BN_CTX_get(ctx)) == NULL) goto err; - - zlen = a->top + b->top + 4; - if (!bn_wexpand(s, zlen)) goto err; - s->top = zlen; - - for (i = 0; i < zlen; i++) s->d[i] = 0; - - for (j = 0; j < b->top; j += 2) - { - y0 = b->d[j]; - y1 = ((j+1) == b->top) ? 0 : b->d[j+1]; - for (i = 0; i < a->top; i += 2) - { - x0 = a->d[i]; - x1 = ((i+1) == a->top) ? 0 : a->d[i+1]; - bn_GF2m_mul_2x2(zz, x1, x0, y1, y0); - for (k = 0; k < 4; k++) s->d[i+j+k] ^= zz[k]; - } - } - - bn_correct_top(s); - if (BN_GF2m_mod_arr(r, s, p)) - ret = 1; - bn_check_top(r); - -err: - BN_CTX_end(ctx); - return ret; - } - -/* Compute the product of two polynomials a and b, reduce modulo p, and store - * the result in r. r could be a or b; a could equal b. - * - * This function calls down to the BN_GF2m_mod_mul_arr implementation; this wrapper - * function is only provided for convenience; for best performance, use the +int BN_GF2m_mod_mul_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const int p[], BN_CTX *ctx) +{ + int zlen, i, j, k, ret = 0; + BIGNUM *s; + BN_ULONG x1, x0, y1, y0, zz[4]; + + bn_check_top(a); + bn_check_top(b); + + if (a == b) { + return BN_GF2m_mod_sqr_arr(r, a, p, ctx); + } + + BN_CTX_start(ctx); + if ((s = BN_CTX_get(ctx)) == NULL) + goto err; + + zlen = a->top + b->top + 4; + if (!bn_wexpand(s, zlen)) + goto err; + s->top = zlen; + + for (i = 0; i < zlen; i++) + s->d[i] = 0; + + for (j = 0; j < b->top; j += 2) { + y0 = b->d[j]; + y1 = ((j + 1) == b->top) ? 0 : b->d[j + 1]; + for (i = 0; i < a->top; i += 2) { + x0 = a->d[i]; + x1 = ((i + 1) == a->top) ? 0 : a->d[i + 1]; + bn_GF2m_mul_2x2(zz, x1, x0, y1, y0); + for (k = 0; k < 4; k++) + s->d[i + j + k] ^= zz[k]; + } + } + + bn_correct_top(s); + if (BN_GF2m_mod_arr(r, s, p)) + ret = 1; + bn_check_top(r); + + err: + BN_CTX_end(ctx); + return ret; +} + +/* + * Compute the product of two polynomials a and b, reduce modulo p, and store + * the result in r. r could be a or b; a could equal b. This function calls + * down to the BN_GF2m_mod_mul_arr implementation; this wrapper function is + * only provided for convenience; for best performance, use the * BN_GF2m_mod_mul_arr function. */ -int BN_GF2m_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *p, BN_CTX *ctx) - { - int ret = 0; - const int max = BN_num_bits(p) + 1; - int *arr=NULL; - bn_check_top(a); - bn_check_top(b); - bn_check_top(p); - if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) goto err; - ret = BN_GF2m_poly2arr(p, arr, max); - if (!ret || ret > max) - { - BNerr(BN_F_BN_GF2M_MOD_MUL,BN_R_INVALID_LENGTH); - goto err; - } - ret = BN_GF2m_mod_mul_arr(r, a, b, arr, ctx); - bn_check_top(r); -err: - if (arr) OPENSSL_free(arr); - return ret; - } - +int BN_GF2m_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *p, BN_CTX *ctx) +{ + int ret = 0; + const int max = BN_num_bits(p) + 1; + int *arr = NULL; + bn_check_top(a); + bn_check_top(b); + bn_check_top(p); + if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) + goto err; + ret = BN_GF2m_poly2arr(p, arr, max); + if (!ret || ret > max) { + BNerr(BN_F_BN_GF2M_MOD_MUL, BN_R_INVALID_LENGTH); + goto err; + } + ret = BN_GF2m_mod_mul_arr(r, a, b, arr, ctx); + bn_check_top(r); + err: + if (arr) + OPENSSL_free(arr); + return ret; +} /* Square a, reduce the result mod p, and store it in a. r could be a. */ -int BN_GF2m_mod_sqr_arr(BIGNUM *r, const BIGNUM *a, const int p[], BN_CTX *ctx) - { - int i, ret = 0; - BIGNUM *s; - - bn_check_top(a); - BN_CTX_start(ctx); - if ((s = BN_CTX_get(ctx)) == NULL) return 0; - if (!bn_wexpand(s, 2 * a->top)) goto err; - - for (i = a->top - 1; i >= 0; i--) - { - s->d[2*i+1] = SQR1(a->d[i]); - s->d[2*i ] = SQR0(a->d[i]); - } - - s->top = 2 * a->top; - bn_correct_top(s); - if (!BN_GF2m_mod_arr(r, s, p)) goto err; - bn_check_top(r); - ret = 1; -err: - BN_CTX_end(ctx); - return ret; - } - -/* Square a, reduce the result mod p, and store it in a. r could be a. - * - * This function calls down to the BN_GF2m_mod_sqr_arr implementation; this wrapper - * function is only provided for convenience; for best performance, use the - * BN_GF2m_mod_sqr_arr function. +int BN_GF2m_mod_sqr_arr(BIGNUM *r, const BIGNUM *a, const int p[], + BN_CTX *ctx) +{ + int i, ret = 0; + BIGNUM *s; + + bn_check_top(a); + BN_CTX_start(ctx); + if ((s = BN_CTX_get(ctx)) == NULL) + return 0; + if (!bn_wexpand(s, 2 * a->top)) + goto err; + + for (i = a->top - 1; i >= 0; i--) { + s->d[2 * i + 1] = SQR1(a->d[i]); + s->d[2 * i] = SQR0(a->d[i]); + } + + s->top = 2 * a->top; + bn_correct_top(s); + if (!BN_GF2m_mod_arr(r, s, p)) + goto err; + bn_check_top(r); + ret = 1; + err: + BN_CTX_end(ctx); + return ret; +} + +/* + * Square a, reduce the result mod p, and store it in a. r could be a. This + * function calls down to the BN_GF2m_mod_sqr_arr implementation; this + * wrapper function is only provided for convenience; for best performance, + * use the BN_GF2m_mod_sqr_arr function. */ -int BN_GF2m_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) - { - int ret = 0; - const int max = BN_num_bits(p) + 1; - int *arr=NULL; - - bn_check_top(a); - bn_check_top(p); - if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) goto err; - ret = BN_GF2m_poly2arr(p, arr, max); - if (!ret || ret > max) - { - BNerr(BN_F_BN_GF2M_MOD_SQR,BN_R_INVALID_LENGTH); - goto err; - } - ret = BN_GF2m_mod_sqr_arr(r, a, arr, ctx); - bn_check_top(r); -err: - if (arr) OPENSSL_free(arr); - return ret; - } - - -/* Invert a, reduce modulo p, and store the result in r. r could be a. - * Uses Modified Almost Inverse Algorithm (Algorithm 10) from - * Hankerson, D., Hernandez, J.L., and Menezes, A. "Software Implementation - * of Elliptic Curve Cryptography Over Binary Fields". +int BN_GF2m_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) +{ + int ret = 0; + const int max = BN_num_bits(p) + 1; + int *arr = NULL; + + bn_check_top(a); + bn_check_top(p); + if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) + goto err; + ret = BN_GF2m_poly2arr(p, arr, max); + if (!ret || ret > max) { + BNerr(BN_F_BN_GF2M_MOD_SQR, BN_R_INVALID_LENGTH); + goto err; + } + ret = BN_GF2m_mod_sqr_arr(r, a, arr, ctx); + bn_check_top(r); + err: + if (arr) + OPENSSL_free(arr); + return ret; +} + +/* + * Invert a, reduce modulo p, and store the result in r. r could be a. Uses + * Modified Almost Inverse Algorithm (Algorithm 10) from Hankerson, D., + * Hernandez, J.L., and Menezes, A. "Software Implementation of Elliptic + * Curve Cryptography Over Binary Fields". */ int BN_GF2m_mod_inv(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) - { - BIGNUM *b, *c = NULL, *u = NULL, *v = NULL, *tmp; - int ret = 0; - - bn_check_top(a); - bn_check_top(p); - - BN_CTX_start(ctx); - - if ((b = BN_CTX_get(ctx))==NULL) goto err; - if ((c = BN_CTX_get(ctx))==NULL) goto err; - if ((u = BN_CTX_get(ctx))==NULL) goto err; - if ((v = BN_CTX_get(ctx))==NULL) goto err; - - if (!BN_GF2m_mod(u, a, p)) goto err; - if (BN_is_zero(u)) goto err; - - if (!BN_copy(v, p)) goto err; -#if 0 - if (!BN_one(b)) goto err; - - while (1) - { - while (!BN_is_odd(u)) - { - if (BN_is_zero(u)) goto err; - if (!BN_rshift1(u, u)) goto err; - if (BN_is_odd(b)) - { - if (!BN_GF2m_add(b, b, p)) goto err; - } - if (!BN_rshift1(b, b)) goto err; - } - - if (BN_abs_is_word(u, 1)) break; - - if (BN_num_bits(u) < BN_num_bits(v)) - { - tmp = u; u = v; v = tmp; - tmp = b; b = c; c = tmp; - } - - if (!BN_GF2m_add(u, u, v)) goto err; - if (!BN_GF2m_add(b, b, c)) goto err; - } -#else - { - int i, ubits = BN_num_bits(u), - vbits = BN_num_bits(v), /* v is copy of p */ - top = p->top; - BN_ULONG *udp,*bdp,*vdp,*cdp; - - bn_wexpand(u,top); udp = u->d; - for (i=u->top;i<top;i++) udp[i] = 0; - u->top = top; - bn_wexpand(b,top); bdp = b->d; - bdp[0] = 1; - for (i=1;i<top;i++) bdp[i] = 0; - b->top = top; - bn_wexpand(c,top); cdp = c->d; - for (i=0;i<top;i++) cdp[i] = 0; - c->top = top; - vdp = v->d; /* It pays off to "cache" *->d pointers, because - * it allows optimizer to be more aggressive. - * But we don't have to "cache" p->d, because *p - * is declared 'const'... */ - while (1) - { - while (ubits && !(udp[0]&1)) - { - BN_ULONG u0,u1,b0,b1,mask; - - u0 = udp[0]; - b0 = bdp[0]; - mask = (BN_ULONG)0-(b0&1); - b0 ^= p->d[0]&mask; - for (i=0;i<top-1;i++) - { - u1 = udp[i+1]; - udp[i] = ((u0>>1)|(u1<<(BN_BITS2-1)))&BN_MASK2; - u0 = u1; - b1 = bdp[i+1]^(p->d[i+1]&mask); - bdp[i] = ((b0>>1)|(b1<<(BN_BITS2-1)))&BN_MASK2; - b0 = b1; - } - udp[i] = u0>>1; - bdp[i] = b0>>1; - ubits--; - } - - if (ubits<=BN_BITS2 && udp[0]==1) break; - - if (ubits<vbits) - { - i = ubits; ubits = vbits; vbits = i; - tmp = u; u = v; v = tmp; - tmp = b; b = c; c = tmp; - udp = vdp; vdp = v->d; - bdp = cdp; cdp = c->d; - } - for(i=0;i<top;i++) - { - udp[i] ^= vdp[i]; - bdp[i] ^= cdp[i]; - } - if (ubits==vbits) - { - BN_ULONG ul; - int utop = (ubits-1)/BN_BITS2; - - while ((ul=udp[utop])==0 && utop) utop--; - ubits = utop*BN_BITS2 + BN_num_bits_word(ul); - } - } - bn_correct_top(b); - } -#endif - - if (!BN_copy(r, b)) goto err; - bn_check_top(r); - ret = 1; - -err: -#ifdef BN_DEBUG /* BN_CTX_end would complain about the expanded form */ - bn_correct_top(c); - bn_correct_top(u); - bn_correct_top(v); -#endif - BN_CTX_end(ctx); - return ret; - } - -/* Invert xx, reduce modulo p, and store the result in r. r could be xx. - * - * This function calls down to the BN_GF2m_mod_inv implementation; this wrapper - * function is only provided for convenience; for best performance, use the - * BN_GF2m_mod_inv function. +{ + BIGNUM *b, *c = NULL, *u = NULL, *v = NULL, *tmp; + int ret = 0; + + bn_check_top(a); + bn_check_top(p); + + BN_CTX_start(ctx); + + if ((b = BN_CTX_get(ctx)) == NULL) + goto err; + if ((c = BN_CTX_get(ctx)) == NULL) + goto err; + if ((u = BN_CTX_get(ctx)) == NULL) + goto err; + if ((v = BN_CTX_get(ctx)) == NULL) + goto err; + + if (!BN_GF2m_mod(u, a, p)) + goto err; + if (BN_is_zero(u)) + goto err; + + if (!BN_copy(v, p)) + goto err; +# if 0 + if (!BN_one(b)) + goto err; + + while (1) { + while (!BN_is_odd(u)) { + if (BN_is_zero(u)) + goto err; + if (!BN_rshift1(u, u)) + goto err; + if (BN_is_odd(b)) { + if (!BN_GF2m_add(b, b, p)) + goto err; + } + if (!BN_rshift1(b, b)) + goto err; + } + + if (BN_abs_is_word(u, 1)) + break; + + if (BN_num_bits(u) < BN_num_bits(v)) { + tmp = u; + u = v; + v = tmp; + tmp = b; + b = c; + c = tmp; + } + + if (!BN_GF2m_add(u, u, v)) + goto err; + if (!BN_GF2m_add(b, b, c)) + goto err; + } +# else + { + int i, ubits = BN_num_bits(u), vbits = BN_num_bits(v), /* v is copy + * of p */ + top = p->top; + BN_ULONG *udp, *bdp, *vdp, *cdp; + + bn_wexpand(u, top); + udp = u->d; + for (i = u->top; i < top; i++) + udp[i] = 0; + u->top = top; + bn_wexpand(b, top); + bdp = b->d; + bdp[0] = 1; + for (i = 1; i < top; i++) + bdp[i] = 0; + b->top = top; + bn_wexpand(c, top); + cdp = c->d; + for (i = 0; i < top; i++) + cdp[i] = 0; + c->top = top; + vdp = v->d; /* It pays off to "cache" *->d pointers, + * because it allows optimizer to be more + * aggressive. But we don't have to "cache" + * p->d, because *p is declared 'const'... */ + while (1) { + while (ubits && !(udp[0] & 1)) { + BN_ULONG u0, u1, b0, b1, mask; + + u0 = udp[0]; + b0 = bdp[0]; + mask = (BN_ULONG)0 - (b0 & 1); + b0 ^= p->d[0] & mask; + for (i = 0; i < top - 1; i++) { + u1 = udp[i + 1]; + udp[i] = ((u0 >> 1) | (u1 << (BN_BITS2 - 1))) & BN_MASK2; + u0 = u1; + b1 = bdp[i + 1] ^ (p->d[i + 1] & mask); + bdp[i] = ((b0 >> 1) | (b1 << (BN_BITS2 - 1))) & BN_MASK2; + b0 = b1; + } + udp[i] = u0 >> 1; + bdp[i] = b0 >> 1; + ubits--; + } + + if (ubits <= BN_BITS2 && udp[0] == 1) + break; + + if (ubits < vbits) { + i = ubits; + ubits = vbits; + vbits = i; + tmp = u; + u = v; + v = tmp; + tmp = b; + b = c; + c = tmp; + udp = vdp; + vdp = v->d; + bdp = cdp; + cdp = c->d; + } + for (i = 0; i < top; i++) { + udp[i] ^= vdp[i]; + bdp[i] ^= cdp[i]; + } + if (ubits == vbits) { + BN_ULONG ul; + int utop = (ubits - 1) / BN_BITS2; + + while ((ul = udp[utop]) == 0 && utop) + utop--; + ubits = utop * BN_BITS2 + BN_num_bits_word(ul); + } + } + bn_correct_top(b); + } +# endif + + if (!BN_copy(r, b)) + goto err; + bn_check_top(r); + ret = 1; + + err: +# ifdef BN_DEBUG /* BN_CTX_end would complain about the + * expanded form */ + bn_correct_top(c); + bn_correct_top(u); + bn_correct_top(v); +# endif + BN_CTX_end(ctx); + return ret; +} + +/* + * Invert xx, reduce modulo p, and store the result in r. r could be xx. + * This function calls down to the BN_GF2m_mod_inv implementation; this + * wrapper function is only provided for convenience; for best performance, + * use the BN_GF2m_mod_inv function. */ -int BN_GF2m_mod_inv_arr(BIGNUM *r, const BIGNUM *xx, const int p[], BN_CTX *ctx) - { - BIGNUM *field; - int ret = 0; - - bn_check_top(xx); - BN_CTX_start(ctx); - if ((field = BN_CTX_get(ctx)) == NULL) goto err; - if (!BN_GF2m_arr2poly(p, field)) goto err; - - ret = BN_GF2m_mod_inv(r, xx, field, ctx); - bn_check_top(r); - -err: - BN_CTX_end(ctx); - return ret; - } - - -#ifndef OPENSSL_SUN_GF2M_DIV -/* Divide y by x, reduce modulo p, and store the result in r. r could be x +int BN_GF2m_mod_inv_arr(BIGNUM *r, const BIGNUM *xx, const int p[], + BN_CTX *ctx) +{ + BIGNUM *field; + int ret = 0; + + bn_check_top(xx); + BN_CTX_start(ctx); + if ((field = BN_CTX_get(ctx)) == NULL) + goto err; + if (!BN_GF2m_arr2poly(p, field)) + goto err; + + ret = BN_GF2m_mod_inv(r, xx, field, ctx); + bn_check_top(r); + + err: + BN_CTX_end(ctx); + return ret; +} + +# ifndef OPENSSL_SUN_GF2M_DIV +/* + * Divide y by x, reduce modulo p, and store the result in r. r could be x * or y, x could equal y. */ -int BN_GF2m_mod_div(BIGNUM *r, const BIGNUM *y, const BIGNUM *x, const BIGNUM *p, BN_CTX *ctx) - { - BIGNUM *xinv = NULL; - int ret = 0; - - bn_check_top(y); - bn_check_top(x); - bn_check_top(p); - - BN_CTX_start(ctx); - xinv = BN_CTX_get(ctx); - if (xinv == NULL) goto err; - - if (!BN_GF2m_mod_inv(xinv, x, p, ctx)) goto err; - if (!BN_GF2m_mod_mul(r, y, xinv, p, ctx)) goto err; - bn_check_top(r); - ret = 1; - -err: - BN_CTX_end(ctx); - return ret; - } -#else -/* Divide y by x, reduce modulo p, and store the result in r. r could be x - * or y, x could equal y. - * Uses algorithm Modular_Division_GF(2^m) from - * Chang-Shantz, S. "From Euclid's GCD to Montgomery Multiplication to - * the Great Divide". +int BN_GF2m_mod_div(BIGNUM *r, const BIGNUM *y, const BIGNUM *x, + const BIGNUM *p, BN_CTX *ctx) +{ + BIGNUM *xinv = NULL; + int ret = 0; + + bn_check_top(y); + bn_check_top(x); + bn_check_top(p); + + BN_CTX_start(ctx); + xinv = BN_CTX_get(ctx); + if (xinv == NULL) + goto err; + + if (!BN_GF2m_mod_inv(xinv, x, p, ctx)) + goto err; + if (!BN_GF2m_mod_mul(r, y, xinv, p, ctx)) + goto err; + bn_check_top(r); + ret = 1; + + err: + BN_CTX_end(ctx); + return ret; +} +# else +/* + * Divide y by x, reduce modulo p, and store the result in r. r could be x + * or y, x could equal y. Uses algorithm Modular_Division_GF(2^m) from + * Chang-Shantz, S. "From Euclid's GCD to Montgomery Multiplication to the + * Great Divide". */ -int BN_GF2m_mod_div(BIGNUM *r, const BIGNUM *y, const BIGNUM *x, const BIGNUM *p, BN_CTX *ctx) - { - BIGNUM *a, *b, *u, *v; - int ret = 0; - - bn_check_top(y); - bn_check_top(x); - bn_check_top(p); - - BN_CTX_start(ctx); - - a = BN_CTX_get(ctx); - b = BN_CTX_get(ctx); - u = BN_CTX_get(ctx); - v = BN_CTX_get(ctx); - if (v == NULL) goto err; - - /* reduce x and y mod p */ - if (!BN_GF2m_mod(u, y, p)) goto err; - if (!BN_GF2m_mod(a, x, p)) goto err; - if (!BN_copy(b, p)) goto err; - - while (!BN_is_odd(a)) - { - if (!BN_rshift1(a, a)) goto err; - if (BN_is_odd(u)) if (!BN_GF2m_add(u, u, p)) goto err; - if (!BN_rshift1(u, u)) goto err; - } - - do - { - if (BN_GF2m_cmp(b, a) > 0) - { - if (!BN_GF2m_add(b, b, a)) goto err; - if (!BN_GF2m_add(v, v, u)) goto err; - do - { - if (!BN_rshift1(b, b)) goto err; - if (BN_is_odd(v)) if (!BN_GF2m_add(v, v, p)) goto err; - if (!BN_rshift1(v, v)) goto err; - } while (!BN_is_odd(b)); - } - else if (BN_abs_is_word(a, 1)) - break; - else - { - if (!BN_GF2m_add(a, a, b)) goto err; - if (!BN_GF2m_add(u, u, v)) goto err; - do - { - if (!BN_rshift1(a, a)) goto err; - if (BN_is_odd(u)) if (!BN_GF2m_add(u, u, p)) goto err; - if (!BN_rshift1(u, u)) goto err; - } while (!BN_is_odd(a)); - } - } while (1); - - if (!BN_copy(r, u)) goto err; - bn_check_top(r); - ret = 1; - -err: - BN_CTX_end(ctx); - return ret; - } -#endif - -/* Divide yy by xx, reduce modulo p, and store the result in r. r could be xx - * or yy, xx could equal yy. - * - * This function calls down to the BN_GF2m_mod_div implementation; this wrapper - * function is only provided for convenience; for best performance, use the - * BN_GF2m_mod_div function. +int BN_GF2m_mod_div(BIGNUM *r, const BIGNUM *y, const BIGNUM *x, + const BIGNUM *p, BN_CTX *ctx) +{ + BIGNUM *a, *b, *u, *v; + int ret = 0; + + bn_check_top(y); + bn_check_top(x); + bn_check_top(p); + + BN_CTX_start(ctx); + + a = BN_CTX_get(ctx); + b = BN_CTX_get(ctx); + u = BN_CTX_get(ctx); + v = BN_CTX_get(ctx); + if (v == NULL) + goto err; + + /* reduce x and y mod p */ + if (!BN_GF2m_mod(u, y, p)) + goto err; + if (!BN_GF2m_mod(a, x, p)) + goto err; + if (!BN_copy(b, p)) + goto err; + + while (!BN_is_odd(a)) { + if (!BN_rshift1(a, a)) + goto err; + if (BN_is_odd(u)) + if (!BN_GF2m_add(u, u, p)) + goto err; + if (!BN_rshift1(u, u)) + goto err; + } + + do { + if (BN_GF2m_cmp(b, a) > 0) { + if (!BN_GF2m_add(b, b, a)) + goto err; + if (!BN_GF2m_add(v, v, u)) + goto err; + do { + if (!BN_rshift1(b, b)) + goto err; + if (BN_is_odd(v)) + if (!BN_GF2m_add(v, v, p)) + goto err; + if (!BN_rshift1(v, v)) + goto err; + } while (!BN_is_odd(b)); + } else if (BN_abs_is_word(a, 1)) + break; + else { + if (!BN_GF2m_add(a, a, b)) + goto err; + if (!BN_GF2m_add(u, u, v)) + goto err; + do { + if (!BN_rshift1(a, a)) + goto err; + if (BN_is_odd(u)) + if (!BN_GF2m_add(u, u, p)) + goto err; + if (!BN_rshift1(u, u)) + goto err; + } while (!BN_is_odd(a)); + } + } while (1); + + if (!BN_copy(r, u)) + goto err; + bn_check_top(r); + ret = 1; + + err: + BN_CTX_end(ctx); + return ret; +} +# endif + +/* + * Divide yy by xx, reduce modulo p, and store the result in r. r could be xx + * * or yy, xx could equal yy. This function calls down to the + * BN_GF2m_mod_div implementation; this wrapper function is only provided for + * convenience; for best performance, use the BN_GF2m_mod_div function. */ -int BN_GF2m_mod_div_arr(BIGNUM *r, const BIGNUM *yy, const BIGNUM *xx, const int p[], BN_CTX *ctx) - { - BIGNUM *field; - int ret = 0; - - bn_check_top(yy); - bn_check_top(xx); - - BN_CTX_start(ctx); - if ((field = BN_CTX_get(ctx)) == NULL) goto err; - if (!BN_GF2m_arr2poly(p, field)) goto err; - - ret = BN_GF2m_mod_div(r, yy, xx, field, ctx); - bn_check_top(r); - -err: - BN_CTX_end(ctx); - return ret; - } - - -/* Compute the bth power of a, reduce modulo p, and store - * the result in r. r could be a. - * Uses simple square-and-multiply algorithm A.5.1 from IEEE P1363. +int BN_GF2m_mod_div_arr(BIGNUM *r, const BIGNUM *yy, const BIGNUM *xx, + const int p[], BN_CTX *ctx) +{ + BIGNUM *field; + int ret = 0; + + bn_check_top(yy); + bn_check_top(xx); + + BN_CTX_start(ctx); + if ((field = BN_CTX_get(ctx)) == NULL) + goto err; + if (!BN_GF2m_arr2poly(p, field)) + goto err; + + ret = BN_GF2m_mod_div(r, yy, xx, field, ctx); + bn_check_top(r); + + err: + BN_CTX_end(ctx); + return ret; +} + +/* + * Compute the bth power of a, reduce modulo p, and store the result in r. r + * could be a. Uses simple square-and-multiply algorithm A.5.1 from IEEE + * P1363. */ -int BN_GF2m_mod_exp_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const int p[], BN_CTX *ctx) - { - int ret = 0, i, n; - BIGNUM *u; - - bn_check_top(a); - bn_check_top(b); - - if (BN_is_zero(b)) - return(BN_one(r)); - - if (BN_abs_is_word(b, 1)) - return (BN_copy(r, a) != NULL); - - BN_CTX_start(ctx); - if ((u = BN_CTX_get(ctx)) == NULL) goto err; - - if (!BN_GF2m_mod_arr(u, a, p)) goto err; - - n = BN_num_bits(b) - 1; - for (i = n - 1; i >= 0; i--) - { - if (!BN_GF2m_mod_sqr_arr(u, u, p, ctx)) goto err; - if (BN_is_bit_set(b, i)) - { - if (!BN_GF2m_mod_mul_arr(u, u, a, p, ctx)) goto err; - } - } - if (!BN_copy(r, u)) goto err; - bn_check_top(r); - ret = 1; -err: - BN_CTX_end(ctx); - return ret; - } - -/* Compute the bth power of a, reduce modulo p, and store - * the result in r. r could be a. - * - * This function calls down to the BN_GF2m_mod_exp_arr implementation; this wrapper - * function is only provided for convenience; for best performance, use the - * BN_GF2m_mod_exp_arr function. +int BN_GF2m_mod_exp_arr(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const int p[], BN_CTX *ctx) +{ + int ret = 0, i, n; + BIGNUM *u; + + bn_check_top(a); + bn_check_top(b); + + if (BN_is_zero(b)) + return (BN_one(r)); + + if (BN_abs_is_word(b, 1)) + return (BN_copy(r, a) != NULL); + + BN_CTX_start(ctx); + if ((u = BN_CTX_get(ctx)) == NULL) + goto err; + + if (!BN_GF2m_mod_arr(u, a, p)) + goto err; + + n = BN_num_bits(b) - 1; + for (i = n - 1; i >= 0; i--) { + if (!BN_GF2m_mod_sqr_arr(u, u, p, ctx)) + goto err; + if (BN_is_bit_set(b, i)) { + if (!BN_GF2m_mod_mul_arr(u, u, a, p, ctx)) + goto err; + } + } + if (!BN_copy(r, u)) + goto err; + bn_check_top(r); + ret = 1; + err: + BN_CTX_end(ctx); + return ret; +} + +/* + * Compute the bth power of a, reduce modulo p, and store the result in r. r + * could be a. This function calls down to the BN_GF2m_mod_exp_arr + * implementation; this wrapper function is only provided for convenience; + * for best performance, use the BN_GF2m_mod_exp_arr function. */ -int BN_GF2m_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *p, BN_CTX *ctx) - { - int ret = 0; - const int max = BN_num_bits(p) + 1; - int *arr=NULL; - bn_check_top(a); - bn_check_top(b); - bn_check_top(p); - if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) goto err; - ret = BN_GF2m_poly2arr(p, arr, max); - if (!ret || ret > max) - { - BNerr(BN_F_BN_GF2M_MOD_EXP,BN_R_INVALID_LENGTH); - goto err; - } - ret = BN_GF2m_mod_exp_arr(r, a, b, arr, ctx); - bn_check_top(r); -err: - if (arr) OPENSSL_free(arr); - return ret; - } - -/* Compute the square root of a, reduce modulo p, and store - * the result in r. r could be a. - * Uses exponentiation as in algorithm A.4.1 from IEEE P1363. +int BN_GF2m_mod_exp(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *p, BN_CTX *ctx) +{ + int ret = 0; + const int max = BN_num_bits(p) + 1; + int *arr = NULL; + bn_check_top(a); + bn_check_top(b); + bn_check_top(p); + if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) + goto err; + ret = BN_GF2m_poly2arr(p, arr, max); + if (!ret || ret > max) { + BNerr(BN_F_BN_GF2M_MOD_EXP, BN_R_INVALID_LENGTH); + goto err; + } + ret = BN_GF2m_mod_exp_arr(r, a, b, arr, ctx); + bn_check_top(r); + err: + if (arr) + OPENSSL_free(arr); + return ret; +} + +/* + * Compute the square root of a, reduce modulo p, and store the result in r. + * r could be a. Uses exponentiation as in algorithm A.4.1 from IEEE P1363. */ -int BN_GF2m_mod_sqrt_arr(BIGNUM *r, const BIGNUM *a, const int p[], BN_CTX *ctx) - { - int ret = 0; - BIGNUM *u; - - bn_check_top(a); - - if (!p[0]) - { - /* reduction mod 1 => return 0 */ - BN_zero(r); - return 1; - } - - BN_CTX_start(ctx); - if ((u = BN_CTX_get(ctx)) == NULL) goto err; - - if (!BN_set_bit(u, p[0] - 1)) goto err; - ret = BN_GF2m_mod_exp_arr(r, a, u, p, ctx); - bn_check_top(r); - -err: - BN_CTX_end(ctx); - return ret; - } - -/* Compute the square root of a, reduce modulo p, and store - * the result in r. r could be a. - * - * This function calls down to the BN_GF2m_mod_sqrt_arr implementation; this wrapper - * function is only provided for convenience; for best performance, use the - * BN_GF2m_mod_sqrt_arr function. +int BN_GF2m_mod_sqrt_arr(BIGNUM *r, const BIGNUM *a, const int p[], + BN_CTX *ctx) +{ + int ret = 0; + BIGNUM *u; + + bn_check_top(a); + + if (!p[0]) { + /* reduction mod 1 => return 0 */ + BN_zero(r); + return 1; + } + + BN_CTX_start(ctx); + if ((u = BN_CTX_get(ctx)) == NULL) + goto err; + + if (!BN_set_bit(u, p[0] - 1)) + goto err; + ret = BN_GF2m_mod_exp_arr(r, a, u, p, ctx); + bn_check_top(r); + + err: + BN_CTX_end(ctx); + return ret; +} + +/* + * Compute the square root of a, reduce modulo p, and store the result in r. + * r could be a. This function calls down to the BN_GF2m_mod_sqrt_arr + * implementation; this wrapper function is only provided for convenience; + * for best performance, use the BN_GF2m_mod_sqrt_arr function. */ int BN_GF2m_mod_sqrt(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) - { - int ret = 0; - const int max = BN_num_bits(p) + 1; - int *arr=NULL; - bn_check_top(a); - bn_check_top(p); - if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) goto err; - ret = BN_GF2m_poly2arr(p, arr, max); - if (!ret || ret > max) - { - BNerr(BN_F_BN_GF2M_MOD_SQRT,BN_R_INVALID_LENGTH); - goto err; - } - ret = BN_GF2m_mod_sqrt_arr(r, a, arr, ctx); - bn_check_top(r); -err: - if (arr) OPENSSL_free(arr); - return ret; - } - -/* Find r such that r^2 + r = a mod p. r could be a. If no r exists returns 0. - * Uses algorithms A.4.7 and A.4.6 from IEEE P1363. +{ + int ret = 0; + const int max = BN_num_bits(p) + 1; + int *arr = NULL; + bn_check_top(a); + bn_check_top(p); + if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) + goto err; + ret = BN_GF2m_poly2arr(p, arr, max); + if (!ret || ret > max) { + BNerr(BN_F_BN_GF2M_MOD_SQRT, BN_R_INVALID_LENGTH); + goto err; + } + ret = BN_GF2m_mod_sqrt_arr(r, a, arr, ctx); + bn_check_top(r); + err: + if (arr) + OPENSSL_free(arr); + return ret; +} + +/* + * Find r such that r^2 + r = a mod p. r could be a. If no r exists returns + * 0. Uses algorithms A.4.7 and A.4.6 from IEEE P1363. */ -int BN_GF2m_mod_solve_quad_arr(BIGNUM *r, const BIGNUM *a_, const int p[], BN_CTX *ctx) - { - int ret = 0, count = 0, j; - BIGNUM *a, *z, *rho, *w, *w2, *tmp; - - bn_check_top(a_); - - if (!p[0]) - { - /* reduction mod 1 => return 0 */ - BN_zero(r); - return 1; - } - - BN_CTX_start(ctx); - a = BN_CTX_get(ctx); - z = BN_CTX_get(ctx); - w = BN_CTX_get(ctx); - if (w == NULL) goto err; - - if (!BN_GF2m_mod_arr(a, a_, p)) goto err; - - if (BN_is_zero(a)) - { - BN_zero(r); - ret = 1; - goto err; - } - - if (p[0] & 0x1) /* m is odd */ - { - /* compute half-trace of a */ - if (!BN_copy(z, a)) goto err; - for (j = 1; j <= (p[0] - 1) / 2; j++) - { - if (!BN_GF2m_mod_sqr_arr(z, z, p, ctx)) goto err; - if (!BN_GF2m_mod_sqr_arr(z, z, p, ctx)) goto err; - if (!BN_GF2m_add(z, z, a)) goto err; - } - - } - else /* m is even */ - { - rho = BN_CTX_get(ctx); - w2 = BN_CTX_get(ctx); - tmp = BN_CTX_get(ctx); - if (tmp == NULL) goto err; - do - { - if (!BN_rand(rho, p[0], 0, 0)) goto err; - if (!BN_GF2m_mod_arr(rho, rho, p)) goto err; - BN_zero(z); - if (!BN_copy(w, rho)) goto err; - for (j = 1; j <= p[0] - 1; j++) - { - if (!BN_GF2m_mod_sqr_arr(z, z, p, ctx)) goto err; - if (!BN_GF2m_mod_sqr_arr(w2, w, p, ctx)) goto err; - if (!BN_GF2m_mod_mul_arr(tmp, w2, a, p, ctx)) goto err; - if (!BN_GF2m_add(z, z, tmp)) goto err; - if (!BN_GF2m_add(w, w2, rho)) goto err; - } - count++; - } while (BN_is_zero(w) && (count < MAX_ITERATIONS)); - if (BN_is_zero(w)) - { - BNerr(BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR,BN_R_TOO_MANY_ITERATIONS); - goto err; - } - } - - if (!BN_GF2m_mod_sqr_arr(w, z, p, ctx)) goto err; - if (!BN_GF2m_add(w, z, w)) goto err; - if (BN_GF2m_cmp(w, a)) - { - BNerr(BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR, BN_R_NO_SOLUTION); - goto err; - } - - if (!BN_copy(r, z)) goto err; - bn_check_top(r); - - ret = 1; - -err: - BN_CTX_end(ctx); - return ret; - } - -/* Find r such that r^2 + r = a mod p. r could be a. If no r exists returns 0. - * - * This function calls down to the BN_GF2m_mod_solve_quad_arr implementation; this wrapper - * function is only provided for convenience; for best performance, use the - * BN_GF2m_mod_solve_quad_arr function. +int BN_GF2m_mod_solve_quad_arr(BIGNUM *r, const BIGNUM *a_, const int p[], + BN_CTX *ctx) +{ + int ret = 0, count = 0, j; + BIGNUM *a, *z, *rho, *w, *w2, *tmp; + + bn_check_top(a_); + + if (!p[0]) { + /* reduction mod 1 => return 0 */ + BN_zero(r); + return 1; + } + + BN_CTX_start(ctx); + a = BN_CTX_get(ctx); + z = BN_CTX_get(ctx); + w = BN_CTX_get(ctx); + if (w == NULL) + goto err; + + if (!BN_GF2m_mod_arr(a, a_, p)) + goto err; + + if (BN_is_zero(a)) { + BN_zero(r); + ret = 1; + goto err; + } + + if (p[0] & 0x1) { /* m is odd */ + /* compute half-trace of a */ + if (!BN_copy(z, a)) + goto err; + for (j = 1; j <= (p[0] - 1) / 2; j++) { + if (!BN_GF2m_mod_sqr_arr(z, z, p, ctx)) + goto err; + if (!BN_GF2m_mod_sqr_arr(z, z, p, ctx)) + goto err; + if (!BN_GF2m_add(z, z, a)) + goto err; + } + + } else { /* m is even */ + + rho = BN_CTX_get(ctx); + w2 = BN_CTX_get(ctx); + tmp = BN_CTX_get(ctx); + if (tmp == NULL) + goto err; + do { + if (!BN_rand(rho, p[0], 0, 0)) + goto err; + if (!BN_GF2m_mod_arr(rho, rho, p)) + goto err; + BN_zero(z); + if (!BN_copy(w, rho)) + goto err; + for (j = 1; j <= p[0] - 1; j++) { + if (!BN_GF2m_mod_sqr_arr(z, z, p, ctx)) + goto err; + if (!BN_GF2m_mod_sqr_arr(w2, w, p, ctx)) + goto err; + if (!BN_GF2m_mod_mul_arr(tmp, w2, a, p, ctx)) + goto err; + if (!BN_GF2m_add(z, z, tmp)) + goto err; + if (!BN_GF2m_add(w, w2, rho)) + goto err; + } + count++; + } while (BN_is_zero(w) && (count < MAX_ITERATIONS)); + if (BN_is_zero(w)) { + BNerr(BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR, BN_R_TOO_MANY_ITERATIONS); + goto err; + } + } + + if (!BN_GF2m_mod_sqr_arr(w, z, p, ctx)) + goto err; + if (!BN_GF2m_add(w, z, w)) + goto err; + if (BN_GF2m_cmp(w, a)) { + BNerr(BN_F_BN_GF2M_MOD_SOLVE_QUAD_ARR, BN_R_NO_SOLUTION); + goto err; + } + + if (!BN_copy(r, z)) + goto err; + bn_check_top(r); + + ret = 1; + + err: + BN_CTX_end(ctx); + return ret; +} + +/* + * Find r such that r^2 + r = a mod p. r could be a. If no r exists returns + * 0. This function calls down to the BN_GF2m_mod_solve_quad_arr + * implementation; this wrapper function is only provided for convenience; + * for best performance, use the BN_GF2m_mod_solve_quad_arr function. */ -int BN_GF2m_mod_solve_quad(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) - { - int ret = 0; - const int max = BN_num_bits(p) + 1; - int *arr=NULL; - bn_check_top(a); - bn_check_top(p); - if ((arr = (int *)OPENSSL_malloc(sizeof(int) * - max)) == NULL) goto err; - ret = BN_GF2m_poly2arr(p, arr, max); - if (!ret || ret > max) - { - BNerr(BN_F_BN_GF2M_MOD_SOLVE_QUAD,BN_R_INVALID_LENGTH); - goto err; - } - ret = BN_GF2m_mod_solve_quad_arr(r, a, arr, ctx); - bn_check_top(r); -err: - if (arr) OPENSSL_free(arr); - return ret; - } - -/* Convert the bit-string representation of a polynomial - * ( \sum_{i=0}^n a_i * x^i) into an array of integers corresponding - * to the bits with non-zero coefficient. Array is terminated with -1. - * Up to max elements of the array will be filled. Return value is total - * number of array elements that would be filled if array was large enough. +int BN_GF2m_mod_solve_quad(BIGNUM *r, const BIGNUM *a, const BIGNUM *p, + BN_CTX *ctx) +{ + int ret = 0; + const int max = BN_num_bits(p) + 1; + int *arr = NULL; + bn_check_top(a); + bn_check_top(p); + if ((arr = (int *)OPENSSL_malloc(sizeof(int) * max)) == NULL) + goto err; + ret = BN_GF2m_poly2arr(p, arr, max); + if (!ret || ret > max) { + BNerr(BN_F_BN_GF2M_MOD_SOLVE_QUAD, BN_R_INVALID_LENGTH); + goto err; + } + ret = BN_GF2m_mod_solve_quad_arr(r, a, arr, ctx); + bn_check_top(r); + err: + if (arr) + OPENSSL_free(arr); + return ret; +} + +/* + * Convert the bit-string representation of a polynomial ( \sum_{i=0}^n a_i * + * x^i) into an array of integers corresponding to the bits with non-zero + * coefficient. Array is terminated with -1. Up to max elements of the array + * will be filled. Return value is total number of array elements that would + * be filled if array was large enough. */ int BN_GF2m_poly2arr(const BIGNUM *a, int p[], int max) - { - int i, j, k = 0; - BN_ULONG mask; - - if (BN_is_zero(a)) - return 0; - - for (i = a->top - 1; i >= 0; i--) - { - if (!a->d[i]) - /* skip word if a->d[i] == 0 */ - continue; - mask = BN_TBIT; - for (j = BN_BITS2 - 1; j >= 0; j--) - { - if (a->d[i] & mask) - { - if (k < max) p[k] = BN_BITS2 * i + j; - k++; - } - mask >>= 1; - } - } - - if (k < max) { - p[k] = -1; - k++; - } - - return k; - } - -/* Convert the coefficient array representation of a polynomial to a +{ + int i, j, k = 0; + BN_ULONG mask; + + if (BN_is_zero(a)) + return 0; + + for (i = a->top - 1; i >= 0; i--) { + if (!a->d[i]) + /* skip word if a->d[i] == 0 */ + continue; + mask = BN_TBIT; + for (j = BN_BITS2 - 1; j >= 0; j--) { + if (a->d[i] & mask) { + if (k < max) + p[k] = BN_BITS2 * i + j; + k++; + } + mask >>= 1; + } + } + + if (k < max) { + p[k] = -1; + k++; + } + + return k; +} + +/* + * Convert the coefficient array representation of a polynomial to a * bit-string. The array must be terminated by -1. */ int BN_GF2m_arr2poly(const int p[], BIGNUM *a) - { - int i; - - bn_check_top(a); - BN_zero(a); - for (i = 0; p[i] != -1; i++) - { - if (BN_set_bit(a, p[i]) == 0) - return 0; - } - bn_check_top(a); - - return 1; - } +{ + int i; + + bn_check_top(a); + BN_zero(a); + for (i = 0; p[i] != -1; i++) { + if (BN_set_bit(a, p[i]) == 0) + return 0; + } + bn_check_top(a); + + return 1; +} #endif diff --git a/openssl/crypto/bn/bn_kron.c b/openssl/crypto/bn/bn_kron.c index 740359b75..88d731ac7 100644 --- a/openssl/crypto/bn/bn_kron.c +++ b/openssl/crypto/bn/bn_kron.c @@ -7,7 +7,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -61,124 +61,126 @@ /* Returns -2 for errors because both -1 and 0 are valid results. */ int BN_kronecker(const BIGNUM *a, const BIGNUM *b, BN_CTX *ctx) - { - int i; - int ret = -2; /* avoid 'uninitialized' warning */ - int err = 0; - BIGNUM *A, *B, *tmp; - /* In 'tab', only odd-indexed entries are relevant: - * For any odd BIGNUM n, - * tab[BN_lsw(n) & 7] - * is $(-1)^{(n^2-1)/8}$ (using TeX notation). - * Note that the sign of n does not matter. - */ - static const int tab[8] = {0, 1, 0, -1, 0, -1, 0, 1}; - - bn_check_top(a); - bn_check_top(b); - - BN_CTX_start(ctx); - A = BN_CTX_get(ctx); - B = BN_CTX_get(ctx); - if (B == NULL) goto end; - - err = !BN_copy(A, a); - if (err) goto end; - err = !BN_copy(B, b); - if (err) goto end; - - /* - * Kronecker symbol, imlemented according to Henri Cohen, - * "A Course in Computational Algebraic Number Theory" - * (algorithm 1.4.10). - */ - - /* Cohen's step 1: */ - - if (BN_is_zero(B)) - { - ret = BN_abs_is_word(A, 1); - goto end; - } - - /* Cohen's step 2: */ - - if (!BN_is_odd(A) && !BN_is_odd(B)) - { - ret = 0; - goto end; - } - - /* now B is non-zero */ - i = 0; - while (!BN_is_bit_set(B, i)) - i++; - err = !BN_rshift(B, B, i); - if (err) goto end; - if (i & 1) - { - /* i is odd */ - /* (thus B was even, thus A must be odd!) */ - - /* set 'ret' to $(-1)^{(A^2-1)/8}$ */ - ret = tab[BN_lsw(A) & 7]; - } - else - { - /* i is even */ - ret = 1; - } - - if (B->neg) - { - B->neg = 0; - if (A->neg) - ret = -ret; - } - - /* now B is positive and odd, so what remains to be done is - * to compute the Jacobi symbol (A/B) and multiply it by 'ret' */ - - while (1) - { - /* Cohen's step 3: */ - - /* B is positive and odd */ - - if (BN_is_zero(A)) - { - ret = BN_is_one(B) ? ret : 0; - goto end; - } - - /* now A is non-zero */ - i = 0; - while (!BN_is_bit_set(A, i)) - i++; - err = !BN_rshift(A, A, i); - if (err) goto end; - if (i & 1) - { - /* i is odd */ - /* multiply 'ret' by $(-1)^{(B^2-1)/8}$ */ - ret = ret * tab[BN_lsw(B) & 7]; - } - - /* Cohen's step 4: */ - /* multiply 'ret' by $(-1)^{(A-1)(B-1)/4}$ */ - if ((A->neg ? ~BN_lsw(A) : BN_lsw(A)) & BN_lsw(B) & 2) - ret = -ret; - - /* (A, B) := (B mod |A|, |A|) */ - err = !BN_nnmod(B, B, A, ctx); - if (err) goto end; - tmp = A; A = B; B = tmp; - tmp->neg = 0; - } -end: - BN_CTX_end(ctx); - if (err) - return -2; - else - return ret; - } +{ + int i; + int ret = -2; /* avoid 'uninitialized' warning */ + int err = 0; + BIGNUM *A, *B, *tmp; + /*- + * In 'tab', only odd-indexed entries are relevant: + * For any odd BIGNUM n, + * tab[BN_lsw(n) & 7] + * is $(-1)^{(n^2-1)/8}$ (using TeX notation). + * Note that the sign of n does not matter. + */ + static const int tab[8] = { 0, 1, 0, -1, 0, -1, 0, 1 }; + + bn_check_top(a); + bn_check_top(b); + + BN_CTX_start(ctx); + A = BN_CTX_get(ctx); + B = BN_CTX_get(ctx); + if (B == NULL) + goto end; + + err = !BN_copy(A, a); + if (err) + goto end; + err = !BN_copy(B, b); + if (err) + goto end; + + /* + * Kronecker symbol, imlemented according to Henri Cohen, + * "A Course in Computational Algebraic Number Theory" + * (algorithm 1.4.10). + */ + + /* Cohen's step 1: */ + + if (BN_is_zero(B)) { + ret = BN_abs_is_word(A, 1); + goto end; + } + + /* Cohen's step 2: */ + + if (!BN_is_odd(A) && !BN_is_odd(B)) { + ret = 0; + goto end; + } + + /* now B is non-zero */ + i = 0; + while (!BN_is_bit_set(B, i)) + i++; + err = !BN_rshift(B, B, i); + if (err) + goto end; + if (i & 1) { + /* i is odd */ + /* (thus B was even, thus A must be odd!) */ + + /* set 'ret' to $(-1)^{(A^2-1)/8}$ */ + ret = tab[BN_lsw(A) & 7]; + } else { + /* i is even */ + ret = 1; + } + + if (B->neg) { + B->neg = 0; + if (A->neg) + ret = -ret; + } + + /* + * now B is positive and odd, so what remains to be done is to compute + * the Jacobi symbol (A/B) and multiply it by 'ret' + */ + + while (1) { + /* Cohen's step 3: */ + + /* B is positive and odd */ + + if (BN_is_zero(A)) { + ret = BN_is_one(B) ? ret : 0; + goto end; + } + + /* now A is non-zero */ + i = 0; + while (!BN_is_bit_set(A, i)) + i++; + err = !BN_rshift(A, A, i); + if (err) + goto end; + if (i & 1) { + /* i is odd */ + /* multiply 'ret' by $(-1)^{(B^2-1)/8}$ */ + ret = ret * tab[BN_lsw(B) & 7]; + } + + /* Cohen's step 4: */ + /* multiply 'ret' by $(-1)^{(A-1)(B-1)/4}$ */ + if ((A->neg ? ~BN_lsw(A) : BN_lsw(A)) & BN_lsw(B) & 2) + ret = -ret; + + /* (A, B) := (B mod |A|, |A|) */ + err = !BN_nnmod(B, B, A, ctx); + if (err) + goto end; + tmp = A; + A = B; + B = tmp; + tmp->neg = 0; + } + end: + BN_CTX_end(ctx); + if (err) + return -2; + else + return ret; +} diff --git a/openssl/crypto/bn/bn_lcl.h b/openssl/crypto/bn/bn_lcl.h index 817c773b6..7cd58830e 100644 --- a/openssl/crypto/bn/bn_lcl.h +++ b/openssl/crypto/bn/bn_lcl.h @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -110,16 +110,15 @@ */ #ifndef HEADER_BN_LCL_H -#define HEADER_BN_LCL_H +# define HEADER_BN_LCL_H -#include <openssl/bn.h> +# include <openssl/bn.h> #ifdef __cplusplus extern "C" { #endif - -/* +/*- * BN_window_bits_for_exponent_size -- macro for sliding window mod_exp functions * * @@ -144,73 +143,86 @@ extern "C" { * (with draws in between). Very small exponents are often selected * with low Hamming weight, so we use w = 1 for b <= 23. */ -#if 1 -#define BN_window_bits_for_exponent_size(b) \ - ((b) > 671 ? 6 : \ - (b) > 239 ? 5 : \ - (b) > 79 ? 4 : \ - (b) > 23 ? 3 : 1) -#else -/* Old SSLeay/OpenSSL table. - * Maximum window size was 5, so this table differs for b==1024; - * but it coincides for other interesting values (b==160, b==512). +# if 1 +# define BN_window_bits_for_exponent_size(b) \ + ((b) > 671 ? 6 : \ + (b) > 239 ? 5 : \ + (b) > 79 ? 4 : \ + (b) > 23 ? 3 : 1) +# else +/* + * Old SSLeay/OpenSSL table. Maximum window size was 5, so this table differs + * for b==1024; but it coincides for other interesting values (b==160, + * b==512). */ -#define BN_window_bits_for_exponent_size(b) \ - ((b) > 255 ? 5 : \ - (b) > 127 ? 4 : \ - (b) > 17 ? 3 : 1) -#endif +# define BN_window_bits_for_exponent_size(b) \ + ((b) > 255 ? 5 : \ + (b) > 127 ? 4 : \ + (b) > 17 ? 3 : 1) +# endif - - -/* BN_mod_exp_mont_conttime is based on the assumption that the - * L1 data cache line width of the target processor is at least - * the following value. +/* + * BN_mod_exp_mont_conttime is based on the assumption that the L1 data cache + * line width of the target processor is at least the following value. */ -#define MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH ( 64 ) -#define MOD_EXP_CTIME_MIN_CACHE_LINE_MASK (MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH - 1) +# define MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH ( 64 ) +# define MOD_EXP_CTIME_MIN_CACHE_LINE_MASK (MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH - 1) -/* Window sizes optimized for fixed window size modular exponentiation - * algorithm (BN_mod_exp_mont_consttime). - * - * To achieve the security goals of BN_mode_exp_mont_consttime, the - * maximum size of the window must not exceed - * log_2(MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH). - * - * Window size thresholds are defined for cache line sizes of 32 and 64, - * cache line sizes where log_2(32)=5 and log_2(64)=6 respectively. A - * window size of 7 should only be used on processors that have a 128 - * byte or greater cache line size. +/* + * Window sizes optimized for fixed window size modular exponentiation + * algorithm (BN_mod_exp_mont_consttime). To achieve the security goals of + * BN_mode_exp_mont_consttime, the maximum size of the window must not exceed + * log_2(MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH). Window size thresholds are + * defined for cache line sizes of 32 and 64, cache line sizes where + * log_2(32)=5 and log_2(64)=6 respectively. A window size of 7 should only be + * used on processors that have a 128 byte or greater cache line size. */ -#if MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH == 64 +# if MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH == 64 # define BN_window_bits_for_ctime_exponent_size(b) \ - ((b) > 937 ? 6 : \ - (b) > 306 ? 5 : \ - (b) > 89 ? 4 : \ - (b) > 22 ? 3 : 1) -# define BN_MAX_WINDOW_BITS_FOR_CTIME_EXPONENT_SIZE (6) + ((b) > 937 ? 6 : \ + (b) > 306 ? 5 : \ + (b) > 89 ? 4 : \ + (b) > 22 ? 3 : 1) +# define BN_MAX_WINDOW_BITS_FOR_CTIME_EXPONENT_SIZE (6) -#elif MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH == 32 +# elif MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH == 32 # define BN_window_bits_for_ctime_exponent_size(b) \ - ((b) > 306 ? 5 : \ - (b) > 89 ? 4 : \ - (b) > 22 ? 3 : 1) -# define BN_MAX_WINDOW_BITS_FOR_CTIME_EXPONENT_SIZE (5) - -#endif + ((b) > 306 ? 5 : \ + (b) > 89 ? 4 : \ + (b) > 22 ? 3 : 1) +# define BN_MAX_WINDOW_BITS_FOR_CTIME_EXPONENT_SIZE (5) +# endif /* Pentium pro 16,16,16,32,64 */ /* Alpha 16,16,16,16.64 */ -#define BN_MULL_SIZE_NORMAL (16) /* 32 */ -#define BN_MUL_RECURSIVE_SIZE_NORMAL (16) /* 32 less than */ -#define BN_SQR_RECURSIVE_SIZE_NORMAL (16) /* 32 */ -#define BN_MUL_LOW_RECURSIVE_SIZE_NORMAL (32) /* 32 */ -#define BN_MONT_CTX_SET_SIZE_WORD (64) /* 32 */ +# define BN_MULL_SIZE_NORMAL (16)/* 32 */ +# define BN_MUL_RECURSIVE_SIZE_NORMAL (16)/* 32 less than */ +# define BN_SQR_RECURSIVE_SIZE_NORMAL (16)/* 32 */ +# define BN_MUL_LOW_RECURSIVE_SIZE_NORMAL (32)/* 32 */ +# define BN_MONT_CTX_SET_SIZE_WORD (64)/* 32 */ -#if !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) && !defined(PEDANTIC) +/* + * 2011-02-22 SMS. In various places, a size_t variable or a type cast to + * size_t was used to perform integer-only operations on pointers. This + * failed on VMS with 64-bit pointers (CC /POINTER_SIZE = 64) because size_t + * is still only 32 bits. What's needed in these cases is an integer type + * with the same size as a pointer, which size_t is not certain to be. The + * only fix here is VMS-specific. + */ +# if defined(OPENSSL_SYS_VMS) +# if __INITIAL_POINTER_SIZE == 64 +# define PTR_SIZE_INT long long +# else /* __INITIAL_POINTER_SIZE == 64 */ +# define PTR_SIZE_INT int +# endif /* __INITIAL_POINTER_SIZE == 64 [else] */ +# elif !defined(PTR_SIZE_INT) /* defined(OPENSSL_SYS_VMS) */ +# define PTR_SIZE_INT size_t +# endif /* defined(OPENSSL_SYS_VMS) [else] */ + +# if !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) && !defined(PEDANTIC) /* * BN_UMULT_HIGH section. * @@ -232,281 +244,291 @@ extern "C" { * exhibiting "native" performance in C. That's what BN_UMULT_HIGH * macro is about:-) * - * <appro@fy.chalmers.se> + * <appro@fy.chalmers.se> */ -# if defined(__alpha) && (defined(SIXTY_FOUR_BIT_LONG) || defined(SIXTY_FOUR_BIT)) -# if defined(__DECC) -# include <c_asm.h> -# define BN_UMULT_HIGH(a,b) (BN_ULONG)asm("umulh %a0,%a1,%v0",(a),(b)) -# elif defined(__GNUC__) && __GNUC__>=2 -# define BN_UMULT_HIGH(a,b) ({ \ - register BN_ULONG ret; \ - asm ("umulh %1,%2,%0" \ - : "=r"(ret) \ - : "r"(a), "r"(b)); \ - ret; }) -# endif /* compiler */ -# elif defined(_ARCH_PPC) && defined(__64BIT__) && defined(SIXTY_FOUR_BIT_LONG) -# if defined(__GNUC__) && __GNUC__>=2 -# define BN_UMULT_HIGH(a,b) ({ \ - register BN_ULONG ret; \ - asm ("mulhdu %0,%1,%2" \ - : "=r"(ret) \ - : "r"(a), "r"(b)); \ - ret; }) -# endif /* compiler */ -# elif (defined(__x86_64) || defined(__x86_64__)) && \ +# if defined(__alpha) && (defined(SIXTY_FOUR_BIT_LONG) || defined(SIXTY_FOUR_BIT)) +# if defined(__DECC) +# include <c_asm.h> +# define BN_UMULT_HIGH(a,b) (BN_ULONG)asm("umulh %a0,%a1,%v0",(a),(b)) +# elif defined(__GNUC__) && __GNUC__>=2 +# define BN_UMULT_HIGH(a,b) ({ \ + register BN_ULONG ret; \ + asm ("umulh %1,%2,%0" \ + : "=r"(ret) \ + : "r"(a), "r"(b)); \ + ret; }) +# endif /* compiler */ +# elif defined(_ARCH_PPC) && defined(__64BIT__) && defined(SIXTY_FOUR_BIT_LONG) +# if defined(__GNUC__) && __GNUC__>=2 +# define BN_UMULT_HIGH(a,b) ({ \ + register BN_ULONG ret; \ + asm ("mulhdu %0,%1,%2" \ + : "=r"(ret) \ + : "r"(a), "r"(b)); \ + ret; }) +# endif /* compiler */ +# elif (defined(__x86_64) || defined(__x86_64__)) && \ (defined(SIXTY_FOUR_BIT_LONG) || defined(SIXTY_FOUR_BIT)) -# if defined(__GNUC__) && __GNUC__>=2 -# define BN_UMULT_HIGH(a,b) ({ \ - register BN_ULONG ret,discard; \ - asm ("mulq %3" \ - : "=a"(discard),"=d"(ret) \ - : "a"(a), "g"(b) \ - : "cc"); \ - ret; }) -# define BN_UMULT_LOHI(low,high,a,b) \ - asm ("mulq %3" \ - : "=a"(low),"=d"(high) \ - : "a"(a),"g"(b) \ - : "cc"); -# endif -# elif (defined(_M_AMD64) || defined(_M_X64)) && defined(SIXTY_FOUR_BIT) -# if defined(_MSC_VER) && _MSC_VER>=1400 - unsigned __int64 __umulh (unsigned __int64 a,unsigned __int64 b); - unsigned __int64 _umul128 (unsigned __int64 a,unsigned __int64 b, - unsigned __int64 *h); -# pragma intrinsic(__umulh,_umul128) -# define BN_UMULT_HIGH(a,b) __umulh((a),(b)) -# define BN_UMULT_LOHI(low,high,a,b) ((low)=_umul128((a),(b),&(high))) -# endif -# elif defined(__mips) && (defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG)) -# if defined(__GNUC__) && __GNUC__>=2 -# if __GNUC__>=4 && __GNUC_MINOR__>=4 /* "h" constraint is no more since 4.4 */ -# define BN_UMULT_HIGH(a,b) (((__uint128_t)(a)*(b))>>64) -# define BN_UMULT_LOHI(low,high,a,b) ({ \ - __uint128_t ret=(__uint128_t)(a)*(b); \ - (high)=ret>>64; (low)=ret; }) -# else -# define BN_UMULT_HIGH(a,b) ({ \ - register BN_ULONG ret; \ - asm ("dmultu %1,%2" \ - : "=h"(ret) \ - : "r"(a), "r"(b) : "l"); \ - ret; }) +# if defined(__GNUC__) && __GNUC__>=2 +# define BN_UMULT_HIGH(a,b) ({ \ + register BN_ULONG ret,discard; \ + asm ("mulq %3" \ + : "=a"(discard),"=d"(ret) \ + : "a"(a), "g"(b) \ + : "cc"); \ + ret; }) +# define BN_UMULT_LOHI(low,high,a,b) \ + asm ("mulq %3" \ + : "=a"(low),"=d"(high) \ + : "a"(a),"g"(b) \ + : "cc"); +# endif +# elif (defined(_M_AMD64) || defined(_M_X64)) && defined(SIXTY_FOUR_BIT) +# if defined(_MSC_VER) && _MSC_VER>=1400 +unsigned __int64 __umulh(unsigned __int64 a, unsigned __int64 b); +unsigned __int64 _umul128(unsigned __int64 a, unsigned __int64 b, + unsigned __int64 *h); +# pragma intrinsic(__umulh,_umul128) +# define BN_UMULT_HIGH(a,b) __umulh((a),(b)) +# define BN_UMULT_LOHI(low,high,a,b) ((low)=_umul128((a),(b),&(high))) +# endif +# elif defined(__mips) && (defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG)) +# if defined(__GNUC__) && __GNUC__>=2 +# if __GNUC__>=4 && __GNUC_MINOR__>=4 + /* "h" constraint is no more since 4.4 */ +# define BN_UMULT_HIGH(a,b) (((__uint128_t)(a)*(b))>>64) +# define BN_UMULT_LOHI(low,high,a,b) ({ \ + __uint128_t ret=(__uint128_t)(a)*(b); \ + (high)=ret>>64; (low)=ret; }) +# else +# define BN_UMULT_HIGH(a,b) ({ \ + register BN_ULONG ret; \ + asm ("dmultu %1,%2" \ + : "=h"(ret) \ + : "r"(a), "r"(b) : "l"); \ + ret; }) # define BN_UMULT_LOHI(low,high,a,b)\ - asm ("dmultu %2,%3" \ - : "=l"(low),"=h"(high) \ - : "r"(a), "r"(b)); + asm ("dmultu %2,%3" \ + : "=l"(low),"=h"(high) \ + : "r"(a), "r"(b)); # endif -# endif -# endif /* cpu */ -#endif /* OPENSSL_NO_ASM */ +# endif +# elif defined(__aarch64__) && defined(SIXTY_FOUR_BIT_LONG) +# if defined(__GNUC__) && __GNUC__>=2 +# define BN_UMULT_HIGH(a,b) ({ \ + register BN_ULONG ret; \ + asm ("umulh %0,%1,%2" \ + : "=r"(ret) \ + : "r"(a), "r"(b)); \ + ret; }) +# endif +# endif /* cpu */ +# endif /* OPENSSL_NO_ASM */ /************************************************************* * Using the long long type */ -#define Lw(t) (((BN_ULONG)(t))&BN_MASK2) -#define Hw(t) (((BN_ULONG)((t)>>BN_BITS2))&BN_MASK2) - -#ifdef BN_DEBUG_RAND -#define bn_clear_top2max(a) \ - { \ - int ind = (a)->dmax - (a)->top; \ - BN_ULONG *ftl = &(a)->d[(a)->top-1]; \ - for (; ind != 0; ind--) \ - *(++ftl) = 0x0; \ - } -#else -#define bn_clear_top2max(a) -#endif - -#ifdef BN_LLONG -#define mul_add(r,a,w,c) { \ - BN_ULLONG t; \ - t=(BN_ULLONG)w * (a) + (r) + (c); \ - (r)= Lw(t); \ - (c)= Hw(t); \ - } - -#define mul(r,a,w,c) { \ - BN_ULLONG t; \ - t=(BN_ULLONG)w * (a) + (c); \ - (r)= Lw(t); \ - (c)= Hw(t); \ - } - -#define sqr(r0,r1,a) { \ - BN_ULLONG t; \ - t=(BN_ULLONG)(a)*(a); \ - (r0)=Lw(t); \ - (r1)=Hw(t); \ - } - -#elif defined(BN_UMULT_LOHI) -#define mul_add(r,a,w,c) { \ - BN_ULONG high,low,ret,tmp=(a); \ - ret = (r); \ - BN_UMULT_LOHI(low,high,w,tmp); \ - ret += (c); \ - (c) = (ret<(c))?1:0; \ - (c) += high; \ - ret += low; \ - (c) += (ret<low)?1:0; \ - (r) = ret; \ - } - -#define mul(r,a,w,c) { \ - BN_ULONG high,low,ret,ta=(a); \ - BN_UMULT_LOHI(low,high,w,ta); \ - ret = low + (c); \ - (c) = high; \ - (c) += (ret<low)?1:0; \ - (r) = ret; \ - } - -#define sqr(r0,r1,a) { \ - BN_ULONG tmp=(a); \ - BN_UMULT_LOHI(r0,r1,tmp,tmp); \ - } - -#elif defined(BN_UMULT_HIGH) -#define mul_add(r,a,w,c) { \ - BN_ULONG high,low,ret,tmp=(a); \ - ret = (r); \ - high= BN_UMULT_HIGH(w,tmp); \ - ret += (c); \ - low = (w) * tmp; \ - (c) = (ret<(c))?1:0; \ - (c) += high; \ - ret += low; \ - (c) += (ret<low)?1:0; \ - (r) = ret; \ - } - -#define mul(r,a,w,c) { \ - BN_ULONG high,low,ret,ta=(a); \ - low = (w) * ta; \ - high= BN_UMULT_HIGH(w,ta); \ - ret = low + (c); \ - (c) = high; \ - (c) += (ret<low)?1:0; \ - (r) = ret; \ - } - -#define sqr(r0,r1,a) { \ - BN_ULONG tmp=(a); \ - (r0) = tmp * tmp; \ - (r1) = BN_UMULT_HIGH(tmp,tmp); \ - } - -#else +# define Lw(t) (((BN_ULONG)(t))&BN_MASK2) +# define Hw(t) (((BN_ULONG)((t)>>BN_BITS2))&BN_MASK2) + +# ifdef BN_DEBUG_RAND +# define bn_clear_top2max(a) \ + { \ + int ind = (a)->dmax - (a)->top; \ + BN_ULONG *ftl = &(a)->d[(a)->top-1]; \ + for (; ind != 0; ind--) \ + *(++ftl) = 0x0; \ + } +# else +# define bn_clear_top2max(a) +# endif + +# ifdef BN_LLONG +# define mul_add(r,a,w,c) { \ + BN_ULLONG t; \ + t=(BN_ULLONG)w * (a) + (r) + (c); \ + (r)= Lw(t); \ + (c)= Hw(t); \ + } + +# define mul(r,a,w,c) { \ + BN_ULLONG t; \ + t=(BN_ULLONG)w * (a) + (c); \ + (r)= Lw(t); \ + (c)= Hw(t); \ + } + +# define sqr(r0,r1,a) { \ + BN_ULLONG t; \ + t=(BN_ULLONG)(a)*(a); \ + (r0)=Lw(t); \ + (r1)=Hw(t); \ + } + +# elif defined(BN_UMULT_LOHI) +# define mul_add(r,a,w,c) { \ + BN_ULONG high,low,ret,tmp=(a); \ + ret = (r); \ + BN_UMULT_LOHI(low,high,w,tmp); \ + ret += (c); \ + (c) = (ret<(c))?1:0; \ + (c) += high; \ + ret += low; \ + (c) += (ret<low)?1:0; \ + (r) = ret; \ + } + +# define mul(r,a,w,c) { \ + BN_ULONG high,low,ret,ta=(a); \ + BN_UMULT_LOHI(low,high,w,ta); \ + ret = low + (c); \ + (c) = high; \ + (c) += (ret<low)?1:0; \ + (r) = ret; \ + } + +# define sqr(r0,r1,a) { \ + BN_ULONG tmp=(a); \ + BN_UMULT_LOHI(r0,r1,tmp,tmp); \ + } + +# elif defined(BN_UMULT_HIGH) +# define mul_add(r,a,w,c) { \ + BN_ULONG high,low,ret,tmp=(a); \ + ret = (r); \ + high= BN_UMULT_HIGH(w,tmp); \ + ret += (c); \ + low = (w) * tmp; \ + (c) = (ret<(c))?1:0; \ + (c) += high; \ + ret += low; \ + (c) += (ret<low)?1:0; \ + (r) = ret; \ + } + +# define mul(r,a,w,c) { \ + BN_ULONG high,low,ret,ta=(a); \ + low = (w) * ta; \ + high= BN_UMULT_HIGH(w,ta); \ + ret = low + (c); \ + (c) = high; \ + (c) += (ret<low)?1:0; \ + (r) = ret; \ + } + +# define sqr(r0,r1,a) { \ + BN_ULONG tmp=(a); \ + (r0) = tmp * tmp; \ + (r1) = BN_UMULT_HIGH(tmp,tmp); \ + } + +# else /************************************************************* * No long long type */ -#define LBITS(a) ((a)&BN_MASK2l) -#define HBITS(a) (((a)>>BN_BITS4)&BN_MASK2l) -#define L2HBITS(a) (((a)<<BN_BITS4)&BN_MASK2) +# define LBITS(a) ((a)&BN_MASK2l) +# define HBITS(a) (((a)>>BN_BITS4)&BN_MASK2l) +# define L2HBITS(a) (((a)<<BN_BITS4)&BN_MASK2) -#define LLBITS(a) ((a)&BN_MASKl) -#define LHBITS(a) (((a)>>BN_BITS2)&BN_MASKl) -#define LL2HBITS(a) ((BN_ULLONG)((a)&BN_MASKl)<<BN_BITS2) +# define LLBITS(a) ((a)&BN_MASKl) +# define LHBITS(a) (((a)>>BN_BITS2)&BN_MASKl) +# define LL2HBITS(a) ((BN_ULLONG)((a)&BN_MASKl)<<BN_BITS2) -#define mul64(l,h,bl,bh) \ - { \ - BN_ULONG m,m1,lt,ht; \ +# define mul64(l,h,bl,bh) \ + { \ + BN_ULONG m,m1,lt,ht; \ \ - lt=l; \ - ht=h; \ - m =(bh)*(lt); \ - lt=(bl)*(lt); \ - m1=(bl)*(ht); \ - ht =(bh)*(ht); \ - m=(m+m1)&BN_MASK2; if (m < m1) ht+=L2HBITS((BN_ULONG)1); \ - ht+=HBITS(m); \ - m1=L2HBITS(m); \ - lt=(lt+m1)&BN_MASK2; if (lt < m1) ht++; \ - (l)=lt; \ - (h)=ht; \ - } - -#define sqr64(lo,ho,in) \ - { \ - BN_ULONG l,h,m; \ + lt=l; \ + ht=h; \ + m =(bh)*(lt); \ + lt=(bl)*(lt); \ + m1=(bl)*(ht); \ + ht =(bh)*(ht); \ + m=(m+m1)&BN_MASK2; if (m < m1) ht+=L2HBITS((BN_ULONG)1); \ + ht+=HBITS(m); \ + m1=L2HBITS(m); \ + lt=(lt+m1)&BN_MASK2; if (lt < m1) ht++; \ + (l)=lt; \ + (h)=ht; \ + } + +# define sqr64(lo,ho,in) \ + { \ + BN_ULONG l,h,m; \ \ - h=(in); \ - l=LBITS(h); \ - h=HBITS(h); \ - m =(l)*(h); \ - l*=l; \ - h*=h; \ - h+=(m&BN_MASK2h1)>>(BN_BITS4-1); \ - m =(m&BN_MASK2l)<<(BN_BITS4+1); \ - l=(l+m)&BN_MASK2; if (l < m) h++; \ - (lo)=l; \ - (ho)=h; \ - } - -#define mul_add(r,a,bl,bh,c) { \ - BN_ULONG l,h; \ + h=(in); \ + l=LBITS(h); \ + h=HBITS(h); \ + m =(l)*(h); \ + l*=l; \ + h*=h; \ + h+=(m&BN_MASK2h1)>>(BN_BITS4-1); \ + m =(m&BN_MASK2l)<<(BN_BITS4+1); \ + l=(l+m)&BN_MASK2; if (l < m) h++; \ + (lo)=l; \ + (ho)=h; \ + } + +# define mul_add(r,a,bl,bh,c) { \ + BN_ULONG l,h; \ \ - h= (a); \ - l=LBITS(h); \ - h=HBITS(h); \ - mul64(l,h,(bl),(bh)); \ + h= (a); \ + l=LBITS(h); \ + h=HBITS(h); \ + mul64(l,h,(bl),(bh)); \ \ - /* non-multiply part */ \ - l=(l+(c))&BN_MASK2; if (l < (c)) h++; \ - (c)=(r); \ - l=(l+(c))&BN_MASK2; if (l < (c)) h++; \ - (c)=h&BN_MASK2; \ - (r)=l; \ - } - -#define mul(r,a,bl,bh,c) { \ - BN_ULONG l,h; \ + /* non-multiply part */ \ + l=(l+(c))&BN_MASK2; if (l < (c)) h++; \ + (c)=(r); \ + l=(l+(c))&BN_MASK2; if (l < (c)) h++; \ + (c)=h&BN_MASK2; \ + (r)=l; \ + } + +# define mul(r,a,bl,bh,c) { \ + BN_ULONG l,h; \ \ - h= (a); \ - l=LBITS(h); \ - h=HBITS(h); \ - mul64(l,h,(bl),(bh)); \ + h= (a); \ + l=LBITS(h); \ + h=HBITS(h); \ + mul64(l,h,(bl),(bh)); \ \ - /* non-multiply part */ \ - l+=(c); if ((l&BN_MASK2) < (c)) h++; \ - (c)=h&BN_MASK2; \ - (r)=l&BN_MASK2; \ - } -#endif /* !BN_LLONG */ - -#if defined(OPENSSL_DOING_MAKEDEPEND) && defined(OPENSSL_FIPS) -#undef bn_div_words -#endif - -void bn_mul_normal(BN_ULONG *r,BN_ULONG *a,int na,BN_ULONG *b,int nb); -void bn_mul_comba8(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b); -void bn_mul_comba4(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b); + /* non-multiply part */ \ + l+=(c); if ((l&BN_MASK2) < (c)) h++; \ + (c)=h&BN_MASK2; \ + (r)=l&BN_MASK2; \ + } +# endif /* !BN_LLONG */ + +# if defined(OPENSSL_DOING_MAKEDEPEND) && defined(OPENSSL_FIPS) +# undef bn_div_words +# endif + +void bn_mul_normal(BN_ULONG *r, BN_ULONG *a, int na, BN_ULONG *b, int nb); +void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b); +void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b); void bn_sqr_normal(BN_ULONG *r, const BN_ULONG *a, int n, BN_ULONG *tmp); -void bn_sqr_comba8(BN_ULONG *r,const BN_ULONG *a); -void bn_sqr_comba4(BN_ULONG *r,const BN_ULONG *a); -int bn_cmp_words(const BN_ULONG *a,const BN_ULONG *b,int n); -int bn_cmp_part_words(const BN_ULONG *a, const BN_ULONG *b, - int cl, int dl); -void bn_mul_recursive(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,int n2, - int dna,int dnb,BN_ULONG *t); -void bn_mul_part_recursive(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b, - int n,int tna,int tnb,BN_ULONG *t); -void bn_sqr_recursive(BN_ULONG *r,const BN_ULONG *a, int n2, BN_ULONG *t); -void bn_mul_low_normal(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b, int n); -void bn_mul_low_recursive(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,int n2, - BN_ULONG *t); -void bn_mul_high(BN_ULONG *r,BN_ULONG *a,BN_ULONG *b,BN_ULONG *l,int n2, - BN_ULONG *t); +void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a); +void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a); +int bn_cmp_words(const BN_ULONG *a, const BN_ULONG *b, int n); +int bn_cmp_part_words(const BN_ULONG *a, const BN_ULONG *b, int cl, int dl); +void bn_mul_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2, + int dna, int dnb, BN_ULONG *t); +void bn_mul_part_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, + int n, int tna, int tnb, BN_ULONG *t); +void bn_sqr_recursive(BN_ULONG *r, const BN_ULONG *a, int n2, BN_ULONG *t); +void bn_mul_low_normal(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n); +void bn_mul_low_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2, + BN_ULONG *t); +void bn_mul_high(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, BN_ULONG *l, int n2, + BN_ULONG *t); BN_ULONG bn_add_part_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, - int cl, int dl); + int cl, int dl); BN_ULONG bn_sub_part_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b, - int cl, int dl); -int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, const BN_ULONG *np,const BN_ULONG *n0, int num); + int cl, int dl); +int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp, + const BN_ULONG *np, const BN_ULONG *n0, int num); #ifdef __cplusplus } diff --git a/openssl/crypto/bn/bn_lib.c b/openssl/crypto/bn/bn_lib.c index d5a211e28..80105fff4 100644 --- a/openssl/crypto/bn/bn_lib.c +++ b/openssl/crypto/bn/bn_lib.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -57,7 +57,7 @@ */ #ifndef BN_DEBUG -# undef NDEBUG /* avoid conflicting definitions */ +# undef NDEBUG /* avoid conflicting definitions */ # define NDEBUG #endif @@ -67,11 +67,12 @@ #include "cryptlib.h" #include "bn_lcl.h" -const char BN_version[]="Big Number" OPENSSL_VERSION_PTEXT; +const char BN_version[] = "Big Number" OPENSSL_VERSION_PTEXT; /* This stuff appears to be completely unused, so is deprecated */ #ifndef OPENSSL_NO_DEPRECATED -/* For a 32 bit machine +/*- + * For a 32 bit machine * 2 - 4 == 128 * 3 - 8 == 256 * 4 - 16 == 512 @@ -80,808 +81,836 @@ const char BN_version[]="Big Number" OPENSSL_VERSION_PTEXT; * 7 - 128 == 4096 * 8 - 256 == 8192 */ -static int bn_limit_bits=0; -static int bn_limit_num=8; /* (1<<bn_limit_bits) */ -static int bn_limit_bits_low=0; -static int bn_limit_num_low=8; /* (1<<bn_limit_bits_low) */ -static int bn_limit_bits_high=0; -static int bn_limit_num_high=8; /* (1<<bn_limit_bits_high) */ -static int bn_limit_bits_mont=0; -static int bn_limit_num_mont=8; /* (1<<bn_limit_bits_mont) */ +static int bn_limit_bits = 0; +static int bn_limit_num = 8; /* (1<<bn_limit_bits) */ +static int bn_limit_bits_low = 0; +static int bn_limit_num_low = 8; /* (1<<bn_limit_bits_low) */ +static int bn_limit_bits_high = 0; +static int bn_limit_num_high = 8; /* (1<<bn_limit_bits_high) */ +static int bn_limit_bits_mont = 0; +static int bn_limit_num_mont = 8; /* (1<<bn_limit_bits_mont) */ void BN_set_params(int mult, int high, int low, int mont) - { - if (mult >= 0) - { - if (mult > (int)(sizeof(int)*8)-1) - mult=sizeof(int)*8-1; - bn_limit_bits=mult; - bn_limit_num=1<<mult; - } - if (high >= 0) - { - if (high > (int)(sizeof(int)*8)-1) - high=sizeof(int)*8-1; - bn_limit_bits_high=high; - bn_limit_num_high=1<<high; - } - if (low >= 0) - { - if (low > (int)(sizeof(int)*8)-1) - low=sizeof(int)*8-1; - bn_limit_bits_low=low; - bn_limit_num_low=1<<low; - } - if (mont >= 0) - { - if (mont > (int)(sizeof(int)*8)-1) - mont=sizeof(int)*8-1; - bn_limit_bits_mont=mont; - bn_limit_num_mont=1<<mont; - } - } +{ + if (mult >= 0) { + if (mult > (int)(sizeof(int) * 8) - 1) + mult = sizeof(int) * 8 - 1; + bn_limit_bits = mult; + bn_limit_num = 1 << mult; + } + if (high >= 0) { + if (high > (int)(sizeof(int) * 8) - 1) + high = sizeof(int) * 8 - 1; + bn_limit_bits_high = high; + bn_limit_num_high = 1 << high; + } + if (low >= 0) { + if (low > (int)(sizeof(int) * 8) - 1) + low = sizeof(int) * 8 - 1; + bn_limit_bits_low = low; + bn_limit_num_low = 1 << low; + } + if (mont >= 0) { + if (mont > (int)(sizeof(int) * 8) - 1) + mont = sizeof(int) * 8 - 1; + bn_limit_bits_mont = mont; + bn_limit_num_mont = 1 << mont; + } +} int BN_get_params(int which) - { - if (which == 0) return(bn_limit_bits); - else if (which == 1) return(bn_limit_bits_high); - else if (which == 2) return(bn_limit_bits_low); - else if (which == 3) return(bn_limit_bits_mont); - else return(0); - } +{ + if (which == 0) + return (bn_limit_bits); + else if (which == 1) + return (bn_limit_bits_high); + else if (which == 2) + return (bn_limit_bits_low); + else if (which == 3) + return (bn_limit_bits_mont); + else + return (0); +} #endif const BIGNUM *BN_value_one(void) - { - static const BN_ULONG data_one=1L; - static const BIGNUM const_one={(BN_ULONG *)&data_one,1,1,0,BN_FLG_STATIC_DATA}; +{ + static const BN_ULONG data_one = 1L; + static const BIGNUM const_one = + { (BN_ULONG *)&data_one, 1, 1, 0, BN_FLG_STATIC_DATA }; - return(&const_one); - } + return (&const_one); +} int BN_num_bits_word(BN_ULONG l) - { - static const unsigned char bits[256]={ - 0,1,2,2,3,3,3,3,4,4,4,4,4,4,4,4, - 5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5, - 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, - 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6, - 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, - 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, - 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, - 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, - }; +{ + static const unsigned char bits[256] = { + 0, 1, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, + 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, + 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, + 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, + }; #if defined(SIXTY_FOUR_BIT_LONG) - if (l & 0xffffffff00000000L) - { - if (l & 0xffff000000000000L) - { - if (l & 0xff00000000000000L) - { - return(bits[(int)(l>>56)]+56); - } - else return(bits[(int)(l>>48)]+48); - } - else - { - if (l & 0x0000ff0000000000L) - { - return(bits[(int)(l>>40)]+40); - } - else return(bits[(int)(l>>32)]+32); - } - } - else + if (l & 0xffffffff00000000L) { + if (l & 0xffff000000000000L) { + if (l & 0xff00000000000000L) { + return (bits[(int)(l >> 56)] + 56); + } else + return (bits[(int)(l >> 48)] + 48); + } else { + if (l & 0x0000ff0000000000L) { + return (bits[(int)(l >> 40)] + 40); + } else + return (bits[(int)(l >> 32)] + 32); + } + } else #else -#ifdef SIXTY_FOUR_BIT - if (l & 0xffffffff00000000LL) - { - if (l & 0xffff000000000000LL) - { - if (l & 0xff00000000000000LL) - { - return(bits[(int)(l>>56)]+56); - } - else return(bits[(int)(l>>48)]+48); - } - else - { - if (l & 0x0000ff0000000000LL) - { - return(bits[(int)(l>>40)]+40); - } - else return(bits[(int)(l>>32)]+32); - } - } - else -#endif +# ifdef SIXTY_FOUR_BIT + if (l & 0xffffffff00000000LL) { + if (l & 0xffff000000000000LL) { + if (l & 0xff00000000000000LL) { + return (bits[(int)(l >> 56)] + 56); + } else + return (bits[(int)(l >> 48)] + 48); + } else { + if (l & 0x0000ff0000000000LL) { + return (bits[(int)(l >> 40)] + 40); + } else + return (bits[(int)(l >> 32)] + 32); + } + } else +# endif #endif - { + { #if defined(THIRTY_TWO_BIT) || defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG) - if (l & 0xffff0000L) - { - if (l & 0xff000000L) - return(bits[(int)(l>>24L)]+24); - else return(bits[(int)(l>>16L)]+16); - } - else + if (l & 0xffff0000L) { + if (l & 0xff000000L) + return (bits[(int)(l >> 24L)] + 24); + else + return (bits[(int)(l >> 16L)] + 16); + } else #endif - { + { #if defined(THIRTY_TWO_BIT) || defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG) - if (l & 0xff00L) - return(bits[(int)(l>>8)]+8); - else + if (l & 0xff00L) + return (bits[(int)(l >> 8)] + 8); + else #endif - return(bits[(int)(l )] ); - } - } - } + return (bits[(int)(l)]); + } + } +} int BN_num_bits(const BIGNUM *a) - { - int i = a->top - 1; - bn_check_top(a); +{ + int i = a->top - 1; + bn_check_top(a); - if (BN_is_zero(a)) return 0; - return ((i*BN_BITS2) + BN_num_bits_word(a->d[i])); - } + if (BN_is_zero(a)) + return 0; + return ((i * BN_BITS2) + BN_num_bits_word(a->d[i])); +} void BN_clear_free(BIGNUM *a) - { - int i; - - if (a == NULL) return; - bn_check_top(a); - if (a->d != NULL) - { - OPENSSL_cleanse(a->d,a->dmax*sizeof(a->d[0])); - if (!(BN_get_flags(a,BN_FLG_STATIC_DATA))) - OPENSSL_free(a->d); - } - i=BN_get_flags(a,BN_FLG_MALLOCED); - OPENSSL_cleanse(a,sizeof(BIGNUM)); - if (i) - OPENSSL_free(a); - } +{ + int i; + + if (a == NULL) + return; + bn_check_top(a); + if (a->d != NULL) { + OPENSSL_cleanse(a->d, a->dmax * sizeof(a->d[0])); + if (!(BN_get_flags(a, BN_FLG_STATIC_DATA))) + OPENSSL_free(a->d); + } + i = BN_get_flags(a, BN_FLG_MALLOCED); + OPENSSL_cleanse(a, sizeof(BIGNUM)); + if (i) + OPENSSL_free(a); +} void BN_free(BIGNUM *a) - { - if (a == NULL) return; - bn_check_top(a); - if ((a->d != NULL) && !(BN_get_flags(a,BN_FLG_STATIC_DATA))) - OPENSSL_free(a->d); - if (a->flags & BN_FLG_MALLOCED) - OPENSSL_free(a); - else - { +{ + if (a == NULL) + return; + bn_check_top(a); + if ((a->d != NULL) && !(BN_get_flags(a, BN_FLG_STATIC_DATA))) + OPENSSL_free(a->d); + if (a->flags & BN_FLG_MALLOCED) + OPENSSL_free(a); + else { #ifndef OPENSSL_NO_DEPRECATED - a->flags|=BN_FLG_FREE; + a->flags |= BN_FLG_FREE; #endif - a->d = NULL; - } - } + a->d = NULL; + } +} void BN_init(BIGNUM *a) - { - memset(a,0,sizeof(BIGNUM)); - bn_check_top(a); - } +{ + memset(a, 0, sizeof(BIGNUM)); + bn_check_top(a); +} BIGNUM *BN_new(void) - { - BIGNUM *ret; - - if ((ret=(BIGNUM *)OPENSSL_malloc(sizeof(BIGNUM))) == NULL) - { - BNerr(BN_F_BN_NEW,ERR_R_MALLOC_FAILURE); - return(NULL); - } - ret->flags=BN_FLG_MALLOCED; - ret->top=0; - ret->neg=0; - ret->dmax=0; - ret->d=NULL; - bn_check_top(ret); - return(ret); - } +{ + BIGNUM *ret; + + if ((ret = (BIGNUM *)OPENSSL_malloc(sizeof(BIGNUM))) == NULL) { + BNerr(BN_F_BN_NEW, ERR_R_MALLOC_FAILURE); + return (NULL); + } + ret->flags = BN_FLG_MALLOCED; + ret->top = 0; + ret->neg = 0; + ret->dmax = 0; + ret->d = NULL; + bn_check_top(ret); + return (ret); +} /* This is used both by bn_expand2() and bn_dup_expand() */ /* The caller MUST check that words > b->dmax before calling this */ static BN_ULONG *bn_expand_internal(const BIGNUM *b, int words) - { - BN_ULONG *A,*a = NULL; - const BN_ULONG *B; - int i; - - bn_check_top(b); - - if (words > (INT_MAX/(4*BN_BITS2))) - { - BNerr(BN_F_BN_EXPAND_INTERNAL,BN_R_BIGNUM_TOO_LONG); - return NULL; - } - if (BN_get_flags(b,BN_FLG_STATIC_DATA)) - { - BNerr(BN_F_BN_EXPAND_INTERNAL,BN_R_EXPAND_ON_STATIC_BIGNUM_DATA); - return(NULL); - } - a=A=(BN_ULONG *)OPENSSL_malloc(sizeof(BN_ULONG)*words); - if (A == NULL) - { - BNerr(BN_F_BN_EXPAND_INTERNAL,ERR_R_MALLOC_FAILURE); - return(NULL); - } +{ + BN_ULONG *A, *a = NULL; + const BN_ULONG *B; + int i; + + bn_check_top(b); + + if (words > (INT_MAX / (4 * BN_BITS2))) { + BNerr(BN_F_BN_EXPAND_INTERNAL, BN_R_BIGNUM_TOO_LONG); + return NULL; + } + if (BN_get_flags(b, BN_FLG_STATIC_DATA)) { + BNerr(BN_F_BN_EXPAND_INTERNAL, BN_R_EXPAND_ON_STATIC_BIGNUM_DATA); + return (NULL); + } + a = A = (BN_ULONG *)OPENSSL_malloc(sizeof(BN_ULONG) * words); + if (A == NULL) { + BNerr(BN_F_BN_EXPAND_INTERNAL, ERR_R_MALLOC_FAILURE); + return (NULL); + } #ifdef PURIFY - /* Valgrind complains in BN_consttime_swap because we process the whole - * array even if it's not initialised yet. This doesn't matter in that - * function - what's important is constant time operation (we're not - * actually going to use the data) - */ - memset(a, 0, sizeof(BN_ULONG)*words); + /* + * Valgrind complains in BN_consttime_swap because we process the whole + * array even if it's not initialised yet. This doesn't matter in that + * function - what's important is constant time operation (we're not + * actually going to use the data) + */ + memset(a, 0, sizeof(BN_ULONG) * words); #endif #if 1 - B=b->d; - /* Check if the previous number needs to be copied */ - if (B != NULL) - { - for (i=b->top>>2; i>0; i--,A+=4,B+=4) - { - /* - * The fact that the loop is unrolled - * 4-wise is a tribute to Intel. It's - * the one that doesn't have enough - * registers to accomodate more data. - * I'd unroll it 8-wise otherwise:-) - * - * <appro@fy.chalmers.se> - */ - BN_ULONG a0,a1,a2,a3; - a0=B[0]; a1=B[1]; a2=B[2]; a3=B[3]; - A[0]=a0; A[1]=a1; A[2]=a2; A[3]=a3; - } - switch (b->top&3) - { - case 3: A[2]=B[2]; - case 2: A[1]=B[1]; - case 1: A[0]=B[0]; - case 0: /* workaround for ultrix cc: without 'case 0', the optimizer does - * the switch table by doing a=top&3; a--; goto jump_table[a]; - * which fails for top== 0 */ - ; - } - } - + B = b->d; + /* Check if the previous number needs to be copied */ + if (B != NULL) { + for (i = b->top >> 2; i > 0; i--, A += 4, B += 4) { + /* + * The fact that the loop is unrolled + * 4-wise is a tribute to Intel. It's + * the one that doesn't have enough + * registers to accomodate more data. + * I'd unroll it 8-wise otherwise:-) + * + * <appro@fy.chalmers.se> + */ + BN_ULONG a0, a1, a2, a3; + a0 = B[0]; + a1 = B[1]; + a2 = B[2]; + a3 = B[3]; + A[0] = a0; + A[1] = a1; + A[2] = a2; + A[3] = a3; + } + /* + * workaround for ultrix cc: without 'case 0', the optimizer does + * the switch table by doing a=top&3; a--; goto jump_table[a]; + * which fails for top== 0 + */ + switch (b->top & 3) { + case 3: + A[2] = B[2]; + case 2: + A[1] = B[1]; + case 1: + A[0] = B[0]; + case 0: + ; + } + } #else - memset(A,0,sizeof(BN_ULONG)*words); - memcpy(A,b->d,sizeof(b->d[0])*b->top); + memset(A, 0, sizeof(BN_ULONG) * words); + memcpy(A, b->d, sizeof(b->d[0]) * b->top); #endif - - return(a); - } - -/* This is an internal function that can be used instead of bn_expand2() - * when there is a need to copy BIGNUMs instead of only expanding the - * data part, while still expanding them. - * Especially useful when needing to expand BIGNUMs that are declared - * 'const' and should therefore not be changed. - * The reason to use this instead of a BN_dup() followed by a bn_expand2() - * is memory allocation overhead. A BN_dup() followed by a bn_expand2() - * will allocate new memory for the BIGNUM data twice, and free it once, - * while bn_dup_expand() makes sure allocation is made only once. + + return (a); +} + +/* + * This is an internal function that can be used instead of bn_expand2() when + * there is a need to copy BIGNUMs instead of only expanding the data part, + * while still expanding them. Especially useful when needing to expand + * BIGNUMs that are declared 'const' and should therefore not be changed. The + * reason to use this instead of a BN_dup() followed by a bn_expand2() is + * memory allocation overhead. A BN_dup() followed by a bn_expand2() will + * allocate new memory for the BIGNUM data twice, and free it once, while + * bn_dup_expand() makes sure allocation is made only once. */ #ifndef OPENSSL_NO_DEPRECATED BIGNUM *bn_dup_expand(const BIGNUM *b, int words) - { - BIGNUM *r = NULL; - - bn_check_top(b); - - /* This function does not work if - * words <= b->dmax && top < words - * because BN_dup() does not preserve 'dmax'! - * (But bn_dup_expand() is not used anywhere yet.) - */ - - if (words > b->dmax) - { - BN_ULONG *a = bn_expand_internal(b, words); - - if (a) - { - r = BN_new(); - if (r) - { - r->top = b->top; - r->dmax = words; - r->neg = b->neg; - r->d = a; - } - else - { - /* r == NULL, BN_new failure */ - OPENSSL_free(a); - } - } - /* If a == NULL, there was an error in allocation in - bn_expand_internal(), and NULL should be returned */ - } - else - { - r = BN_dup(b); - } - - bn_check_top(r); - return r; - } +{ + BIGNUM *r = NULL; + + bn_check_top(b); + + /* + * This function does not work if words <= b->dmax && top < words because + * BN_dup() does not preserve 'dmax'! (But bn_dup_expand() is not used + * anywhere yet.) + */ + + if (words > b->dmax) { + BN_ULONG *a = bn_expand_internal(b, words); + + if (a) { + r = BN_new(); + if (r) { + r->top = b->top; + r->dmax = words; + r->neg = b->neg; + r->d = a; + } else { + /* r == NULL, BN_new failure */ + OPENSSL_free(a); + } + } + /* + * If a == NULL, there was an error in allocation in + * bn_expand_internal(), and NULL should be returned + */ + } else { + r = BN_dup(b); + } + + bn_check_top(r); + return r; +} #endif -/* This is an internal function that should not be used in applications. - * It ensures that 'b' has enough room for a 'words' word number - * and initialises any unused part of b->d with leading zeros. - * It is mostly used by the various BIGNUM routines. If there is an error, - * NULL is returned. If not, 'b' is returned. */ +/* + * This is an internal function that should not be used in applications. It + * ensures that 'b' has enough room for a 'words' word number and initialises + * any unused part of b->d with leading zeros. It is mostly used by the + * various BIGNUM routines. If there is an error, NULL is returned. If not, + * 'b' is returned. + */ BIGNUM *bn_expand2(BIGNUM *b, int words) - { - bn_check_top(b); - - if (words > b->dmax) - { - BN_ULONG *a = bn_expand_internal(b, words); - if(!a) return NULL; - if(b->d) OPENSSL_free(b->d); - b->d=a; - b->dmax=words; - } +{ + bn_check_top(b); + + if (words > b->dmax) { + BN_ULONG *a = bn_expand_internal(b, words); + if (!a) + return NULL; + if (b->d) + OPENSSL_free(b->d); + b->d = a; + b->dmax = words; + } /* None of this should be necessary because of what b->top means! */ #if 0 - /* NB: bn_wexpand() calls this only if the BIGNUM really has to grow */ - if (b->top < b->dmax) - { - int i; - BN_ULONG *A = &(b->d[b->top]); - for (i=(b->dmax - b->top)>>3; i>0; i--,A+=8) - { - A[0]=0; A[1]=0; A[2]=0; A[3]=0; - A[4]=0; A[5]=0; A[6]=0; A[7]=0; - } - for (i=(b->dmax - b->top)&7; i>0; i--,A++) - A[0]=0; - assert(A == &(b->d[b->dmax])); - } + /* + * NB: bn_wexpand() calls this only if the BIGNUM really has to grow + */ + if (b->top < b->dmax) { + int i; + BN_ULONG *A = &(b->d[b->top]); + for (i = (b->dmax - b->top) >> 3; i > 0; i--, A += 8) { + A[0] = 0; + A[1] = 0; + A[2] = 0; + A[3] = 0; + A[4] = 0; + A[5] = 0; + A[6] = 0; + A[7] = 0; + } + for (i = (b->dmax - b->top) & 7; i > 0; i--, A++) + A[0] = 0; + assert(A == &(b->d[b->dmax])); + } #endif - bn_check_top(b); - return b; - } + bn_check_top(b); + return b; +} BIGNUM *BN_dup(const BIGNUM *a) - { - BIGNUM *t; - - if (a == NULL) return NULL; - bn_check_top(a); - - t = BN_new(); - if (t == NULL) return NULL; - if(!BN_copy(t, a)) - { - BN_free(t); - return NULL; - } - bn_check_top(t); - return t; - } +{ + BIGNUM *t; + + if (a == NULL) + return NULL; + bn_check_top(a); + + t = BN_new(); + if (t == NULL) + return NULL; + if (!BN_copy(t, a)) { + BN_free(t); + return NULL; + } + bn_check_top(t); + return t; +} BIGNUM *BN_copy(BIGNUM *a, const BIGNUM *b) - { - int i; - BN_ULONG *A; - const BN_ULONG *B; +{ + int i; + BN_ULONG *A; + const BN_ULONG *B; - bn_check_top(b); + bn_check_top(b); - if (a == b) return(a); - if (bn_wexpand(a,b->top) == NULL) return(NULL); + if (a == b) + return (a); + if (bn_wexpand(a, b->top) == NULL) + return (NULL); #if 1 - A=a->d; - B=b->d; - for (i=b->top>>2; i>0; i--,A+=4,B+=4) - { - BN_ULONG a0,a1,a2,a3; - a0=B[0]; a1=B[1]; a2=B[2]; a3=B[3]; - A[0]=a0; A[1]=a1; A[2]=a2; A[3]=a3; - } - switch (b->top&3) - { - case 3: A[2]=B[2]; - case 2: A[1]=B[1]; - case 1: A[0]=B[0]; - case 0: ; /* ultrix cc workaround, see comments in bn_expand_internal */ - } + A = a->d; + B = b->d; + for (i = b->top >> 2; i > 0; i--, A += 4, B += 4) { + BN_ULONG a0, a1, a2, a3; + a0 = B[0]; + a1 = B[1]; + a2 = B[2]; + a3 = B[3]; + A[0] = a0; + A[1] = a1; + A[2] = a2; + A[3] = a3; + } + /* ultrix cc workaround, see comments in bn_expand_internal */ + switch (b->top & 3) { + case 3: + A[2] = B[2]; + case 2: + A[1] = B[1]; + case 1: + A[0] = B[0]; + case 0:; + } #else - memcpy(a->d,b->d,sizeof(b->d[0])*b->top); + memcpy(a->d, b->d, sizeof(b->d[0]) * b->top); #endif - a->top=b->top; - a->neg=b->neg; - bn_check_top(a); - return(a); - } + a->top = b->top; + a->neg = b->neg; + bn_check_top(a); + return (a); +} void BN_swap(BIGNUM *a, BIGNUM *b) - { - int flags_old_a, flags_old_b; - BN_ULONG *tmp_d; - int tmp_top, tmp_dmax, tmp_neg; - - bn_check_top(a); - bn_check_top(b); - - flags_old_a = a->flags; - flags_old_b = b->flags; - - tmp_d = a->d; - tmp_top = a->top; - tmp_dmax = a->dmax; - tmp_neg = a->neg; - - a->d = b->d; - a->top = b->top; - a->dmax = b->dmax; - a->neg = b->neg; - - b->d = tmp_d; - b->top = tmp_top; - b->dmax = tmp_dmax; - b->neg = tmp_neg; - - a->flags = (flags_old_a & BN_FLG_MALLOCED) | (flags_old_b & BN_FLG_STATIC_DATA); - b->flags = (flags_old_b & BN_FLG_MALLOCED) | (flags_old_a & BN_FLG_STATIC_DATA); - bn_check_top(a); - bn_check_top(b); - } +{ + int flags_old_a, flags_old_b; + BN_ULONG *tmp_d; + int tmp_top, tmp_dmax, tmp_neg; + + bn_check_top(a); + bn_check_top(b); + + flags_old_a = a->flags; + flags_old_b = b->flags; + + tmp_d = a->d; + tmp_top = a->top; + tmp_dmax = a->dmax; + tmp_neg = a->neg; + + a->d = b->d; + a->top = b->top; + a->dmax = b->dmax; + a->neg = b->neg; + + b->d = tmp_d; + b->top = tmp_top; + b->dmax = tmp_dmax; + b->neg = tmp_neg; + + a->flags = + (flags_old_a & BN_FLG_MALLOCED) | (flags_old_b & BN_FLG_STATIC_DATA); + b->flags = + (flags_old_b & BN_FLG_MALLOCED) | (flags_old_a & BN_FLG_STATIC_DATA); + bn_check_top(a); + bn_check_top(b); +} void BN_clear(BIGNUM *a) - { - bn_check_top(a); - if (a->d != NULL) - memset(a->d,0,a->dmax*sizeof(a->d[0])); - a->top=0; - a->neg=0; - } +{ + bn_check_top(a); + if (a->d != NULL) + memset(a->d, 0, a->dmax * sizeof(a->d[0])); + a->top = 0; + a->neg = 0; +} BN_ULONG BN_get_word(const BIGNUM *a) - { - if (a->top > 1) - return BN_MASK2; - else if (a->top == 1) - return a->d[0]; - /* a->top == 0 */ - return 0; - } +{ + if (a->top > 1) + return BN_MASK2; + else if (a->top == 1) + return a->d[0]; + /* a->top == 0 */ + return 0; +} int BN_set_word(BIGNUM *a, BN_ULONG w) - { - bn_check_top(a); - if (bn_expand(a,(int)sizeof(BN_ULONG)*8) == NULL) return(0); - a->neg = 0; - a->d[0] = w; - a->top = (w ? 1 : 0); - bn_check_top(a); - return(1); - } +{ + bn_check_top(a); + if (bn_expand(a, (int)sizeof(BN_ULONG) * 8) == NULL) + return (0); + a->neg = 0; + a->d[0] = w; + a->top = (w ? 1 : 0); + bn_check_top(a); + return (1); +} BIGNUM *BN_bin2bn(const unsigned char *s, int len, BIGNUM *ret) - { - unsigned int i,m; - unsigned int n; - BN_ULONG l; - BIGNUM *bn = NULL; - - if (ret == NULL) - ret = bn = BN_new(); - if (ret == NULL) return(NULL); - bn_check_top(ret); - l=0; - n=len; - if (n == 0) - { - ret->top=0; - return(ret); - } - i=((n-1)/BN_BYTES)+1; - m=((n-1)%(BN_BYTES)); - if (bn_wexpand(ret, (int)i) == NULL) - { - if (bn) BN_free(bn); - return NULL; - } - ret->top=i; - ret->neg=0; - while (n--) - { - l=(l<<8L)| *(s++); - if (m-- == 0) - { - ret->d[--i]=l; - l=0; - m=BN_BYTES-1; - } - } - /* need to call this due to clear byte at top if avoiding - * having the top bit set (-ve number) */ - bn_correct_top(ret); - return(ret); - } +{ + unsigned int i, m; + unsigned int n; + BN_ULONG l; + BIGNUM *bn = NULL; + + if (ret == NULL) + ret = bn = BN_new(); + if (ret == NULL) + return (NULL); + bn_check_top(ret); + l = 0; + n = len; + if (n == 0) { + ret->top = 0; + return (ret); + } + i = ((n - 1) / BN_BYTES) + 1; + m = ((n - 1) % (BN_BYTES)); + if (bn_wexpand(ret, (int)i) == NULL) { + if (bn) + BN_free(bn); + return NULL; + } + ret->top = i; + ret->neg = 0; + while (n--) { + l = (l << 8L) | *(s++); + if (m-- == 0) { + ret->d[--i] = l; + l = 0; + m = BN_BYTES - 1; + } + } + /* + * need to call this due to clear byte at top if avoiding having the top + * bit set (-ve number) + */ + bn_correct_top(ret); + return (ret); +} /* ignore negative */ int BN_bn2bin(const BIGNUM *a, unsigned char *to) - { - int n,i; - BN_ULONG l; - - bn_check_top(a); - n=i=BN_num_bytes(a); - while (i--) - { - l=a->d[i/BN_BYTES]; - *(to++)=(unsigned char)(l>>(8*(i%BN_BYTES)))&0xff; - } - return(n); - } +{ + int n, i; + BN_ULONG l; + + bn_check_top(a); + n = i = BN_num_bytes(a); + while (i--) { + l = a->d[i / BN_BYTES]; + *(to++) = (unsigned char)(l >> (8 * (i % BN_BYTES))) & 0xff; + } + return (n); +} int BN_ucmp(const BIGNUM *a, const BIGNUM *b) - { - int i; - BN_ULONG t1,t2,*ap,*bp; - - bn_check_top(a); - bn_check_top(b); - - i=a->top-b->top; - if (i != 0) return(i); - ap=a->d; - bp=b->d; - for (i=a->top-1; i>=0; i--) - { - t1= ap[i]; - t2= bp[i]; - if (t1 != t2) - return((t1 > t2) ? 1 : -1); - } - return(0); - } +{ + int i; + BN_ULONG t1, t2, *ap, *bp; + + bn_check_top(a); + bn_check_top(b); + + i = a->top - b->top; + if (i != 0) + return (i); + ap = a->d; + bp = b->d; + for (i = a->top - 1; i >= 0; i--) { + t1 = ap[i]; + t2 = bp[i]; + if (t1 != t2) + return ((t1 > t2) ? 1 : -1); + } + return (0); +} int BN_cmp(const BIGNUM *a, const BIGNUM *b) - { - int i; - int gt,lt; - BN_ULONG t1,t2; - - if ((a == NULL) || (b == NULL)) - { - if (a != NULL) - return(-1); - else if (b != NULL) - return(1); - else - return(0); - } - - bn_check_top(a); - bn_check_top(b); - - if (a->neg != b->neg) - { - if (a->neg) - return(-1); - else return(1); - } - if (a->neg == 0) - { gt=1; lt= -1; } - else { gt= -1; lt=1; } - - if (a->top > b->top) return(gt); - if (a->top < b->top) return(lt); - for (i=a->top-1; i>=0; i--) - { - t1=a->d[i]; - t2=b->d[i]; - if (t1 > t2) return(gt); - if (t1 < t2) return(lt); - } - return(0); - } +{ + int i; + int gt, lt; + BN_ULONG t1, t2; + + if ((a == NULL) || (b == NULL)) { + if (a != NULL) + return (-1); + else if (b != NULL) + return (1); + else + return (0); + } + + bn_check_top(a); + bn_check_top(b); + + if (a->neg != b->neg) { + if (a->neg) + return (-1); + else + return (1); + } + if (a->neg == 0) { + gt = 1; + lt = -1; + } else { + gt = -1; + lt = 1; + } + + if (a->top > b->top) + return (gt); + if (a->top < b->top) + return (lt); + for (i = a->top - 1; i >= 0; i--) { + t1 = a->d[i]; + t2 = b->d[i]; + if (t1 > t2) + return (gt); + if (t1 < t2) + return (lt); + } + return (0); +} int BN_set_bit(BIGNUM *a, int n) - { - int i,j,k; - - if (n < 0) - return 0; - - i=n/BN_BITS2; - j=n%BN_BITS2; - if (a->top <= i) - { - if (bn_wexpand(a,i+1) == NULL) return(0); - for(k=a->top; k<i+1; k++) - a->d[k]=0; - a->top=i+1; - } - - a->d[i]|=(((BN_ULONG)1)<<j); - bn_check_top(a); - return(1); - } +{ + int i, j, k; + + if (n < 0) + return 0; + + i = n / BN_BITS2; + j = n % BN_BITS2; + if (a->top <= i) { + if (bn_wexpand(a, i + 1) == NULL) + return (0); + for (k = a->top; k < i + 1; k++) + a->d[k] = 0; + a->top = i + 1; + } + + a->d[i] |= (((BN_ULONG)1) << j); + bn_check_top(a); + return (1); +} int BN_clear_bit(BIGNUM *a, int n) - { - int i,j; +{ + int i, j; - bn_check_top(a); - if (n < 0) return 0; + bn_check_top(a); + if (n < 0) + return 0; - i=n/BN_BITS2; - j=n%BN_BITS2; - if (a->top <= i) return(0); + i = n / BN_BITS2; + j = n % BN_BITS2; + if (a->top <= i) + return (0); - a->d[i]&=(~(((BN_ULONG)1)<<j)); - bn_correct_top(a); - return(1); - } + a->d[i] &= (~(((BN_ULONG)1) << j)); + bn_correct_top(a); + return (1); +} int BN_is_bit_set(const BIGNUM *a, int n) - { - int i,j; - - bn_check_top(a); - if (n < 0) return 0; - i=n/BN_BITS2; - j=n%BN_BITS2; - if (a->top <= i) return 0; - return (int)(((a->d[i])>>j)&((BN_ULONG)1)); - } +{ + int i, j; + + bn_check_top(a); + if (n < 0) + return 0; + i = n / BN_BITS2; + j = n % BN_BITS2; + if (a->top <= i) + return 0; + return (int)(((a->d[i]) >> j) & ((BN_ULONG)1)); +} int BN_mask_bits(BIGNUM *a, int n) - { - int b,w; - - bn_check_top(a); - if (n < 0) return 0; - - w=n/BN_BITS2; - b=n%BN_BITS2; - if (w >= a->top) return 0; - if (b == 0) - a->top=w; - else - { - a->top=w+1; - a->d[w]&= ~(BN_MASK2<<b); - } - bn_correct_top(a); - return(1); - } +{ + int b, w; + + bn_check_top(a); + if (n < 0) + return 0; + + w = n / BN_BITS2; + b = n % BN_BITS2; + if (w >= a->top) + return 0; + if (b == 0) + a->top = w; + else { + a->top = w + 1; + a->d[w] &= ~(BN_MASK2 << b); + } + bn_correct_top(a); + return (1); +} void BN_set_negative(BIGNUM *a, int b) - { - if (b && !BN_is_zero(a)) - a->neg = 1; - else - a->neg = 0; - } +{ + if (b && !BN_is_zero(a)) + a->neg = 1; + else + a->neg = 0; +} int bn_cmp_words(const BN_ULONG *a, const BN_ULONG *b, int n) - { - int i; - BN_ULONG aa,bb; - - aa=a[n-1]; - bb=b[n-1]; - if (aa != bb) return((aa > bb)?1:-1); - for (i=n-2; i>=0; i--) - { - aa=a[i]; - bb=b[i]; - if (aa != bb) return((aa > bb)?1:-1); - } - return(0); - } - -/* Here follows a specialised variants of bn_cmp_words(). It has the - property of performing the operation on arrays of different sizes. - The sizes of those arrays is expressed through cl, which is the - common length ( basicall, min(len(a),len(b)) ), and dl, which is the - delta between the two lengths, calculated as len(a)-len(b). - All lengths are the number of BN_ULONGs... */ - -int bn_cmp_part_words(const BN_ULONG *a, const BN_ULONG *b, - int cl, int dl) - { - int n,i; - n = cl-1; - - if (dl < 0) - { - for (i=dl; i<0; i++) - { - if (b[n-i] != 0) - return -1; /* a < b */ - } - } - if (dl > 0) - { - for (i=dl; i>0; i--) - { - if (a[n+i] != 0) - return 1; /* a > b */ - } - } - return bn_cmp_words(a,b,cl); - } - -/* - * Constant-time conditional swap of a and b. +{ + int i; + BN_ULONG aa, bb; + + aa = a[n - 1]; + bb = b[n - 1]; + if (aa != bb) + return ((aa > bb) ? 1 : -1); + for (i = n - 2; i >= 0; i--) { + aa = a[i]; + bb = b[i]; + if (aa != bb) + return ((aa > bb) ? 1 : -1); + } + return (0); +} + +/* + * Here follows a specialised variants of bn_cmp_words(). It has the + * property of performing the operation on arrays of different sizes. The + * sizes of those arrays is expressed through cl, which is the common length + * ( basicall, min(len(a),len(b)) ), and dl, which is the delta between the + * two lengths, calculated as len(a)-len(b). All lengths are the number of + * BN_ULONGs... + */ + +int bn_cmp_part_words(const BN_ULONG *a, const BN_ULONG *b, int cl, int dl) +{ + int n, i; + n = cl - 1; + + if (dl < 0) { + for (i = dl; i < 0; i++) { + if (b[n - i] != 0) + return -1; /* a < b */ + } + } + if (dl > 0) { + for (i = dl; i > 0; i--) { + if (a[n + i] != 0) + return 1; /* a > b */ + } + } + return bn_cmp_words(a, b, cl); +} + +/* + * Constant-time conditional swap of a and b. * a and b are swapped if condition is not 0. The code assumes that at most one bit of condition is set. * nwords is the number of words to swap. The code assumes that at least nwords are allocated in both a and b, * and that no more than nwords are used by either a or b. * a and b cannot be the same number */ void BN_consttime_swap(BN_ULONG condition, BIGNUM *a, BIGNUM *b, int nwords) - { - BN_ULONG t; - int i; +{ + BN_ULONG t; + int i; - bn_wcheck_size(a, nwords); - bn_wcheck_size(b, nwords); + bn_wcheck_size(a, nwords); + bn_wcheck_size(b, nwords); - assert(a != b); - assert((condition & (condition - 1)) == 0); - assert(sizeof(BN_ULONG) >= sizeof(int)); + assert(a != b); + assert((condition & (condition - 1)) == 0); + assert(sizeof(BN_ULONG) >= sizeof(int)); - condition = ((condition - 1) >> (BN_BITS2 - 1)) - 1; + condition = ((condition - 1) >> (BN_BITS2 - 1)) - 1; - t = (a->top^b->top) & condition; - a->top ^= t; - b->top ^= t; + t = (a->top ^ b->top) & condition; + a->top ^= t; + b->top ^= t; #define BN_CONSTTIME_SWAP(ind) \ - do { \ - t = (a->d[ind] ^ b->d[ind]) & condition; \ - a->d[ind] ^= t; \ - b->d[ind] ^= t; \ - } while (0) - - - switch (nwords) { - default: - for (i = 10; i < nwords; i++) - BN_CONSTTIME_SWAP(i); - /* Fallthrough */ - case 10: BN_CONSTTIME_SWAP(9); /* Fallthrough */ - case 9: BN_CONSTTIME_SWAP(8); /* Fallthrough */ - case 8: BN_CONSTTIME_SWAP(7); /* Fallthrough */ - case 7: BN_CONSTTIME_SWAP(6); /* Fallthrough */ - case 6: BN_CONSTTIME_SWAP(5); /* Fallthrough */ - case 5: BN_CONSTTIME_SWAP(4); /* Fallthrough */ - case 4: BN_CONSTTIME_SWAP(3); /* Fallthrough */ - case 3: BN_CONSTTIME_SWAP(2); /* Fallthrough */ - case 2: BN_CONSTTIME_SWAP(1); /* Fallthrough */ - case 1: BN_CONSTTIME_SWAP(0); - } + do { \ + t = (a->d[ind] ^ b->d[ind]) & condition; \ + a->d[ind] ^= t; \ + b->d[ind] ^= t; \ + } while (0) + + switch (nwords) { + default: + for (i = 10; i < nwords; i++) + BN_CONSTTIME_SWAP(i); + /* Fallthrough */ + case 10: + BN_CONSTTIME_SWAP(9); /* Fallthrough */ + case 9: + BN_CONSTTIME_SWAP(8); /* Fallthrough */ + case 8: + BN_CONSTTIME_SWAP(7); /* Fallthrough */ + case 7: + BN_CONSTTIME_SWAP(6); /* Fallthrough */ + case 6: + BN_CONSTTIME_SWAP(5); /* Fallthrough */ + case 5: + BN_CONSTTIME_SWAP(4); /* Fallthrough */ + case 4: + BN_CONSTTIME_SWAP(3); /* Fallthrough */ + case 3: + BN_CONSTTIME_SWAP(2); /* Fallthrough */ + case 2: + BN_CONSTTIME_SWAP(1); /* Fallthrough */ + case 1: + BN_CONSTTIME_SWAP(0); + } #undef BN_CONSTTIME_SWAP } diff --git a/openssl/crypto/bn/bn_mod.c b/openssl/crypto/bn/bn_mod.c index 77d6ddb91..ffbce890c 100644 --- a/openssl/crypto/bn/bn_mod.c +++ b/openssl/crypto/bn/bn_mod.c @@ -1,6 +1,8 @@ /* crypto/bn/bn_mod.c */ -/* Includes code written by Lenka Fibikova <fibikova@exp-math.uni-essen.de> - * for the OpenSSL project. */ +/* + * Includes code written by Lenka Fibikova <fibikova@exp-math.uni-essen.de> + * for the OpenSSL project. + */ /* ==================================================================== * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved. * @@ -9,7 +11,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -60,21 +62,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -89,10 +91,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -104,7 +106,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -114,188 +116,201 @@ #include "cryptlib.h" #include "bn_lcl.h" - -#if 0 /* now just a #define */ +#if 0 /* now just a #define */ int BN_mod(BIGNUM *rem, const BIGNUM *m, const BIGNUM *d, BN_CTX *ctx) - { - return(BN_div(NULL,rem,m,d,ctx)); - /* note that rem->neg == m->neg (unless the remainder is zero) */ - } +{ + return (BN_div(NULL, rem, m, d, ctx)); + /* note that rem->neg == m->neg (unless the remainder is zero) */ +} #endif - int BN_nnmod(BIGNUM *r, const BIGNUM *m, const BIGNUM *d, BN_CTX *ctx) - { - /* like BN_mod, but returns non-negative remainder - * (i.e., 0 <= r < |d| always holds) */ - - if (!(BN_mod(r,m,d,ctx))) - return 0; - if (!r->neg) - return 1; - /* now -|d| < r < 0, so we have to set r := r + |d| */ - return (d->neg ? BN_sub : BN_add)(r, r, d); +{ + /* + * like BN_mod, but returns non-negative remainder (i.e., 0 <= r < |d| + * always holds) + */ + + if (!(BN_mod(r, m, d, ctx))) + return 0; + if (!r->neg) + return 1; + /* now -|d| < r < 0, so we have to set r := r + |d| */ + return (d->neg ? BN_sub : BN_add) (r, r, d); } +int BN_mod_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, + BN_CTX *ctx) +{ + if (!BN_add(r, a, b)) + return 0; + return BN_nnmod(r, r, m, ctx); +} -int BN_mod_add(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, BN_CTX *ctx) - { - if (!BN_add(r, a, b)) return 0; - return BN_nnmod(r, r, m, ctx); - } - - -/* BN_mod_add variant that may be used if both a and b are non-negative - * and less than m */ -int BN_mod_add_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m) - { - if (!BN_uadd(r, a, b)) return 0; - if (BN_ucmp(r, m) >= 0) - return BN_usub(r, r, m); - return 1; - } - - -int BN_mod_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, BN_CTX *ctx) - { - if (!BN_sub(r, a, b)) return 0; - return BN_nnmod(r, r, m, ctx); - } - +/* + * BN_mod_add variant that may be used if both a and b are non-negative and + * less than m + */ +int BN_mod_add_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *m) +{ + if (!BN_uadd(r, a, b)) + return 0; + if (BN_ucmp(r, m) >= 0) + return BN_usub(r, r, m); + return 1; +} -/* BN_mod_sub variant that may be used if both a and b are non-negative - * and less than m */ -int BN_mod_sub_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m) - { - if (!BN_sub(r, a, b)) return 0; - if (r->neg) - return BN_add(r, r, m); - return 1; - } +int BN_mod_sub(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, + BN_CTX *ctx) +{ + if (!BN_sub(r, a, b)) + return 0; + return BN_nnmod(r, r, m, ctx); +} +/* + * BN_mod_sub variant that may be used if both a and b are non-negative and + * less than m + */ +int BN_mod_sub_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, + const BIGNUM *m) +{ + if (!BN_sub(r, a, b)) + return 0; + if (r->neg) + return BN_add(r, r, m); + return 1; +} /* slow but works */ int BN_mod_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, const BIGNUM *m, - BN_CTX *ctx) - { - BIGNUM *t; - int ret=0; - - bn_check_top(a); - bn_check_top(b); - bn_check_top(m); - - BN_CTX_start(ctx); - if ((t = BN_CTX_get(ctx)) == NULL) goto err; - if (a == b) - { if (!BN_sqr(t,a,ctx)) goto err; } - else - { if (!BN_mul(t,a,b,ctx)) goto err; } - if (!BN_nnmod(r,t,m,ctx)) goto err; - bn_check_top(r); - ret=1; -err: - BN_CTX_end(ctx); - return(ret); - } - + BN_CTX *ctx) +{ + BIGNUM *t; + int ret = 0; + + bn_check_top(a); + bn_check_top(b); + bn_check_top(m); + + BN_CTX_start(ctx); + if ((t = BN_CTX_get(ctx)) == NULL) + goto err; + if (a == b) { + if (!BN_sqr(t, a, ctx)) + goto err; + } else { + if (!BN_mul(t, a, b, ctx)) + goto err; + } + if (!BN_nnmod(r, t, m, ctx)) + goto err; + bn_check_top(r); + ret = 1; + err: + BN_CTX_end(ctx); + return (ret); +} int BN_mod_sqr(BIGNUM *r, const BIGNUM *a, const BIGNUM *m, BN_CTX *ctx) - { - if (!BN_sqr(r, a, ctx)) return 0; - /* r->neg == 0, thus we don't need BN_nnmod */ - return BN_mod(r, r, m, ctx); - } - +{ + if (!BN_sqr(r, a, ctx)) + return 0; + /* r->neg == 0, thus we don't need BN_nnmod */ + return BN_mod(r, r, m, ctx); +} int BN_mod_lshift1(BIGNUM *r, const BIGNUM *a, const BIGNUM *m, BN_CTX *ctx) - { - if (!BN_lshift1(r, a)) return 0; - bn_check_top(r); - return BN_nnmod(r, r, m, ctx); - } - +{ + if (!BN_lshift1(r, a)) + return 0; + bn_check_top(r); + return BN_nnmod(r, r, m, ctx); +} -/* BN_mod_lshift1 variant that may be used if a is non-negative - * and less than m */ +/* + * BN_mod_lshift1 variant that may be used if a is non-negative and less than + * m + */ int BN_mod_lshift1_quick(BIGNUM *r, const BIGNUM *a, const BIGNUM *m) - { - if (!BN_lshift1(r, a)) return 0; - bn_check_top(r); - if (BN_cmp(r, m) >= 0) - return BN_sub(r, r, m); - return 1; - } - +{ + if (!BN_lshift1(r, a)) + return 0; + bn_check_top(r); + if (BN_cmp(r, m) >= 0) + return BN_sub(r, r, m); + return 1; +} -int BN_mod_lshift(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m, BN_CTX *ctx) - { - BIGNUM *abs_m = NULL; - int ret; +int BN_mod_lshift(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m, + BN_CTX *ctx) +{ + BIGNUM *abs_m = NULL; + int ret; - if (!BN_nnmod(r, a, m, ctx)) return 0; + if (!BN_nnmod(r, a, m, ctx)) + return 0; - if (m->neg) - { - abs_m = BN_dup(m); - if (abs_m == NULL) return 0; - abs_m->neg = 0; - } - - ret = BN_mod_lshift_quick(r, r, n, (abs_m ? abs_m : m)); - bn_check_top(r); + if (m->neg) { + abs_m = BN_dup(m); + if (abs_m == NULL) + return 0; + abs_m->neg = 0; + } - if (abs_m) - BN_free(abs_m); - return ret; - } + ret = BN_mod_lshift_quick(r, r, n, (abs_m ? abs_m : m)); + bn_check_top(r); + if (abs_m) + BN_free(abs_m); + return ret; +} -/* BN_mod_lshift variant that may be used if a is non-negative - * and less than m */ +/* + * BN_mod_lshift variant that may be used if a is non-negative and less than + * m + */ int BN_mod_lshift_quick(BIGNUM *r, const BIGNUM *a, int n, const BIGNUM *m) - { - if (r != a) - { - if (BN_copy(r, a) == NULL) return 0; - } - - while (n > 0) - { - int max_shift; - - /* 0 < r < m */ - max_shift = BN_num_bits(m) - BN_num_bits(r); - /* max_shift >= 0 */ - - if (max_shift < 0) - { - BNerr(BN_F_BN_MOD_LSHIFT_QUICK, BN_R_INPUT_NOT_REDUCED); - return 0; - } - - if (max_shift > n) - max_shift = n; - - if (max_shift) - { - if (!BN_lshift(r, r, max_shift)) return 0; - n -= max_shift; - } - else - { - if (!BN_lshift1(r, r)) return 0; - --n; - } - - /* BN_num_bits(r) <= BN_num_bits(m) */ - - if (BN_cmp(r, m) >= 0) - { - if (!BN_sub(r, r, m)) return 0; - } - } - bn_check_top(r); - - return 1; - } +{ + if (r != a) { + if (BN_copy(r, a) == NULL) + return 0; + } + + while (n > 0) { + int max_shift; + + /* 0 < r < m */ + max_shift = BN_num_bits(m) - BN_num_bits(r); + /* max_shift >= 0 */ + + if (max_shift < 0) { + BNerr(BN_F_BN_MOD_LSHIFT_QUICK, BN_R_INPUT_NOT_REDUCED); + return 0; + } + + if (max_shift > n) + max_shift = n; + + if (max_shift) { + if (!BN_lshift(r, r, max_shift)) + return 0; + n -= max_shift; + } else { + if (!BN_lshift1(r, r)) + return 0; + --n; + } + + /* BN_num_bits(r) <= BN_num_bits(m) */ + + if (BN_cmp(r, m) >= 0) { + if (!BN_sub(r, r, m)) + return 0; + } + } + bn_check_top(r); + + return 1; +} diff --git a/openssl/crypto/bn/bn_mont.c b/openssl/crypto/bn/bn_mont.c index ee8532c7d..aadd5db1d 100644 --- a/openssl/crypto/bn/bn_mont.c +++ b/openssl/crypto/bn/bn_mont.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -120,396 +120,436 @@ #include "cryptlib.h" #include "bn_lcl.h" -#define MONT_WORD /* use the faster word-based algorithm */ +#define MONT_WORD /* use the faster word-based algorithm */ #ifdef MONT_WORD static int BN_from_montgomery_word(BIGNUM *ret, BIGNUM *r, BN_MONT_CTX *mont); #endif int BN_mod_mul_montgomery(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, - BN_MONT_CTX *mont, BN_CTX *ctx) - { - BIGNUM *tmp; - int ret=0; + BN_MONT_CTX *mont, BN_CTX *ctx) +{ + BIGNUM *tmp; + int ret = 0; #if defined(OPENSSL_BN_ASM_MONT) && defined(MONT_WORD) - int num = mont->N.top; - - if (num>1 && a->top==num && b->top==num) - { - if (bn_wexpand(r,num) == NULL) return(0); - if (bn_mul_mont(r->d,a->d,b->d,mont->N.d,mont->n0,num)) - { - r->neg = a->neg^b->neg; - r->top = num; - bn_correct_top(r); - return(1); - } - } + int num = mont->N.top; + + if (num > 1 && a->top == num && b->top == num) { + if (bn_wexpand(r, num) == NULL) + return (0); + if (bn_mul_mont(r->d, a->d, b->d, mont->N.d, mont->n0, num)) { + r->neg = a->neg ^ b->neg; + r->top = num; + bn_correct_top(r); + return (1); + } + } #endif - BN_CTX_start(ctx); - tmp = BN_CTX_get(ctx); - if (tmp == NULL) goto err; - - bn_check_top(tmp); - if (a == b) - { - if (!BN_sqr(tmp,a,ctx)) goto err; - } - else - { - if (!BN_mul(tmp,a,b,ctx)) goto err; - } - /* reduce from aRR to aR */ + BN_CTX_start(ctx); + tmp = BN_CTX_get(ctx); + if (tmp == NULL) + goto err; + + bn_check_top(tmp); + if (a == b) { + if (!BN_sqr(tmp, a, ctx)) + goto err; + } else { + if (!BN_mul(tmp, a, b, ctx)) + goto err; + } + /* reduce from aRR to aR */ #ifdef MONT_WORD - if (!BN_from_montgomery_word(r,tmp,mont)) goto err; + if (!BN_from_montgomery_word(r, tmp, mont)) + goto err; #else - if (!BN_from_montgomery(r,tmp,mont,ctx)) goto err; + if (!BN_from_montgomery(r, tmp, mont, ctx)) + goto err; #endif - bn_check_top(r); - ret=1; -err: - BN_CTX_end(ctx); - return(ret); - } + bn_check_top(r); + ret = 1; + err: + BN_CTX_end(ctx); + return (ret); +} #ifdef MONT_WORD static int BN_from_montgomery_word(BIGNUM *ret, BIGNUM *r, BN_MONT_CTX *mont) - { - BIGNUM *n; - BN_ULONG *ap,*np,*rp,n0,v,carry; - int nl,max,i; - - n= &(mont->N); - nl=n->top; - if (nl == 0) { ret->top=0; return(1); } - - max=(2*nl); /* carry is stored separately */ - if (bn_wexpand(r,max) == NULL) return(0); - - r->neg^=n->neg; - np=n->d; - rp=r->d; - - /* clear the top words of T */ -#if 1 - for (i=r->top; i<max; i++) /* memset? XXX */ - rp[i]=0; -#else - memset(&(rp[r->top]),0,(max-r->top)*sizeof(BN_ULONG)); -#endif - - r->top=max; - n0=mont->n0[0]; - -#ifdef BN_COUNT - fprintf(stderr,"word BN_from_montgomery_word %d * %d\n",nl,nl); -#endif - for (carry=0, i=0; i<nl; i++, rp++) - { -#ifdef __TANDEM - { - long long t1; - long long t2; - long long t3; - t1 = rp[0] * (n0 & 0177777); - t2 = 037777600000l; - t2 = n0 & t2; - t3 = rp[0] & 0177777; - t2 = (t3 * t2) & BN_MASK2; - t1 = t1 + t2; - v=bn_mul_add_words(rp,np,nl,(BN_ULONG) t1); - } -#else - v=bn_mul_add_words(rp,np,nl,(rp[0]*n0)&BN_MASK2); -#endif - v = (v+carry+rp[nl])&BN_MASK2; - carry |= (v != rp[nl]); - carry &= (v <= rp[nl]); - rp[nl]=v; - } - - if (bn_wexpand(ret,nl) == NULL) return(0); - ret->top=nl; - ret->neg=r->neg; - - rp=ret->d; - ap=&(r->d[nl]); - -#define BRANCH_FREE 1 -#if BRANCH_FREE - { - BN_ULONG *nrp; - size_t m; - - v=bn_sub_words(rp,ap,np,nl)-carry; - /* if subtraction result is real, then - * trick unconditional memcpy below to perform in-place - * "refresh" instead of actual copy. */ - m=(0-(size_t)v); - nrp=(BN_ULONG *)(((PTR_SIZE_INT)rp&~m)|((PTR_SIZE_INT)ap&m)); - - for (i=0,nl-=4; i<nl; i+=4) - { - BN_ULONG t1,t2,t3,t4; - - t1=nrp[i+0]; - t2=nrp[i+1]; - t3=nrp[i+2]; ap[i+0]=0; - t4=nrp[i+3]; ap[i+1]=0; - rp[i+0]=t1; ap[i+2]=0; - rp[i+1]=t2; ap[i+3]=0; - rp[i+2]=t3; - rp[i+3]=t4; - } - for (nl+=4; i<nl; i++) - rp[i]=nrp[i], ap[i]=0; - } -#else - if (bn_sub_words (rp,ap,np,nl)-carry) - memcpy(rp,ap,nl*sizeof(BN_ULONG)); -#endif - bn_correct_top(r); - bn_correct_top(ret); - bn_check_top(ret); - - return(1); - } -#endif /* MONT_WORD */ +{ + BIGNUM *n; + BN_ULONG *ap, *np, *rp, n0, v, carry; + int nl, max, i; + + n = &(mont->N); + nl = n->top; + if (nl == 0) { + ret->top = 0; + return (1); + } + + max = (2 * nl); /* carry is stored separately */ + if (bn_wexpand(r, max) == NULL) + return (0); + + r->neg ^= n->neg; + np = n->d; + rp = r->d; + + /* clear the top words of T */ +# if 1 + for (i = r->top; i < max; i++) /* memset? XXX */ + rp[i] = 0; +# else + memset(&(rp[r->top]), 0, (max - r->top) * sizeof(BN_ULONG)); +# endif + + r->top = max; + n0 = mont->n0[0]; + +# ifdef BN_COUNT + fprintf(stderr, "word BN_from_montgomery_word %d * %d\n", nl, nl); +# endif + for (carry = 0, i = 0; i < nl; i++, rp++) { +# ifdef __TANDEM + { + long long t1; + long long t2; + long long t3; + t1 = rp[0] * (n0 & 0177777); + t2 = 037777600000l; + t2 = n0 & t2; + t3 = rp[0] & 0177777; + t2 = (t3 * t2) & BN_MASK2; + t1 = t1 + t2; + v = bn_mul_add_words(rp, np, nl, (BN_ULONG)t1); + } +# else + v = bn_mul_add_words(rp, np, nl, (rp[0] * n0) & BN_MASK2); +# endif + v = (v + carry + rp[nl]) & BN_MASK2; + carry |= (v != rp[nl]); + carry &= (v <= rp[nl]); + rp[nl] = v; + } + + if (bn_wexpand(ret, nl) == NULL) + return (0); + ret->top = nl; + ret->neg = r->neg; + + rp = ret->d; + ap = &(r->d[nl]); + +# define BRANCH_FREE 1 +# if BRANCH_FREE + { + BN_ULONG *nrp; + size_t m; + + v = bn_sub_words(rp, ap, np, nl) - carry; + /* + * if subtraction result is real, then trick unconditional memcpy + * below to perform in-place "refresh" instead of actual copy. + */ + m = (0 - (size_t)v); + nrp = + (BN_ULONG *)(((PTR_SIZE_INT) rp & ~m) | ((PTR_SIZE_INT) ap & m)); + + for (i = 0, nl -= 4; i < nl; i += 4) { + BN_ULONG t1, t2, t3, t4; + + t1 = nrp[i + 0]; + t2 = nrp[i + 1]; + t3 = nrp[i + 2]; + ap[i + 0] = 0; + t4 = nrp[i + 3]; + ap[i + 1] = 0; + rp[i + 0] = t1; + ap[i + 2] = 0; + rp[i + 1] = t2; + ap[i + 3] = 0; + rp[i + 2] = t3; + rp[i + 3] = t4; + } + for (nl += 4; i < nl; i++) + rp[i] = nrp[i], ap[i] = 0; + } +# else + if (bn_sub_words(rp, ap, np, nl) - carry) + memcpy(rp, ap, nl * sizeof(BN_ULONG)); +# endif + bn_correct_top(r); + bn_correct_top(ret); + bn_check_top(ret); + + return (1); +} +#endif /* MONT_WORD */ int BN_from_montgomery(BIGNUM *ret, const BIGNUM *a, BN_MONT_CTX *mont, - BN_CTX *ctx) - { - int retn=0; + BN_CTX *ctx) +{ + int retn = 0; #ifdef MONT_WORD - BIGNUM *t; - - BN_CTX_start(ctx); - if ((t = BN_CTX_get(ctx)) && BN_copy(t,a)) - retn = BN_from_montgomery_word(ret,t,mont); - BN_CTX_end(ctx); -#else /* !MONT_WORD */ - BIGNUM *t1,*t2; - - BN_CTX_start(ctx); - t1 = BN_CTX_get(ctx); - t2 = BN_CTX_get(ctx); - if (t1 == NULL || t2 == NULL) goto err; - - if (!BN_copy(t1,a)) goto err; - BN_mask_bits(t1,mont->ri); - - if (!BN_mul(t2,t1,&mont->Ni,ctx)) goto err; - BN_mask_bits(t2,mont->ri); - - if (!BN_mul(t1,t2,&mont->N,ctx)) goto err; - if (!BN_add(t2,a,t1)) goto err; - if (!BN_rshift(ret,t2,mont->ri)) goto err; - - if (BN_ucmp(ret, &(mont->N)) >= 0) - { - if (!BN_usub(ret,ret,&(mont->N))) goto err; - } - retn=1; - bn_check_top(ret); + BIGNUM *t; + + BN_CTX_start(ctx); + if ((t = BN_CTX_get(ctx)) && BN_copy(t, a)) + retn = BN_from_montgomery_word(ret, t, mont); + BN_CTX_end(ctx); +#else /* !MONT_WORD */ + BIGNUM *t1, *t2; + + BN_CTX_start(ctx); + t1 = BN_CTX_get(ctx); + t2 = BN_CTX_get(ctx); + if (t1 == NULL || t2 == NULL) + goto err; + + if (!BN_copy(t1, a)) + goto err; + BN_mask_bits(t1, mont->ri); + + if (!BN_mul(t2, t1, &mont->Ni, ctx)) + goto err; + BN_mask_bits(t2, mont->ri); + + if (!BN_mul(t1, t2, &mont->N, ctx)) + goto err; + if (!BN_add(t2, a, t1)) + goto err; + if (!BN_rshift(ret, t2, mont->ri)) + goto err; + + if (BN_ucmp(ret, &(mont->N)) >= 0) { + if (!BN_usub(ret, ret, &(mont->N))) + goto err; + } + retn = 1; + bn_check_top(ret); err: - BN_CTX_end(ctx); -#endif /* MONT_WORD */ - return(retn); - } + BN_CTX_end(ctx); +#endif /* MONT_WORD */ + return (retn); +} BN_MONT_CTX *BN_MONT_CTX_new(void) - { - BN_MONT_CTX *ret; +{ + BN_MONT_CTX *ret; - if ((ret=(BN_MONT_CTX *)OPENSSL_malloc(sizeof(BN_MONT_CTX))) == NULL) - return(NULL); + if ((ret = (BN_MONT_CTX *)OPENSSL_malloc(sizeof(BN_MONT_CTX))) == NULL) + return (NULL); - BN_MONT_CTX_init(ret); - ret->flags=BN_FLG_MALLOCED; - return(ret); - } + BN_MONT_CTX_init(ret); + ret->flags = BN_FLG_MALLOCED; + return (ret); +} void BN_MONT_CTX_init(BN_MONT_CTX *ctx) - { - ctx->ri=0; - BN_init(&(ctx->RR)); - BN_init(&(ctx->N)); - BN_init(&(ctx->Ni)); - ctx->n0[0] = ctx->n0[1] = 0; - ctx->flags=0; - } +{ + ctx->ri = 0; + BN_init(&(ctx->RR)); + BN_init(&(ctx->N)); + BN_init(&(ctx->Ni)); + ctx->n0[0] = ctx->n0[1] = 0; + ctx->flags = 0; +} void BN_MONT_CTX_free(BN_MONT_CTX *mont) - { - if(mont == NULL) - return; +{ + if (mont == NULL) + return; - BN_free(&(mont->RR)); - BN_free(&(mont->N)); - BN_free(&(mont->Ni)); - if (mont->flags & BN_FLG_MALLOCED) - OPENSSL_free(mont); - } + BN_free(&(mont->RR)); + BN_free(&(mont->N)); + BN_free(&(mont->Ni)); + if (mont->flags & BN_FLG_MALLOCED) + OPENSSL_free(mont); +} int BN_MONT_CTX_set(BN_MONT_CTX *mont, const BIGNUM *mod, BN_CTX *ctx) - { - int ret = 0; - BIGNUM *Ri,*R; - - BN_CTX_start(ctx); - if((Ri = BN_CTX_get(ctx)) == NULL) goto err; - R= &(mont->RR); /* grab RR as a temp */ - if (!BN_copy(&(mont->N),mod)) goto err; /* Set N */ - mont->N.neg = 0; +{ + int ret = 0; + BIGNUM *Ri, *R; + + BN_CTX_start(ctx); + if ((Ri = BN_CTX_get(ctx)) == NULL) + goto err; + R = &(mont->RR); /* grab RR as a temp */ + if (!BN_copy(&(mont->N), mod)) + goto err; /* Set N */ + mont->N.neg = 0; #ifdef MONT_WORD - { - BIGNUM tmod; - BN_ULONG buf[2]; - - BN_init(&tmod); - tmod.d=buf; - tmod.dmax=2; - tmod.neg=0; - - mont->ri=(BN_num_bits(mod)+(BN_BITS2-1))/BN_BITS2*BN_BITS2; - -#if defined(OPENSSL_BN_ASM_MONT) && (BN_BITS2<=32) - /* Only certain BN_BITS2<=32 platforms actually make use of - * n0[1], and we could use the #else case (with a shorter R - * value) for the others. However, currently only the assembler - * files do know which is which. */ - - BN_zero(R); - if (!(BN_set_bit(R,2*BN_BITS2))) goto err; - - tmod.top=0; - if ((buf[0] = mod->d[0])) tmod.top=1; - if ((buf[1] = mod->top>1 ? mod->d[1] : 0)) tmod.top=2; - - if ((BN_mod_inverse(Ri,R,&tmod,ctx)) == NULL) - goto err; - if (!BN_lshift(Ri,Ri,2*BN_BITS2)) goto err; /* R*Ri */ - if (!BN_is_zero(Ri)) - { - if (!BN_sub_word(Ri,1)) goto err; - } - else /* if N mod word size == 1 */ - { - if (bn_expand(Ri,(int)sizeof(BN_ULONG)*2) == NULL) - goto err; - /* Ri-- (mod double word size) */ - Ri->neg=0; - Ri->d[0]=BN_MASK2; - Ri->d[1]=BN_MASK2; - Ri->top=2; - } - if (!BN_div(Ri,NULL,Ri,&tmod,ctx)) goto err; - /* Ni = (R*Ri-1)/N, - * keep only couple of least significant words: */ - mont->n0[0] = (Ri->top > 0) ? Ri->d[0] : 0; - mont->n0[1] = (Ri->top > 1) ? Ri->d[1] : 0; -#else - BN_zero(R); - if (!(BN_set_bit(R,BN_BITS2))) goto err; /* R */ - - buf[0]=mod->d[0]; /* tmod = N mod word size */ - buf[1]=0; - tmod.top = buf[0] != 0 ? 1 : 0; - /* Ri = R^-1 mod N*/ - if ((BN_mod_inverse(Ri,R,&tmod,ctx)) == NULL) - goto err; - if (!BN_lshift(Ri,Ri,BN_BITS2)) goto err; /* R*Ri */ - if (!BN_is_zero(Ri)) - { - if (!BN_sub_word(Ri,1)) goto err; - } - else /* if N mod word size == 1 */ - { - if (!BN_set_word(Ri,BN_MASK2)) goto err; /* Ri-- (mod word size) */ - } - if (!BN_div(Ri,NULL,Ri,&tmod,ctx)) goto err; - /* Ni = (R*Ri-1)/N, - * keep only least significant word: */ - mont->n0[0] = (Ri->top > 0) ? Ri->d[0] : 0; - mont->n0[1] = 0; -#endif - } -#else /* !MONT_WORD */ - { /* bignum version */ - mont->ri=BN_num_bits(&mont->N); - BN_zero(R); - if (!BN_set_bit(R,mont->ri)) goto err; /* R = 2^ri */ - /* Ri = R^-1 mod N*/ - if ((BN_mod_inverse(Ri,R,&mont->N,ctx)) == NULL) - goto err; - if (!BN_lshift(Ri,Ri,mont->ri)) goto err; /* R*Ri */ - if (!BN_sub_word(Ri,1)) goto err; - /* Ni = (R*Ri-1) / N */ - if (!BN_div(&(mont->Ni),NULL,Ri,&mont->N,ctx)) goto err; - } + { + BIGNUM tmod; + BN_ULONG buf[2]; + + BN_init(&tmod); + tmod.d = buf; + tmod.dmax = 2; + tmod.neg = 0; + + mont->ri = (BN_num_bits(mod) + (BN_BITS2 - 1)) / BN_BITS2 * BN_BITS2; + +# if defined(OPENSSL_BN_ASM_MONT) && (BN_BITS2<=32) + /* + * Only certain BN_BITS2<=32 platforms actually make use of n0[1], + * and we could use the #else case (with a shorter R value) for the + * others. However, currently only the assembler files do know which + * is which. + */ + + BN_zero(R); + if (!(BN_set_bit(R, 2 * BN_BITS2))) + goto err; + + tmod.top = 0; + if ((buf[0] = mod->d[0])) + tmod.top = 1; + if ((buf[1] = mod->top > 1 ? mod->d[1] : 0)) + tmod.top = 2; + + if ((BN_mod_inverse(Ri, R, &tmod, ctx)) == NULL) + goto err; + if (!BN_lshift(Ri, Ri, 2 * BN_BITS2)) + goto err; /* R*Ri */ + if (!BN_is_zero(Ri)) { + if (!BN_sub_word(Ri, 1)) + goto err; + } else { /* if N mod word size == 1 */ + + if (bn_expand(Ri, (int)sizeof(BN_ULONG) * 2) == NULL) + goto err; + /* Ri-- (mod double word size) */ + Ri->neg = 0; + Ri->d[0] = BN_MASK2; + Ri->d[1] = BN_MASK2; + Ri->top = 2; + } + if (!BN_div(Ri, NULL, Ri, &tmod, ctx)) + goto err; + /* + * Ni = (R*Ri-1)/N, keep only couple of least significant words: + */ + mont->n0[0] = (Ri->top > 0) ? Ri->d[0] : 0; + mont->n0[1] = (Ri->top > 1) ? Ri->d[1] : 0; +# else + BN_zero(R); + if (!(BN_set_bit(R, BN_BITS2))) + goto err; /* R */ + + buf[0] = mod->d[0]; /* tmod = N mod word size */ + buf[1] = 0; + tmod.top = buf[0] != 0 ? 1 : 0; + /* Ri = R^-1 mod N */ + if ((BN_mod_inverse(Ri, R, &tmod, ctx)) == NULL) + goto err; + if (!BN_lshift(Ri, Ri, BN_BITS2)) + goto err; /* R*Ri */ + if (!BN_is_zero(Ri)) { + if (!BN_sub_word(Ri, 1)) + goto err; + } else { /* if N mod word size == 1 */ + + if (!BN_set_word(Ri, BN_MASK2)) + goto err; /* Ri-- (mod word size) */ + } + if (!BN_div(Ri, NULL, Ri, &tmod, ctx)) + goto err; + /* + * Ni = (R*Ri-1)/N, keep only least significant word: + */ + mont->n0[0] = (Ri->top > 0) ? Ri->d[0] : 0; + mont->n0[1] = 0; +# endif + } +#else /* !MONT_WORD */ + { /* bignum version */ + mont->ri = BN_num_bits(&mont->N); + BN_zero(R); + if (!BN_set_bit(R, mont->ri)) + goto err; /* R = 2^ri */ + /* Ri = R^-1 mod N */ + if ((BN_mod_inverse(Ri, R, &mont->N, ctx)) == NULL) + goto err; + if (!BN_lshift(Ri, Ri, mont->ri)) + goto err; /* R*Ri */ + if (!BN_sub_word(Ri, 1)) + goto err; + /* + * Ni = (R*Ri-1) / N + */ + if (!BN_div(&(mont->Ni), NULL, Ri, &mont->N, ctx)) + goto err; + } #endif - /* setup RR for conversions */ - BN_zero(&(mont->RR)); - if (!BN_set_bit(&(mont->RR),mont->ri*2)) goto err; - if (!BN_mod(&(mont->RR),&(mont->RR),&(mont->N),ctx)) goto err; + /* setup RR for conversions */ + BN_zero(&(mont->RR)); + if (!BN_set_bit(&(mont->RR), mont->ri * 2)) + goto err; + if (!BN_mod(&(mont->RR), &(mont->RR), &(mont->N), ctx)) + goto err; - ret = 1; -err: - BN_CTX_end(ctx); - return ret; - } + ret = 1; + err: + BN_CTX_end(ctx); + return ret; +} BN_MONT_CTX *BN_MONT_CTX_copy(BN_MONT_CTX *to, BN_MONT_CTX *from) - { - if (to == from) return(to); - - if (!BN_copy(&(to->RR),&(from->RR))) return NULL; - if (!BN_copy(&(to->N),&(from->N))) return NULL; - if (!BN_copy(&(to->Ni),&(from->Ni))) return NULL; - to->ri=from->ri; - to->n0[0]=from->n0[0]; - to->n0[1]=from->n0[1]; - return(to); - } +{ + if (to == from) + return (to); + + if (!BN_copy(&(to->RR), &(from->RR))) + return NULL; + if (!BN_copy(&(to->N), &(from->N))) + return NULL; + if (!BN_copy(&(to->Ni), &(from->Ni))) + return NULL; + to->ri = from->ri; + to->n0[0] = from->n0[0]; + to->n0[1] = from->n0[1]; + return (to); +} BN_MONT_CTX *BN_MONT_CTX_set_locked(BN_MONT_CTX **pmont, int lock, - const BIGNUM *mod, BN_CTX *ctx) - { - BN_MONT_CTX *ret; - - CRYPTO_r_lock(lock); - ret = *pmont; - CRYPTO_r_unlock(lock); - if (ret) - return ret; - - /* We don't want to serialise globally while doing our lazy-init math in - * BN_MONT_CTX_set. That punishes threads that are doing independent - * things. Instead, punish the case where more than one thread tries to - * lazy-init the same 'pmont', by having each do the lazy-init math work - * independently and only use the one from the thread that wins the race - * (the losers throw away the work they've done). */ - ret = BN_MONT_CTX_new(); - if (!ret) - return NULL; - if (!BN_MONT_CTX_set(ret, mod, ctx)) - { - BN_MONT_CTX_free(ret); - return NULL; - } - - /* The locked compare-and-set, after the local work is done. */ - CRYPTO_w_lock(lock); - if (*pmont) - { - BN_MONT_CTX_free(ret); - ret = *pmont; - } - else - *pmont = ret; - CRYPTO_w_unlock(lock); - return ret; - } + const BIGNUM *mod, BN_CTX *ctx) +{ + BN_MONT_CTX *ret; + + CRYPTO_r_lock(lock); + ret = *pmont; + CRYPTO_r_unlock(lock); + if (ret) + return ret; + + /* + * We don't want to serialise globally while doing our lazy-init math in + * BN_MONT_CTX_set. That punishes threads that are doing independent + * things. Instead, punish the case where more than one thread tries to + * lazy-init the same 'pmont', by having each do the lazy-init math work + * independently and only use the one from the thread that wins the race + * (the losers throw away the work they've done). + */ + ret = BN_MONT_CTX_new(); + if (!ret) + return NULL; + if (!BN_MONT_CTX_set(ret, mod, ctx)) { + BN_MONT_CTX_free(ret); + return NULL; + } + + /* The locked compare-and-set, after the local work is done. */ + CRYPTO_w_lock(lock); + if (*pmont) { + BN_MONT_CTX_free(ret); + ret = *pmont; + } else + *pmont = ret; + CRYPTO_w_unlock(lock); + return ret; +} diff --git a/openssl/crypto/bn/bn_mpi.c b/openssl/crypto/bn/bn_mpi.c index a054d21ae..3bd40bbd2 100644 --- a/openssl/crypto/bn/bn_mpi.c +++ b/openssl/crypto/bn/bn_mpi.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -61,70 +61,68 @@ #include "bn_lcl.h" int BN_bn2mpi(const BIGNUM *a, unsigned char *d) - { - int bits; - int num=0; - int ext=0; - long l; +{ + int bits; + int num = 0; + int ext = 0; + long l; - bits=BN_num_bits(a); - num=(bits+7)/8; - if (bits > 0) - { - ext=((bits & 0x07) == 0); - } - if (d == NULL) - return(num+4+ext); + bits = BN_num_bits(a); + num = (bits + 7) / 8; + if (bits > 0) { + ext = ((bits & 0x07) == 0); + } + if (d == NULL) + return (num + 4 + ext); - l=num+ext; - d[0]=(unsigned char)(l>>24)&0xff; - d[1]=(unsigned char)(l>>16)&0xff; - d[2]=(unsigned char)(l>> 8)&0xff; - d[3]=(unsigned char)(l )&0xff; - if (ext) d[4]=0; - num=BN_bn2bin(a,&(d[4+ext])); - if (a->neg) - d[4]|=0x80; - return(num+4+ext); - } + l = num + ext; + d[0] = (unsigned char)(l >> 24) & 0xff; + d[1] = (unsigned char)(l >> 16) & 0xff; + d[2] = (unsigned char)(l >> 8) & 0xff; + d[3] = (unsigned char)(l) & 0xff; + if (ext) + d[4] = 0; + num = BN_bn2bin(a, &(d[4 + ext])); + if (a->neg) + d[4] |= 0x80; + return (num + 4 + ext); +} BIGNUM *BN_mpi2bn(const unsigned char *d, int n, BIGNUM *a) - { - long len; - int neg=0; - - if (n < 4) - { - BNerr(BN_F_BN_MPI2BN,BN_R_INVALID_LENGTH); - return(NULL); - } - len=((long)d[0]<<24)|((long)d[1]<<16)|((int)d[2]<<8)|(int)d[3]; - if ((len+4) != n) - { - BNerr(BN_F_BN_MPI2BN,BN_R_ENCODING_ERROR); - return(NULL); - } +{ + long len; + int neg = 0; - if (a == NULL) a=BN_new(); - if (a == NULL) return(NULL); + if (n < 4) { + BNerr(BN_F_BN_MPI2BN, BN_R_INVALID_LENGTH); + return (NULL); + } + len = ((long)d[0] << 24) | ((long)d[1] << 16) | ((int)d[2] << 8) | (int) + d[3]; + if ((len + 4) != n) { + BNerr(BN_F_BN_MPI2BN, BN_R_ENCODING_ERROR); + return (NULL); + } - if (len == 0) - { - a->neg=0; - a->top=0; - return(a); - } - d+=4; - if ((*d) & 0x80) - neg=1; - if (BN_bin2bn(d,(int)len,a) == NULL) - return(NULL); - a->neg=neg; - if (neg) - { - BN_clear_bit(a,BN_num_bits(a)-1); - } - bn_check_top(a); - return(a); - } + if (a == NULL) + a = BN_new(); + if (a == NULL) + return (NULL); + if (len == 0) { + a->neg = 0; + a->top = 0; + return (a); + } + d += 4; + if ((*d) & 0x80) + neg = 1; + if (BN_bin2bn(d, (int)len, a) == NULL) + return (NULL); + a->neg = neg; + if (neg) { + BN_clear_bit(a, BN_num_bits(a) - 1); + } + bn_check_top(a); + return (a); +} diff --git a/openssl/crypto/bn/bn_mul.c b/openssl/crypto/bn/bn_mul.c index 12e5be80e..b174850b6 100644 --- a/openssl/crypto/bn/bn_mul.c +++ b/openssl/crypto/bn/bn_mul.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -57,7 +57,7 @@ */ #ifndef BN_DEBUG -# undef NDEBUG /* avoid conflicting definitions */ +# undef NDEBUG /* avoid conflicting definitions */ # define NDEBUG #endif @@ -67,319 +67,353 @@ #include "bn_lcl.h" #if defined(OPENSSL_NO_ASM) || !defined(OPENSSL_BN_ASM_PART_WORDS) -/* Here follows specialised variants of bn_add_words() and - bn_sub_words(). They have the property performing operations on - arrays of different sizes. The sizes of those arrays is expressed through - cl, which is the common length ( basicall, min(len(a),len(b)) ), and dl, - which is the delta between the two lengths, calculated as len(a)-len(b). - All lengths are the number of BN_ULONGs... For the operations that require - a result array as parameter, it must have the length cl+abs(dl). - These functions should probably end up in bn_asm.c as soon as there are - assembler counterparts for the systems that use assembler files. */ +/* + * Here follows specialised variants of bn_add_words() and bn_sub_words(). + * They have the property performing operations on arrays of different sizes. + * The sizes of those arrays is expressed through cl, which is the common + * length ( basicall, min(len(a),len(b)) ), and dl, which is the delta + * between the two lengths, calculated as len(a)-len(b). All lengths are the + * number of BN_ULONGs... For the operations that require a result array as + * parameter, it must have the length cl+abs(dl). These functions should + * probably end up in bn_asm.c as soon as there are assembler counterparts + * for the systems that use assembler files. + */ BN_ULONG bn_sub_part_words(BN_ULONG *r, - const BN_ULONG *a, const BN_ULONG *b, - int cl, int dl) - { - BN_ULONG c, t; + const BN_ULONG *a, const BN_ULONG *b, + int cl, int dl) +{ + BN_ULONG c, t; - assert(cl >= 0); - c = bn_sub_words(r, a, b, cl); + assert(cl >= 0); + c = bn_sub_words(r, a, b, cl); - if (dl == 0) - return c; + if (dl == 0) + return c; - r += cl; - a += cl; - b += cl; + r += cl; + a += cl; + b += cl; - if (dl < 0) - { -#ifdef BN_COUNT - fprintf(stderr, " bn_sub_part_words %d + %d (dl < 0, c = %d)\n", cl, dl, c); -#endif - for (;;) - { - t = b[0]; - r[0] = (0-t-c)&BN_MASK2; - if (t != 0) c=1; - if (++dl >= 0) break; - - t = b[1]; - r[1] = (0-t-c)&BN_MASK2; - if (t != 0) c=1; - if (++dl >= 0) break; - - t = b[2]; - r[2] = (0-t-c)&BN_MASK2; - if (t != 0) c=1; - if (++dl >= 0) break; - - t = b[3]; - r[3] = (0-t-c)&BN_MASK2; - if (t != 0) c=1; - if (++dl >= 0) break; - - b += 4; - r += 4; - } - } - else - { - int save_dl = dl; -#ifdef BN_COUNT - fprintf(stderr, " bn_sub_part_words %d + %d (dl > 0, c = %d)\n", cl, dl, c); -#endif - while(c) - { - t = a[0]; - r[0] = (t-c)&BN_MASK2; - if (t != 0) c=0; - if (--dl <= 0) break; - - t = a[1]; - r[1] = (t-c)&BN_MASK2; - if (t != 0) c=0; - if (--dl <= 0) break; - - t = a[2]; - r[2] = (t-c)&BN_MASK2; - if (t != 0) c=0; - if (--dl <= 0) break; - - t = a[3]; - r[3] = (t-c)&BN_MASK2; - if (t != 0) c=0; - if (--dl <= 0) break; - - save_dl = dl; - a += 4; - r += 4; - } - if (dl > 0) - { -#ifdef BN_COUNT - fprintf(stderr, " bn_sub_part_words %d + %d (dl > 0, c == 0)\n", cl, dl); -#endif - if (save_dl > dl) - { - switch (save_dl - dl) - { - case 1: - r[1] = a[1]; - if (--dl <= 0) break; - case 2: - r[2] = a[2]; - if (--dl <= 0) break; - case 3: - r[3] = a[3]; - if (--dl <= 0) break; - } - a += 4; - r += 4; - } - } - if (dl > 0) - { -#ifdef BN_COUNT - fprintf(stderr, " bn_sub_part_words %d + %d (dl > 0, copy)\n", cl, dl); -#endif - for(;;) - { - r[0] = a[0]; - if (--dl <= 0) break; - r[1] = a[1]; - if (--dl <= 0) break; - r[2] = a[2]; - if (--dl <= 0) break; - r[3] = a[3]; - if (--dl <= 0) break; - - a += 4; - r += 4; - } - } - } - return c; - } + if (dl < 0) { +# ifdef BN_COUNT + fprintf(stderr, " bn_sub_part_words %d + %d (dl < 0, c = %d)\n", cl, + dl, c); +# endif + for (;;) { + t = b[0]; + r[0] = (0 - t - c) & BN_MASK2; + if (t != 0) + c = 1; + if (++dl >= 0) + break; + + t = b[1]; + r[1] = (0 - t - c) & BN_MASK2; + if (t != 0) + c = 1; + if (++dl >= 0) + break; + + t = b[2]; + r[2] = (0 - t - c) & BN_MASK2; + if (t != 0) + c = 1; + if (++dl >= 0) + break; + + t = b[3]; + r[3] = (0 - t - c) & BN_MASK2; + if (t != 0) + c = 1; + if (++dl >= 0) + break; + + b += 4; + r += 4; + } + } else { + int save_dl = dl; +# ifdef BN_COUNT + fprintf(stderr, " bn_sub_part_words %d + %d (dl > 0, c = %d)\n", cl, + dl, c); +# endif + while (c) { + t = a[0]; + r[0] = (t - c) & BN_MASK2; + if (t != 0) + c = 0; + if (--dl <= 0) + break; + + t = a[1]; + r[1] = (t - c) & BN_MASK2; + if (t != 0) + c = 0; + if (--dl <= 0) + break; + + t = a[2]; + r[2] = (t - c) & BN_MASK2; + if (t != 0) + c = 0; + if (--dl <= 0) + break; + + t = a[3]; + r[3] = (t - c) & BN_MASK2; + if (t != 0) + c = 0; + if (--dl <= 0) + break; + + save_dl = dl; + a += 4; + r += 4; + } + if (dl > 0) { +# ifdef BN_COUNT + fprintf(stderr, " bn_sub_part_words %d + %d (dl > 0, c == 0)\n", + cl, dl); +# endif + if (save_dl > dl) { + switch (save_dl - dl) { + case 1: + r[1] = a[1]; + if (--dl <= 0) + break; + case 2: + r[2] = a[2]; + if (--dl <= 0) + break; + case 3: + r[3] = a[3]; + if (--dl <= 0) + break; + } + a += 4; + r += 4; + } + } + if (dl > 0) { +# ifdef BN_COUNT + fprintf(stderr, " bn_sub_part_words %d + %d (dl > 0, copy)\n", + cl, dl); +# endif + for (;;) { + r[0] = a[0]; + if (--dl <= 0) + break; + r[1] = a[1]; + if (--dl <= 0) + break; + r[2] = a[2]; + if (--dl <= 0) + break; + r[3] = a[3]; + if (--dl <= 0) + break; + + a += 4; + r += 4; + } + } + } + return c; +} #endif BN_ULONG bn_add_part_words(BN_ULONG *r, - const BN_ULONG *a, const BN_ULONG *b, - int cl, int dl) - { - BN_ULONG c, l, t; + const BN_ULONG *a, const BN_ULONG *b, + int cl, int dl) +{ + BN_ULONG c, l, t; - assert(cl >= 0); - c = bn_add_words(r, a, b, cl); + assert(cl >= 0); + c = bn_add_words(r, a, b, cl); - if (dl == 0) - return c; + if (dl == 0) + return c; - r += cl; - a += cl; - b += cl; + r += cl; + a += cl; + b += cl; - if (dl < 0) - { - int save_dl = dl; + if (dl < 0) { + int save_dl = dl; #ifdef BN_COUNT - fprintf(stderr, " bn_add_part_words %d + %d (dl < 0, c = %d)\n", cl, dl, c); + fprintf(stderr, " bn_add_part_words %d + %d (dl < 0, c = %d)\n", cl, + dl, c); #endif - while (c) - { - l=(c+b[0])&BN_MASK2; - c=(l < c); - r[0]=l; - if (++dl >= 0) break; - - l=(c+b[1])&BN_MASK2; - c=(l < c); - r[1]=l; - if (++dl >= 0) break; - - l=(c+b[2])&BN_MASK2; - c=(l < c); - r[2]=l; - if (++dl >= 0) break; - - l=(c+b[3])&BN_MASK2; - c=(l < c); - r[3]=l; - if (++dl >= 0) break; - - save_dl = dl; - b+=4; - r+=4; - } - if (dl < 0) - { + while (c) { + l = (c + b[0]) & BN_MASK2; + c = (l < c); + r[0] = l; + if (++dl >= 0) + break; + + l = (c + b[1]) & BN_MASK2; + c = (l < c); + r[1] = l; + if (++dl >= 0) + break; + + l = (c + b[2]) & BN_MASK2; + c = (l < c); + r[2] = l; + if (++dl >= 0) + break; + + l = (c + b[3]) & BN_MASK2; + c = (l < c); + r[3] = l; + if (++dl >= 0) + break; + + save_dl = dl; + b += 4; + r += 4; + } + if (dl < 0) { #ifdef BN_COUNT - fprintf(stderr, " bn_add_part_words %d + %d (dl < 0, c == 0)\n", cl, dl); + fprintf(stderr, " bn_add_part_words %d + %d (dl < 0, c == 0)\n", + cl, dl); #endif - if (save_dl < dl) - { - switch (dl - save_dl) - { - case 1: - r[1] = b[1]; - if (++dl >= 0) break; - case 2: - r[2] = b[2]; - if (++dl >= 0) break; - case 3: - r[3] = b[3]; - if (++dl >= 0) break; - } - b += 4; - r += 4; - } - } - if (dl < 0) - { + if (save_dl < dl) { + switch (dl - save_dl) { + case 1: + r[1] = b[1]; + if (++dl >= 0) + break; + case 2: + r[2] = b[2]; + if (++dl >= 0) + break; + case 3: + r[3] = b[3]; + if (++dl >= 0) + break; + } + b += 4; + r += 4; + } + } + if (dl < 0) { #ifdef BN_COUNT - fprintf(stderr, " bn_add_part_words %d + %d (dl < 0, copy)\n", cl, dl); + fprintf(stderr, " bn_add_part_words %d + %d (dl < 0, copy)\n", + cl, dl); #endif - for(;;) - { - r[0] = b[0]; - if (++dl >= 0) break; - r[1] = b[1]; - if (++dl >= 0) break; - r[2] = b[2]; - if (++dl >= 0) break; - r[3] = b[3]; - if (++dl >= 0) break; - - b += 4; - r += 4; - } - } - } - else - { - int save_dl = dl; + for (;;) { + r[0] = b[0]; + if (++dl >= 0) + break; + r[1] = b[1]; + if (++dl >= 0) + break; + r[2] = b[2]; + if (++dl >= 0) + break; + r[3] = b[3]; + if (++dl >= 0) + break; + + b += 4; + r += 4; + } + } + } else { + int save_dl = dl; #ifdef BN_COUNT - fprintf(stderr, " bn_add_part_words %d + %d (dl > 0)\n", cl, dl); + fprintf(stderr, " bn_add_part_words %d + %d (dl > 0)\n", cl, dl); #endif - while (c) - { - t=(a[0]+c)&BN_MASK2; - c=(t < c); - r[0]=t; - if (--dl <= 0) break; - - t=(a[1]+c)&BN_MASK2; - c=(t < c); - r[1]=t; - if (--dl <= 0) break; - - t=(a[2]+c)&BN_MASK2; - c=(t < c); - r[2]=t; - if (--dl <= 0) break; - - t=(a[3]+c)&BN_MASK2; - c=(t < c); - r[3]=t; - if (--dl <= 0) break; - - save_dl = dl; - a+=4; - r+=4; - } + while (c) { + t = (a[0] + c) & BN_MASK2; + c = (t < c); + r[0] = t; + if (--dl <= 0) + break; + + t = (a[1] + c) & BN_MASK2; + c = (t < c); + r[1] = t; + if (--dl <= 0) + break; + + t = (a[2] + c) & BN_MASK2; + c = (t < c); + r[2] = t; + if (--dl <= 0) + break; + + t = (a[3] + c) & BN_MASK2; + c = (t < c); + r[3] = t; + if (--dl <= 0) + break; + + save_dl = dl; + a += 4; + r += 4; + } #ifdef BN_COUNT - fprintf(stderr, " bn_add_part_words %d + %d (dl > 0, c == 0)\n", cl, dl); + fprintf(stderr, " bn_add_part_words %d + %d (dl > 0, c == 0)\n", cl, + dl); #endif - if (dl > 0) - { - if (save_dl > dl) - { - switch (save_dl - dl) - { - case 1: - r[1] = a[1]; - if (--dl <= 0) break; - case 2: - r[2] = a[2]; - if (--dl <= 0) break; - case 3: - r[3] = a[3]; - if (--dl <= 0) break; - } - a += 4; - r += 4; - } - } - if (dl > 0) - { + if (dl > 0) { + if (save_dl > dl) { + switch (save_dl - dl) { + case 1: + r[1] = a[1]; + if (--dl <= 0) + break; + case 2: + r[2] = a[2]; + if (--dl <= 0) + break; + case 3: + r[3] = a[3]; + if (--dl <= 0) + break; + } + a += 4; + r += 4; + } + } + if (dl > 0) { #ifdef BN_COUNT - fprintf(stderr, " bn_add_part_words %d + %d (dl > 0, copy)\n", cl, dl); + fprintf(stderr, " bn_add_part_words %d + %d (dl > 0, copy)\n", + cl, dl); #endif - for(;;) - { - r[0] = a[0]; - if (--dl <= 0) break; - r[1] = a[1]; - if (--dl <= 0) break; - r[2] = a[2]; - if (--dl <= 0) break; - r[3] = a[3]; - if (--dl <= 0) break; - - a += 4; - r += 4; - } - } - } - return c; - } + for (;;) { + r[0] = a[0]; + if (--dl <= 0) + break; + r[1] = a[1]; + if (--dl <= 0) + break; + r[2] = a[2]; + if (--dl <= 0) + break; + r[3] = a[3]; + if (--dl <= 0) + break; + + a += 4; + r += 4; + } + } + } + return c; +} #ifdef BN_RECURSION -/* Karatsuba recursive multiplication algorithm - * (cf. Knuth, The Art of Computer Programming, Vol. 2) */ +/* + * Karatsuba recursive multiplication algorithm (cf. Knuth, The Art of + * Computer Programming, Vol. 2) + */ -/* r is 2*n2 words in size, +/*- + * r is 2*n2 words in size, * a and b are both n2 words in size. * n2 must be a power of 2. * We multiply and return the result. @@ -391,776 +425,740 @@ BN_ULONG bn_add_part_words(BN_ULONG *r, */ /* dnX may not be positive, but n2/2+dnX has to be */ void bn_mul_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2, - int dna, int dnb, BN_ULONG *t) - { - int n=n2/2,c1,c2; - int tna=n+dna, tnb=n+dnb; - unsigned int neg,zero; - BN_ULONG ln,lo,*p; + int dna, int dnb, BN_ULONG *t) +{ + int n = n2 / 2, c1, c2; + int tna = n + dna, tnb = n + dnb; + unsigned int neg, zero; + BN_ULONG ln, lo, *p; # ifdef BN_COUNT - fprintf(stderr," bn_mul_recursive %d%+d * %d%+d\n",n2,dna,n2,dnb); + fprintf(stderr, " bn_mul_recursive %d%+d * %d%+d\n", n2, dna, n2, dnb); # endif # ifdef BN_MUL_COMBA # if 0 - if (n2 == 4) - { - bn_mul_comba4(r,a,b); - return; - } + if (n2 == 4) { + bn_mul_comba4(r, a, b); + return; + } # endif - /* Only call bn_mul_comba 8 if n2 == 8 and the - * two arrays are complete [steve] - */ - if (n2 == 8 && dna == 0 && dnb == 0) - { - bn_mul_comba8(r,a,b); - return; - } -# endif /* BN_MUL_COMBA */ - /* Else do normal multiply */ - if (n2 < BN_MUL_RECURSIVE_SIZE_NORMAL) - { - bn_mul_normal(r,a,n2+dna,b,n2+dnb); - if ((dna + dnb) < 0) - memset(&r[2*n2 + dna + dnb], 0, - sizeof(BN_ULONG) * -(dna + dnb)); - return; - } - /* r=(a[0]-a[1])*(b[1]-b[0]) */ - c1=bn_cmp_part_words(a,&(a[n]),tna,n-tna); - c2=bn_cmp_part_words(&(b[n]),b,tnb,tnb-n); - zero=neg=0; - switch (c1*3+c2) - { - case -4: - bn_sub_part_words(t, &(a[n]),a, tna,tna-n); /* - */ - bn_sub_part_words(&(t[n]),b, &(b[n]),tnb,n-tnb); /* - */ - break; - case -3: - zero=1; - break; - case -2: - bn_sub_part_words(t, &(a[n]),a, tna,tna-n); /* - */ - bn_sub_part_words(&(t[n]),&(b[n]),b, tnb,tnb-n); /* + */ - neg=1; - break; - case -1: - case 0: - case 1: - zero=1; - break; - case 2: - bn_sub_part_words(t, a, &(a[n]),tna,n-tna); /* + */ - bn_sub_part_words(&(t[n]),b, &(b[n]),tnb,n-tnb); /* - */ - neg=1; - break; - case 3: - zero=1; - break; - case 4: - bn_sub_part_words(t, a, &(a[n]),tna,n-tna); - bn_sub_part_words(&(t[n]),&(b[n]),b, tnb,tnb-n); - break; - } + /* + * Only call bn_mul_comba 8 if n2 == 8 and the two arrays are complete + * [steve] + */ + if (n2 == 8 && dna == 0 && dnb == 0) { + bn_mul_comba8(r, a, b); + return; + } +# endif /* BN_MUL_COMBA */ + /* Else do normal multiply */ + if (n2 < BN_MUL_RECURSIVE_SIZE_NORMAL) { + bn_mul_normal(r, a, n2 + dna, b, n2 + dnb); + if ((dna + dnb) < 0) + memset(&r[2 * n2 + dna + dnb], 0, + sizeof(BN_ULONG) * -(dna + dnb)); + return; + } + /* r=(a[0]-a[1])*(b[1]-b[0]) */ + c1 = bn_cmp_part_words(a, &(a[n]), tna, n - tna); + c2 = bn_cmp_part_words(&(b[n]), b, tnb, tnb - n); + zero = neg = 0; + switch (c1 * 3 + c2) { + case -4: + bn_sub_part_words(t, &(a[n]), a, tna, tna - n); /* - */ + bn_sub_part_words(&(t[n]), b, &(b[n]), tnb, n - tnb); /* - */ + break; + case -3: + zero = 1; + break; + case -2: + bn_sub_part_words(t, &(a[n]), a, tna, tna - n); /* - */ + bn_sub_part_words(&(t[n]), &(b[n]), b, tnb, tnb - n); /* + */ + neg = 1; + break; + case -1: + case 0: + case 1: + zero = 1; + break; + case 2: + bn_sub_part_words(t, a, &(a[n]), tna, n - tna); /* + */ + bn_sub_part_words(&(t[n]), b, &(b[n]), tnb, n - tnb); /* - */ + neg = 1; + break; + case 3: + zero = 1; + break; + case 4: + bn_sub_part_words(t, a, &(a[n]), tna, n - tna); + bn_sub_part_words(&(t[n]), &(b[n]), b, tnb, tnb - n); + break; + } # ifdef BN_MUL_COMBA - if (n == 4 && dna == 0 && dnb == 0) /* XXX: bn_mul_comba4 could take - extra args to do this well */ - { - if (!zero) - bn_mul_comba4(&(t[n2]),t,&(t[n])); - else - memset(&(t[n2]),0,8*sizeof(BN_ULONG)); - - bn_mul_comba4(r,a,b); - bn_mul_comba4(&(r[n2]),&(a[n]),&(b[n])); - } - else if (n == 8 && dna == 0 && dnb == 0) /* XXX: bn_mul_comba8 could - take extra args to do this - well */ - { - if (!zero) - bn_mul_comba8(&(t[n2]),t,&(t[n])); - else - memset(&(t[n2]),0,16*sizeof(BN_ULONG)); - - bn_mul_comba8(r,a,b); - bn_mul_comba8(&(r[n2]),&(a[n]),&(b[n])); - } - else -# endif /* BN_MUL_COMBA */ - { - p= &(t[n2*2]); - if (!zero) - bn_mul_recursive(&(t[n2]),t,&(t[n]),n,0,0,p); - else - memset(&(t[n2]),0,n2*sizeof(BN_ULONG)); - bn_mul_recursive(r,a,b,n,0,0,p); - bn_mul_recursive(&(r[n2]),&(a[n]),&(b[n]),n,dna,dnb,p); - } - - /* t[32] holds (a[0]-a[1])*(b[1]-b[0]), c1 is the sign - * r[10] holds (a[0]*b[0]) - * r[32] holds (b[1]*b[1]) - */ - - c1=(int)(bn_add_words(t,r,&(r[n2]),n2)); - - if (neg) /* if t[32] is negative */ - { - c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2)); - } - else - { - /* Might have a carry */ - c1+=(int)(bn_add_words(&(t[n2]),&(t[n2]),t,n2)); - } - - /* t[32] holds (a[0]-a[1])*(b[1]-b[0])+(a[0]*b[0])+(a[1]*b[1]) - * r[10] holds (a[0]*b[0]) - * r[32] holds (b[1]*b[1]) - * c1 holds the carry bits - */ - c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2)); - if (c1) - { - p= &(r[n+n2]); - lo= *p; - ln=(lo+c1)&BN_MASK2; - *p=ln; - - /* The overflow will stop before we over write - * words we should not overwrite */ - if (ln < (BN_ULONG)c1) - { - do { - p++; - lo= *p; - ln=(lo+1)&BN_MASK2; - *p=ln; - } while (ln == 0); - } - } - } - -/* n+tn is the word length - * t needs to be n*4 is size, as does r */ + if (n == 4 && dna == 0 && dnb == 0) { /* XXX: bn_mul_comba4 could take + * extra args to do this well */ + if (!zero) + bn_mul_comba4(&(t[n2]), t, &(t[n])); + else + memset(&(t[n2]), 0, 8 * sizeof(BN_ULONG)); + + bn_mul_comba4(r, a, b); + bn_mul_comba4(&(r[n2]), &(a[n]), &(b[n])); + } else if (n == 8 && dna == 0 && dnb == 0) { /* XXX: bn_mul_comba8 could + * take extra args to do + * this well */ + if (!zero) + bn_mul_comba8(&(t[n2]), t, &(t[n])); + else + memset(&(t[n2]), 0, 16 * sizeof(BN_ULONG)); + + bn_mul_comba8(r, a, b); + bn_mul_comba8(&(r[n2]), &(a[n]), &(b[n])); + } else +# endif /* BN_MUL_COMBA */ + { + p = &(t[n2 * 2]); + if (!zero) + bn_mul_recursive(&(t[n2]), t, &(t[n]), n, 0, 0, p); + else + memset(&(t[n2]), 0, n2 * sizeof(BN_ULONG)); + bn_mul_recursive(r, a, b, n, 0, 0, p); + bn_mul_recursive(&(r[n2]), &(a[n]), &(b[n]), n, dna, dnb, p); + } + + /*- + * t[32] holds (a[0]-a[1])*(b[1]-b[0]), c1 is the sign + * r[10] holds (a[0]*b[0]) + * r[32] holds (b[1]*b[1]) + */ + + c1 = (int)(bn_add_words(t, r, &(r[n2]), n2)); + + if (neg) { /* if t[32] is negative */ + c1 -= (int)(bn_sub_words(&(t[n2]), t, &(t[n2]), n2)); + } else { + /* Might have a carry */ + c1 += (int)(bn_add_words(&(t[n2]), &(t[n2]), t, n2)); + } + + /*- + * t[32] holds (a[0]-a[1])*(b[1]-b[0])+(a[0]*b[0])+(a[1]*b[1]) + * r[10] holds (a[0]*b[0]) + * r[32] holds (b[1]*b[1]) + * c1 holds the carry bits + */ + c1 += (int)(bn_add_words(&(r[n]), &(r[n]), &(t[n2]), n2)); + if (c1) { + p = &(r[n + n2]); + lo = *p; + ln = (lo + c1) & BN_MASK2; + *p = ln; + + /* + * The overflow will stop before we over write words we should not + * overwrite + */ + if (ln < (BN_ULONG)c1) { + do { + p++; + lo = *p; + ln = (lo + 1) & BN_MASK2; + *p = ln; + } while (ln == 0); + } + } +} + +/* + * n+tn is the word length t needs to be n*4 is size, as does r + */ /* tnX may not be negative but less than n */ void bn_mul_part_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n, - int tna, int tnb, BN_ULONG *t) - { - int i,j,n2=n*2; - int c1,c2,neg; - BN_ULONG ln,lo,*p; + int tna, int tnb, BN_ULONG *t) +{ + int i, j, n2 = n * 2; + int c1, c2, neg; + BN_ULONG ln, lo, *p; # ifdef BN_COUNT - fprintf(stderr," bn_mul_part_recursive (%d%+d) * (%d%+d)\n", - n, tna, n, tnb); + fprintf(stderr, " bn_mul_part_recursive (%d%+d) * (%d%+d)\n", + n, tna, n, tnb); # endif - if (n < 8) - { - bn_mul_normal(r,a,n+tna,b,n+tnb); - return; - } - - /* r=(a[0]-a[1])*(b[1]-b[0]) */ - c1=bn_cmp_part_words(a,&(a[n]),tna,n-tna); - c2=bn_cmp_part_words(&(b[n]),b,tnb,tnb-n); - neg=0; - switch (c1*3+c2) - { - case -4: - bn_sub_part_words(t, &(a[n]),a, tna,tna-n); /* - */ - bn_sub_part_words(&(t[n]),b, &(b[n]),tnb,n-tnb); /* - */ - break; - case -3: - /* break; */ - case -2: - bn_sub_part_words(t, &(a[n]),a, tna,tna-n); /* - */ - bn_sub_part_words(&(t[n]),&(b[n]),b, tnb,tnb-n); /* + */ - neg=1; - break; - case -1: - case 0: - case 1: - /* break; */ - case 2: - bn_sub_part_words(t, a, &(a[n]),tna,n-tna); /* + */ - bn_sub_part_words(&(t[n]),b, &(b[n]),tnb,n-tnb); /* - */ - neg=1; - break; - case 3: - /* break; */ - case 4: - bn_sub_part_words(t, a, &(a[n]),tna,n-tna); - bn_sub_part_words(&(t[n]),&(b[n]),b, tnb,tnb-n); - break; - } - /* The zero case isn't yet implemented here. The speedup - would probably be negligible. */ + if (n < 8) { + bn_mul_normal(r, a, n + tna, b, n + tnb); + return; + } + + /* r=(a[0]-a[1])*(b[1]-b[0]) */ + c1 = bn_cmp_part_words(a, &(a[n]), tna, n - tna); + c2 = bn_cmp_part_words(&(b[n]), b, tnb, tnb - n); + neg = 0; + switch (c1 * 3 + c2) { + case -4: + bn_sub_part_words(t, &(a[n]), a, tna, tna - n); /* - */ + bn_sub_part_words(&(t[n]), b, &(b[n]), tnb, n - tnb); /* - */ + break; + case -3: + /* break; */ + case -2: + bn_sub_part_words(t, &(a[n]), a, tna, tna - n); /* - */ + bn_sub_part_words(&(t[n]), &(b[n]), b, tnb, tnb - n); /* + */ + neg = 1; + break; + case -1: + case 0: + case 1: + /* break; */ + case 2: + bn_sub_part_words(t, a, &(a[n]), tna, n - tna); /* + */ + bn_sub_part_words(&(t[n]), b, &(b[n]), tnb, n - tnb); /* - */ + neg = 1; + break; + case 3: + /* break; */ + case 4: + bn_sub_part_words(t, a, &(a[n]), tna, n - tna); + bn_sub_part_words(&(t[n]), &(b[n]), b, tnb, tnb - n); + break; + } + /* + * The zero case isn't yet implemented here. The speedup would probably + * be negligible. + */ # if 0 - if (n == 4) - { - bn_mul_comba4(&(t[n2]),t,&(t[n])); - bn_mul_comba4(r,a,b); - bn_mul_normal(&(r[n2]),&(a[n]),tn,&(b[n]),tn); - memset(&(r[n2+tn*2]),0,sizeof(BN_ULONG)*(n2-tn*2)); - } - else + if (n == 4) { + bn_mul_comba4(&(t[n2]), t, &(t[n])); + bn_mul_comba4(r, a, b); + bn_mul_normal(&(r[n2]), &(a[n]), tn, &(b[n]), tn); + memset(&(r[n2 + tn * 2]), 0, sizeof(BN_ULONG) * (n2 - tn * 2)); + } else # endif - if (n == 8) - { - bn_mul_comba8(&(t[n2]),t,&(t[n])); - bn_mul_comba8(r,a,b); - bn_mul_normal(&(r[n2]),&(a[n]),tna,&(b[n]),tnb); - memset(&(r[n2+tna+tnb]),0,sizeof(BN_ULONG)*(n2-tna-tnb)); - } - else - { - p= &(t[n2*2]); - bn_mul_recursive(&(t[n2]),t,&(t[n]),n,0,0,p); - bn_mul_recursive(r,a,b,n,0,0,p); - i=n/2; - /* If there is only a bottom half to the number, - * just do it */ - if (tna > tnb) - j = tna - i; - else - j = tnb - i; - if (j == 0) - { - bn_mul_recursive(&(r[n2]),&(a[n]),&(b[n]), - i,tna-i,tnb-i,p); - memset(&(r[n2+i*2]),0,sizeof(BN_ULONG)*(n2-i*2)); - } - else if (j > 0) /* eg, n == 16, i == 8 and tn == 11 */ - { - bn_mul_part_recursive(&(r[n2]),&(a[n]),&(b[n]), - i,tna-i,tnb-i,p); - memset(&(r[n2+tna+tnb]),0, - sizeof(BN_ULONG)*(n2-tna-tnb)); - } - else /* (j < 0) eg, n == 16, i == 8 and tn == 5 */ - { - memset(&(r[n2]),0,sizeof(BN_ULONG)*n2); - if (tna < BN_MUL_RECURSIVE_SIZE_NORMAL - && tnb < BN_MUL_RECURSIVE_SIZE_NORMAL) - { - bn_mul_normal(&(r[n2]),&(a[n]),tna,&(b[n]),tnb); - } - else - { - for (;;) - { - i/=2; - /* these simplified conditions work - * exclusively because difference - * between tna and tnb is 1 or 0 */ - if (i < tna || i < tnb) - { - bn_mul_part_recursive(&(r[n2]), - &(a[n]),&(b[n]), - i,tna-i,tnb-i,p); - break; - } - else if (i == tna || i == tnb) - { - bn_mul_recursive(&(r[n2]), - &(a[n]),&(b[n]), - i,tna-i,tnb-i,p); - break; - } - } - } - } - } - - /* t[32] holds (a[0]-a[1])*(b[1]-b[0]), c1 is the sign - * r[10] holds (a[0]*b[0]) - * r[32] holds (b[1]*b[1]) - */ - - c1=(int)(bn_add_words(t,r,&(r[n2]),n2)); - - if (neg) /* if t[32] is negative */ - { - c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2)); - } - else - { - /* Might have a carry */ - c1+=(int)(bn_add_words(&(t[n2]),&(t[n2]),t,n2)); - } - - /* t[32] holds (a[0]-a[1])*(b[1]-b[0])+(a[0]*b[0])+(a[1]*b[1]) - * r[10] holds (a[0]*b[0]) - * r[32] holds (b[1]*b[1]) - * c1 holds the carry bits - */ - c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2)); - if (c1) - { - p= &(r[n+n2]); - lo= *p; - ln=(lo+c1)&BN_MASK2; - *p=ln; - - /* The overflow will stop before we over write - * words we should not overwrite */ - if (ln < (BN_ULONG)c1) - { - do { - p++; - lo= *p; - ln=(lo+1)&BN_MASK2; - *p=ln; - } while (ln == 0); - } - } - } - -/* a and b must be the same size, which is n2. + if (n == 8) { + bn_mul_comba8(&(t[n2]), t, &(t[n])); + bn_mul_comba8(r, a, b); + bn_mul_normal(&(r[n2]), &(a[n]), tna, &(b[n]), tnb); + memset(&(r[n2 + tna + tnb]), 0, sizeof(BN_ULONG) * (n2 - tna - tnb)); + } else { + p = &(t[n2 * 2]); + bn_mul_recursive(&(t[n2]), t, &(t[n]), n, 0, 0, p); + bn_mul_recursive(r, a, b, n, 0, 0, p); + i = n / 2; + /* + * If there is only a bottom half to the number, just do it + */ + if (tna > tnb) + j = tna - i; + else + j = tnb - i; + if (j == 0) { + bn_mul_recursive(&(r[n2]), &(a[n]), &(b[n]), + i, tna - i, tnb - i, p); + memset(&(r[n2 + i * 2]), 0, sizeof(BN_ULONG) * (n2 - i * 2)); + } else if (j > 0) { /* eg, n == 16, i == 8 and tn == 11 */ + bn_mul_part_recursive(&(r[n2]), &(a[n]), &(b[n]), + i, tna - i, tnb - i, p); + memset(&(r[n2 + tna + tnb]), 0, + sizeof(BN_ULONG) * (n2 - tna - tnb)); + } else { /* (j < 0) eg, n == 16, i == 8 and tn == 5 */ + + memset(&(r[n2]), 0, sizeof(BN_ULONG) * n2); + if (tna < BN_MUL_RECURSIVE_SIZE_NORMAL + && tnb < BN_MUL_RECURSIVE_SIZE_NORMAL) { + bn_mul_normal(&(r[n2]), &(a[n]), tna, &(b[n]), tnb); + } else { + for (;;) { + i /= 2; + /* + * these simplified conditions work exclusively because + * difference between tna and tnb is 1 or 0 + */ + if (i < tna || i < tnb) { + bn_mul_part_recursive(&(r[n2]), + &(a[n]), &(b[n]), + i, tna - i, tnb - i, p); + break; + } else if (i == tna || i == tnb) { + bn_mul_recursive(&(r[n2]), + &(a[n]), &(b[n]), + i, tna - i, tnb - i, p); + break; + } + } + } + } + } + + /*- + * t[32] holds (a[0]-a[1])*(b[1]-b[0]), c1 is the sign + * r[10] holds (a[0]*b[0]) + * r[32] holds (b[1]*b[1]) + */ + + c1 = (int)(bn_add_words(t, r, &(r[n2]), n2)); + + if (neg) { /* if t[32] is negative */ + c1 -= (int)(bn_sub_words(&(t[n2]), t, &(t[n2]), n2)); + } else { + /* Might have a carry */ + c1 += (int)(bn_add_words(&(t[n2]), &(t[n2]), t, n2)); + } + + /*- + * t[32] holds (a[0]-a[1])*(b[1]-b[0])+(a[0]*b[0])+(a[1]*b[1]) + * r[10] holds (a[0]*b[0]) + * r[32] holds (b[1]*b[1]) + * c1 holds the carry bits + */ + c1 += (int)(bn_add_words(&(r[n]), &(r[n]), &(t[n2]), n2)); + if (c1) { + p = &(r[n + n2]); + lo = *p; + ln = (lo + c1) & BN_MASK2; + *p = ln; + + /* + * The overflow will stop before we over write words we should not + * overwrite + */ + if (ln < (BN_ULONG)c1) { + do { + p++; + lo = *p; + ln = (lo + 1) & BN_MASK2; + *p = ln; + } while (ln == 0); + } + } +} + +/*- + * a and b must be the same size, which is n2. * r needs to be n2 words and t needs to be n2*2 */ void bn_mul_low_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2, - BN_ULONG *t) - { - int n=n2/2; + BN_ULONG *t) +{ + int n = n2 / 2; # ifdef BN_COUNT - fprintf(stderr," bn_mul_low_recursive %d * %d\n",n2,n2); + fprintf(stderr, " bn_mul_low_recursive %d * %d\n", n2, n2); # endif - bn_mul_recursive(r,a,b,n,0,0,&(t[0])); - if (n >= BN_MUL_LOW_RECURSIVE_SIZE_NORMAL) - { - bn_mul_low_recursive(&(t[0]),&(a[0]),&(b[n]),n,&(t[n2])); - bn_add_words(&(r[n]),&(r[n]),&(t[0]),n); - bn_mul_low_recursive(&(t[0]),&(a[n]),&(b[0]),n,&(t[n2])); - bn_add_words(&(r[n]),&(r[n]),&(t[0]),n); - } - else - { - bn_mul_low_normal(&(t[0]),&(a[0]),&(b[n]),n); - bn_mul_low_normal(&(t[n]),&(a[n]),&(b[0]),n); - bn_add_words(&(r[n]),&(r[n]),&(t[0]),n); - bn_add_words(&(r[n]),&(r[n]),&(t[n]),n); - } - } - -/* a and b must be the same size, which is n2. + bn_mul_recursive(r, a, b, n, 0, 0, &(t[0])); + if (n >= BN_MUL_LOW_RECURSIVE_SIZE_NORMAL) { + bn_mul_low_recursive(&(t[0]), &(a[0]), &(b[n]), n, &(t[n2])); + bn_add_words(&(r[n]), &(r[n]), &(t[0]), n); + bn_mul_low_recursive(&(t[0]), &(a[n]), &(b[0]), n, &(t[n2])); + bn_add_words(&(r[n]), &(r[n]), &(t[0]), n); + } else { + bn_mul_low_normal(&(t[0]), &(a[0]), &(b[n]), n); + bn_mul_low_normal(&(t[n]), &(a[n]), &(b[0]), n); + bn_add_words(&(r[n]), &(r[n]), &(t[0]), n); + bn_add_words(&(r[n]), &(r[n]), &(t[n]), n); + } +} + +/*- + * a and b must be the same size, which is n2. * r needs to be n2 words and t needs to be n2*2 * l is the low words of the output. * t needs to be n2*3 */ void bn_mul_high(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, BN_ULONG *l, int n2, - BN_ULONG *t) - { - int i,n; - int c1,c2; - int neg,oneg,zero; - BN_ULONG ll,lc,*lp,*mp; + BN_ULONG *t) +{ + int i, n; + int c1, c2; + int neg, oneg, zero; + BN_ULONG ll, lc, *lp, *mp; # ifdef BN_COUNT - fprintf(stderr," bn_mul_high %d * %d\n",n2,n2); + fprintf(stderr, " bn_mul_high %d * %d\n", n2, n2); # endif - n=n2/2; - - /* Calculate (al-ah)*(bh-bl) */ - neg=zero=0; - c1=bn_cmp_words(&(a[0]),&(a[n]),n); - c2=bn_cmp_words(&(b[n]),&(b[0]),n); - switch (c1*3+c2) - { - case -4: - bn_sub_words(&(r[0]),&(a[n]),&(a[0]),n); - bn_sub_words(&(r[n]),&(b[0]),&(b[n]),n); - break; - case -3: - zero=1; - break; - case -2: - bn_sub_words(&(r[0]),&(a[n]),&(a[0]),n); - bn_sub_words(&(r[n]),&(b[n]),&(b[0]),n); - neg=1; - break; - case -1: - case 0: - case 1: - zero=1; - break; - case 2: - bn_sub_words(&(r[0]),&(a[0]),&(a[n]),n); - bn_sub_words(&(r[n]),&(b[0]),&(b[n]),n); - neg=1; - break; - case 3: - zero=1; - break; - case 4: - bn_sub_words(&(r[0]),&(a[0]),&(a[n]),n); - bn_sub_words(&(r[n]),&(b[n]),&(b[0]),n); - break; - } - - oneg=neg; - /* t[10] = (a[0]-a[1])*(b[1]-b[0]) */ - /* r[10] = (a[1]*b[1]) */ + n = n2 / 2; + + /* Calculate (al-ah)*(bh-bl) */ + neg = zero = 0; + c1 = bn_cmp_words(&(a[0]), &(a[n]), n); + c2 = bn_cmp_words(&(b[n]), &(b[0]), n); + switch (c1 * 3 + c2) { + case -4: + bn_sub_words(&(r[0]), &(a[n]), &(a[0]), n); + bn_sub_words(&(r[n]), &(b[0]), &(b[n]), n); + break; + case -3: + zero = 1; + break; + case -2: + bn_sub_words(&(r[0]), &(a[n]), &(a[0]), n); + bn_sub_words(&(r[n]), &(b[n]), &(b[0]), n); + neg = 1; + break; + case -1: + case 0: + case 1: + zero = 1; + break; + case 2: + bn_sub_words(&(r[0]), &(a[0]), &(a[n]), n); + bn_sub_words(&(r[n]), &(b[0]), &(b[n]), n); + neg = 1; + break; + case 3: + zero = 1; + break; + case 4: + bn_sub_words(&(r[0]), &(a[0]), &(a[n]), n); + bn_sub_words(&(r[n]), &(b[n]), &(b[0]), n); + break; + } + + oneg = neg; + /* t[10] = (a[0]-a[1])*(b[1]-b[0]) */ + /* r[10] = (a[1]*b[1]) */ # ifdef BN_MUL_COMBA - if (n == 8) - { - bn_mul_comba8(&(t[0]),&(r[0]),&(r[n])); - bn_mul_comba8(r,&(a[n]),&(b[n])); - } - else + if (n == 8) { + bn_mul_comba8(&(t[0]), &(r[0]), &(r[n])); + bn_mul_comba8(r, &(a[n]), &(b[n])); + } else # endif - { - bn_mul_recursive(&(t[0]),&(r[0]),&(r[n]),n,0,0,&(t[n2])); - bn_mul_recursive(r,&(a[n]),&(b[n]),n,0,0,&(t[n2])); - } - - /* s0 == low(al*bl) - * s1 == low(ah*bh)+low((al-ah)*(bh-bl))+low(al*bl)+high(al*bl) - * We know s0 and s1 so the only unknown is high(al*bl) - * high(al*bl) == s1 - low(ah*bh+s0+(al-ah)*(bh-bl)) - * high(al*bl) == s1 - (r[0]+l[0]+t[0]) - */ - if (l != NULL) - { - lp= &(t[n2+n]); - c1=(int)(bn_add_words(lp,&(r[0]),&(l[0]),n)); - } - else - { - c1=0; - lp= &(r[0]); - } - - if (neg) - neg=(int)(bn_sub_words(&(t[n2]),lp,&(t[0]),n)); - else - { - bn_add_words(&(t[n2]),lp,&(t[0]),n); - neg=0; - } - - if (l != NULL) - { - bn_sub_words(&(t[n2+n]),&(l[n]),&(t[n2]),n); - } - else - { - lp= &(t[n2+n]); - mp= &(t[n2]); - for (i=0; i<n; i++) - lp[i]=((~mp[i])+1)&BN_MASK2; - } - - /* s[0] = low(al*bl) - * t[3] = high(al*bl) - * t[10] = (a[0]-a[1])*(b[1]-b[0]) neg is the sign - * r[10] = (a[1]*b[1]) - */ - /* R[10] = al*bl - * R[21] = al*bl + ah*bh + (a[0]-a[1])*(b[1]-b[0]) - * R[32] = ah*bh - */ - /* R[1]=t[3]+l[0]+r[0](+-)t[0] (have carry/borrow) - * R[2]=r[0]+t[3]+r[1](+-)t[1] (have carry/borrow) - * R[3]=r[1]+(carry/borrow) - */ - if (l != NULL) - { - lp= &(t[n2]); - c1= (int)(bn_add_words(lp,&(t[n2+n]),&(l[0]),n)); - } - else - { - lp= &(t[n2+n]); - c1=0; - } - c1+=(int)(bn_add_words(&(t[n2]),lp, &(r[0]),n)); - if (oneg) - c1-=(int)(bn_sub_words(&(t[n2]),&(t[n2]),&(t[0]),n)); - else - c1+=(int)(bn_add_words(&(t[n2]),&(t[n2]),&(t[0]),n)); - - c2 =(int)(bn_add_words(&(r[0]),&(r[0]),&(t[n2+n]),n)); - c2+=(int)(bn_add_words(&(r[0]),&(r[0]),&(r[n]),n)); - if (oneg) - c2-=(int)(bn_sub_words(&(r[0]),&(r[0]),&(t[n]),n)); - else - c2+=(int)(bn_add_words(&(r[0]),&(r[0]),&(t[n]),n)); - - if (c1 != 0) /* Add starting at r[0], could be +ve or -ve */ - { - i=0; - if (c1 > 0) - { - lc=c1; - do { - ll=(r[i]+lc)&BN_MASK2; - r[i++]=ll; - lc=(lc > ll); - } while (lc); - } - else - { - lc= -c1; - do { - ll=r[i]; - r[i++]=(ll-lc)&BN_MASK2; - lc=(lc > ll); - } while (lc); - } - } - if (c2 != 0) /* Add starting at r[1] */ - { - i=n; - if (c2 > 0) - { - lc=c2; - do { - ll=(r[i]+lc)&BN_MASK2; - r[i++]=ll; - lc=(lc > ll); - } while (lc); - } - else - { - lc= -c2; - do { - ll=r[i]; - r[i++]=(ll-lc)&BN_MASK2; - lc=(lc > ll); - } while (lc); - } - } - } -#endif /* BN_RECURSION */ + { + bn_mul_recursive(&(t[0]), &(r[0]), &(r[n]), n, 0, 0, &(t[n2])); + bn_mul_recursive(r, &(a[n]), &(b[n]), n, 0, 0, &(t[n2])); + } + + /*- + * s0 == low(al*bl) + * s1 == low(ah*bh)+low((al-ah)*(bh-bl))+low(al*bl)+high(al*bl) + * We know s0 and s1 so the only unknown is high(al*bl) + * high(al*bl) == s1 - low(ah*bh+s0+(al-ah)*(bh-bl)) + * high(al*bl) == s1 - (r[0]+l[0]+t[0]) + */ + if (l != NULL) { + lp = &(t[n2 + n]); + c1 = (int)(bn_add_words(lp, &(r[0]), &(l[0]), n)); + } else { + c1 = 0; + lp = &(r[0]); + } + + if (neg) + neg = (int)(bn_sub_words(&(t[n2]), lp, &(t[0]), n)); + else { + bn_add_words(&(t[n2]), lp, &(t[0]), n); + neg = 0; + } + + if (l != NULL) { + bn_sub_words(&(t[n2 + n]), &(l[n]), &(t[n2]), n); + } else { + lp = &(t[n2 + n]); + mp = &(t[n2]); + for (i = 0; i < n; i++) + lp[i] = ((~mp[i]) + 1) & BN_MASK2; + } + + /*- + * s[0] = low(al*bl) + * t[3] = high(al*bl) + * t[10] = (a[0]-a[1])*(b[1]-b[0]) neg is the sign + * r[10] = (a[1]*b[1]) + */ + /*- + * R[10] = al*bl + * R[21] = al*bl + ah*bh + (a[0]-a[1])*(b[1]-b[0]) + * R[32] = ah*bh + */ + /*- + * R[1]=t[3]+l[0]+r[0](+-)t[0] (have carry/borrow) + * R[2]=r[0]+t[3]+r[1](+-)t[1] (have carry/borrow) + * R[3]=r[1]+(carry/borrow) + */ + if (l != NULL) { + lp = &(t[n2]); + c1 = (int)(bn_add_words(lp, &(t[n2 + n]), &(l[0]), n)); + } else { + lp = &(t[n2 + n]); + c1 = 0; + } + c1 += (int)(bn_add_words(&(t[n2]), lp, &(r[0]), n)); + if (oneg) + c1 -= (int)(bn_sub_words(&(t[n2]), &(t[n2]), &(t[0]), n)); + else + c1 += (int)(bn_add_words(&(t[n2]), &(t[n2]), &(t[0]), n)); + + c2 = (int)(bn_add_words(&(r[0]), &(r[0]), &(t[n2 + n]), n)); + c2 += (int)(bn_add_words(&(r[0]), &(r[0]), &(r[n]), n)); + if (oneg) + c2 -= (int)(bn_sub_words(&(r[0]), &(r[0]), &(t[n]), n)); + else + c2 += (int)(bn_add_words(&(r[0]), &(r[0]), &(t[n]), n)); + + if (c1 != 0) { /* Add starting at r[0], could be +ve or -ve */ + i = 0; + if (c1 > 0) { + lc = c1; + do { + ll = (r[i] + lc) & BN_MASK2; + r[i++] = ll; + lc = (lc > ll); + } while (lc); + } else { + lc = -c1; + do { + ll = r[i]; + r[i++] = (ll - lc) & BN_MASK2; + lc = (lc > ll); + } while (lc); + } + } + if (c2 != 0) { /* Add starting at r[1] */ + i = n; + if (c2 > 0) { + lc = c2; + do { + ll = (r[i] + lc) & BN_MASK2; + r[i++] = ll; + lc = (lc > ll); + } while (lc); + } else { + lc = -c2; + do { + ll = r[i]; + r[i++] = (ll - lc) & BN_MASK2; + lc = (lc > ll); + } while (lc); + } + } +} +#endif /* BN_RECURSION */ int BN_mul(BIGNUM *r, const BIGNUM *a, const BIGNUM *b, BN_CTX *ctx) - { - int ret=0; - int top,al,bl; - BIGNUM *rr; +{ + int ret = 0; + int top, al, bl; + BIGNUM *rr; #if defined(BN_MUL_COMBA) || defined(BN_RECURSION) - int i; + int i; #endif #ifdef BN_RECURSION - BIGNUM *t=NULL; - int j=0,k; + BIGNUM *t = NULL; + int j = 0, k; #endif #ifdef BN_COUNT - fprintf(stderr,"BN_mul %d * %d\n",a->top,b->top); + fprintf(stderr, "BN_mul %d * %d\n", a->top, b->top); #endif - bn_check_top(a); - bn_check_top(b); - bn_check_top(r); - - al=a->top; - bl=b->top; - - if ((al == 0) || (bl == 0)) - { - BN_zero(r); - return(1); - } - top=al+bl; - - BN_CTX_start(ctx); - if ((r == a) || (r == b)) - { - if ((rr = BN_CTX_get(ctx)) == NULL) goto err; - } - else - rr = r; - rr->neg=a->neg^b->neg; + bn_check_top(a); + bn_check_top(b); + bn_check_top(r); + + al = a->top; + bl = b->top; + + if ((al == 0) || (bl == 0)) { + BN_zero(r); + return (1); + } + top = al + bl; + + BN_CTX_start(ctx); + if ((r == a) || (r == b)) { + if ((rr = BN_CTX_get(ctx)) == NULL) + goto err; + } else + rr = r; + rr->neg = a->neg ^ b->neg; #if defined(BN_MUL_COMBA) || defined(BN_RECURSION) - i = al-bl; + i = al - bl; #endif #ifdef BN_MUL_COMBA - if (i == 0) - { + if (i == 0) { # if 0 - if (al == 4) - { - if (bn_wexpand(rr,8) == NULL) goto err; - rr->top=8; - bn_mul_comba4(rr->d,a->d,b->d); - goto end; - } + if (al == 4) { + if (bn_wexpand(rr, 8) == NULL) + goto err; + rr->top = 8; + bn_mul_comba4(rr->d, a->d, b->d); + goto end; + } # endif - if (al == 8) - { - if (bn_wexpand(rr,16) == NULL) goto err; - rr->top=16; - bn_mul_comba8(rr->d,a->d,b->d); - goto end; - } - } -#endif /* BN_MUL_COMBA */ + if (al == 8) { + if (bn_wexpand(rr, 16) == NULL) + goto err; + rr->top = 16; + bn_mul_comba8(rr->d, a->d, b->d); + goto end; + } + } +#endif /* BN_MUL_COMBA */ #ifdef BN_RECURSION - if ((al >= BN_MULL_SIZE_NORMAL) && (bl >= BN_MULL_SIZE_NORMAL)) - { - if (i >= -1 && i <= 1) - { - /* Find out the power of two lower or equal - to the longest of the two numbers */ - if (i >= 0) - { - j = BN_num_bits_word((BN_ULONG)al); - } - if (i == -1) - { - j = BN_num_bits_word((BN_ULONG)bl); - } - j = 1<<(j-1); - assert(j <= al || j <= bl); - k = j+j; - t = BN_CTX_get(ctx); - if (t == NULL) - goto err; - if (al > j || bl > j) - { - if (bn_wexpand(t,k*4) == NULL) goto err; - if (bn_wexpand(rr,k*4) == NULL) goto err; - bn_mul_part_recursive(rr->d,a->d,b->d, - j,al-j,bl-j,t->d); - } - else /* al <= j || bl <= j */ - { - if (bn_wexpand(t,k*2) == NULL) goto err; - if (bn_wexpand(rr,k*2) == NULL) goto err; - bn_mul_recursive(rr->d,a->d,b->d, - j,al-j,bl-j,t->d); - } - rr->top=top; - goto end; - } -#if 0 - if (i == 1 && !BN_get_flags(b,BN_FLG_STATIC_DATA)) - { - BIGNUM *tmp_bn = (BIGNUM *)b; - if (bn_wexpand(tmp_bn,al) == NULL) goto err; - tmp_bn->d[bl]=0; - bl++; - i--; - } - else if (i == -1 && !BN_get_flags(a,BN_FLG_STATIC_DATA)) - { - BIGNUM *tmp_bn = (BIGNUM *)a; - if (bn_wexpand(tmp_bn,bl) == NULL) goto err; - tmp_bn->d[al]=0; - al++; - i++; - } - if (i == 0) - { - /* symmetric and > 4 */ - /* 16 or larger */ - j=BN_num_bits_word((BN_ULONG)al); - j=1<<(j-1); - k=j+j; - t = BN_CTX_get(ctx); - if (al == j) /* exact multiple */ - { - if (bn_wexpand(t,k*2) == NULL) goto err; - if (bn_wexpand(rr,k*2) == NULL) goto err; - bn_mul_recursive(rr->d,a->d,b->d,al,t->d); - } - else - { - if (bn_wexpand(t,k*4) == NULL) goto err; - if (bn_wexpand(rr,k*4) == NULL) goto err; - bn_mul_part_recursive(rr->d,a->d,b->d,al-j,j,t->d); - } - rr->top=top; - goto end; - } -#endif - } -#endif /* BN_RECURSION */ - if (bn_wexpand(rr,top) == NULL) goto err; - rr->top=top; - bn_mul_normal(rr->d,a->d,al,b->d,bl); + if ((al >= BN_MULL_SIZE_NORMAL) && (bl >= BN_MULL_SIZE_NORMAL)) { + if (i >= -1 && i <= 1) { + /* + * Find out the power of two lower or equal to the longest of the + * two numbers + */ + if (i >= 0) { + j = BN_num_bits_word((BN_ULONG)al); + } + if (i == -1) { + j = BN_num_bits_word((BN_ULONG)bl); + } + j = 1 << (j - 1); + assert(j <= al || j <= bl); + k = j + j; + t = BN_CTX_get(ctx); + if (t == NULL) + goto err; + if (al > j || bl > j) { + if (bn_wexpand(t, k * 4) == NULL) + goto err; + if (bn_wexpand(rr, k * 4) == NULL) + goto err; + bn_mul_part_recursive(rr->d, a->d, b->d, + j, al - j, bl - j, t->d); + } else { /* al <= j || bl <= j */ + + if (bn_wexpand(t, k * 2) == NULL) + goto err; + if (bn_wexpand(rr, k * 2) == NULL) + goto err; + bn_mul_recursive(rr->d, a->d, b->d, j, al - j, bl - j, t->d); + } + rr->top = top; + goto end; + } +# if 0 + if (i == 1 && !BN_get_flags(b, BN_FLG_STATIC_DATA)) { + BIGNUM *tmp_bn = (BIGNUM *)b; + if (bn_wexpand(tmp_bn, al) == NULL) + goto err; + tmp_bn->d[bl] = 0; + bl++; + i--; + } else if (i == -1 && !BN_get_flags(a, BN_FLG_STATIC_DATA)) { + BIGNUM *tmp_bn = (BIGNUM *)a; + if (bn_wexpand(tmp_bn, bl) == NULL) + goto err; + tmp_bn->d[al] = 0; + al++; + i++; + } + if (i == 0) { + /* symmetric and > 4 */ + /* 16 or larger */ + j = BN_num_bits_word((BN_ULONG)al); + j = 1 << (j - 1); + k = j + j; + t = BN_CTX_get(ctx); + if (al == j) { /* exact multiple */ + if (bn_wexpand(t, k * 2) == NULL) + goto err; + if (bn_wexpand(rr, k * 2) == NULL) + goto err; + bn_mul_recursive(rr->d, a->d, b->d, al, t->d); + } else { + if (bn_wexpand(t, k * 4) == NULL) + goto err; + if (bn_wexpand(rr, k * 4) == NULL) + goto err; + bn_mul_part_recursive(rr->d, a->d, b->d, al - j, j, t->d); + } + rr->top = top; + goto end; + } +# endif + } +#endif /* BN_RECURSION */ + if (bn_wexpand(rr, top) == NULL) + goto err; + rr->top = top; + bn_mul_normal(rr->d, a->d, al, b->d, bl); #if defined(BN_MUL_COMBA) || defined(BN_RECURSION) -end: + end: #endif - bn_correct_top(rr); - if (r != rr) BN_copy(r,rr); - ret=1; -err: - bn_check_top(r); - BN_CTX_end(ctx); - return(ret); - } + bn_correct_top(rr); + if (r != rr) + BN_copy(r, rr); + ret = 1; + err: + bn_check_top(r); + BN_CTX_end(ctx); + return (ret); +} void bn_mul_normal(BN_ULONG *r, BN_ULONG *a, int na, BN_ULONG *b, int nb) - { - BN_ULONG *rr; +{ + BN_ULONG *rr; #ifdef BN_COUNT - fprintf(stderr," bn_mul_normal %d * %d\n",na,nb); + fprintf(stderr, " bn_mul_normal %d * %d\n", na, nb); #endif - if (na < nb) - { - int itmp; - BN_ULONG *ltmp; - - itmp=na; na=nb; nb=itmp; - ltmp=a; a=b; b=ltmp; - - } - rr= &(r[na]); - if (nb <= 0) - { - (void)bn_mul_words(r,a,na,0); - return; - } - else - rr[0]=bn_mul_words(r,a,na,b[0]); - - for (;;) - { - if (--nb <= 0) return; - rr[1]=bn_mul_add_words(&(r[1]),a,na,b[1]); - if (--nb <= 0) return; - rr[2]=bn_mul_add_words(&(r[2]),a,na,b[2]); - if (--nb <= 0) return; - rr[3]=bn_mul_add_words(&(r[3]),a,na,b[3]); - if (--nb <= 0) return; - rr[4]=bn_mul_add_words(&(r[4]),a,na,b[4]); - rr+=4; - r+=4; - b+=4; - } - } + if (na < nb) { + int itmp; + BN_ULONG *ltmp; + + itmp = na; + na = nb; + nb = itmp; + ltmp = a; + a = b; + b = ltmp; + + } + rr = &(r[na]); + if (nb <= 0) { + (void)bn_mul_words(r, a, na, 0); + return; + } else + rr[0] = bn_mul_words(r, a, na, b[0]); + + for (;;) { + if (--nb <= 0) + return; + rr[1] = bn_mul_add_words(&(r[1]), a, na, b[1]); + if (--nb <= 0) + return; + rr[2] = bn_mul_add_words(&(r[2]), a, na, b[2]); + if (--nb <= 0) + return; + rr[3] = bn_mul_add_words(&(r[3]), a, na, b[3]); + if (--nb <= 0) + return; + rr[4] = bn_mul_add_words(&(r[4]), a, na, b[4]); + rr += 4; + r += 4; + b += 4; + } +} void bn_mul_low_normal(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n) - { +{ #ifdef BN_COUNT - fprintf(stderr," bn_mul_low_normal %d * %d\n",n,n); + fprintf(stderr, " bn_mul_low_normal %d * %d\n", n, n); #endif - bn_mul_words(r,a,n,b[0]); - - for (;;) - { - if (--n <= 0) return; - bn_mul_add_words(&(r[1]),a,n,b[1]); - if (--n <= 0) return; - bn_mul_add_words(&(r[2]),a,n,b[2]); - if (--n <= 0) return; - bn_mul_add_words(&(r[3]),a,n,b[3]); - if (--n <= 0) return; - bn_mul_add_words(&(r[4]),a,n,b[4]); - r+=4; - b+=4; - } - } + bn_mul_words(r, a, n, b[0]); + + for (;;) { + if (--n <= 0) + return; + bn_mul_add_words(&(r[1]), a, n, b[1]); + if (--n <= 0) + return; + bn_mul_add_words(&(r[2]), a, n, b[2]); + if (--n <= 0) + return; + bn_mul_add_words(&(r[3]), a, n, b[3]); + if (--n <= 0) + return; + bn_mul_add_words(&(r[4]), a, n, b[4]); + r += 4; + b += 4; + } +} diff --git a/openssl/crypto/bn/bn_nist.c b/openssl/crypto/bn/bn_nist.c index abb157085..4a45404c6 100644 --- a/openssl/crypto/bn/bn_nist.c +++ b/openssl/crypto/bn/bn_nist.c @@ -10,7 +10,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -59,265 +59,277 @@ #include "bn_lcl.h" #include "cryptlib.h" - -#define BN_NIST_192_TOP (192+BN_BITS2-1)/BN_BITS2 -#define BN_NIST_224_TOP (224+BN_BITS2-1)/BN_BITS2 -#define BN_NIST_256_TOP (256+BN_BITS2-1)/BN_BITS2 -#define BN_NIST_384_TOP (384+BN_BITS2-1)/BN_BITS2 -#define BN_NIST_521_TOP (521+BN_BITS2-1)/BN_BITS2 +#define BN_NIST_192_TOP (192+BN_BITS2-1)/BN_BITS2 +#define BN_NIST_224_TOP (224+BN_BITS2-1)/BN_BITS2 +#define BN_NIST_256_TOP (256+BN_BITS2-1)/BN_BITS2 +#define BN_NIST_384_TOP (384+BN_BITS2-1)/BN_BITS2 +#define BN_NIST_521_TOP (521+BN_BITS2-1)/BN_BITS2 /* pre-computed tables are "carry-less" values of modulus*(i+1) */ #if BN_BITS2 == 64 static const BN_ULONG _nist_p_192[][BN_NIST_192_TOP] = { - {0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFFULL}, - {0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFFULL}, - {0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFCULL,0xFFFFFFFFFFFFFFFFULL} - }; + {0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFEULL, 0xFFFFFFFFFFFFFFFFULL}, + {0xFFFFFFFFFFFFFFFEULL, 0xFFFFFFFFFFFFFFFDULL, 0xFFFFFFFFFFFFFFFFULL}, + {0xFFFFFFFFFFFFFFFDULL, 0xFFFFFFFFFFFFFFFCULL, 0xFFFFFFFFFFFFFFFFULL} +}; + static const BN_ULONG _nist_p_192_sqr[] = { - 0x0000000000000001ULL,0x0000000000000002ULL,0x0000000000000001ULL, - 0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFFULL - }; + 0x0000000000000001ULL, 0x0000000000000002ULL, 0x0000000000000001ULL, + 0xFFFFFFFFFFFFFFFEULL, 0xFFFFFFFFFFFFFFFDULL, 0xFFFFFFFFFFFFFFFFULL +}; + static const BN_ULONG _nist_p_224[][BN_NIST_224_TOP] = { - {0x0000000000000001ULL,0xFFFFFFFF00000000ULL, - 0xFFFFFFFFFFFFFFFFULL,0x00000000FFFFFFFFULL}, - {0x0000000000000002ULL,0xFFFFFFFE00000000ULL, - 0xFFFFFFFFFFFFFFFFULL,0x00000001FFFFFFFFULL} /* this one is "carry-full" */ - }; + {0x0000000000000001ULL, 0xFFFFFFFF00000000ULL, + 0xFFFFFFFFFFFFFFFFULL, 0x00000000FFFFFFFFULL}, + {0x0000000000000002ULL, 0xFFFFFFFE00000000ULL, + 0xFFFFFFFFFFFFFFFFULL, 0x00000001FFFFFFFFULL} /* this one is + * "carry-full" */ +}; + static const BN_ULONG _nist_p_224_sqr[] = { - 0x0000000000000001ULL,0xFFFFFFFE00000000ULL, - 0xFFFFFFFFFFFFFFFFULL,0x0000000200000000ULL, - 0x0000000000000000ULL,0xFFFFFFFFFFFFFFFEULL, - 0xFFFFFFFFFFFFFFFFULL - }; + 0x0000000000000001ULL, 0xFFFFFFFE00000000ULL, + 0xFFFFFFFFFFFFFFFFULL, 0x0000000200000000ULL, + 0x0000000000000000ULL, 0xFFFFFFFFFFFFFFFEULL, + 0xFFFFFFFFFFFFFFFFULL +}; + static const BN_ULONG _nist_p_256[][BN_NIST_256_TOP] = { - {0xFFFFFFFFFFFFFFFFULL,0x00000000FFFFFFFFULL, - 0x0000000000000000ULL,0xFFFFFFFF00000001ULL}, - {0xFFFFFFFFFFFFFFFEULL,0x00000001FFFFFFFFULL, - 0x0000000000000000ULL,0xFFFFFFFE00000002ULL}, - {0xFFFFFFFFFFFFFFFDULL,0x00000002FFFFFFFFULL, - 0x0000000000000000ULL,0xFFFFFFFD00000003ULL}, - {0xFFFFFFFFFFFFFFFCULL,0x00000003FFFFFFFFULL, - 0x0000000000000000ULL,0xFFFFFFFC00000004ULL}, - {0xFFFFFFFFFFFFFFFBULL,0x00000004FFFFFFFFULL, - 0x0000000000000000ULL,0xFFFFFFFB00000005ULL}, - }; + {0xFFFFFFFFFFFFFFFFULL, 0x00000000FFFFFFFFULL, + 0x0000000000000000ULL, 0xFFFFFFFF00000001ULL}, + {0xFFFFFFFFFFFFFFFEULL, 0x00000001FFFFFFFFULL, + 0x0000000000000000ULL, 0xFFFFFFFE00000002ULL}, + {0xFFFFFFFFFFFFFFFDULL, 0x00000002FFFFFFFFULL, + 0x0000000000000000ULL, 0xFFFFFFFD00000003ULL}, + {0xFFFFFFFFFFFFFFFCULL, 0x00000003FFFFFFFFULL, + 0x0000000000000000ULL, 0xFFFFFFFC00000004ULL}, + {0xFFFFFFFFFFFFFFFBULL, 0x00000004FFFFFFFFULL, + 0x0000000000000000ULL, 0xFFFFFFFB00000005ULL}, +}; + static const BN_ULONG _nist_p_256_sqr[] = { - 0x0000000000000001ULL,0xFFFFFFFE00000000ULL, - 0xFFFFFFFFFFFFFFFFULL,0x00000001FFFFFFFEULL, - 0x00000001FFFFFFFEULL,0x00000001FFFFFFFEULL, - 0xFFFFFFFE00000001ULL,0xFFFFFFFE00000002ULL - }; + 0x0000000000000001ULL, 0xFFFFFFFE00000000ULL, + 0xFFFFFFFFFFFFFFFFULL, 0x00000001FFFFFFFEULL, + 0x00000001FFFFFFFEULL, 0x00000001FFFFFFFEULL, + 0xFFFFFFFE00000001ULL, 0xFFFFFFFE00000002ULL +}; + static const BN_ULONG _nist_p_384[][BN_NIST_384_TOP] = { - {0x00000000FFFFFFFFULL,0xFFFFFFFF00000000ULL,0xFFFFFFFFFFFFFFFEULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL}, - {0x00000001FFFFFFFEULL,0xFFFFFFFE00000000ULL,0xFFFFFFFFFFFFFFFDULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL}, - {0x00000002FFFFFFFDULL,0xFFFFFFFD00000000ULL,0xFFFFFFFFFFFFFFFCULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL}, - {0x00000003FFFFFFFCULL,0xFFFFFFFC00000000ULL,0xFFFFFFFFFFFFFFFBULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL}, - {0x00000004FFFFFFFBULL,0xFFFFFFFB00000000ULL,0xFFFFFFFFFFFFFFFAULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL}, - }; + {0x00000000FFFFFFFFULL, 0xFFFFFFFF00000000ULL, 0xFFFFFFFFFFFFFFFEULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL}, + {0x00000001FFFFFFFEULL, 0xFFFFFFFE00000000ULL, 0xFFFFFFFFFFFFFFFDULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL}, + {0x00000002FFFFFFFDULL, 0xFFFFFFFD00000000ULL, 0xFFFFFFFFFFFFFFFCULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL}, + {0x00000003FFFFFFFCULL, 0xFFFFFFFC00000000ULL, 0xFFFFFFFFFFFFFFFBULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL}, + {0x00000004FFFFFFFBULL, 0xFFFFFFFB00000000ULL, 0xFFFFFFFFFFFFFFFAULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL}, +}; + static const BN_ULONG _nist_p_384_sqr[] = { - 0xFFFFFFFE00000001ULL,0x0000000200000000ULL,0xFFFFFFFE00000000ULL, - 0x0000000200000000ULL,0x0000000000000001ULL,0x0000000000000000ULL, - 0x00000001FFFFFFFEULL,0xFFFFFFFE00000000ULL,0xFFFFFFFFFFFFFFFDULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL - }; + 0xFFFFFFFE00000001ULL, 0x0000000200000000ULL, 0xFFFFFFFE00000000ULL, + 0x0000000200000000ULL, 0x0000000000000001ULL, 0x0000000000000000ULL, + 0x00000001FFFFFFFEULL, 0xFFFFFFFE00000000ULL, 0xFFFFFFFFFFFFFFFDULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL +}; + static const BN_ULONG _nist_p_521[] = - {0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL, - 0x00000000000001FFULL}; + { 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, + 0x00000000000001FFULL +}; + static const BN_ULONG _nist_p_521_sqr[] = { - 0x0000000000000001ULL,0x0000000000000000ULL,0x0000000000000000ULL, - 0x0000000000000000ULL,0x0000000000000000ULL,0x0000000000000000ULL, - 0x0000000000000000ULL,0x0000000000000000ULL,0xFFFFFFFFFFFFFC00ULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL, - 0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL, - 0xFFFFFFFFFFFFFFFFULL,0x000000000003FFFFULL - }; + 0x0000000000000001ULL, 0x0000000000000000ULL, 0x0000000000000000ULL, + 0x0000000000000000ULL, 0x0000000000000000ULL, 0x0000000000000000ULL, + 0x0000000000000000ULL, 0x0000000000000000ULL, 0xFFFFFFFFFFFFFC00ULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, + 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, 0xFFFFFFFFFFFFFFFFULL, + 0xFFFFFFFFFFFFFFFFULL, 0x000000000003FFFFULL +}; #elif BN_BITS2 == 32 static const BN_ULONG _nist_p_192[][BN_NIST_192_TOP] = { - {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFC,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF} - }; + {0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFE, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0xFFFFFFFE, 0xFFFFFFFF, 0xFFFFFFFD, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0xFFFFFFFD, 0xFFFFFFFF, 0xFFFFFFFC, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF} +}; + static const BN_ULONG _nist_p_192_sqr[] = { - 0x00000001,0x00000000,0x00000002,0x00000000,0x00000001,0x00000000, - 0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF - }; + 0x00000001, 0x00000000, 0x00000002, 0x00000000, 0x00000001, 0x00000000, + 0xFFFFFFFE, 0xFFFFFFFF, 0xFFFFFFFD, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF +}; + static const BN_ULONG _nist_p_224[][BN_NIST_224_TOP] = { - {0x00000001,0x00000000,0x00000000,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0x00000002,0x00000000,0x00000000,0xFFFFFFFE, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF} - }; + {0x00000001, 0x00000000, 0x00000000, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0x00000002, 0x00000000, 0x00000000, 0xFFFFFFFE, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF} +}; + static const BN_ULONG _nist_p_224_sqr[] = { - 0x00000001,0x00000000,0x00000000,0xFFFFFFFE, - 0xFFFFFFFF,0xFFFFFFFF,0x00000000,0x00000002, - 0x00000000,0x00000000,0xFFFFFFFE,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF - }; + 0x00000001, 0x00000000, 0x00000000, 0xFFFFFFFE, + 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, 0x00000002, + 0x00000000, 0x00000000, 0xFFFFFFFE, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF +}; + static const BN_ULONG _nist_p_256[][BN_NIST_256_TOP] = { - {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0x00000000, - 0x00000000,0x00000000,0x00000001,0xFFFFFFFF}, - {0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFF,0x00000001, - 0x00000000,0x00000000,0x00000002,0xFFFFFFFE}, - {0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0x00000002, - 0x00000000,0x00000000,0x00000003,0xFFFFFFFD}, - {0xFFFFFFFC,0xFFFFFFFF,0xFFFFFFFF,0x00000003, - 0x00000000,0x00000000,0x00000004,0xFFFFFFFC}, - {0xFFFFFFFB,0xFFFFFFFF,0xFFFFFFFF,0x00000004, - 0x00000000,0x00000000,0x00000005,0xFFFFFFFB}, - }; + {0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000000, + 0x00000000, 0x00000000, 0x00000001, 0xFFFFFFFF}, + {0xFFFFFFFE, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000001, + 0x00000000, 0x00000000, 0x00000002, 0xFFFFFFFE}, + {0xFFFFFFFD, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000002, + 0x00000000, 0x00000000, 0x00000003, 0xFFFFFFFD}, + {0xFFFFFFFC, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000003, + 0x00000000, 0x00000000, 0x00000004, 0xFFFFFFFC}, + {0xFFFFFFFB, 0xFFFFFFFF, 0xFFFFFFFF, 0x00000004, + 0x00000000, 0x00000000, 0x00000005, 0xFFFFFFFB}, +}; + static const BN_ULONG _nist_p_256_sqr[] = { - 0x00000001,0x00000000,0x00000000,0xFFFFFFFE, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFE,0x00000001, - 0xFFFFFFFE,0x00000001,0xFFFFFFFE,0x00000001, - 0x00000001,0xFFFFFFFE,0x00000002,0xFFFFFFFE - }; + 0x00000001, 0x00000000, 0x00000000, 0xFFFFFFFE, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFE, 0x00000001, + 0xFFFFFFFE, 0x00000001, 0xFFFFFFFE, 0x00000001, + 0x00000001, 0xFFFFFFFE, 0x00000002, 0xFFFFFFFE +}; + static const BN_ULONG _nist_p_384[][BN_NIST_384_TOP] = { - {0xFFFFFFFF,0x00000000,0x00000000,0xFFFFFFFF,0xFFFFFFFE,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0xFFFFFFFE,0x00000001,0x00000000,0xFFFFFFFE,0xFFFFFFFD,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0xFFFFFFFD,0x00000002,0x00000000,0xFFFFFFFD,0xFFFFFFFC,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0xFFFFFFFC,0x00000003,0x00000000,0xFFFFFFFC,0xFFFFFFFB,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - {0xFFFFFFFB,0x00000004,0x00000000,0xFFFFFFFB,0xFFFFFFFA,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}, - }; + {0xFFFFFFFF, 0x00000000, 0x00000000, 0xFFFFFFFF, 0xFFFFFFFE, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0xFFFFFFFE, 0x00000001, 0x00000000, 0xFFFFFFFE, 0xFFFFFFFD, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0xFFFFFFFD, 0x00000002, 0x00000000, 0xFFFFFFFD, 0xFFFFFFFC, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0xFFFFFFFC, 0x00000003, 0x00000000, 0xFFFFFFFC, 0xFFFFFFFB, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, + {0xFFFFFFFB, 0x00000004, 0x00000000, 0xFFFFFFFB, 0xFFFFFFFA, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF}, +}; + static const BN_ULONG _nist_p_384_sqr[] = { - 0x00000001,0xFFFFFFFE,0x00000000,0x00000002,0x00000000,0xFFFFFFFE, - 0x00000000,0x00000002,0x00000001,0x00000000,0x00000000,0x00000000, - 0xFFFFFFFE,0x00000001,0x00000000,0xFFFFFFFE,0xFFFFFFFD,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF - }; -static const BN_ULONG _nist_p_521[] = {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF, - 0xFFFFFFFF,0x000001FF}; + 0x00000001, 0xFFFFFFFE, 0x00000000, 0x00000002, 0x00000000, 0xFFFFFFFE, + 0x00000000, 0x00000002, 0x00000001, 0x00000000, 0x00000000, 0x00000000, + 0xFFFFFFFE, 0x00000001, 0x00000000, 0xFFFFFFFE, 0xFFFFFFFD, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF +}; + +static const BN_ULONG _nist_p_521[] = { 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0x000001FF +}; + static const BN_ULONG _nist_p_521_sqr[] = { - 0x00000001,0x00000000,0x00000000,0x00000000,0x00000000,0x00000000, - 0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,0x00000000, - 0x00000000,0x00000000,0x00000000,0x00000000,0xFFFFFC00,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF, - 0xFFFFFFFF,0xFFFFFFFF,0x0003FFFF - }; + 0x00000001, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, + 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0xFFFFFC00, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, + 0xFFFFFFFF, 0xFFFFFFFF, 0x0003FFFF +}; #else -#error "unsupported BN_BITS2" +# error "unsupported BN_BITS2" #endif - -static const BIGNUM _bignum_nist_p_192 = - { - (BN_ULONG *)_nist_p_192[0], - BN_NIST_192_TOP, - BN_NIST_192_TOP, - 0, - BN_FLG_STATIC_DATA - }; - -static const BIGNUM _bignum_nist_p_224 = - { - (BN_ULONG *)_nist_p_224[0], - BN_NIST_224_TOP, - BN_NIST_224_TOP, - 0, - BN_FLG_STATIC_DATA - }; - -static const BIGNUM _bignum_nist_p_256 = - { - (BN_ULONG *)_nist_p_256[0], - BN_NIST_256_TOP, - BN_NIST_256_TOP, - 0, - BN_FLG_STATIC_DATA - }; - -static const BIGNUM _bignum_nist_p_384 = - { - (BN_ULONG *)_nist_p_384[0], - BN_NIST_384_TOP, - BN_NIST_384_TOP, - 0, - BN_FLG_STATIC_DATA - }; - -static const BIGNUM _bignum_nist_p_521 = - { - (BN_ULONG *)_nist_p_521, - BN_NIST_521_TOP, - BN_NIST_521_TOP, - 0, - BN_FLG_STATIC_DATA - }; - +static const BIGNUM _bignum_nist_p_192 = { + (BN_ULONG *)_nist_p_192[0], + BN_NIST_192_TOP, + BN_NIST_192_TOP, + 0, + BN_FLG_STATIC_DATA +}; + +static const BIGNUM _bignum_nist_p_224 = { + (BN_ULONG *)_nist_p_224[0], + BN_NIST_224_TOP, + BN_NIST_224_TOP, + 0, + BN_FLG_STATIC_DATA +}; + +static const BIGNUM _bignum_nist_p_256 = { + (BN_ULONG *)_nist_p_256[0], + BN_NIST_256_TOP, + BN_NIST_256_TOP, + 0, + BN_FLG_STATIC_DATA +}; + +static const BIGNUM _bignum_nist_p_384 = { + (BN_ULONG *)_nist_p_384[0], + BN_NIST_384_TOP, + BN_NIST_384_TOP, + 0, + BN_FLG_STATIC_DATA +}; + +static const BIGNUM _bignum_nist_p_521 = { + (BN_ULONG *)_nist_p_521, + BN_NIST_521_TOP, + BN_NIST_521_TOP, + 0, + BN_FLG_STATIC_DATA +}; const BIGNUM *BN_get0_nist_prime_192(void) - { - return &_bignum_nist_p_192; - } +{ + return &_bignum_nist_p_192; +} const BIGNUM *BN_get0_nist_prime_224(void) - { - return &_bignum_nist_p_224; - } +{ + return &_bignum_nist_p_224; +} const BIGNUM *BN_get0_nist_prime_256(void) - { - return &_bignum_nist_p_256; - } +{ + return &_bignum_nist_p_256; +} const BIGNUM *BN_get0_nist_prime_384(void) - { - return &_bignum_nist_p_384; - } +{ + return &_bignum_nist_p_384; +} const BIGNUM *BN_get0_nist_prime_521(void) - { - return &_bignum_nist_p_521; - } - +{ + return &_bignum_nist_p_521; +} static void nist_cp_bn_0(BN_ULONG *dst, const BN_ULONG *src, int top, int max) - { - int i; +{ + int i; #ifdef BN_DEBUG - OPENSSL_assert(top <= max); + OPENSSL_assert(top <= max); #endif - for (i = 0; i < top; i++) - dst[i] = src[i]; - for (; i < max; i++) - dst[i] = 0; - } + for (i = 0; i < top; i++) + dst[i] = src[i]; + for (; i < max; i++) + dst[i] = 0; +} static void nist_cp_bn(BN_ULONG *dst, const BN_ULONG *src, int top) - { - int i; +{ + int i; - for (i = 0; i < top; i++) - dst[i] = src[i]; - } + for (i = 0; i < top; i++) + dst[i] = src[i]; +} #if BN_BITS2 == 64 -#define bn_cp_64(to, n, from, m) (to)[n] = (m>=0)?((from)[m]):0; -#define bn_64_set_0(to, n) (to)[n] = (BN_ULONG)0; +# define bn_cp_64(to, n, from, m) (to)[n] = (m>=0)?((from)[m]):0; +# define bn_64_set_0(to, n) (to)[n] = (BN_ULONG)0; /* * two following macros are implemented under assumption that they * are called in a sequence with *ascending* n, i.e. as they are... */ -#define bn_cp_32_naked(to, n, from, m) (((n)&1)?(to[(n)/2]|=((m)&1)?(from[(m)/2]&BN_MASK2h):(from[(m)/2]<<32))\ - :(to[(n)/2] =((m)&1)?(from[(m)/2]>>32):(from[(m)/2]&BN_MASK2l))) -#define bn_32_set_0(to, n) (((n)&1)?(to[(n)/2]&=BN_MASK2l):(to[(n)/2]=0)); -#define bn_cp_32(to,n,from,m) ((m)>=0)?bn_cp_32_naked(to,n,from,m):bn_32_set_0(to,n) +# define bn_cp_32_naked(to, n, from, m) (((n)&1)?(to[(n)/2]|=((m)&1)?(from[(m)/2]&BN_MASK2h):(from[(m)/2]<<32))\ + :(to[(n)/2] =((m)&1)?(from[(m)/2]>>32):(from[(m)/2]&BN_MASK2l))) +# define bn_32_set_0(to, n) (((n)&1)?(to[(n)/2]&=BN_MASK2l):(to[(n)/2]=0)); +# define bn_cp_32(to,n,from,m) ((m)>=0)?bn_cp_32_naked(to,n,from,m):bn_32_set_0(to,n) # if defined(L_ENDIAN) # if defined(__arch64__) # define NIST_INT64 long @@ -326,784 +338,925 @@ static void nist_cp_bn(BN_ULONG *dst, const BN_ULONG *src, int top) # endif # endif #else -#define bn_cp_64(to, n, from, m) \ - { \ - bn_cp_32(to, (n)*2, from, (m)*2); \ - bn_cp_32(to, (n)*2+1, from, (m)*2+1); \ - } -#define bn_64_set_0(to, n) \ - { \ - bn_32_set_0(to, (n)*2); \ - bn_32_set_0(to, (n)*2+1); \ - } -#define bn_cp_32(to, n, from, m) (to)[n] = (m>=0)?((from)[m]):0; -#define bn_32_set_0(to, n) (to)[n] = (BN_ULONG)0; +# define bn_cp_64(to, n, from, m) \ + { \ + bn_cp_32(to, (n)*2, from, (m)*2); \ + bn_cp_32(to, (n)*2+1, from, (m)*2+1); \ + } +# define bn_64_set_0(to, n) \ + { \ + bn_32_set_0(to, (n)*2); \ + bn_32_set_0(to, (n)*2+1); \ + } +# define bn_cp_32(to, n, from, m) (to)[n] = (m>=0)?((from)[m]):0; +# define bn_32_set_0(to, n) (to)[n] = (BN_ULONG)0; # if defined(_WIN32) && !defined(__GNUC__) # define NIST_INT64 __int64 # elif defined(BN_LLONG) # define NIST_INT64 long long # endif -#endif /* BN_BITS2 != 64 */ +#endif /* BN_BITS2 != 64 */ #define nist_set_192(to, from, a1, a2, a3) \ - { \ - bn_cp_64(to, 0, from, (a3) - 3) \ - bn_cp_64(to, 1, from, (a2) - 3) \ - bn_cp_64(to, 2, from, (a1) - 3) \ - } + { \ + bn_cp_64(to, 0, from, (a3) - 3) \ + bn_cp_64(to, 1, from, (a2) - 3) \ + bn_cp_64(to, 2, from, (a1) - 3) \ + } int BN_nist_mod_192(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, - BN_CTX *ctx) - { - int top = a->top, i; - int carry; - register BN_ULONG *r_d, *a_d = a->d; - union { - BN_ULONG bn[BN_NIST_192_TOP]; - unsigned int ui[BN_NIST_192_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)]; - } buf; - BN_ULONG c_d[BN_NIST_192_TOP], - *res; - PTR_SIZE_INT mask; - static const BIGNUM _bignum_nist_p_192_sqr = { - (BN_ULONG *)_nist_p_192_sqr, - sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]), - sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]), - 0,BN_FLG_STATIC_DATA }; - - field = &_bignum_nist_p_192; /* just to make sure */ - - if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_192_sqr)>=0) - return BN_nnmod(r, a, field, ctx); - - i = BN_ucmp(field, a); - if (i == 0) - { - BN_zero(r); - return 1; - } - else if (i > 0) - return (r == a) ? 1 : (BN_copy(r ,a) != NULL); - - if (r != a) - { - if (!bn_wexpand(r, BN_NIST_192_TOP)) - return 0; - r_d = r->d; - nist_cp_bn(r_d, a_d, BN_NIST_192_TOP); - } - else - r_d = a_d; - - nist_cp_bn_0(buf.bn, a_d + BN_NIST_192_TOP, top - BN_NIST_192_TOP, BN_NIST_192_TOP); + BN_CTX *ctx) +{ + int top = a->top, i; + int carry; + register BN_ULONG *r_d, *a_d = a->d; + union { + BN_ULONG bn[BN_NIST_192_TOP]; + unsigned int ui[BN_NIST_192_TOP * sizeof(BN_ULONG) / + sizeof(unsigned int)]; + } buf; + BN_ULONG c_d[BN_NIST_192_TOP], *res; + PTR_SIZE_INT mask; + static const BIGNUM _bignum_nist_p_192_sqr = { + (BN_ULONG *)_nist_p_192_sqr, + sizeof(_nist_p_192_sqr) / sizeof(_nist_p_192_sqr[0]), + sizeof(_nist_p_192_sqr) / sizeof(_nist_p_192_sqr[0]), + 0, BN_FLG_STATIC_DATA + }; + + field = &_bignum_nist_p_192; /* just to make sure */ + + if (BN_is_negative(a) || BN_ucmp(a, &_bignum_nist_p_192_sqr) >= 0) + return BN_nnmod(r, a, field, ctx); + + i = BN_ucmp(field, a); + if (i == 0) { + BN_zero(r); + return 1; + } else if (i > 0) + return (r == a) ? 1 : (BN_copy(r, a) != NULL); + + if (r != a) { + if (!bn_wexpand(r, BN_NIST_192_TOP)) + return 0; + r_d = r->d; + nist_cp_bn(r_d, a_d, BN_NIST_192_TOP); + } else + r_d = a_d; + + nist_cp_bn_0(buf.bn, a_d + BN_NIST_192_TOP, top - BN_NIST_192_TOP, + BN_NIST_192_TOP); #if defined(NIST_INT64) - { - NIST_INT64 acc; /* accumulator */ - unsigned int *rp=(unsigned int *)r_d; - const unsigned int *bp=(const unsigned int *)buf.ui; - - acc = rp[0]; acc += bp[3*2-6]; - acc += bp[5*2-6]; rp[0] = (unsigned int)acc; acc >>= 32; - - acc += rp[1]; acc += bp[3*2-5]; - acc += bp[5*2-5]; rp[1] = (unsigned int)acc; acc >>= 32; - - acc += rp[2]; acc += bp[3*2-6]; - acc += bp[4*2-6]; - acc += bp[5*2-6]; rp[2] = (unsigned int)acc; acc >>= 32; - - acc += rp[3]; acc += bp[3*2-5]; - acc += bp[4*2-5]; - acc += bp[5*2-5]; rp[3] = (unsigned int)acc; acc >>= 32; - - acc += rp[4]; acc += bp[4*2-6]; - acc += bp[5*2-6]; rp[4] = (unsigned int)acc; acc >>= 32; - - acc += rp[5]; acc += bp[4*2-5]; - acc += bp[5*2-5]; rp[5] = (unsigned int)acc; - - carry = (int)(acc>>32); - } + { + NIST_INT64 acc; /* accumulator */ + unsigned int *rp = (unsigned int *)r_d; + const unsigned int *bp = (const unsigned int *)buf.ui; + + acc = rp[0]; + acc += bp[3 * 2 - 6]; + acc += bp[5 * 2 - 6]; + rp[0] = (unsigned int)acc; + acc >>= 32; + + acc += rp[1]; + acc += bp[3 * 2 - 5]; + acc += bp[5 * 2 - 5]; + rp[1] = (unsigned int)acc; + acc >>= 32; + + acc += rp[2]; + acc += bp[3 * 2 - 6]; + acc += bp[4 * 2 - 6]; + acc += bp[5 * 2 - 6]; + rp[2] = (unsigned int)acc; + acc >>= 32; + + acc += rp[3]; + acc += bp[3 * 2 - 5]; + acc += bp[4 * 2 - 5]; + acc += bp[5 * 2 - 5]; + rp[3] = (unsigned int)acc; + acc >>= 32; + + acc += rp[4]; + acc += bp[4 * 2 - 6]; + acc += bp[5 * 2 - 6]; + rp[4] = (unsigned int)acc; + acc >>= 32; + + acc += rp[5]; + acc += bp[4 * 2 - 5]; + acc += bp[5 * 2 - 5]; + rp[5] = (unsigned int)acc; + + carry = (int)(acc >> 32); + } #else - { - BN_ULONG t_d[BN_NIST_192_TOP]; - - nist_set_192(t_d, buf.bn, 0, 3, 3); - carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP); - nist_set_192(t_d, buf.bn, 4, 4, 0); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP); - nist_set_192(t_d, buf.bn, 5, 5, 5) - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP); - } + { + BN_ULONG t_d[BN_NIST_192_TOP]; + + nist_set_192(t_d, buf.bn, 0, 3, 3); + carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP); + nist_set_192(t_d, buf.bn, 4, 4, 0); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP); + nist_set_192(t_d, buf.bn, 5, 5, 5) + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP); + } #endif - if (carry > 0) - carry = (int)bn_sub_words(r_d,r_d,_nist_p_192[carry-1],BN_NIST_192_TOP); - else - carry = 1; - - /* - * we need 'if (carry==0 || result>=modulus) result-=modulus;' - * as comparison implies subtraction, we can write - * 'tmp=result-modulus; if (!carry || !borrow) result=tmp;' - * this is what happens below, but without explicit if:-) a. - */ - mask = 0-(PTR_SIZE_INT)bn_sub_words(c_d,r_d,_nist_p_192[0],BN_NIST_192_TOP); - mask &= 0-(PTR_SIZE_INT)carry; - res = c_d; - res = (BN_ULONG *) - (((PTR_SIZE_INT)res&~mask) | ((PTR_SIZE_INT)r_d&mask)); - nist_cp_bn(r_d, res, BN_NIST_192_TOP); - r->top = BN_NIST_192_TOP; - bn_correct_top(r); - - return 1; - } - -typedef BN_ULONG (*bn_addsub_f)(BN_ULONG *,const BN_ULONG *,const BN_ULONG *,int); + if (carry > 0) + carry = + (int)bn_sub_words(r_d, r_d, _nist_p_192[carry - 1], + BN_NIST_192_TOP); + else + carry = 1; + + /* + * we need 'if (carry==0 || result>=modulus) result-=modulus;' + * as comparison implies subtraction, we can write + * 'tmp=result-modulus; if (!carry || !borrow) result=tmp;' + * this is what happens below, but without explicit if:-) a. + */ + mask = + 0 - (PTR_SIZE_INT) bn_sub_words(c_d, r_d, _nist_p_192[0], + BN_NIST_192_TOP); + mask &= 0 - (PTR_SIZE_INT) carry; + res = c_d; + res = (BN_ULONG *) + (((PTR_SIZE_INT) res & ~mask) | ((PTR_SIZE_INT) r_d & mask)); + nist_cp_bn(r_d, res, BN_NIST_192_TOP); + r->top = BN_NIST_192_TOP; + bn_correct_top(r); + + return 1; +} + +typedef BN_ULONG (*bn_addsub_f) (BN_ULONG *, const BN_ULONG *, + const BN_ULONG *, int); #define nist_set_224(to, from, a1, a2, a3, a4, a5, a6, a7) \ - { \ - bn_cp_32(to, 0, from, (a7) - 7) \ - bn_cp_32(to, 1, from, (a6) - 7) \ - bn_cp_32(to, 2, from, (a5) - 7) \ - bn_cp_32(to, 3, from, (a4) - 7) \ - bn_cp_32(to, 4, from, (a3) - 7) \ - bn_cp_32(to, 5, from, (a2) - 7) \ - bn_cp_32(to, 6, from, (a1) - 7) \ - } + { \ + bn_cp_32(to, 0, from, (a7) - 7) \ + bn_cp_32(to, 1, from, (a6) - 7) \ + bn_cp_32(to, 2, from, (a5) - 7) \ + bn_cp_32(to, 3, from, (a4) - 7) \ + bn_cp_32(to, 4, from, (a3) - 7) \ + bn_cp_32(to, 5, from, (a2) - 7) \ + bn_cp_32(to, 6, from, (a1) - 7) \ + } int BN_nist_mod_224(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, - BN_CTX *ctx) - { - int top = a->top, i; - int carry; - BN_ULONG *r_d, *a_d = a->d; - union { - BN_ULONG bn[BN_NIST_224_TOP]; - unsigned int ui[BN_NIST_224_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)]; - } buf; - BN_ULONG c_d[BN_NIST_224_TOP], - *res; - PTR_SIZE_INT mask; - union { bn_addsub_f f; PTR_SIZE_INT p; } u; - static const BIGNUM _bignum_nist_p_224_sqr = { - (BN_ULONG *)_nist_p_224_sqr, - sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]), - sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]), - 0,BN_FLG_STATIC_DATA }; - - - field = &_bignum_nist_p_224; /* just to make sure */ - - if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_224_sqr)>=0) - return BN_nnmod(r, a, field, ctx); - - i = BN_ucmp(field, a); - if (i == 0) - { - BN_zero(r); - return 1; - } - else if (i > 0) - return (r == a)? 1 : (BN_copy(r ,a) != NULL); - - if (r != a) - { - if (!bn_wexpand(r, BN_NIST_224_TOP)) - return 0; - r_d = r->d; - nist_cp_bn(r_d, a_d, BN_NIST_224_TOP); - } - else - r_d = a_d; + BN_CTX *ctx) +{ + int top = a->top, i; + int carry; + BN_ULONG *r_d, *a_d = a->d; + union { + BN_ULONG bn[BN_NIST_224_TOP]; + unsigned int ui[BN_NIST_224_TOP * sizeof(BN_ULONG) / + sizeof(unsigned int)]; + } buf; + BN_ULONG c_d[BN_NIST_224_TOP], *res; + PTR_SIZE_INT mask; + union { + bn_addsub_f f; + PTR_SIZE_INT p; + } u; + static const BIGNUM _bignum_nist_p_224_sqr = { + (BN_ULONG *)_nist_p_224_sqr, + sizeof(_nist_p_224_sqr) / sizeof(_nist_p_224_sqr[0]), + sizeof(_nist_p_224_sqr) / sizeof(_nist_p_224_sqr[0]), + 0, BN_FLG_STATIC_DATA + }; + + field = &_bignum_nist_p_224; /* just to make sure */ + + if (BN_is_negative(a) || BN_ucmp(a, &_bignum_nist_p_224_sqr) >= 0) + return BN_nnmod(r, a, field, ctx); + + i = BN_ucmp(field, a); + if (i == 0) { + BN_zero(r); + return 1; + } else if (i > 0) + return (r == a) ? 1 : (BN_copy(r, a) != NULL); + + if (r != a) { + if (!bn_wexpand(r, BN_NIST_224_TOP)) + return 0; + r_d = r->d; + nist_cp_bn(r_d, a_d, BN_NIST_224_TOP); + } else + r_d = a_d; #if BN_BITS2==64 - /* copy upper 256 bits of 448 bit number ... */ - nist_cp_bn_0(c_d, a_d + (BN_NIST_224_TOP-1), top - (BN_NIST_224_TOP-1), BN_NIST_224_TOP); - /* ... and right shift by 32 to obtain upper 224 bits */ - nist_set_224(buf.bn, c_d, 14, 13, 12, 11, 10, 9, 8); - /* truncate lower part to 224 bits too */ - r_d[BN_NIST_224_TOP-1] &= BN_MASK2l; + /* copy upper 256 bits of 448 bit number ... */ + nist_cp_bn_0(c_d, a_d + (BN_NIST_224_TOP - 1), + top - (BN_NIST_224_TOP - 1), BN_NIST_224_TOP); + /* ... and right shift by 32 to obtain upper 224 bits */ + nist_set_224(buf.bn, c_d, 14, 13, 12, 11, 10, 9, 8); + /* truncate lower part to 224 bits too */ + r_d[BN_NIST_224_TOP - 1] &= BN_MASK2l; #else - nist_cp_bn_0(buf.bn, a_d + BN_NIST_224_TOP, top - BN_NIST_224_TOP, BN_NIST_224_TOP); + nist_cp_bn_0(buf.bn, a_d + BN_NIST_224_TOP, top - BN_NIST_224_TOP, + BN_NIST_224_TOP); #endif #if defined(NIST_INT64) && BN_BITS2!=64 - { - NIST_INT64 acc; /* accumulator */ - unsigned int *rp=(unsigned int *)r_d; - const unsigned int *bp=(const unsigned int *)buf.ui; - - acc = rp[0]; acc -= bp[7-7]; - acc -= bp[11-7]; rp[0] = (unsigned int)acc; acc >>= 32; - - acc += rp[1]; acc -= bp[8-7]; - acc -= bp[12-7]; rp[1] = (unsigned int)acc; acc >>= 32; - - acc += rp[2]; acc -= bp[9-7]; - acc -= bp[13-7]; rp[2] = (unsigned int)acc; acc >>= 32; - - acc += rp[3]; acc += bp[7-7]; - acc += bp[11-7]; - acc -= bp[10-7]; rp[3] = (unsigned int)acc; acc>>= 32; - - acc += rp[4]; acc += bp[8-7]; - acc += bp[12-7]; - acc -= bp[11-7]; rp[4] = (unsigned int)acc; acc >>= 32; - - acc += rp[5]; acc += bp[9-7]; - acc += bp[13-7]; - acc -= bp[12-7]; rp[5] = (unsigned int)acc; acc >>= 32; - - acc += rp[6]; acc += bp[10-7]; - acc -= bp[13-7]; rp[6] = (unsigned int)acc; - - carry = (int)(acc>>32); + { + NIST_INT64 acc; /* accumulator */ + unsigned int *rp = (unsigned int *)r_d; + const unsigned int *bp = (const unsigned int *)buf.ui; + + acc = rp[0]; + acc -= bp[7 - 7]; + acc -= bp[11 - 7]; + rp[0] = (unsigned int)acc; + acc >>= 32; + + acc += rp[1]; + acc -= bp[8 - 7]; + acc -= bp[12 - 7]; + rp[1] = (unsigned int)acc; + acc >>= 32; + + acc += rp[2]; + acc -= bp[9 - 7]; + acc -= bp[13 - 7]; + rp[2] = (unsigned int)acc; + acc >>= 32; + + acc += rp[3]; + acc += bp[7 - 7]; + acc += bp[11 - 7]; + acc -= bp[10 - 7]; + rp[3] = (unsigned int)acc; + acc >>= 32; + + acc += rp[4]; + acc += bp[8 - 7]; + acc += bp[12 - 7]; + acc -= bp[11 - 7]; + rp[4] = (unsigned int)acc; + acc >>= 32; + + acc += rp[5]; + acc += bp[9 - 7]; + acc += bp[13 - 7]; + acc -= bp[12 - 7]; + rp[5] = (unsigned int)acc; + acc >>= 32; + + acc += rp[6]; + acc += bp[10 - 7]; + acc -= bp[13 - 7]; + rp[6] = (unsigned int)acc; + + carry = (int)(acc >> 32); # if BN_BITS2==64 - rp[7] = carry; + rp[7] = carry; # endif - } + } #else - { - BN_ULONG t_d[BN_NIST_224_TOP]; - - nist_set_224(t_d, buf.bn, 10, 9, 8, 7, 0, 0, 0); - carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP); - nist_set_224(t_d, buf.bn, 0, 13, 12, 11, 0, 0, 0); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP); - nist_set_224(t_d, buf.bn, 13, 12, 11, 10, 9, 8, 7); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP); - nist_set_224(t_d, buf.bn, 0, 0, 0, 0, 13, 12, 11); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP); + { + BN_ULONG t_d[BN_NIST_224_TOP]; + + nist_set_224(t_d, buf.bn, 10, 9, 8, 7, 0, 0, 0); + carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP); + nist_set_224(t_d, buf.bn, 0, 13, 12, 11, 0, 0, 0); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP); + nist_set_224(t_d, buf.bn, 13, 12, 11, 10, 9, 8, 7); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP); + nist_set_224(t_d, buf.bn, 0, 0, 0, 0, 13, 12, 11); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP); -#if BN_BITS2==64 - carry = (int)(r_d[BN_NIST_224_TOP-1]>>32); -#endif - } +# if BN_BITS2==64 + carry = (int)(r_d[BN_NIST_224_TOP - 1] >> 32); +# endif + } #endif - u.f = bn_sub_words; - if (carry > 0) - { - carry = (int)bn_sub_words(r_d,r_d,_nist_p_224[carry-1],BN_NIST_224_TOP); + u.f = bn_sub_words; + if (carry > 0) { + carry = + (int)bn_sub_words(r_d, r_d, _nist_p_224[carry - 1], + BN_NIST_224_TOP); #if BN_BITS2==64 - carry=(int)(~(r_d[BN_NIST_224_TOP-1]>>32))&1; + carry = (int)(~(r_d[BN_NIST_224_TOP - 1] >> 32)) & 1; #endif - } - else if (carry < 0) - { - /* it's a bit more comlicated logic in this case. - * if bn_add_words yields no carry, then result - * has to be adjusted by unconditionally *adding* - * the modulus. but if it does, then result has - * to be compared to the modulus and conditionally - * adjusted by *subtracting* the latter. */ - carry = (int)bn_add_words(r_d,r_d,_nist_p_224[-carry-1],BN_NIST_224_TOP); - mask = 0-(PTR_SIZE_INT)carry; - u.p = ((PTR_SIZE_INT)bn_sub_words&mask) | - ((PTR_SIZE_INT)bn_add_words&~mask); - } - else - carry = 1; - - /* otherwise it's effectively same as in BN_nist_mod_192... */ - mask = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_224[0],BN_NIST_224_TOP); - mask &= 0-(PTR_SIZE_INT)carry; - res = c_d; - res = (BN_ULONG *)(((PTR_SIZE_INT)res&~mask) | - ((PTR_SIZE_INT)r_d&mask)); - nist_cp_bn(r_d, res, BN_NIST_224_TOP); - r->top = BN_NIST_224_TOP; - bn_correct_top(r); - - return 1; - } + } else if (carry < 0) { + /* + * it's a bit more comlicated logic in this case. if bn_add_words + * yields no carry, then result has to be adjusted by unconditionally + * *adding* the modulus. but if it does, then result has to be + * compared to the modulus and conditionally adjusted by + * *subtracting* the latter. + */ + carry = + (int)bn_add_words(r_d, r_d, _nist_p_224[-carry - 1], + BN_NIST_224_TOP); + mask = 0 - (PTR_SIZE_INT) carry; + u.p = ((PTR_SIZE_INT) bn_sub_words & mask) | + ((PTR_SIZE_INT) bn_add_words & ~mask); + } else + carry = 1; + + /* otherwise it's effectively same as in BN_nist_mod_192... */ + mask = + 0 - (PTR_SIZE_INT) (*u.f) (c_d, r_d, _nist_p_224[0], BN_NIST_224_TOP); + mask &= 0 - (PTR_SIZE_INT) carry; + res = c_d; + res = (BN_ULONG *)(((PTR_SIZE_INT) res & ~mask) | + ((PTR_SIZE_INT) r_d & mask)); + nist_cp_bn(r_d, res, BN_NIST_224_TOP); + r->top = BN_NIST_224_TOP; + bn_correct_top(r); + + return 1; +} #define nist_set_256(to, from, a1, a2, a3, a4, a5, a6, a7, a8) \ - { \ - bn_cp_32(to, 0, from, (a8) - 8) \ - bn_cp_32(to, 1, from, (a7) - 8) \ - bn_cp_32(to, 2, from, (a6) - 8) \ - bn_cp_32(to, 3, from, (a5) - 8) \ - bn_cp_32(to, 4, from, (a4) - 8) \ - bn_cp_32(to, 5, from, (a3) - 8) \ - bn_cp_32(to, 6, from, (a2) - 8) \ - bn_cp_32(to, 7, from, (a1) - 8) \ - } + { \ + bn_cp_32(to, 0, from, (a8) - 8) \ + bn_cp_32(to, 1, from, (a7) - 8) \ + bn_cp_32(to, 2, from, (a6) - 8) \ + bn_cp_32(to, 3, from, (a5) - 8) \ + bn_cp_32(to, 4, from, (a4) - 8) \ + bn_cp_32(to, 5, from, (a3) - 8) \ + bn_cp_32(to, 6, from, (a2) - 8) \ + bn_cp_32(to, 7, from, (a1) - 8) \ + } int BN_nist_mod_256(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, - BN_CTX *ctx) - { - int i, top = a->top; - int carry = 0; - register BN_ULONG *a_d = a->d, *r_d; - union { - BN_ULONG bn[BN_NIST_256_TOP]; - unsigned int ui[BN_NIST_256_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)]; - } buf; - BN_ULONG c_d[BN_NIST_256_TOP], - *res; - PTR_SIZE_INT mask; - union { bn_addsub_f f; PTR_SIZE_INT p; } u; - static const BIGNUM _bignum_nist_p_256_sqr = { - (BN_ULONG *)_nist_p_256_sqr, - sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]), - sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]), - 0,BN_FLG_STATIC_DATA }; - - field = &_bignum_nist_p_256; /* just to make sure */ - - if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_256_sqr)>=0) - return BN_nnmod(r, a, field, ctx); - - i = BN_ucmp(field, a); - if (i == 0) - { - BN_zero(r); - return 1; - } - else if (i > 0) - return (r == a)? 1 : (BN_copy(r ,a) != NULL); - - if (r != a) - { - if (!bn_wexpand(r, BN_NIST_256_TOP)) - return 0; - r_d = r->d; - nist_cp_bn(r_d, a_d, BN_NIST_256_TOP); - } - else - r_d = a_d; - - nist_cp_bn_0(buf.bn, a_d + BN_NIST_256_TOP, top - BN_NIST_256_TOP, BN_NIST_256_TOP); + BN_CTX *ctx) +{ + int i, top = a->top; + int carry = 0; + register BN_ULONG *a_d = a->d, *r_d; + union { + BN_ULONG bn[BN_NIST_256_TOP]; + unsigned int ui[BN_NIST_256_TOP * sizeof(BN_ULONG) / + sizeof(unsigned int)]; + } buf; + BN_ULONG c_d[BN_NIST_256_TOP], *res; + PTR_SIZE_INT mask; + union { + bn_addsub_f f; + PTR_SIZE_INT p; + } u; + static const BIGNUM _bignum_nist_p_256_sqr = { + (BN_ULONG *)_nist_p_256_sqr, + sizeof(_nist_p_256_sqr) / sizeof(_nist_p_256_sqr[0]), + sizeof(_nist_p_256_sqr) / sizeof(_nist_p_256_sqr[0]), + 0, BN_FLG_STATIC_DATA + }; + + field = &_bignum_nist_p_256; /* just to make sure */ + + if (BN_is_negative(a) || BN_ucmp(a, &_bignum_nist_p_256_sqr) >= 0) + return BN_nnmod(r, a, field, ctx); + + i = BN_ucmp(field, a); + if (i == 0) { + BN_zero(r); + return 1; + } else if (i > 0) + return (r == a) ? 1 : (BN_copy(r, a) != NULL); + + if (r != a) { + if (!bn_wexpand(r, BN_NIST_256_TOP)) + return 0; + r_d = r->d; + nist_cp_bn(r_d, a_d, BN_NIST_256_TOP); + } else + r_d = a_d; + + nist_cp_bn_0(buf.bn, a_d + BN_NIST_256_TOP, top - BN_NIST_256_TOP, + BN_NIST_256_TOP); #if defined(NIST_INT64) - { - NIST_INT64 acc; /* accumulator */ - unsigned int *rp=(unsigned int *)r_d; - const unsigned int *bp=(const unsigned int *)buf.ui; - - acc = rp[0]; acc += bp[8-8]; - acc += bp[9-8]; - acc -= bp[11-8]; - acc -= bp[12-8]; - acc -= bp[13-8]; - acc -= bp[14-8]; rp[0] = (unsigned int)acc; acc >>= 32; - - acc += rp[1]; acc += bp[9-8]; - acc += bp[10-8]; - acc -= bp[12-8]; - acc -= bp[13-8]; - acc -= bp[14-8]; - acc -= bp[15-8]; rp[1] = (unsigned int)acc; acc >>= 32; - - acc += rp[2]; acc += bp[10-8]; - acc += bp[11-8]; - acc -= bp[13-8]; - acc -= bp[14-8]; - acc -= bp[15-8]; rp[2] = (unsigned int)acc; acc >>= 32; - - acc += rp[3]; acc += bp[11-8]; - acc += bp[11-8]; - acc += bp[12-8]; - acc += bp[12-8]; - acc += bp[13-8]; - acc -= bp[15-8]; - acc -= bp[8-8]; - acc -= bp[9-8]; rp[3] = (unsigned int)acc; acc >>= 32; - - acc += rp[4]; acc += bp[12-8]; - acc += bp[12-8]; - acc += bp[13-8]; - acc += bp[13-8]; - acc += bp[14-8]; - acc -= bp[9-8]; - acc -= bp[10-8]; rp[4] = (unsigned int)acc; acc >>= 32; - - acc += rp[5]; acc += bp[13-8]; - acc += bp[13-8]; - acc += bp[14-8]; - acc += bp[14-8]; - acc += bp[15-8]; - acc -= bp[10-8]; - acc -= bp[11-8]; rp[5] = (unsigned int)acc; acc >>= 32; - - acc += rp[6]; acc += bp[14-8]; - acc += bp[14-8]; - acc += bp[15-8]; - acc += bp[15-8]; - acc += bp[14-8]; - acc += bp[13-8]; - acc -= bp[8-8]; - acc -= bp[9-8]; rp[6] = (unsigned int)acc; acc >>= 32; - - acc += rp[7]; acc += bp[15-8]; - acc += bp[15-8]; - acc += bp[15-8]; - acc += bp[8 -8]; - acc -= bp[10-8]; - acc -= bp[11-8]; - acc -= bp[12-8]; - acc -= bp[13-8]; rp[7] = (unsigned int)acc; - - carry = (int)(acc>>32); - } + { + NIST_INT64 acc; /* accumulator */ + unsigned int *rp = (unsigned int *)r_d; + const unsigned int *bp = (const unsigned int *)buf.ui; + + acc = rp[0]; + acc += bp[8 - 8]; + acc += bp[9 - 8]; + acc -= bp[11 - 8]; + acc -= bp[12 - 8]; + acc -= bp[13 - 8]; + acc -= bp[14 - 8]; + rp[0] = (unsigned int)acc; + acc >>= 32; + + acc += rp[1]; + acc += bp[9 - 8]; + acc += bp[10 - 8]; + acc -= bp[12 - 8]; + acc -= bp[13 - 8]; + acc -= bp[14 - 8]; + acc -= bp[15 - 8]; + rp[1] = (unsigned int)acc; + acc >>= 32; + + acc += rp[2]; + acc += bp[10 - 8]; + acc += bp[11 - 8]; + acc -= bp[13 - 8]; + acc -= bp[14 - 8]; + acc -= bp[15 - 8]; + rp[2] = (unsigned int)acc; + acc >>= 32; + + acc += rp[3]; + acc += bp[11 - 8]; + acc += bp[11 - 8]; + acc += bp[12 - 8]; + acc += bp[12 - 8]; + acc += bp[13 - 8]; + acc -= bp[15 - 8]; + acc -= bp[8 - 8]; + acc -= bp[9 - 8]; + rp[3] = (unsigned int)acc; + acc >>= 32; + + acc += rp[4]; + acc += bp[12 - 8]; + acc += bp[12 - 8]; + acc += bp[13 - 8]; + acc += bp[13 - 8]; + acc += bp[14 - 8]; + acc -= bp[9 - 8]; + acc -= bp[10 - 8]; + rp[4] = (unsigned int)acc; + acc >>= 32; + + acc += rp[5]; + acc += bp[13 - 8]; + acc += bp[13 - 8]; + acc += bp[14 - 8]; + acc += bp[14 - 8]; + acc += bp[15 - 8]; + acc -= bp[10 - 8]; + acc -= bp[11 - 8]; + rp[5] = (unsigned int)acc; + acc >>= 32; + + acc += rp[6]; + acc += bp[14 - 8]; + acc += bp[14 - 8]; + acc += bp[15 - 8]; + acc += bp[15 - 8]; + acc += bp[14 - 8]; + acc += bp[13 - 8]; + acc -= bp[8 - 8]; + acc -= bp[9 - 8]; + rp[6] = (unsigned int)acc; + acc >>= 32; + + acc += rp[7]; + acc += bp[15 - 8]; + acc += bp[15 - 8]; + acc += bp[15 - 8]; + acc += bp[8 - 8]; + acc -= bp[10 - 8]; + acc -= bp[11 - 8]; + acc -= bp[12 - 8]; + acc -= bp[13 - 8]; + rp[7] = (unsigned int)acc; + + carry = (int)(acc >> 32); + } #else - { - BN_ULONG t_d[BN_NIST_256_TOP]; - - /*S1*/ - nist_set_256(t_d, buf.bn, 15, 14, 13, 12, 11, 0, 0, 0); - /*S2*/ - nist_set_256(c_d, buf.bn, 0, 15, 14, 13, 12, 0, 0, 0); - carry = (int)bn_add_words(t_d, t_d, c_d, BN_NIST_256_TOP); - /* left shift */ - { - register BN_ULONG *ap,t,c; - ap = t_d; - c=0; - for (i = BN_NIST_256_TOP; i != 0; --i) - { - t= *ap; - *(ap++)=((t<<1)|c)&BN_MASK2; - c=(t & BN_TBIT)?1:0; - } - carry <<= 1; - carry |= c; - } - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP); - /*S3*/ - nist_set_256(t_d, buf.bn, 15, 14, 0, 0, 0, 10, 9, 8); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP); - /*S4*/ - nist_set_256(t_d, buf.bn, 8, 13, 15, 14, 13, 11, 10, 9); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP); - /*D1*/ - nist_set_256(t_d, buf.bn, 10, 8, 0, 0, 0, 13, 12, 11); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); - /*D2*/ - nist_set_256(t_d, buf.bn, 11, 9, 0, 0, 15, 14, 13, 12); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); - /*D3*/ - nist_set_256(t_d, buf.bn, 12, 0, 10, 9, 8, 15, 14, 13); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); - /*D4*/ - nist_set_256(t_d, buf.bn, 13, 0, 11, 10, 9, 0, 15, 14); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); - - } + { + BN_ULONG t_d[BN_NIST_256_TOP]; + + /* + * S1 + */ + nist_set_256(t_d, buf.bn, 15, 14, 13, 12, 11, 0, 0, 0); + /* + * S2 + */ + nist_set_256(c_d, buf.bn, 0, 15, 14, 13, 12, 0, 0, 0); + carry = (int)bn_add_words(t_d, t_d, c_d, BN_NIST_256_TOP); + /* left shift */ + { + register BN_ULONG *ap, t, c; + ap = t_d; + c = 0; + for (i = BN_NIST_256_TOP; i != 0; --i) { + t = *ap; + *(ap++) = ((t << 1) | c) & BN_MASK2; + c = (t & BN_TBIT) ? 1 : 0; + } + carry <<= 1; + carry |= c; + } + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP); + /* + * S3 + */ + nist_set_256(t_d, buf.bn, 15, 14, 0, 0, 0, 10, 9, 8); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP); + /* + * S4 + */ + nist_set_256(t_d, buf.bn, 8, 13, 15, 14, 13, 11, 10, 9); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP); + /* + * D1 + */ + nist_set_256(t_d, buf.bn, 10, 8, 0, 0, 0, 13, 12, 11); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); + /* + * D2 + */ + nist_set_256(t_d, buf.bn, 11, 9, 0, 0, 15, 14, 13, 12); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); + /* + * D3 + */ + nist_set_256(t_d, buf.bn, 12, 0, 10, 9, 8, 15, 14, 13); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); + /* + * D4 + */ + nist_set_256(t_d, buf.bn, 13, 0, 11, 10, 9, 0, 15, 14); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP); + + } #endif - /* see BN_nist_mod_224 for explanation */ - u.f = bn_sub_words; - if (carry > 0) - carry = (int)bn_sub_words(r_d,r_d,_nist_p_256[carry-1],BN_NIST_256_TOP); - else if (carry < 0) - { - carry = (int)bn_add_words(r_d,r_d,_nist_p_256[-carry-1],BN_NIST_256_TOP); - mask = 0-(PTR_SIZE_INT)carry; - u.p = ((PTR_SIZE_INT)bn_sub_words&mask) | - ((PTR_SIZE_INT)bn_add_words&~mask); - } - else - carry = 1; - - mask = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_256[0],BN_NIST_256_TOP); - mask &= 0-(PTR_SIZE_INT)carry; - res = c_d; - res = (BN_ULONG *)(((PTR_SIZE_INT)res&~mask) | - ((PTR_SIZE_INT)r_d&mask)); - nist_cp_bn(r_d, res, BN_NIST_256_TOP); - r->top = BN_NIST_256_TOP; - bn_correct_top(r); - - return 1; - } + /* see BN_nist_mod_224 for explanation */ + u.f = bn_sub_words; + if (carry > 0) + carry = + (int)bn_sub_words(r_d, r_d, _nist_p_256[carry - 1], + BN_NIST_256_TOP); + else if (carry < 0) { + carry = + (int)bn_add_words(r_d, r_d, _nist_p_256[-carry - 1], + BN_NIST_256_TOP); + mask = 0 - (PTR_SIZE_INT) carry; + u.p = ((PTR_SIZE_INT) bn_sub_words & mask) | + ((PTR_SIZE_INT) bn_add_words & ~mask); + } else + carry = 1; + + mask = + 0 - (PTR_SIZE_INT) (*u.f) (c_d, r_d, _nist_p_256[0], BN_NIST_256_TOP); + mask &= 0 - (PTR_SIZE_INT) carry; + res = c_d; + res = (BN_ULONG *)(((PTR_SIZE_INT) res & ~mask) | + ((PTR_SIZE_INT) r_d & mask)); + nist_cp_bn(r_d, res, BN_NIST_256_TOP); + r->top = BN_NIST_256_TOP; + bn_correct_top(r); + + return 1; +} #define nist_set_384(to,from,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12) \ - { \ - bn_cp_32(to, 0, from, (a12) - 12) \ - bn_cp_32(to, 1, from, (a11) - 12) \ - bn_cp_32(to, 2, from, (a10) - 12) \ - bn_cp_32(to, 3, from, (a9) - 12) \ - bn_cp_32(to, 4, from, (a8) - 12) \ - bn_cp_32(to, 5, from, (a7) - 12) \ - bn_cp_32(to, 6, from, (a6) - 12) \ - bn_cp_32(to, 7, from, (a5) - 12) \ - bn_cp_32(to, 8, from, (a4) - 12) \ - bn_cp_32(to, 9, from, (a3) - 12) \ - bn_cp_32(to, 10, from, (a2) - 12) \ - bn_cp_32(to, 11, from, (a1) - 12) \ - } + { \ + bn_cp_32(to, 0, from, (a12) - 12) \ + bn_cp_32(to, 1, from, (a11) - 12) \ + bn_cp_32(to, 2, from, (a10) - 12) \ + bn_cp_32(to, 3, from, (a9) - 12) \ + bn_cp_32(to, 4, from, (a8) - 12) \ + bn_cp_32(to, 5, from, (a7) - 12) \ + bn_cp_32(to, 6, from, (a6) - 12) \ + bn_cp_32(to, 7, from, (a5) - 12) \ + bn_cp_32(to, 8, from, (a4) - 12) \ + bn_cp_32(to, 9, from, (a3) - 12) \ + bn_cp_32(to, 10, from, (a2) - 12) \ + bn_cp_32(to, 11, from, (a1) - 12) \ + } int BN_nist_mod_384(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, - BN_CTX *ctx) - { - int i, top = a->top; - int carry = 0; - register BN_ULONG *r_d, *a_d = a->d; - union { - BN_ULONG bn[BN_NIST_384_TOP]; - unsigned int ui[BN_NIST_384_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)]; - } buf; - BN_ULONG c_d[BN_NIST_384_TOP], - *res; - PTR_SIZE_INT mask; - union { bn_addsub_f f; PTR_SIZE_INT p; } u; - static const BIGNUM _bignum_nist_p_384_sqr = { - (BN_ULONG *)_nist_p_384_sqr, - sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]), - sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]), - 0,BN_FLG_STATIC_DATA }; - - - field = &_bignum_nist_p_384; /* just to make sure */ - - if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_384_sqr)>=0) - return BN_nnmod(r, a, field, ctx); - - i = BN_ucmp(field, a); - if (i == 0) - { - BN_zero(r); - return 1; - } - else if (i > 0) - return (r == a)? 1 : (BN_copy(r ,a) != NULL); - - if (r != a) - { - if (!bn_wexpand(r, BN_NIST_384_TOP)) - return 0; - r_d = r->d; - nist_cp_bn(r_d, a_d, BN_NIST_384_TOP); - } - else - r_d = a_d; - - nist_cp_bn_0(buf.bn, a_d + BN_NIST_384_TOP, top - BN_NIST_384_TOP, BN_NIST_384_TOP); + BN_CTX *ctx) +{ + int i, top = a->top; + int carry = 0; + register BN_ULONG *r_d, *a_d = a->d; + union { + BN_ULONG bn[BN_NIST_384_TOP]; + unsigned int ui[BN_NIST_384_TOP * sizeof(BN_ULONG) / + sizeof(unsigned int)]; + } buf; + BN_ULONG c_d[BN_NIST_384_TOP], *res; + PTR_SIZE_INT mask; + union { + bn_addsub_f f; + PTR_SIZE_INT p; + } u; + static const BIGNUM _bignum_nist_p_384_sqr = { + (BN_ULONG *)_nist_p_384_sqr, + sizeof(_nist_p_384_sqr) / sizeof(_nist_p_384_sqr[0]), + sizeof(_nist_p_384_sqr) / sizeof(_nist_p_384_sqr[0]), + 0, BN_FLG_STATIC_DATA + }; + + field = &_bignum_nist_p_384; /* just to make sure */ + + if (BN_is_negative(a) || BN_ucmp(a, &_bignum_nist_p_384_sqr) >= 0) + return BN_nnmod(r, a, field, ctx); + + i = BN_ucmp(field, a); + if (i == 0) { + BN_zero(r); + return 1; + } else if (i > 0) + return (r == a) ? 1 : (BN_copy(r, a) != NULL); + + if (r != a) { + if (!bn_wexpand(r, BN_NIST_384_TOP)) + return 0; + r_d = r->d; + nist_cp_bn(r_d, a_d, BN_NIST_384_TOP); + } else + r_d = a_d; + + nist_cp_bn_0(buf.bn, a_d + BN_NIST_384_TOP, top - BN_NIST_384_TOP, + BN_NIST_384_TOP); #if defined(NIST_INT64) - { - NIST_INT64 acc; /* accumulator */ - unsigned int *rp=(unsigned int *)r_d; - const unsigned int *bp=(const unsigned int *)buf.ui; - - acc = rp[0]; acc += bp[12-12]; - acc += bp[21-12]; - acc += bp[20-12]; - acc -= bp[23-12]; rp[0] = (unsigned int)acc; acc >>= 32; - - acc += rp[1]; acc += bp[13-12]; - acc += bp[22-12]; - acc += bp[23-12]; - acc -= bp[12-12]; - acc -= bp[20-12]; rp[1] = (unsigned int)acc; acc >>= 32; - - acc += rp[2]; acc += bp[14-12]; - acc += bp[23-12]; - acc -= bp[13-12]; - acc -= bp[21-12]; rp[2] = (unsigned int)acc; acc >>= 32; - - acc += rp[3]; acc += bp[15-12]; - acc += bp[12-12]; - acc += bp[20-12]; - acc += bp[21-12]; - acc -= bp[14-12]; - acc -= bp[22-12]; - acc -= bp[23-12]; rp[3] = (unsigned int)acc; acc >>= 32; - - acc += rp[4]; acc += bp[21-12]; - acc += bp[21-12]; - acc += bp[16-12]; - acc += bp[13-12]; - acc += bp[12-12]; - acc += bp[20-12]; - acc += bp[22-12]; - acc -= bp[15-12]; - acc -= bp[23-12]; - acc -= bp[23-12]; rp[4] = (unsigned int)acc; acc >>= 32; - - acc += rp[5]; acc += bp[22-12]; - acc += bp[22-12]; - acc += bp[17-12]; - acc += bp[14-12]; - acc += bp[13-12]; - acc += bp[21-12]; - acc += bp[23-12]; - acc -= bp[16-12]; rp[5] = (unsigned int)acc; acc >>= 32; - - acc += rp[6]; acc += bp[23-12]; - acc += bp[23-12]; - acc += bp[18-12]; - acc += bp[15-12]; - acc += bp[14-12]; - acc += bp[22-12]; - acc -= bp[17-12]; rp[6] = (unsigned int)acc; acc >>= 32; - - acc += rp[7]; acc += bp[19-12]; - acc += bp[16-12]; - acc += bp[15-12]; - acc += bp[23-12]; - acc -= bp[18-12]; rp[7] = (unsigned int)acc; acc >>= 32; - - acc += rp[8]; acc += bp[20-12]; - acc += bp[17-12]; - acc += bp[16-12]; - acc -= bp[19-12]; rp[8] = (unsigned int)acc; acc >>= 32; - - acc += rp[9]; acc += bp[21-12]; - acc += bp[18-12]; - acc += bp[17-12]; - acc -= bp[20-12]; rp[9] = (unsigned int)acc; acc >>= 32; - - acc += rp[10]; acc += bp[22-12]; - acc += bp[19-12]; - acc += bp[18-12]; - acc -= bp[21-12]; rp[10] = (unsigned int)acc; acc >>= 32; - - acc += rp[11]; acc += bp[23-12]; - acc += bp[20-12]; - acc += bp[19-12]; - acc -= bp[22-12]; rp[11] = (unsigned int)acc; - - carry = (int)(acc>>32); - } + { + NIST_INT64 acc; /* accumulator */ + unsigned int *rp = (unsigned int *)r_d; + const unsigned int *bp = (const unsigned int *)buf.ui; + + acc = rp[0]; + acc += bp[12 - 12]; + acc += bp[21 - 12]; + acc += bp[20 - 12]; + acc -= bp[23 - 12]; + rp[0] = (unsigned int)acc; + acc >>= 32; + + acc += rp[1]; + acc += bp[13 - 12]; + acc += bp[22 - 12]; + acc += bp[23 - 12]; + acc -= bp[12 - 12]; + acc -= bp[20 - 12]; + rp[1] = (unsigned int)acc; + acc >>= 32; + + acc += rp[2]; + acc += bp[14 - 12]; + acc += bp[23 - 12]; + acc -= bp[13 - 12]; + acc -= bp[21 - 12]; + rp[2] = (unsigned int)acc; + acc >>= 32; + + acc += rp[3]; + acc += bp[15 - 12]; + acc += bp[12 - 12]; + acc += bp[20 - 12]; + acc += bp[21 - 12]; + acc -= bp[14 - 12]; + acc -= bp[22 - 12]; + acc -= bp[23 - 12]; + rp[3] = (unsigned int)acc; + acc >>= 32; + + acc += rp[4]; + acc += bp[21 - 12]; + acc += bp[21 - 12]; + acc += bp[16 - 12]; + acc += bp[13 - 12]; + acc += bp[12 - 12]; + acc += bp[20 - 12]; + acc += bp[22 - 12]; + acc -= bp[15 - 12]; + acc -= bp[23 - 12]; + acc -= bp[23 - 12]; + rp[4] = (unsigned int)acc; + acc >>= 32; + + acc += rp[5]; + acc += bp[22 - 12]; + acc += bp[22 - 12]; + acc += bp[17 - 12]; + acc += bp[14 - 12]; + acc += bp[13 - 12]; + acc += bp[21 - 12]; + acc += bp[23 - 12]; + acc -= bp[16 - 12]; + rp[5] = (unsigned int)acc; + acc >>= 32; + + acc += rp[6]; + acc += bp[23 - 12]; + acc += bp[23 - 12]; + acc += bp[18 - 12]; + acc += bp[15 - 12]; + acc += bp[14 - 12]; + acc += bp[22 - 12]; + acc -= bp[17 - 12]; + rp[6] = (unsigned int)acc; + acc >>= 32; + + acc += rp[7]; + acc += bp[19 - 12]; + acc += bp[16 - 12]; + acc += bp[15 - 12]; + acc += bp[23 - 12]; + acc -= bp[18 - 12]; + rp[7] = (unsigned int)acc; + acc >>= 32; + + acc += rp[8]; + acc += bp[20 - 12]; + acc += bp[17 - 12]; + acc += bp[16 - 12]; + acc -= bp[19 - 12]; + rp[8] = (unsigned int)acc; + acc >>= 32; + + acc += rp[9]; + acc += bp[21 - 12]; + acc += bp[18 - 12]; + acc += bp[17 - 12]; + acc -= bp[20 - 12]; + rp[9] = (unsigned int)acc; + acc >>= 32; + + acc += rp[10]; + acc += bp[22 - 12]; + acc += bp[19 - 12]; + acc += bp[18 - 12]; + acc -= bp[21 - 12]; + rp[10] = (unsigned int)acc; + acc >>= 32; + + acc += rp[11]; + acc += bp[23 - 12]; + acc += bp[20 - 12]; + acc += bp[19 - 12]; + acc -= bp[22 - 12]; + rp[11] = (unsigned int)acc; + + carry = (int)(acc >> 32); + } #else - { - BN_ULONG t_d[BN_NIST_384_TOP]; - - /*S1*/ - nist_set_256(t_d, buf.bn, 0, 0, 0, 0, 0, 23-4, 22-4, 21-4); - /* left shift */ - { - register BN_ULONG *ap,t,c; - ap = t_d; - c=0; - for (i = 3; i != 0; --i) - { - t= *ap; - *(ap++)=((t<<1)|c)&BN_MASK2; - c=(t & BN_TBIT)?1:0; - } - *ap=c; - } - carry = (int)bn_add_words(r_d+(128/BN_BITS2), r_d+(128/BN_BITS2), - t_d, BN_NIST_256_TOP); - /*S2 */ - carry += (int)bn_add_words(r_d, r_d, buf.bn, BN_NIST_384_TOP); - /*S3*/ - nist_set_384(t_d,buf.bn,20,19,18,17,16,15,14,13,12,23,22,21); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); - /*S4*/ - nist_set_384(t_d,buf.bn,19,18,17,16,15,14,13,12,20,0,23,0); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); - /*S5*/ - nist_set_384(t_d, buf.bn,0,0,0,0,23,22,21,20,0,0,0,0); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); - /*S6*/ - nist_set_384(t_d,buf.bn,0,0,0,0,0,0,23,22,21,0,0,20); - carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); - /*D1*/ - nist_set_384(t_d,buf.bn,22,21,20,19,18,17,16,15,14,13,12,23); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP); - /*D2*/ - nist_set_384(t_d,buf.bn,0,0,0,0,0,0,0,23,22,21,20,0); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP); - /*D3*/ - nist_set_384(t_d,buf.bn,0,0,0,0,0,0,0,23,23,0,0,0); - carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP); - - } + { + BN_ULONG t_d[BN_NIST_384_TOP]; + + /* + * S1 + */ + nist_set_256(t_d, buf.bn, 0, 0, 0, 0, 0, 23 - 4, 22 - 4, 21 - 4); + /* left shift */ + { + register BN_ULONG *ap, t, c; + ap = t_d; + c = 0; + for (i = 3; i != 0; --i) { + t = *ap; + *(ap++) = ((t << 1) | c) & BN_MASK2; + c = (t & BN_TBIT) ? 1 : 0; + } + *ap = c; + } + carry = + (int)bn_add_words(r_d + (128 / BN_BITS2), r_d + (128 / BN_BITS2), + t_d, BN_NIST_256_TOP); + /* + * S2 + */ + carry += (int)bn_add_words(r_d, r_d, buf.bn, BN_NIST_384_TOP); + /* + * S3 + */ + nist_set_384(t_d, buf.bn, 20, 19, 18, 17, 16, 15, 14, 13, 12, 23, 22, + 21); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); + /* + * S4 + */ + nist_set_384(t_d, buf.bn, 19, 18, 17, 16, 15, 14, 13, 12, 20, 0, 23, + 0); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); + /* + * S5 + */ + nist_set_384(t_d, buf.bn, 0, 0, 0, 0, 23, 22, 21, 20, 0, 0, 0, 0); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); + /* + * S6 + */ + nist_set_384(t_d, buf.bn, 0, 0, 0, 0, 0, 0, 23, 22, 21, 0, 0, 20); + carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP); + /* + * D1 + */ + nist_set_384(t_d, buf.bn, 22, 21, 20, 19, 18, 17, 16, 15, 14, 13, 12, + 23); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP); + /* + * D2 + */ + nist_set_384(t_d, buf.bn, 0, 0, 0, 0, 0, 0, 0, 23, 22, 21, 20, 0); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP); + /* + * D3 + */ + nist_set_384(t_d, buf.bn, 0, 0, 0, 0, 0, 0, 0, 23, 23, 0, 0, 0); + carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP); + + } #endif - /* see BN_nist_mod_224 for explanation */ - u.f = bn_sub_words; - if (carry > 0) - carry = (int)bn_sub_words(r_d,r_d,_nist_p_384[carry-1],BN_NIST_384_TOP); - else if (carry < 0) - { - carry = (int)bn_add_words(r_d,r_d,_nist_p_384[-carry-1],BN_NIST_384_TOP); - mask = 0-(PTR_SIZE_INT)carry; - u.p = ((PTR_SIZE_INT)bn_sub_words&mask) | - ((PTR_SIZE_INT)bn_add_words&~mask); - } - else - carry = 1; - - mask = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_384[0],BN_NIST_384_TOP); - mask &= 0-(PTR_SIZE_INT)carry; - res = c_d; - res = (BN_ULONG *)(((PTR_SIZE_INT)res&~mask) | - ((PTR_SIZE_INT)r_d&mask)); - nist_cp_bn(r_d, res, BN_NIST_384_TOP); - r->top = BN_NIST_384_TOP; - bn_correct_top(r); - - return 1; - } - -#define BN_NIST_521_RSHIFT (521%BN_BITS2) -#define BN_NIST_521_LSHIFT (BN_BITS2-BN_NIST_521_RSHIFT) -#define BN_NIST_521_TOP_MASK ((BN_ULONG)BN_MASK2>>BN_NIST_521_LSHIFT) + /* see BN_nist_mod_224 for explanation */ + u.f = bn_sub_words; + if (carry > 0) + carry = + (int)bn_sub_words(r_d, r_d, _nist_p_384[carry - 1], + BN_NIST_384_TOP); + else if (carry < 0) { + carry = + (int)bn_add_words(r_d, r_d, _nist_p_384[-carry - 1], + BN_NIST_384_TOP); + mask = 0 - (PTR_SIZE_INT) carry; + u.p = ((PTR_SIZE_INT) bn_sub_words & mask) | + ((PTR_SIZE_INT) bn_add_words & ~mask); + } else + carry = 1; + + mask = + 0 - (PTR_SIZE_INT) (*u.f) (c_d, r_d, _nist_p_384[0], BN_NIST_384_TOP); + mask &= 0 - (PTR_SIZE_INT) carry; + res = c_d; + res = (BN_ULONG *)(((PTR_SIZE_INT) res & ~mask) | + ((PTR_SIZE_INT) r_d & mask)); + nist_cp_bn(r_d, res, BN_NIST_384_TOP); + r->top = BN_NIST_384_TOP; + bn_correct_top(r); + + return 1; +} + +#define BN_NIST_521_RSHIFT (521%BN_BITS2) +#define BN_NIST_521_LSHIFT (BN_BITS2-BN_NIST_521_RSHIFT) +#define BN_NIST_521_TOP_MASK ((BN_ULONG)BN_MASK2>>BN_NIST_521_LSHIFT) int BN_nist_mod_521(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, - BN_CTX *ctx) - { - int top = a->top, i; - BN_ULONG *r_d, *a_d = a->d, - t_d[BN_NIST_521_TOP], - val,tmp,*res; - PTR_SIZE_INT mask; - static const BIGNUM _bignum_nist_p_521_sqr = { - (BN_ULONG *)_nist_p_521_sqr, - sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]), - sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]), - 0,BN_FLG_STATIC_DATA }; - - field = &_bignum_nist_p_521; /* just to make sure */ - - if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_521_sqr)>=0) - return BN_nnmod(r, a, field, ctx); - - i = BN_ucmp(field, a); - if (i == 0) - { - BN_zero(r); - return 1; - } - else if (i > 0) - return (r == a)? 1 : (BN_copy(r ,a) != NULL); - - if (r != a) - { - if (!bn_wexpand(r,BN_NIST_521_TOP)) - return 0; - r_d = r->d; - nist_cp_bn(r_d,a_d, BN_NIST_521_TOP); - } - else - r_d = a_d; - - /* upper 521 bits, copy ... */ - nist_cp_bn_0(t_d,a_d + (BN_NIST_521_TOP-1), top - (BN_NIST_521_TOP-1),BN_NIST_521_TOP); - /* ... and right shift */ - for (val=t_d[0],i=0; i<BN_NIST_521_TOP-1; i++) - { - t_d[i] = ( val>>BN_NIST_521_RSHIFT | - (tmp=t_d[i+1])<<BN_NIST_521_LSHIFT ) & BN_MASK2; - val=tmp; - } - t_d[i] = val>>BN_NIST_521_RSHIFT; - /* lower 521 bits */ - r_d[i] &= BN_NIST_521_TOP_MASK; - - bn_add_words(r_d,r_d,t_d,BN_NIST_521_TOP); - mask = 0-(PTR_SIZE_INT)bn_sub_words(t_d,r_d,_nist_p_521,BN_NIST_521_TOP); - res = t_d; - res = (BN_ULONG *)(((PTR_SIZE_INT)res&~mask) | - ((PTR_SIZE_INT)r_d&mask)); - nist_cp_bn(r_d,res,BN_NIST_521_TOP); - r->top = BN_NIST_521_TOP; - bn_correct_top(r); - - return 1; - } + BN_CTX *ctx) +{ + int top = a->top, i; + BN_ULONG *r_d, *a_d = a->d, t_d[BN_NIST_521_TOP], val, tmp, *res; + PTR_SIZE_INT mask; + static const BIGNUM _bignum_nist_p_521_sqr = { + (BN_ULONG *)_nist_p_521_sqr, + sizeof(_nist_p_521_sqr) / sizeof(_nist_p_521_sqr[0]), + sizeof(_nist_p_521_sqr) / sizeof(_nist_p_521_sqr[0]), + 0, BN_FLG_STATIC_DATA + }; + + field = &_bignum_nist_p_521; /* just to make sure */ + + if (BN_is_negative(a) || BN_ucmp(a, &_bignum_nist_p_521_sqr) >= 0) + return BN_nnmod(r, a, field, ctx); + + i = BN_ucmp(field, a); + if (i == 0) { + BN_zero(r); + return 1; + } else if (i > 0) + return (r == a) ? 1 : (BN_copy(r, a) != NULL); + + if (r != a) { + if (!bn_wexpand(r, BN_NIST_521_TOP)) + return 0; + r_d = r->d; + nist_cp_bn(r_d, a_d, BN_NIST_521_TOP); + } else + r_d = a_d; + + /* upper 521 bits, copy ... */ + nist_cp_bn_0(t_d, a_d + (BN_NIST_521_TOP - 1), + top - (BN_NIST_521_TOP - 1), BN_NIST_521_TOP); + /* ... and right shift */ + for (val = t_d[0], i = 0; i < BN_NIST_521_TOP - 1; i++) { + t_d[i] = (val >> BN_NIST_521_RSHIFT | + (tmp = t_d[i + 1]) << BN_NIST_521_LSHIFT) & BN_MASK2; + val = tmp; + } + t_d[i] = val >> BN_NIST_521_RSHIFT; + /* lower 521 bits */ + r_d[i] &= BN_NIST_521_TOP_MASK; + + bn_add_words(r_d, r_d, t_d, BN_NIST_521_TOP); + mask = + 0 - (PTR_SIZE_INT) bn_sub_words(t_d, r_d, _nist_p_521, + BN_NIST_521_TOP); + res = t_d; + res = (BN_ULONG *)(((PTR_SIZE_INT) res & ~mask) | + ((PTR_SIZE_INT) r_d & mask)); + nist_cp_bn(r_d, res, BN_NIST_521_TOP); + r->top = BN_NIST_521_TOP; + bn_correct_top(r); + + return 1; +} diff --git a/openssl/crypto/bn/bn_prime.c b/openssl/crypto/bn/bn_prime.c index 7b25979dd..1d256874c 100644 --- a/openssl/crypto/bn/bn_prime.c +++ b/openssl/crypto/bn/bn_prime.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -115,380 +115,401 @@ #include "bn_lcl.h" #include <openssl/rand.h> -/* NB: these functions have been "upgraded", the deprecated versions (which are - * compatibility wrappers using these functions) are in bn_depr.c. - * - Geoff +/* + * NB: these functions have been "upgraded", the deprecated versions (which + * are compatibility wrappers using these functions) are in bn_depr.c. - + * Geoff */ -/* The quick sieve algorithm approach to weeding out primes is - * Philip Zimmermann's, as implemented in PGP. I have had a read of - * his comments and implemented my own version. +/* + * The quick sieve algorithm approach to weeding out primes is Philip + * Zimmermann's, as implemented in PGP. I have had a read of his comments + * and implemented my own version. */ #include "bn_prime.h" static int witness(BIGNUM *w, const BIGNUM *a, const BIGNUM *a1, - const BIGNUM *a1_odd, int k, BN_CTX *ctx, BN_MONT_CTX *mont); + const BIGNUM *a1_odd, int k, BN_CTX *ctx, + BN_MONT_CTX *mont); static int probable_prime(BIGNUM *rnd, int bits); static int probable_prime_dh(BIGNUM *rnd, int bits, - const BIGNUM *add, const BIGNUM *rem, BN_CTX *ctx); -static int probable_prime_dh_safe(BIGNUM *rnd, int bits, - const BIGNUM *add, const BIGNUM *rem, BN_CTX *ctx); + const BIGNUM *add, const BIGNUM *rem, + BN_CTX *ctx); +static int probable_prime_dh_safe(BIGNUM *rnd, int bits, const BIGNUM *add, + const BIGNUM *rem, BN_CTX *ctx); int BN_GENCB_call(BN_GENCB *cb, int a, int b) - { - /* No callback means continue */ - if(!cb) return 1; - switch(cb->ver) - { - case 1: - /* Deprecated-style callbacks */ - if(!cb->cb.cb_1) - return 1; - cb->cb.cb_1(a, b, cb->arg); - return 1; - case 2: - /* New-style callbacks */ - return cb->cb.cb_2(a, b, cb); - default: - break; - } - /* Unrecognised callback type */ - return 0; - } +{ + /* No callback means continue */ + if (!cb) + return 1; + switch (cb->ver) { + case 1: + /* Deprecated-style callbacks */ + if (!cb->cb.cb_1) + return 1; + cb->cb.cb_1(a, b, cb->arg); + return 1; + case 2: + /* New-style callbacks */ + return cb->cb.cb_2(a, b, cb); + default: + break; + } + /* Unrecognised callback type */ + return 0; +} int BN_generate_prime_ex(BIGNUM *ret, int bits, int safe, - const BIGNUM *add, const BIGNUM *rem, BN_GENCB *cb) - { - BIGNUM *t; - int found=0; - int i,j,c1=0; - BN_CTX *ctx; - int checks = BN_prime_checks_for_size(bits); - - ctx=BN_CTX_new(); - if (ctx == NULL) goto err; - BN_CTX_start(ctx); - t = BN_CTX_get(ctx); - if(!t) goto err; -loop: - /* make a random number and set the top and bottom bits */ - if (add == NULL) - { - if (!probable_prime(ret,bits)) goto err; - } - else - { - if (safe) - { - if (!probable_prime_dh_safe(ret,bits,add,rem,ctx)) - goto err; - } - else - { - if (!probable_prime_dh(ret,bits,add,rem,ctx)) - goto err; - } - } - /* if (BN_mod_word(ret,(BN_ULONG)3) == 1) goto loop; */ - if(!BN_GENCB_call(cb, 0, c1++)) - /* aborted */ - goto err; - - if (!safe) - { - i=BN_is_prime_fasttest_ex(ret,checks,ctx,0,cb); - if (i == -1) goto err; - if (i == 0) goto loop; - } - else - { - /* for "safe prime" generation, - * check that (p-1)/2 is prime. - * Since a prime is odd, We just - * need to divide by 2 */ - if (!BN_rshift1(t,ret)) goto err; - - for (i=0; i<checks; i++) - { - j=BN_is_prime_fasttest_ex(ret,1,ctx,0,cb); - if (j == -1) goto err; - if (j == 0) goto loop; - - j=BN_is_prime_fasttest_ex(t,1,ctx,0,cb); - if (j == -1) goto err; - if (j == 0) goto loop; - - if(!BN_GENCB_call(cb, 2, c1-1)) - goto err; - /* We have a safe prime test pass */ - } - } - /* we have a prime :-) */ - found = 1; -err: - if (ctx != NULL) - { - BN_CTX_end(ctx); - BN_CTX_free(ctx); - } - bn_check_top(ret); - return found; - } - -int BN_is_prime_ex(const BIGNUM *a, int checks, BN_CTX *ctx_passed, BN_GENCB *cb) - { - return BN_is_prime_fasttest_ex(a, checks, ctx_passed, 0, cb); - } + const BIGNUM *add, const BIGNUM *rem, BN_GENCB *cb) +{ + BIGNUM *t; + int found = 0; + int i, j, c1 = 0; + BN_CTX *ctx; + int checks = BN_prime_checks_for_size(bits); + + ctx = BN_CTX_new(); + if (ctx == NULL) + goto err; + BN_CTX_start(ctx); + t = BN_CTX_get(ctx); + if (!t) + goto err; + loop: + /* make a random number and set the top and bottom bits */ + if (add == NULL) { + if (!probable_prime(ret, bits)) + goto err; + } else { + if (safe) { + if (!probable_prime_dh_safe(ret, bits, add, rem, ctx)) + goto err; + } else { + if (!probable_prime_dh(ret, bits, add, rem, ctx)) + goto err; + } + } + /* if (BN_mod_word(ret,(BN_ULONG)3) == 1) goto loop; */ + if (!BN_GENCB_call(cb, 0, c1++)) + /* aborted */ + goto err; + + if (!safe) { + i = BN_is_prime_fasttest_ex(ret, checks, ctx, 0, cb); + if (i == -1) + goto err; + if (i == 0) + goto loop; + } else { + /* + * for "safe prime" generation, check that (p-1)/2 is prime. Since a + * prime is odd, We just need to divide by 2 + */ + if (!BN_rshift1(t, ret)) + goto err; + + for (i = 0; i < checks; i++) { + j = BN_is_prime_fasttest_ex(ret, 1, ctx, 0, cb); + if (j == -1) + goto err; + if (j == 0) + goto loop; + + j = BN_is_prime_fasttest_ex(t, 1, ctx, 0, cb); + if (j == -1) + goto err; + if (j == 0) + goto loop; + + if (!BN_GENCB_call(cb, 2, c1 - 1)) + goto err; + /* We have a safe prime test pass */ + } + } + /* we have a prime :-) */ + found = 1; + err: + if (ctx != NULL) { + BN_CTX_end(ctx); + BN_CTX_free(ctx); + } + bn_check_top(ret); + return found; +} + +int BN_is_prime_ex(const BIGNUM *a, int checks, BN_CTX *ctx_passed, + BN_GENCB *cb) +{ + return BN_is_prime_fasttest_ex(a, checks, ctx_passed, 0, cb); +} int BN_is_prime_fasttest_ex(const BIGNUM *a, int checks, BN_CTX *ctx_passed, - int do_trial_division, BN_GENCB *cb) - { - int i, j, ret = -1; - int k; - BN_CTX *ctx = NULL; - BIGNUM *A1, *A1_odd, *check; /* taken from ctx */ - BN_MONT_CTX *mont = NULL; - const BIGNUM *A = NULL; - - if (BN_cmp(a, BN_value_one()) <= 0) - return 0; - - if (checks == BN_prime_checks) - checks = BN_prime_checks_for_size(BN_num_bits(a)); - - /* first look for small factors */ - if (!BN_is_odd(a)) - /* a is even => a is prime if and only if a == 2 */ - return BN_is_word(a, 2); - if (do_trial_division) - { - for (i = 1; i < NUMPRIMES; i++) - if (BN_mod_word(a, primes[i]) == 0) - return 0; - if(!BN_GENCB_call(cb, 1, -1)) - goto err; - } - - if (ctx_passed != NULL) - ctx = ctx_passed; - else - if ((ctx=BN_CTX_new()) == NULL) - goto err; - BN_CTX_start(ctx); - - /* A := abs(a) */ - if (a->neg) - { - BIGNUM *t; - if ((t = BN_CTX_get(ctx)) == NULL) goto err; - BN_copy(t, a); - t->neg = 0; - A = t; - } - else - A = a; - A1 = BN_CTX_get(ctx); - A1_odd = BN_CTX_get(ctx); - check = BN_CTX_get(ctx); - if (check == NULL) goto err; - - /* compute A1 := A - 1 */ - if (!BN_copy(A1, A)) - goto err; - if (!BN_sub_word(A1, 1)) - goto err; - if (BN_is_zero(A1)) - { - ret = 0; - goto err; - } - - /* write A1 as A1_odd * 2^k */ - k = 1; - while (!BN_is_bit_set(A1, k)) - k++; - if (!BN_rshift(A1_odd, A1, k)) - goto err; - - /* Montgomery setup for computations mod A */ - mont = BN_MONT_CTX_new(); - if (mont == NULL) - goto err; - if (!BN_MONT_CTX_set(mont, A, ctx)) - goto err; - - for (i = 0; i < checks; i++) - { - if (!BN_pseudo_rand_range(check, A1)) - goto err; - if (!BN_add_word(check, 1)) - goto err; - /* now 1 <= check < A */ - - j = witness(check, A, A1, A1_odd, k, ctx, mont); - if (j == -1) goto err; - if (j) - { - ret=0; - goto err; - } - if(!BN_GENCB_call(cb, 1, i)) - goto err; - } - ret=1; -err: - if (ctx != NULL) - { - BN_CTX_end(ctx); - if (ctx_passed == NULL) - BN_CTX_free(ctx); - } - if (mont != NULL) - BN_MONT_CTX_free(mont); - - return(ret); - } + int do_trial_division, BN_GENCB *cb) +{ + int i, j, ret = -1; + int k; + BN_CTX *ctx = NULL; + BIGNUM *A1, *A1_odd, *check; /* taken from ctx */ + BN_MONT_CTX *mont = NULL; + const BIGNUM *A = NULL; + + if (BN_cmp(a, BN_value_one()) <= 0) + return 0; + + if (checks == BN_prime_checks) + checks = BN_prime_checks_for_size(BN_num_bits(a)); + + /* first look for small factors */ + if (!BN_is_odd(a)) + /* a is even => a is prime if and only if a == 2 */ + return BN_is_word(a, 2); + if (do_trial_division) { + for (i = 1; i < NUMPRIMES; i++) + if (BN_mod_word(a, primes[i]) == 0) + return 0; + if (!BN_GENCB_call(cb, 1, -1)) + goto err; + } + + if (ctx_passed != NULL) + ctx = ctx_passed; + else if ((ctx = BN_CTX_new()) == NULL) + goto err; + BN_CTX_start(ctx); + + /* A := abs(a) */ + if (a->neg) { + BIGNUM *t; + if ((t = BN_CTX_get(ctx)) == NULL) + goto err; + BN_copy(t, a); + t->neg = 0; + A = t; + } else + A = a; + A1 = BN_CTX_get(ctx); + A1_odd = BN_CTX_get(ctx); + check = BN_CTX_get(ctx); + if (check == NULL) + goto err; + + /* compute A1 := A - 1 */ + if (!BN_copy(A1, A)) + goto err; + if (!BN_sub_word(A1, 1)) + goto err; + if (BN_is_zero(A1)) { + ret = 0; + goto err; + } + + /* write A1 as A1_odd * 2^k */ + k = 1; + while (!BN_is_bit_set(A1, k)) + k++; + if (!BN_rshift(A1_odd, A1, k)) + goto err; + + /* Montgomery setup for computations mod A */ + mont = BN_MONT_CTX_new(); + if (mont == NULL) + goto err; + if (!BN_MONT_CTX_set(mont, A, ctx)) + goto err; + + for (i = 0; i < checks; i++) { + if (!BN_pseudo_rand_range(check, A1)) + goto err; + if (!BN_add_word(check, 1)) + goto err; + /* now 1 <= check < A */ + + j = witness(check, A, A1, A1_odd, k, ctx, mont); + if (j == -1) + goto err; + if (j) { + ret = 0; + goto err; + } + if (!BN_GENCB_call(cb, 1, i)) + goto err; + } + ret = 1; + err: + if (ctx != NULL) { + BN_CTX_end(ctx); + if (ctx_passed == NULL) + BN_CTX_free(ctx); + } + if (mont != NULL) + BN_MONT_CTX_free(mont); + + return (ret); +} static int witness(BIGNUM *w, const BIGNUM *a, const BIGNUM *a1, - const BIGNUM *a1_odd, int k, BN_CTX *ctx, BN_MONT_CTX *mont) - { - if (!BN_mod_exp_mont(w, w, a1_odd, a, ctx, mont)) /* w := w^a1_odd mod a */ - return -1; - if (BN_is_one(w)) - return 0; /* probably prime */ - if (BN_cmp(w, a1) == 0) - return 0; /* w == -1 (mod a), 'a' is probably prime */ - while (--k) - { - if (!BN_mod_mul(w, w, w, a, ctx)) /* w := w^2 mod a */ - return -1; - if (BN_is_one(w)) - return 1; /* 'a' is composite, otherwise a previous 'w' would - * have been == -1 (mod 'a') */ - if (BN_cmp(w, a1) == 0) - return 0; /* w == -1 (mod a), 'a' is probably prime */ - } - /* If we get here, 'w' is the (a-1)/2-th power of the original 'w', - * and it is neither -1 nor +1 -- so 'a' cannot be prime */ - bn_check_top(w); - return 1; - } + const BIGNUM *a1_odd, int k, BN_CTX *ctx, + BN_MONT_CTX *mont) +{ + if (!BN_mod_exp_mont(w, w, a1_odd, a, ctx, mont)) /* w := w^a1_odd mod a */ + return -1; + if (BN_is_one(w)) + return 0; /* probably prime */ + if (BN_cmp(w, a1) == 0) + return 0; /* w == -1 (mod a), 'a' is probably prime */ + while (--k) { + if (!BN_mod_mul(w, w, w, a, ctx)) /* w := w^2 mod a */ + return -1; + if (BN_is_one(w)) + return 1; /* 'a' is composite, otherwise a previous 'w' + * would have been == -1 (mod 'a') */ + if (BN_cmp(w, a1) == 0) + return 0; /* w == -1 (mod a), 'a' is probably prime */ + } + /* + * If we get here, 'w' is the (a-1)/2-th power of the original 'w', and + * it is neither -1 nor +1 -- so 'a' cannot be prime + */ + bn_check_top(w); + return 1; +} static int probable_prime(BIGNUM *rnd, int bits) - { - int i; - prime_t mods[NUMPRIMES]; - BN_ULONG delta,maxdelta; - -again: - if (!BN_rand(rnd,bits,1,1)) return(0); - /* we now have a random number 'rand' to test. */ - for (i=1; i<NUMPRIMES; i++) - mods[i]=(prime_t)BN_mod_word(rnd,(BN_ULONG)primes[i]); - maxdelta=BN_MASK2 - primes[NUMPRIMES-1]; - delta=0; - loop: for (i=1; i<NUMPRIMES; i++) - { - /* check that rnd is not a prime and also - * that gcd(rnd-1,primes) == 1 (except for 2) */ - if (((mods[i]+delta)%primes[i]) <= 1) - { - delta+=2; - if (delta > maxdelta) goto again; - goto loop; - } - } - if (!BN_add_word(rnd,delta)) return(0); - bn_check_top(rnd); - return(1); - } +{ + int i; + prime_t mods[NUMPRIMES]; + BN_ULONG delta, maxdelta; + + again: + if (!BN_rand(rnd, bits, 1, 1)) + return (0); + /* we now have a random number 'rand' to test. */ + for (i = 1; i < NUMPRIMES; i++) + mods[i] = (prime_t) BN_mod_word(rnd, (BN_ULONG)primes[i]); + maxdelta = BN_MASK2 - primes[NUMPRIMES - 1]; + delta = 0; + loop:for (i = 1; i < NUMPRIMES; i++) { + /* + * check that rnd is not a prime and also that gcd(rnd-1,primes) == 1 + * (except for 2) + */ + if (((mods[i] + delta) % primes[i]) <= 1) { + delta += 2; + if (delta > maxdelta) + goto again; + goto loop; + } + } + if (!BN_add_word(rnd, delta)) + return (0); + bn_check_top(rnd); + return (1); +} static int probable_prime_dh(BIGNUM *rnd, int bits, - const BIGNUM *add, const BIGNUM *rem, BN_CTX *ctx) - { - int i,ret=0; - BIGNUM *t1; - - BN_CTX_start(ctx); - if ((t1 = BN_CTX_get(ctx)) == NULL) goto err; - - if (!BN_rand(rnd,bits,0,1)) goto err; - - /* we need ((rnd-rem) % add) == 0 */ - - if (!BN_mod(t1,rnd,add,ctx)) goto err; - if (!BN_sub(rnd,rnd,t1)) goto err; - if (rem == NULL) - { if (!BN_add_word(rnd,1)) goto err; } - else - { if (!BN_add(rnd,rnd,rem)) goto err; } - - /* we now have a random number 'rand' to test. */ - - loop: for (i=1; i<NUMPRIMES; i++) - { - /* check that rnd is a prime */ - if (BN_mod_word(rnd,(BN_ULONG)primes[i]) <= 1) - { - if (!BN_add(rnd,rnd,add)) goto err; - goto loop; - } - } - ret=1; -err: - BN_CTX_end(ctx); - bn_check_top(rnd); - return(ret); - } + const BIGNUM *add, const BIGNUM *rem, + BN_CTX *ctx) +{ + int i, ret = 0; + BIGNUM *t1; + + BN_CTX_start(ctx); + if ((t1 = BN_CTX_get(ctx)) == NULL) + goto err; + + if (!BN_rand(rnd, bits, 0, 1)) + goto err; + + /* we need ((rnd-rem) % add) == 0 */ + + if (!BN_mod(t1, rnd, add, ctx)) + goto err; + if (!BN_sub(rnd, rnd, t1)) + goto err; + if (rem == NULL) { + if (!BN_add_word(rnd, 1)) + goto err; + } else { + if (!BN_add(rnd, rnd, rem)) + goto err; + } + + /* we now have a random number 'rand' to test. */ + + loop:for (i = 1; i < NUMPRIMES; i++) { + /* check that rnd is a prime */ + if (BN_mod_word(rnd, (BN_ULONG)primes[i]) <= 1) { + if (!BN_add(rnd, rnd, add)) + goto err; + goto loop; + } + } + ret = 1; + err: + BN_CTX_end(ctx); + bn_check_top(rnd); + return (ret); +} static int probable_prime_dh_safe(BIGNUM *p, int bits, const BIGNUM *padd, - const BIGNUM *rem, BN_CTX *ctx) - { - int i,ret=0; - BIGNUM *t1,*qadd,*q; - - bits--; - BN_CTX_start(ctx); - t1 = BN_CTX_get(ctx); - q = BN_CTX_get(ctx); - qadd = BN_CTX_get(ctx); - if (qadd == NULL) goto err; - - if (!BN_rshift1(qadd,padd)) goto err; - - if (!BN_rand(q,bits,0,1)) goto err; - - /* we need ((rnd-rem) % add) == 0 */ - if (!BN_mod(t1,q,qadd,ctx)) goto err; - if (!BN_sub(q,q,t1)) goto err; - if (rem == NULL) - { if (!BN_add_word(q,1)) goto err; } - else - { - if (!BN_rshift1(t1,rem)) goto err; - if (!BN_add(q,q,t1)) goto err; - } - - /* we now have a random number 'rand' to test. */ - if (!BN_lshift1(p,q)) goto err; - if (!BN_add_word(p,1)) goto err; - - loop: for (i=1; i<NUMPRIMES; i++) - { - /* check that p and q are prime */ - /* check that for p and q - * gcd(p-1,primes) == 1 (except for 2) */ - if ( (BN_mod_word(p,(BN_ULONG)primes[i]) == 0) || - (BN_mod_word(q,(BN_ULONG)primes[i]) == 0)) - { - if (!BN_add(p,p,padd)) goto err; - if (!BN_add(q,q,qadd)) goto err; - goto loop; - } - } - ret=1; -err: - BN_CTX_end(ctx); - bn_check_top(p); - return(ret); - } + const BIGNUM *rem, BN_CTX *ctx) +{ + int i, ret = 0; + BIGNUM *t1, *qadd, *q; + + bits--; + BN_CTX_start(ctx); + t1 = BN_CTX_get(ctx); + q = BN_CTX_get(ctx); + qadd = BN_CTX_get(ctx); + if (qadd == NULL) + goto err; + + if (!BN_rshift1(qadd, padd)) + goto err; + + if (!BN_rand(q, bits, 0, 1)) + goto err; + + /* we need ((rnd-rem) % add) == 0 */ + if (!BN_mod(t1, q, qadd, ctx)) + goto err; + if (!BN_sub(q, q, t1)) + goto err; + if (rem == NULL) { + if (!BN_add_word(q, 1)) + goto err; + } else { + if (!BN_rshift1(t1, rem)) + goto err; + if (!BN_add(q, q, t1)) + goto err; + } + + /* we now have a random number 'rand' to test. */ + if (!BN_lshift1(p, q)) + goto err; + if (!BN_add_word(p, 1)) + goto err; + + loop:for (i = 1; i < NUMPRIMES; i++) { + /* check that p and q are prime */ + /* + * check that for p and q gcd(p-1,primes) == 1 (except for 2) + */ + if ((BN_mod_word(p, (BN_ULONG)primes[i]) == 0) || + (BN_mod_word(q, (BN_ULONG)primes[i]) == 0)) { + if (!BN_add(p, p, padd)) + goto err; + if (!BN_add(q, q, qadd)) + goto err; + goto loop; + } + } + ret = 1; + err: + BN_CTX_end(ctx); + bn_check_top(p); + return (ret); +} diff --git a/openssl/crypto/bn/bn_prime.h b/openssl/crypto/bn/bn_prime.h index 51d2194fe..5cf0de169 100644 --- a/openssl/crypto/bn/bn_prime.h +++ b/openssl/crypto/bn/bn_prime.h @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -57,271 +57,270 @@ */ #ifndef EIGHT_BIT -#define NUMPRIMES 2048 +# define NUMPRIMES 2048 typedef unsigned short prime_t; #else -#define NUMPRIMES 54 +# define NUMPRIMES 54 typedef unsigned char prime_t; #endif -static const prime_t primes[NUMPRIMES]= - { - 2, 3, 5, 7, 11, 13, 17, 19, - 23, 29, 31, 37, 41, 43, 47, 53, - 59, 61, 67, 71, 73, 79, 83, 89, - 97, 101, 103, 107, 109, 113, 127, 131, - 137, 139, 149, 151, 157, 163, 167, 173, - 179, 181, 191, 193, 197, 199, 211, 223, - 227, 229, 233, 239, 241, 251, +static const prime_t primes[NUMPRIMES] = { + 2, 3, 5, 7, 11, 13, 17, 19, + 23, 29, 31, 37, 41, 43, 47, 53, + 59, 61, 67, 71, 73, 79, 83, 89, + 97, 101, 103, 107, 109, 113, 127, 131, + 137, 139, 149, 151, 157, 163, 167, 173, + 179, 181, 191, 193, 197, 199, 211, 223, + 227, 229, 233, 239, 241, 251, #ifndef EIGHT_BIT - 257, 263, - 269, 271, 277, 281, 283, 293, 307, 311, - 313, 317, 331, 337, 347, 349, 353, 359, - 367, 373, 379, 383, 389, 397, 401, 409, - 419, 421, 431, 433, 439, 443, 449, 457, - 461, 463, 467, 479, 487, 491, 499, 503, - 509, 521, 523, 541, 547, 557, 563, 569, - 571, 577, 587, 593, 599, 601, 607, 613, - 617, 619, 631, 641, 643, 647, 653, 659, - 661, 673, 677, 683, 691, 701, 709, 719, - 727, 733, 739, 743, 751, 757, 761, 769, - 773, 787, 797, 809, 811, 821, 823, 827, - 829, 839, 853, 857, 859, 863, 877, 881, - 883, 887, 907, 911, 919, 929, 937, 941, - 947, 953, 967, 971, 977, 983, 991, 997, - 1009,1013,1019,1021,1031,1033,1039,1049, - 1051,1061,1063,1069,1087,1091,1093,1097, - 1103,1109,1117,1123,1129,1151,1153,1163, - 1171,1181,1187,1193,1201,1213,1217,1223, - 1229,1231,1237,1249,1259,1277,1279,1283, - 1289,1291,1297,1301,1303,1307,1319,1321, - 1327,1361,1367,1373,1381,1399,1409,1423, - 1427,1429,1433,1439,1447,1451,1453,1459, - 1471,1481,1483,1487,1489,1493,1499,1511, - 1523,1531,1543,1549,1553,1559,1567,1571, - 1579,1583,1597,1601,1607,1609,1613,1619, - 1621,1627,1637,1657,1663,1667,1669,1693, - 1697,1699,1709,1721,1723,1733,1741,1747, - 1753,1759,1777,1783,1787,1789,1801,1811, - 1823,1831,1847,1861,1867,1871,1873,1877, - 1879,1889,1901,1907,1913,1931,1933,1949, - 1951,1973,1979,1987,1993,1997,1999,2003, - 2011,2017,2027,2029,2039,2053,2063,2069, - 2081,2083,2087,2089,2099,2111,2113,2129, - 2131,2137,2141,2143,2153,2161,2179,2203, - 2207,2213,2221,2237,2239,2243,2251,2267, - 2269,2273,2281,2287,2293,2297,2309,2311, - 2333,2339,2341,2347,2351,2357,2371,2377, - 2381,2383,2389,2393,2399,2411,2417,2423, - 2437,2441,2447,2459,2467,2473,2477,2503, - 2521,2531,2539,2543,2549,2551,2557,2579, - 2591,2593,2609,2617,2621,2633,2647,2657, - 2659,2663,2671,2677,2683,2687,2689,2693, - 2699,2707,2711,2713,2719,2729,2731,2741, - 2749,2753,2767,2777,2789,2791,2797,2801, - 2803,2819,2833,2837,2843,2851,2857,2861, - 2879,2887,2897,2903,2909,2917,2927,2939, - 2953,2957,2963,2969,2971,2999,3001,3011, - 3019,3023,3037,3041,3049,3061,3067,3079, - 3083,3089,3109,3119,3121,3137,3163,3167, - 3169,3181,3187,3191,3203,3209,3217,3221, - 3229,3251,3253,3257,3259,3271,3299,3301, - 3307,3313,3319,3323,3329,3331,3343,3347, - 3359,3361,3371,3373,3389,3391,3407,3413, - 3433,3449,3457,3461,3463,3467,3469,3491, - 3499,3511,3517,3527,3529,3533,3539,3541, - 3547,3557,3559,3571,3581,3583,3593,3607, - 3613,3617,3623,3631,3637,3643,3659,3671, - 3673,3677,3691,3697,3701,3709,3719,3727, - 3733,3739,3761,3767,3769,3779,3793,3797, - 3803,3821,3823,3833,3847,3851,3853,3863, - 3877,3881,3889,3907,3911,3917,3919,3923, - 3929,3931,3943,3947,3967,3989,4001,4003, - 4007,4013,4019,4021,4027,4049,4051,4057, - 4073,4079,4091,4093,4099,4111,4127,4129, - 4133,4139,4153,4157,4159,4177,4201,4211, - 4217,4219,4229,4231,4241,4243,4253,4259, - 4261,4271,4273,4283,4289,4297,4327,4337, - 4339,4349,4357,4363,4373,4391,4397,4409, - 4421,4423,4441,4447,4451,4457,4463,4481, - 4483,4493,4507,4513,4517,4519,4523,4547, - 4549,4561,4567,4583,4591,4597,4603,4621, - 4637,4639,4643,4649,4651,4657,4663,4673, - 4679,4691,4703,4721,4723,4729,4733,4751, - 4759,4783,4787,4789,4793,4799,4801,4813, - 4817,4831,4861,4871,4877,4889,4903,4909, - 4919,4931,4933,4937,4943,4951,4957,4967, - 4969,4973,4987,4993,4999,5003,5009,5011, - 5021,5023,5039,5051,5059,5077,5081,5087, - 5099,5101,5107,5113,5119,5147,5153,5167, - 5171,5179,5189,5197,5209,5227,5231,5233, - 5237,5261,5273,5279,5281,5297,5303,5309, - 5323,5333,5347,5351,5381,5387,5393,5399, - 5407,5413,5417,5419,5431,5437,5441,5443, - 5449,5471,5477,5479,5483,5501,5503,5507, - 5519,5521,5527,5531,5557,5563,5569,5573, - 5581,5591,5623,5639,5641,5647,5651,5653, - 5657,5659,5669,5683,5689,5693,5701,5711, - 5717,5737,5741,5743,5749,5779,5783,5791, - 5801,5807,5813,5821,5827,5839,5843,5849, - 5851,5857,5861,5867,5869,5879,5881,5897, - 5903,5923,5927,5939,5953,5981,5987,6007, - 6011,6029,6037,6043,6047,6053,6067,6073, - 6079,6089,6091,6101,6113,6121,6131,6133, - 6143,6151,6163,6173,6197,6199,6203,6211, - 6217,6221,6229,6247,6257,6263,6269,6271, - 6277,6287,6299,6301,6311,6317,6323,6329, - 6337,6343,6353,6359,6361,6367,6373,6379, - 6389,6397,6421,6427,6449,6451,6469,6473, - 6481,6491,6521,6529,6547,6551,6553,6563, - 6569,6571,6577,6581,6599,6607,6619,6637, - 6653,6659,6661,6673,6679,6689,6691,6701, - 6703,6709,6719,6733,6737,6761,6763,6779, - 6781,6791,6793,6803,6823,6827,6829,6833, - 6841,6857,6863,6869,6871,6883,6899,6907, - 6911,6917,6947,6949,6959,6961,6967,6971, - 6977,6983,6991,6997,7001,7013,7019,7027, - 7039,7043,7057,7069,7079,7103,7109,7121, - 7127,7129,7151,7159,7177,7187,7193,7207, - 7211,7213,7219,7229,7237,7243,7247,7253, - 7283,7297,7307,7309,7321,7331,7333,7349, - 7351,7369,7393,7411,7417,7433,7451,7457, - 7459,7477,7481,7487,7489,7499,7507,7517, - 7523,7529,7537,7541,7547,7549,7559,7561, - 7573,7577,7583,7589,7591,7603,7607,7621, - 7639,7643,7649,7669,7673,7681,7687,7691, - 7699,7703,7717,7723,7727,7741,7753,7757, - 7759,7789,7793,7817,7823,7829,7841,7853, - 7867,7873,7877,7879,7883,7901,7907,7919, - 7927,7933,7937,7949,7951,7963,7993,8009, - 8011,8017,8039,8053,8059,8069,8081,8087, - 8089,8093,8101,8111,8117,8123,8147,8161, - 8167,8171,8179,8191,8209,8219,8221,8231, - 8233,8237,8243,8263,8269,8273,8287,8291, - 8293,8297,8311,8317,8329,8353,8363,8369, - 8377,8387,8389,8419,8423,8429,8431,8443, - 8447,8461,8467,8501,8513,8521,8527,8537, - 8539,8543,8563,8573,8581,8597,8599,8609, - 8623,8627,8629,8641,8647,8663,8669,8677, - 8681,8689,8693,8699,8707,8713,8719,8731, - 8737,8741,8747,8753,8761,8779,8783,8803, - 8807,8819,8821,8831,8837,8839,8849,8861, - 8863,8867,8887,8893,8923,8929,8933,8941, - 8951,8963,8969,8971,8999,9001,9007,9011, - 9013,9029,9041,9043,9049,9059,9067,9091, - 9103,9109,9127,9133,9137,9151,9157,9161, - 9173,9181,9187,9199,9203,9209,9221,9227, - 9239,9241,9257,9277,9281,9283,9293,9311, - 9319,9323,9337,9341,9343,9349,9371,9377, - 9391,9397,9403,9413,9419,9421,9431,9433, - 9437,9439,9461,9463,9467,9473,9479,9491, - 9497,9511,9521,9533,9539,9547,9551,9587, - 9601,9613,9619,9623,9629,9631,9643,9649, - 9661,9677,9679,9689,9697,9719,9721,9733, - 9739,9743,9749,9767,9769,9781,9787,9791, - 9803,9811,9817,9829,9833,9839,9851,9857, - 9859,9871,9883,9887,9901,9907,9923,9929, - 9931,9941,9949,9967,9973,10007,10009,10037, - 10039,10061,10067,10069,10079,10091,10093,10099, - 10103,10111,10133,10139,10141,10151,10159,10163, - 10169,10177,10181,10193,10211,10223,10243,10247, - 10253,10259,10267,10271,10273,10289,10301,10303, - 10313,10321,10331,10333,10337,10343,10357,10369, - 10391,10399,10427,10429,10433,10453,10457,10459, - 10463,10477,10487,10499,10501,10513,10529,10531, - 10559,10567,10589,10597,10601,10607,10613,10627, - 10631,10639,10651,10657,10663,10667,10687,10691, - 10709,10711,10723,10729,10733,10739,10753,10771, - 10781,10789,10799,10831,10837,10847,10853,10859, - 10861,10867,10883,10889,10891,10903,10909,10937, - 10939,10949,10957,10973,10979,10987,10993,11003, - 11027,11047,11057,11059,11069,11071,11083,11087, - 11093,11113,11117,11119,11131,11149,11159,11161, - 11171,11173,11177,11197,11213,11239,11243,11251, - 11257,11261,11273,11279,11287,11299,11311,11317, - 11321,11329,11351,11353,11369,11383,11393,11399, - 11411,11423,11437,11443,11447,11467,11471,11483, - 11489,11491,11497,11503,11519,11527,11549,11551, - 11579,11587,11593,11597,11617,11621,11633,11657, - 11677,11681,11689,11699,11701,11717,11719,11731, - 11743,11777,11779,11783,11789,11801,11807,11813, - 11821,11827,11831,11833,11839,11863,11867,11887, - 11897,11903,11909,11923,11927,11933,11939,11941, - 11953,11959,11969,11971,11981,11987,12007,12011, - 12037,12041,12043,12049,12071,12073,12097,12101, - 12107,12109,12113,12119,12143,12149,12157,12161, - 12163,12197,12203,12211,12227,12239,12241,12251, - 12253,12263,12269,12277,12281,12289,12301,12323, - 12329,12343,12347,12373,12377,12379,12391,12401, - 12409,12413,12421,12433,12437,12451,12457,12473, - 12479,12487,12491,12497,12503,12511,12517,12527, - 12539,12541,12547,12553,12569,12577,12583,12589, - 12601,12611,12613,12619,12637,12641,12647,12653, - 12659,12671,12689,12697,12703,12713,12721,12739, - 12743,12757,12763,12781,12791,12799,12809,12821, - 12823,12829,12841,12853,12889,12893,12899,12907, - 12911,12917,12919,12923,12941,12953,12959,12967, - 12973,12979,12983,13001,13003,13007,13009,13033, - 13037,13043,13049,13063,13093,13099,13103,13109, - 13121,13127,13147,13151,13159,13163,13171,13177, - 13183,13187,13217,13219,13229,13241,13249,13259, - 13267,13291,13297,13309,13313,13327,13331,13337, - 13339,13367,13381,13397,13399,13411,13417,13421, - 13441,13451,13457,13463,13469,13477,13487,13499, - 13513,13523,13537,13553,13567,13577,13591,13597, - 13613,13619,13627,13633,13649,13669,13679,13681, - 13687,13691,13693,13697,13709,13711,13721,13723, - 13729,13751,13757,13759,13763,13781,13789,13799, - 13807,13829,13831,13841,13859,13873,13877,13879, - 13883,13901,13903,13907,13913,13921,13931,13933, - 13963,13967,13997,13999,14009,14011,14029,14033, - 14051,14057,14071,14081,14083,14087,14107,14143, - 14149,14153,14159,14173,14177,14197,14207,14221, - 14243,14249,14251,14281,14293,14303,14321,14323, - 14327,14341,14347,14369,14387,14389,14401,14407, - 14411,14419,14423,14431,14437,14447,14449,14461, - 14479,14489,14503,14519,14533,14537,14543,14549, - 14551,14557,14561,14563,14591,14593,14621,14627, - 14629,14633,14639,14653,14657,14669,14683,14699, - 14713,14717,14723,14731,14737,14741,14747,14753, - 14759,14767,14771,14779,14783,14797,14813,14821, - 14827,14831,14843,14851,14867,14869,14879,14887, - 14891,14897,14923,14929,14939,14947,14951,14957, - 14969,14983,15013,15017,15031,15053,15061,15073, - 15077,15083,15091,15101,15107,15121,15131,15137, - 15139,15149,15161,15173,15187,15193,15199,15217, - 15227,15233,15241,15259,15263,15269,15271,15277, - 15287,15289,15299,15307,15313,15319,15329,15331, - 15349,15359,15361,15373,15377,15383,15391,15401, - 15413,15427,15439,15443,15451,15461,15467,15473, - 15493,15497,15511,15527,15541,15551,15559,15569, - 15581,15583,15601,15607,15619,15629,15641,15643, - 15647,15649,15661,15667,15671,15679,15683,15727, - 15731,15733,15737,15739,15749,15761,15767,15773, - 15787,15791,15797,15803,15809,15817,15823,15859, - 15877,15881,15887,15889,15901,15907,15913,15919, - 15923,15937,15959,15971,15973,15991,16001,16007, - 16033,16057,16061,16063,16067,16069,16073,16087, - 16091,16097,16103,16111,16127,16139,16141,16183, - 16187,16189,16193,16217,16223,16229,16231,16249, - 16253,16267,16273,16301,16319,16333,16339,16349, - 16361,16363,16369,16381,16411,16417,16421,16427, - 16433,16447,16451,16453,16477,16481,16487,16493, - 16519,16529,16547,16553,16561,16567,16573,16603, - 16607,16619,16631,16633,16649,16651,16657,16661, - 16673,16691,16693,16699,16703,16729,16741,16747, - 16759,16763,16787,16811,16823,16829,16831,16843, - 16871,16879,16883,16889,16901,16903,16921,16927, - 16931,16937,16943,16963,16979,16981,16987,16993, - 17011,17021,17027,17029,17033,17041,17047,17053, - 17077,17093,17099,17107,17117,17123,17137,17159, - 17167,17183,17189,17191,17203,17207,17209,17231, - 17239,17257,17291,17293,17299,17317,17321,17327, - 17333,17341,17351,17359,17377,17383,17387,17389, - 17393,17401,17417,17419,17431,17443,17449,17467, - 17471,17477,17483,17489,17491,17497,17509,17519, - 17539,17551,17569,17573,17579,17581,17597,17599, - 17609,17623,17627,17657,17659,17669,17681,17683, - 17707,17713,17729,17737,17747,17749,17761,17783, - 17789,17791,17807,17827,17837,17839,17851,17863, + 257, 263, + 269, 271, 277, 281, 283, 293, 307, 311, + 313, 317, 331, 337, 347, 349, 353, 359, + 367, 373, 379, 383, 389, 397, 401, 409, + 419, 421, 431, 433, 439, 443, 449, 457, + 461, 463, 467, 479, 487, 491, 499, 503, + 509, 521, 523, 541, 547, 557, 563, 569, + 571, 577, 587, 593, 599, 601, 607, 613, + 617, 619, 631, 641, 643, 647, 653, 659, + 661, 673, 677, 683, 691, 701, 709, 719, + 727, 733, 739, 743, 751, 757, 761, 769, + 773, 787, 797, 809, 811, 821, 823, 827, + 829, 839, 853, 857, 859, 863, 877, 881, + 883, 887, 907, 911, 919, 929, 937, 941, + 947, 953, 967, 971, 977, 983, 991, 997, + 1009, 1013, 1019, 1021, 1031, 1033, 1039, 1049, + 1051, 1061, 1063, 1069, 1087, 1091, 1093, 1097, + 1103, 1109, 1117, 1123, 1129, 1151, 1153, 1163, + 1171, 1181, 1187, 1193, 1201, 1213, 1217, 1223, + 1229, 1231, 1237, 1249, 1259, 1277, 1279, 1283, + 1289, 1291, 1297, 1301, 1303, 1307, 1319, 1321, + 1327, 1361, 1367, 1373, 1381, 1399, 1409, 1423, + 1427, 1429, 1433, 1439, 1447, 1451, 1453, 1459, + 1471, 1481, 1483, 1487, 1489, 1493, 1499, 1511, + 1523, 1531, 1543, 1549, 1553, 1559, 1567, 1571, + 1579, 1583, 1597, 1601, 1607, 1609, 1613, 1619, + 1621, 1627, 1637, 1657, 1663, 1667, 1669, 1693, + 1697, 1699, 1709, 1721, 1723, 1733, 1741, 1747, + 1753, 1759, 1777, 1783, 1787, 1789, 1801, 1811, + 1823, 1831, 1847, 1861, 1867, 1871, 1873, 1877, + 1879, 1889, 1901, 1907, 1913, 1931, 1933, 1949, + 1951, 1973, 1979, 1987, 1993, 1997, 1999, 2003, + 2011, 2017, 2027, 2029, 2039, 2053, 2063, 2069, + 2081, 2083, 2087, 2089, 2099, 2111, 2113, 2129, + 2131, 2137, 2141, 2143, 2153, 2161, 2179, 2203, + 2207, 2213, 2221, 2237, 2239, 2243, 2251, 2267, + 2269, 2273, 2281, 2287, 2293, 2297, 2309, 2311, + 2333, 2339, 2341, 2347, 2351, 2357, 2371, 2377, + 2381, 2383, 2389, 2393, 2399, 2411, 2417, 2423, + 2437, 2441, 2447, 2459, 2467, 2473, 2477, 2503, + 2521, 2531, 2539, 2543, 2549, 2551, 2557, 2579, + 2591, 2593, 2609, 2617, 2621, 2633, 2647, 2657, + 2659, 2663, 2671, 2677, 2683, 2687, 2689, 2693, + 2699, 2707, 2711, 2713, 2719, 2729, 2731, 2741, + 2749, 2753, 2767, 2777, 2789, 2791, 2797, 2801, + 2803, 2819, 2833, 2837, 2843, 2851, 2857, 2861, + 2879, 2887, 2897, 2903, 2909, 2917, 2927, 2939, + 2953, 2957, 2963, 2969, 2971, 2999, 3001, 3011, + 3019, 3023, 3037, 3041, 3049, 3061, 3067, 3079, + 3083, 3089, 3109, 3119, 3121, 3137, 3163, 3167, + 3169, 3181, 3187, 3191, 3203, 3209, 3217, 3221, + 3229, 3251, 3253, 3257, 3259, 3271, 3299, 3301, + 3307, 3313, 3319, 3323, 3329, 3331, 3343, 3347, + 3359, 3361, 3371, 3373, 3389, 3391, 3407, 3413, + 3433, 3449, 3457, 3461, 3463, 3467, 3469, 3491, + 3499, 3511, 3517, 3527, 3529, 3533, 3539, 3541, + 3547, 3557, 3559, 3571, 3581, 3583, 3593, 3607, + 3613, 3617, 3623, 3631, 3637, 3643, 3659, 3671, + 3673, 3677, 3691, 3697, 3701, 3709, 3719, 3727, + 3733, 3739, 3761, 3767, 3769, 3779, 3793, 3797, + 3803, 3821, 3823, 3833, 3847, 3851, 3853, 3863, + 3877, 3881, 3889, 3907, 3911, 3917, 3919, 3923, + 3929, 3931, 3943, 3947, 3967, 3989, 4001, 4003, + 4007, 4013, 4019, 4021, 4027, 4049, 4051, 4057, + 4073, 4079, 4091, 4093, 4099, 4111, 4127, 4129, + 4133, 4139, 4153, 4157, 4159, 4177, 4201, 4211, + 4217, 4219, 4229, 4231, 4241, 4243, 4253, 4259, + 4261, 4271, 4273, 4283, 4289, 4297, 4327, 4337, + 4339, 4349, 4357, 4363, 4373, 4391, 4397, 4409, + 4421, 4423, 4441, 4447, 4451, 4457, 4463, 4481, + 4483, 4493, 4507, 4513, 4517, 4519, 4523, 4547, + 4549, 4561, 4567, 4583, 4591, 4597, 4603, 4621, + 4637, 4639, 4643, 4649, 4651, 4657, 4663, 4673, + 4679, 4691, 4703, 4721, 4723, 4729, 4733, 4751, + 4759, 4783, 4787, 4789, 4793, 4799, 4801, 4813, + 4817, 4831, 4861, 4871, 4877, 4889, 4903, 4909, + 4919, 4931, 4933, 4937, 4943, 4951, 4957, 4967, + 4969, 4973, 4987, 4993, 4999, 5003, 5009, 5011, + 5021, 5023, 5039, 5051, 5059, 5077, 5081, 5087, + 5099, 5101, 5107, 5113, 5119, 5147, 5153, 5167, + 5171, 5179, 5189, 5197, 5209, 5227, 5231, 5233, + 5237, 5261, 5273, 5279, 5281, 5297, 5303, 5309, + 5323, 5333, 5347, 5351, 5381, 5387, 5393, 5399, + 5407, 5413, 5417, 5419, 5431, 5437, 5441, 5443, + 5449, 5471, 5477, 5479, 5483, 5501, 5503, 5507, + 5519, 5521, 5527, 5531, 5557, 5563, 5569, 5573, + 5581, 5591, 5623, 5639, 5641, 5647, 5651, 5653, + 5657, 5659, 5669, 5683, 5689, 5693, 5701, 5711, + 5717, 5737, 5741, 5743, 5749, 5779, 5783, 5791, + 5801, 5807, 5813, 5821, 5827, 5839, 5843, 5849, + 5851, 5857, 5861, 5867, 5869, 5879, 5881, 5897, + 5903, 5923, 5927, 5939, 5953, 5981, 5987, 6007, + 6011, 6029, 6037, 6043, 6047, 6053, 6067, 6073, + 6079, 6089, 6091, 6101, 6113, 6121, 6131, 6133, + 6143, 6151, 6163, 6173, 6197, 6199, 6203, 6211, + 6217, 6221, 6229, 6247, 6257, 6263, 6269, 6271, + 6277, 6287, 6299, 6301, 6311, 6317, 6323, 6329, + 6337, 6343, 6353, 6359, 6361, 6367, 6373, 6379, + 6389, 6397, 6421, 6427, 6449, 6451, 6469, 6473, + 6481, 6491, 6521, 6529, 6547, 6551, 6553, 6563, + 6569, 6571, 6577, 6581, 6599, 6607, 6619, 6637, + 6653, 6659, 6661, 6673, 6679, 6689, 6691, 6701, + 6703, 6709, 6719, 6733, 6737, 6761, 6763, 6779, + 6781, 6791, 6793, 6803, 6823, 6827, 6829, 6833, + 6841, 6857, 6863, 6869, 6871, 6883, 6899, 6907, + 6911, 6917, 6947, 6949, 6959, 6961, 6967, 6971, + 6977, 6983, 6991, 6997, 7001, 7013, 7019, 7027, + 7039, 7043, 7057, 7069, 7079, 7103, 7109, 7121, + 7127, 7129, 7151, 7159, 7177, 7187, 7193, 7207, + 7211, 7213, 7219, 7229, 7237, 7243, 7247, 7253, + 7283, 7297, 7307, 7309, 7321, 7331, 7333, 7349, + 7351, 7369, 7393, 7411, 7417, 7433, 7451, 7457, + 7459, 7477, 7481, 7487, 7489, 7499, 7507, 7517, + 7523, 7529, 7537, 7541, 7547, 7549, 7559, 7561, + 7573, 7577, 7583, 7589, 7591, 7603, 7607, 7621, + 7639, 7643, 7649, 7669, 7673, 7681, 7687, 7691, + 7699, 7703, 7717, 7723, 7727, 7741, 7753, 7757, + 7759, 7789, 7793, 7817, 7823, 7829, 7841, 7853, + 7867, 7873, 7877, 7879, 7883, 7901, 7907, 7919, + 7927, 7933, 7937, 7949, 7951, 7963, 7993, 8009, + 8011, 8017, 8039, 8053, 8059, 8069, 8081, 8087, + 8089, 8093, 8101, 8111, 8117, 8123, 8147, 8161, + 8167, 8171, 8179, 8191, 8209, 8219, 8221, 8231, + 8233, 8237, 8243, 8263, 8269, 8273, 8287, 8291, + 8293, 8297, 8311, 8317, 8329, 8353, 8363, 8369, + 8377, 8387, 8389, 8419, 8423, 8429, 8431, 8443, + 8447, 8461, 8467, 8501, 8513, 8521, 8527, 8537, + 8539, 8543, 8563, 8573, 8581, 8597, 8599, 8609, + 8623, 8627, 8629, 8641, 8647, 8663, 8669, 8677, + 8681, 8689, 8693, 8699, 8707, 8713, 8719, 8731, + 8737, 8741, 8747, 8753, 8761, 8779, 8783, 8803, + 8807, 8819, 8821, 8831, 8837, 8839, 8849, 8861, + 8863, 8867, 8887, 8893, 8923, 8929, 8933, 8941, + 8951, 8963, 8969, 8971, 8999, 9001, 9007, 9011, + 9013, 9029, 9041, 9043, 9049, 9059, 9067, 9091, + 9103, 9109, 9127, 9133, 9137, 9151, 9157, 9161, + 9173, 9181, 9187, 9199, 9203, 9209, 9221, 9227, + 9239, 9241, 9257, 9277, 9281, 9283, 9293, 9311, + 9319, 9323, 9337, 9341, 9343, 9349, 9371, 9377, + 9391, 9397, 9403, 9413, 9419, 9421, 9431, 9433, + 9437, 9439, 9461, 9463, 9467, 9473, 9479, 9491, + 9497, 9511, 9521, 9533, 9539, 9547, 9551, 9587, + 9601, 9613, 9619, 9623, 9629, 9631, 9643, 9649, + 9661, 9677, 9679, 9689, 9697, 9719, 9721, 9733, + 9739, 9743, 9749, 9767, 9769, 9781, 9787, 9791, + 9803, 9811, 9817, 9829, 9833, 9839, 9851, 9857, + 9859, 9871, 9883, 9887, 9901, 9907, 9923, 9929, + 9931, 9941, 9949, 9967, 9973, 10007, 10009, 10037, + 10039, 10061, 10067, 10069, 10079, 10091, 10093, 10099, + 10103, 10111, 10133, 10139, 10141, 10151, 10159, 10163, + 10169, 10177, 10181, 10193, 10211, 10223, 10243, 10247, + 10253, 10259, 10267, 10271, 10273, 10289, 10301, 10303, + 10313, 10321, 10331, 10333, 10337, 10343, 10357, 10369, + 10391, 10399, 10427, 10429, 10433, 10453, 10457, 10459, + 10463, 10477, 10487, 10499, 10501, 10513, 10529, 10531, + 10559, 10567, 10589, 10597, 10601, 10607, 10613, 10627, + 10631, 10639, 10651, 10657, 10663, 10667, 10687, 10691, + 10709, 10711, 10723, 10729, 10733, 10739, 10753, 10771, + 10781, 10789, 10799, 10831, 10837, 10847, 10853, 10859, + 10861, 10867, 10883, 10889, 10891, 10903, 10909, 10937, + 10939, 10949, 10957, 10973, 10979, 10987, 10993, 11003, + 11027, 11047, 11057, 11059, 11069, 11071, 11083, 11087, + 11093, 11113, 11117, 11119, 11131, 11149, 11159, 11161, + 11171, 11173, 11177, 11197, 11213, 11239, 11243, 11251, + 11257, 11261, 11273, 11279, 11287, 11299, 11311, 11317, + 11321, 11329, 11351, 11353, 11369, 11383, 11393, 11399, + 11411, 11423, 11437, 11443, 11447, 11467, 11471, 11483, + 11489, 11491, 11497, 11503, 11519, 11527, 11549, 11551, + 11579, 11587, 11593, 11597, 11617, 11621, 11633, 11657, + 11677, 11681, 11689, 11699, 11701, 11717, 11719, 11731, + 11743, 11777, 11779, 11783, 11789, 11801, 11807, 11813, + 11821, 11827, 11831, 11833, 11839, 11863, 11867, 11887, + 11897, 11903, 11909, 11923, 11927, 11933, 11939, 11941, + 11953, 11959, 11969, 11971, 11981, 11987, 12007, 12011, + 12037, 12041, 12043, 12049, 12071, 12073, 12097, 12101, + 12107, 12109, 12113, 12119, 12143, 12149, 12157, 12161, + 12163, 12197, 12203, 12211, 12227, 12239, 12241, 12251, + 12253, 12263, 12269, 12277, 12281, 12289, 12301, 12323, + 12329, 12343, 12347, 12373, 12377, 12379, 12391, 12401, + 12409, 12413, 12421, 12433, 12437, 12451, 12457, 12473, + 12479, 12487, 12491, 12497, 12503, 12511, 12517, 12527, + 12539, 12541, 12547, 12553, 12569, 12577, 12583, 12589, + 12601, 12611, 12613, 12619, 12637, 12641, 12647, 12653, + 12659, 12671, 12689, 12697, 12703, 12713, 12721, 12739, + 12743, 12757, 12763, 12781, 12791, 12799, 12809, 12821, + 12823, 12829, 12841, 12853, 12889, 12893, 12899, 12907, + 12911, 12917, 12919, 12923, 12941, 12953, 12959, 12967, + 12973, 12979, 12983, 13001, 13003, 13007, 13009, 13033, + 13037, 13043, 13049, 13063, 13093, 13099, 13103, 13109, + 13121, 13127, 13147, 13151, 13159, 13163, 13171, 13177, + 13183, 13187, 13217, 13219, 13229, 13241, 13249, 13259, + 13267, 13291, 13297, 13309, 13313, 13327, 13331, 13337, + 13339, 13367, 13381, 13397, 13399, 13411, 13417, 13421, + 13441, 13451, 13457, 13463, 13469, 13477, 13487, 13499, + 13513, 13523, 13537, 13553, 13567, 13577, 13591, 13597, + 13613, 13619, 13627, 13633, 13649, 13669, 13679, 13681, + 13687, 13691, 13693, 13697, 13709, 13711, 13721, 13723, + 13729, 13751, 13757, 13759, 13763, 13781, 13789, 13799, + 13807, 13829, 13831, 13841, 13859, 13873, 13877, 13879, + 13883, 13901, 13903, 13907, 13913, 13921, 13931, 13933, + 13963, 13967, 13997, 13999, 14009, 14011, 14029, 14033, + 14051, 14057, 14071, 14081, 14083, 14087, 14107, 14143, + 14149, 14153, 14159, 14173, 14177, 14197, 14207, 14221, + 14243, 14249, 14251, 14281, 14293, 14303, 14321, 14323, + 14327, 14341, 14347, 14369, 14387, 14389, 14401, 14407, + 14411, 14419, 14423, 14431, 14437, 14447, 14449, 14461, + 14479, 14489, 14503, 14519, 14533, 14537, 14543, 14549, + 14551, 14557, 14561, 14563, 14591, 14593, 14621, 14627, + 14629, 14633, 14639, 14653, 14657, 14669, 14683, 14699, + 14713, 14717, 14723, 14731, 14737, 14741, 14747, 14753, + 14759, 14767, 14771, 14779, 14783, 14797, 14813, 14821, + 14827, 14831, 14843, 14851, 14867, 14869, 14879, 14887, + 14891, 14897, 14923, 14929, 14939, 14947, 14951, 14957, + 14969, 14983, 15013, 15017, 15031, 15053, 15061, 15073, + 15077, 15083, 15091, 15101, 15107, 15121, 15131, 15137, + 15139, 15149, 15161, 15173, 15187, 15193, 15199, 15217, + 15227, 15233, 15241, 15259, 15263, 15269, 15271, 15277, + 15287, 15289, 15299, 15307, 15313, 15319, 15329, 15331, + 15349, 15359, 15361, 15373, 15377, 15383, 15391, 15401, + 15413, 15427, 15439, 15443, 15451, 15461, 15467, 15473, + 15493, 15497, 15511, 15527, 15541, 15551, 15559, 15569, + 15581, 15583, 15601, 15607, 15619, 15629, 15641, 15643, + 15647, 15649, 15661, 15667, 15671, 15679, 15683, 15727, + 15731, 15733, 15737, 15739, 15749, 15761, 15767, 15773, + 15787, 15791, 15797, 15803, 15809, 15817, 15823, 15859, + 15877, 15881, 15887, 15889, 15901, 15907, 15913, 15919, + 15923, 15937, 15959, 15971, 15973, 15991, 16001, 16007, + 16033, 16057, 16061, 16063, 16067, 16069, 16073, 16087, + 16091, 16097, 16103, 16111, 16127, 16139, 16141, 16183, + 16187, 16189, 16193, 16217, 16223, 16229, 16231, 16249, + 16253, 16267, 16273, 16301, 16319, 16333, 16339, 16349, + 16361, 16363, 16369, 16381, 16411, 16417, 16421, 16427, + 16433, 16447, 16451, 16453, 16477, 16481, 16487, 16493, + 16519, 16529, 16547, 16553, 16561, 16567, 16573, 16603, + 16607, 16619, 16631, 16633, 16649, 16651, 16657, 16661, + 16673, 16691, 16693, 16699, 16703, 16729, 16741, 16747, + 16759, 16763, 16787, 16811, 16823, 16829, 16831, 16843, + 16871, 16879, 16883, 16889, 16901, 16903, 16921, 16927, + 16931, 16937, 16943, 16963, 16979, 16981, 16987, 16993, + 17011, 17021, 17027, 17029, 17033, 17041, 17047, 17053, + 17077, 17093, 17099, 17107, 17117, 17123, 17137, 17159, + 17167, 17183, 17189, 17191, 17203, 17207, 17209, 17231, + 17239, 17257, 17291, 17293, 17299, 17317, 17321, 17327, + 17333, 17341, 17351, 17359, 17377, 17383, 17387, 17389, + 17393, 17401, 17417, 17419, 17431, 17443, 17449, 17467, + 17471, 17477, 17483, 17489, 17491, 17497, 17509, 17519, + 17539, 17551, 17569, 17573, 17579, 17581, 17597, 17599, + 17609, 17623, 17627, 17657, 17659, 17669, 17681, 17683, + 17707, 17713, 17729, 17737, 17747, 17749, 17761, 17783, + 17789, 17791, 17807, 17827, 17837, 17839, 17851, 17863, #endif - }; +}; diff --git a/openssl/crypto/bn/bn_print.c b/openssl/crypto/bn/bn_print.c index 1743b6a7e..4dcaae32b 100644 --- a/openssl/crypto/bn/bn_print.c +++ b/openssl/crypto/bn/bn_print.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -62,317 +62,322 @@ #include <openssl/buffer.h> #include "bn_lcl.h" -static const char Hex[]="0123456789ABCDEF"; +static const char Hex[] = "0123456789ABCDEF"; /* Must 'OPENSSL_free' the returned data */ char *BN_bn2hex(const BIGNUM *a) - { - int i,j,v,z=0; - char *buf; - char *p; - - buf=(char *)OPENSSL_malloc(a->top*BN_BYTES*2+2); - if (buf == NULL) - { - BNerr(BN_F_BN_BN2HEX,ERR_R_MALLOC_FAILURE); - goto err; - } - p=buf; - if (a->neg) *(p++)='-'; - if (BN_is_zero(a)) *(p++)='0'; - for (i=a->top-1; i >=0; i--) - { - for (j=BN_BITS2-8; j >= 0; j-=8) - { - /* strip leading zeros */ - v=((int)(a->d[i]>>(long)j))&0xff; - if (z || (v != 0)) - { - *(p++)=Hex[v>>4]; - *(p++)=Hex[v&0x0f]; - z=1; - } - } - } - *p='\0'; -err: - return(buf); - } +{ + int i, j, v, z = 0; + char *buf; + char *p; + + buf = (char *)OPENSSL_malloc(a->top * BN_BYTES * 2 + 2); + if (buf == NULL) { + BNerr(BN_F_BN_BN2HEX, ERR_R_MALLOC_FAILURE); + goto err; + } + p = buf; + if (a->neg) + *(p++) = '-'; + if (BN_is_zero(a)) + *(p++) = '0'; + for (i = a->top - 1; i >= 0; i--) { + for (j = BN_BITS2 - 8; j >= 0; j -= 8) { + /* strip leading zeros */ + v = ((int)(a->d[i] >> (long)j)) & 0xff; + if (z || (v != 0)) { + *(p++) = Hex[v >> 4]; + *(p++) = Hex[v & 0x0f]; + z = 1; + } + } + } + *p = '\0'; + err: + return (buf); +} /* Must 'OPENSSL_free' the returned data */ char *BN_bn2dec(const BIGNUM *a) - { - int i=0,num, ok = 0; - char *buf=NULL; - char *p; - BIGNUM *t=NULL; - BN_ULONG *bn_data=NULL,*lp; - - /* get an upper bound for the length of the decimal integer - * num <= (BN_num_bits(a) + 1) * log(2) - * <= 3 * BN_num_bits(a) * 0.1001 + log(2) + 1 (rounding error) - * <= BN_num_bits(a)/10 + BN_num_bits/1000 + 1 + 1 - */ - i=BN_num_bits(a)*3; - num=(i/10+i/1000+1)+1; - bn_data=(BN_ULONG *)OPENSSL_malloc((num/BN_DEC_NUM+1)*sizeof(BN_ULONG)); - buf=(char *)OPENSSL_malloc(num+3); - if ((buf == NULL) || (bn_data == NULL)) - { - BNerr(BN_F_BN_BN2DEC,ERR_R_MALLOC_FAILURE); - goto err; - } - if ((t=BN_dup(a)) == NULL) goto err; +{ + int i = 0, num, ok = 0; + char *buf = NULL; + char *p; + BIGNUM *t = NULL; + BN_ULONG *bn_data = NULL, *lp; + + /*- + * get an upper bound for the length of the decimal integer + * num <= (BN_num_bits(a) + 1) * log(2) + * <= 3 * BN_num_bits(a) * 0.1001 + log(2) + 1 (rounding error) + * <= BN_num_bits(a)/10 + BN_num_bits/1000 + 1 + 1 + */ + i = BN_num_bits(a) * 3; + num = (i / 10 + i / 1000 + 1) + 1; + bn_data = + (BN_ULONG *)OPENSSL_malloc((num / BN_DEC_NUM + 1) * sizeof(BN_ULONG)); + buf = (char *)OPENSSL_malloc(num + 3); + if ((buf == NULL) || (bn_data == NULL)) { + BNerr(BN_F_BN_BN2DEC, ERR_R_MALLOC_FAILURE); + goto err; + } + if ((t = BN_dup(a)) == NULL) + goto err; #define BUF_REMAIN (num+3 - (size_t)(p - buf)) - p=buf; - lp=bn_data; - if (BN_is_zero(t)) - { - *(p++)='0'; - *(p++)='\0'; - } - else - { - if (BN_is_negative(t)) - *p++ = '-'; - - i=0; - while (!BN_is_zero(t)) - { - *lp=BN_div_word(t,BN_DEC_CONV); - lp++; - } - lp--; - /* We now have a series of blocks, BN_DEC_NUM chars - * in length, where the last one needs truncation. - * The blocks need to be reversed in order. */ - BIO_snprintf(p,BUF_REMAIN,BN_DEC_FMT1,*lp); - while (*p) p++; - while (lp != bn_data) - { - lp--; - BIO_snprintf(p,BUF_REMAIN,BN_DEC_FMT2,*lp); - while (*p) p++; - } - } - ok = 1; -err: - if (bn_data != NULL) OPENSSL_free(bn_data); - if (t != NULL) BN_free(t); - if (!ok && buf) - { - OPENSSL_free(buf); - buf = NULL; - } - - return(buf); - } + p = buf; + lp = bn_data; + if (BN_is_zero(t)) { + *(p++) = '0'; + *(p++) = '\0'; + } else { + if (BN_is_negative(t)) + *p++ = '-'; + + i = 0; + while (!BN_is_zero(t)) { + *lp = BN_div_word(t, BN_DEC_CONV); + lp++; + } + lp--; + /* + * We now have a series of blocks, BN_DEC_NUM chars in length, where + * the last one needs truncation. The blocks need to be reversed in + * order. + */ + BIO_snprintf(p, BUF_REMAIN, BN_DEC_FMT1, *lp); + while (*p) + p++; + while (lp != bn_data) { + lp--; + BIO_snprintf(p, BUF_REMAIN, BN_DEC_FMT2, *lp); + while (*p) + p++; + } + } + ok = 1; + err: + if (bn_data != NULL) + OPENSSL_free(bn_data); + if (t != NULL) + BN_free(t); + if (!ok && buf) { + OPENSSL_free(buf); + buf = NULL; + } + + return (buf); +} int BN_hex2bn(BIGNUM **bn, const char *a) - { - BIGNUM *ret=NULL; - BN_ULONG l=0; - int neg=0,h,m,i,j,k,c; - int num; - - if ((a == NULL) || (*a == '\0')) return(0); - - if (*a == '-') { neg=1; a++; } - - for (i=0; isxdigit((unsigned char) a[i]); i++) - ; - - num=i+neg; - if (bn == NULL) return(num); - - /* a is the start of the hex digits, and it is 'i' long */ - if (*bn == NULL) - { - if ((ret=BN_new()) == NULL) return(0); - } - else - { - ret= *bn; - BN_zero(ret); - } - - /* i is the number of hex digests; */ - if (bn_expand(ret,i*4) == NULL) goto err; - - j=i; /* least significant 'hex' */ - m=0; - h=0; - while (j > 0) - { - m=((BN_BYTES*2) <= j)?(BN_BYTES*2):j; - l=0; - for (;;) - { - c=a[j-m]; - if ((c >= '0') && (c <= '9')) k=c-'0'; - else if ((c >= 'a') && (c <= 'f')) k=c-'a'+10; - else if ((c >= 'A') && (c <= 'F')) k=c-'A'+10; - else k=0; /* paranoia */ - l=(l<<4)|k; - - if (--m <= 0) - { - ret->d[h++]=l; - break; - } - } - j-=(BN_BYTES*2); - } - ret->top=h; - bn_correct_top(ret); - ret->neg=neg; - - *bn=ret; - bn_check_top(ret); - return(num); -err: - if (*bn == NULL) BN_free(ret); - return(0); - } +{ + BIGNUM *ret = NULL; + BN_ULONG l = 0; + int neg = 0, h, m, i, j, k, c; + int num; + + if ((a == NULL) || (*a == '\0')) + return (0); + + if (*a == '-') { + neg = 1; + a++; + } + + for (i = 0; isxdigit((unsigned char)a[i]); i++) ; + + num = i + neg; + if (bn == NULL) + return (num); + + /* a is the start of the hex digits, and it is 'i' long */ + if (*bn == NULL) { + if ((ret = BN_new()) == NULL) + return (0); + } else { + ret = *bn; + BN_zero(ret); + } + + /* i is the number of hex digests; */ + if (bn_expand(ret, i * 4) == NULL) + goto err; + + j = i; /* least significant 'hex' */ + m = 0; + h = 0; + while (j > 0) { + m = ((BN_BYTES * 2) <= j) ? (BN_BYTES * 2) : j; + l = 0; + for (;;) { + c = a[j - m]; + if ((c >= '0') && (c <= '9')) + k = c - '0'; + else if ((c >= 'a') && (c <= 'f')) + k = c - 'a' + 10; + else if ((c >= 'A') && (c <= 'F')) + k = c - 'A' + 10; + else + k = 0; /* paranoia */ + l = (l << 4) | k; + + if (--m <= 0) { + ret->d[h++] = l; + break; + } + } + j -= (BN_BYTES * 2); + } + ret->top = h; + bn_correct_top(ret); + ret->neg = neg; + + *bn = ret; + bn_check_top(ret); + return (num); + err: + if (*bn == NULL) + BN_free(ret); + return (0); +} int BN_dec2bn(BIGNUM **bn, const char *a) - { - BIGNUM *ret=NULL; - BN_ULONG l=0; - int neg=0,i,j; - int num; - - if ((a == NULL) || (*a == '\0')) return(0); - if (*a == '-') { neg=1; a++; } - - for (i=0; isdigit((unsigned char) a[i]); i++) - ; - - num=i+neg; - if (bn == NULL) return(num); - - /* a is the start of the digits, and it is 'i' long. - * We chop it into BN_DEC_NUM digits at a time */ - if (*bn == NULL) - { - if ((ret=BN_new()) == NULL) return(0); - } - else - { - ret= *bn; - BN_zero(ret); - } - - /* i is the number of digests, a bit of an over expand; */ - if (bn_expand(ret,i*4) == NULL) goto err; - - j=BN_DEC_NUM-(i%BN_DEC_NUM); - if (j == BN_DEC_NUM) j=0; - l=0; - while (*a) - { - l*=10; - l+= *a-'0'; - a++; - if (++j == BN_DEC_NUM) - { - BN_mul_word(ret,BN_DEC_CONV); - BN_add_word(ret,l); - l=0; - j=0; - } - } - ret->neg=neg; - - bn_correct_top(ret); - *bn=ret; - bn_check_top(ret); - return(num); -err: - if (*bn == NULL) BN_free(ret); - return(0); - } +{ + BIGNUM *ret = NULL; + BN_ULONG l = 0; + int neg = 0, i, j; + int num; + + if ((a == NULL) || (*a == '\0')) + return (0); + if (*a == '-') { + neg = 1; + a++; + } + + for (i = 0; isdigit((unsigned char)a[i]); i++) ; + + num = i + neg; + if (bn == NULL) + return (num); + + /* + * a is the start of the digits, and it is 'i' long. We chop it into + * BN_DEC_NUM digits at a time + */ + if (*bn == NULL) { + if ((ret = BN_new()) == NULL) + return (0); + } else { + ret = *bn; + BN_zero(ret); + } + + /* i is the number of digests, a bit of an over expand; */ + if (bn_expand(ret, i * 4) == NULL) + goto err; + + j = BN_DEC_NUM - (i % BN_DEC_NUM); + if (j == BN_DEC_NUM) + j = 0; + l = 0; + while (*a) { + l *= 10; + l += *a - '0'; + a++; + if (++j == BN_DEC_NUM) { + BN_mul_word(ret, BN_DEC_CONV); + BN_add_word(ret, l); + l = 0; + j = 0; + } + } + ret->neg = neg; + + bn_correct_top(ret); + *bn = ret; + bn_check_top(ret); + return (num); + err: + if (*bn == NULL) + BN_free(ret); + return (0); +} int BN_asc2bn(BIGNUM **bn, const char *a) - { - const char *p = a; - if (*p == '-') - p++; - - if (p[0] == '0' && (p[1] == 'X' || p[1] == 'x')) - { - if (!BN_hex2bn(bn, p + 2)) - return 0; - } - else - { - if (!BN_dec2bn(bn, p)) - return 0; - } - if (*a == '-') - (*bn)->neg = 1; - return 1; - } +{ + const char *p = a; + if (*p == '-') + p++; + + if (p[0] == '0' && (p[1] == 'X' || p[1] == 'x')) { + if (!BN_hex2bn(bn, p + 2)) + return 0; + } else { + if (!BN_dec2bn(bn, p)) + return 0; + } + if (*a == '-') + (*bn)->neg = 1; + return 1; +} #ifndef OPENSSL_NO_BIO -#ifndef OPENSSL_NO_FP_API +# ifndef OPENSSL_NO_FP_API int BN_print_fp(FILE *fp, const BIGNUM *a) - { - BIO *b; - int ret; - - if ((b=BIO_new(BIO_s_file())) == NULL) - return(0); - BIO_set_fp(b,fp,BIO_NOCLOSE); - ret=BN_print(b,a); - BIO_free(b); - return(ret); - } -#endif +{ + BIO *b; + int ret; + + if ((b = BIO_new(BIO_s_file())) == NULL) + return (0); + BIO_set_fp(b, fp, BIO_NOCLOSE); + ret = BN_print(b, a); + BIO_free(b); + return (ret); +} +# endif int BN_print(BIO *bp, const BIGNUM *a) - { - int i,j,v,z=0; - int ret=0; - - if ((a->neg) && (BIO_write(bp,"-",1) != 1)) goto end; - if (BN_is_zero(a) && (BIO_write(bp,"0",1) != 1)) goto end; - for (i=a->top-1; i >=0; i--) - { - for (j=BN_BITS2-4; j >= 0; j-=4) - { - /* strip leading zeros */ - v=((int)(a->d[i]>>(long)j))&0x0f; - if (z || (v != 0)) - { - if (BIO_write(bp,&(Hex[v]),1) != 1) - goto end; - z=1; - } - } - } - ret=1; -end: - return(ret); - } +{ + int i, j, v, z = 0; + int ret = 0; + + if ((a->neg) && (BIO_write(bp, "-", 1) != 1)) + goto end; + if (BN_is_zero(a) && (BIO_write(bp, "0", 1) != 1)) + goto end; + for (i = a->top - 1; i >= 0; i--) { + for (j = BN_BITS2 - 4; j >= 0; j -= 4) { + /* strip leading zeros */ + v = ((int)(a->d[i] >> (long)j)) & 0x0f; + if (z || (v != 0)) { + if (BIO_write(bp, &(Hex[v]), 1) != 1) + goto end; + z = 1; + } + } + } + ret = 1; + end: + return (ret); +} #endif char *BN_options(void) - { - static int init=0; - static char data[16]; +{ + static int init = 0; + static char data[16]; - if (!init) - { - init++; + if (!init) { + init++; #ifdef BN_LLONG - BIO_snprintf(data,sizeof data,"bn(%d,%d)", - (int)sizeof(BN_ULLONG)*8,(int)sizeof(BN_ULONG)*8); + BIO_snprintf(data, sizeof data, "bn(%d,%d)", + (int)sizeof(BN_ULLONG) * 8, (int)sizeof(BN_ULONG) * 8); #else - BIO_snprintf(data,sizeof data,"bn(%d,%d)", - (int)sizeof(BN_ULONG)*8,(int)sizeof(BN_ULONG)*8); + BIO_snprintf(data, sizeof data, "bn(%d,%d)", + (int)sizeof(BN_ULONG) * 8, (int)sizeof(BN_ULONG) * 8); #endif - } - return(data); - } + } + return (data); +} diff --git a/openssl/crypto/bn/bn_rand.c b/openssl/crypto/bn/bn_rand.c index b376c28ff..7ac71ec8e 100644 --- a/openssl/crypto/bn/bn_rand.c +++ b/openssl/crypto/bn/bn_rand.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -63,7 +63,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -116,190 +116,174 @@ #include <openssl/rand.h> static int bnrand(int pseudorand, BIGNUM *rnd, int bits, int top, int bottom) - { - unsigned char *buf=NULL; - int ret=0,bit,bytes,mask; - time_t tim; +{ + unsigned char *buf = NULL; + int ret = 0, bit, bytes, mask; + time_t tim; - if (bits == 0) - { - BN_zero(rnd); - return 1; - } + if (bits == 0) { + BN_zero(rnd); + return 1; + } - bytes=(bits+7)/8; - bit=(bits-1)%8; - mask=0xff<<(bit+1); + bytes = (bits + 7) / 8; + bit = (bits - 1) % 8; + mask = 0xff << (bit + 1); - buf=(unsigned char *)OPENSSL_malloc(bytes); - if (buf == NULL) - { - BNerr(BN_F_BNRAND,ERR_R_MALLOC_FAILURE); - goto err; - } + buf = (unsigned char *)OPENSSL_malloc(bytes); + if (buf == NULL) { + BNerr(BN_F_BNRAND, ERR_R_MALLOC_FAILURE); + goto err; + } - /* make a random number and set the top and bottom bits */ - time(&tim); - RAND_add(&tim,sizeof(tim),0.0); + /* make a random number and set the top and bottom bits */ + time(&tim); + RAND_add(&tim, sizeof(tim), 0.0); - if (pseudorand) - { - if (RAND_pseudo_bytes(buf, bytes) == -1) - goto err; - } - else - { - if (RAND_bytes(buf, bytes) <= 0) - goto err; - } + if (pseudorand) { + if (RAND_pseudo_bytes(buf, bytes) == -1) + goto err; + } else { + if (RAND_bytes(buf, bytes) <= 0) + goto err; + } #if 1 - if (pseudorand == 2) - { - /* generate patterns that are more likely to trigger BN - library bugs */ - int i; - unsigned char c; + if (pseudorand == 2) { + /* + * generate patterns that are more likely to trigger BN library bugs + */ + int i; + unsigned char c; - for (i = 0; i < bytes; i++) - { - RAND_pseudo_bytes(&c, 1); - if (c >= 128 && i > 0) - buf[i] = buf[i-1]; - else if (c < 42) - buf[i] = 0; - else if (c < 84) - buf[i] = 255; - } - } + for (i = 0; i < bytes; i++) { + RAND_pseudo_bytes(&c, 1); + if (c >= 128 && i > 0) + buf[i] = buf[i - 1]; + else if (c < 42) + buf[i] = 0; + else if (c < 84) + buf[i] = 255; + } + } #endif - if (top != -1) - { - if (top) - { - if (bit == 0) - { - buf[0]=1; - buf[1]|=0x80; - } - else - { - buf[0]|=(3<<(bit-1)); - } - } - else - { - buf[0]|=(1<<bit); - } - } - buf[0] &= ~mask; - if (bottom) /* set bottom bit if requested */ - buf[bytes-1]|=1; - if (!BN_bin2bn(buf,bytes,rnd)) goto err; - ret=1; -err: - if (buf != NULL) - { - OPENSSL_cleanse(buf,bytes); - OPENSSL_free(buf); - } - bn_check_top(rnd); - return(ret); - } + if (top != -1) { + if (top) { + if (bit == 0) { + buf[0] = 1; + buf[1] |= 0x80; + } else { + buf[0] |= (3 << (bit - 1)); + } + } else { + buf[0] |= (1 << bit); + } + } + buf[0] &= ~mask; + if (bottom) /* set bottom bit if requested */ + buf[bytes - 1] |= 1; + if (!BN_bin2bn(buf, bytes, rnd)) + goto err; + ret = 1; + err: + if (buf != NULL) { + OPENSSL_cleanse(buf, bytes); + OPENSSL_free(buf); + } + bn_check_top(rnd); + return (ret); +} -int BN_rand(BIGNUM *rnd, int bits, int top, int bottom) - { - return bnrand(0, rnd, bits, top, bottom); - } +int BN_rand(BIGNUM *rnd, int bits, int top, int bottom) +{ + return bnrand(0, rnd, bits, top, bottom); +} -int BN_pseudo_rand(BIGNUM *rnd, int bits, int top, int bottom) - { - return bnrand(1, rnd, bits, top, bottom); - } +int BN_pseudo_rand(BIGNUM *rnd, int bits, int top, int bottom) +{ + return bnrand(1, rnd, bits, top, bottom); +} #if 1 -int BN_bntest_rand(BIGNUM *rnd, int bits, int top, int bottom) - { - return bnrand(2, rnd, bits, top, bottom); - } +int BN_bntest_rand(BIGNUM *rnd, int bits, int top, int bottom) +{ + return bnrand(2, rnd, bits, top, bottom); +} #endif - /* random number r: 0 <= r < range */ static int bn_rand_range(int pseudo, BIGNUM *r, const BIGNUM *range) - { - int (*bn_rand)(BIGNUM *, int, int, int) = pseudo ? BN_pseudo_rand : BN_rand; - int n; - int count = 100; +{ + int (*bn_rand) (BIGNUM *, int, int, int) = + pseudo ? BN_pseudo_rand : BN_rand; + int n; + int count = 100; - if (range->neg || BN_is_zero(range)) - { - BNerr(BN_F_BN_RAND_RANGE, BN_R_INVALID_RANGE); - return 0; - } + if (range->neg || BN_is_zero(range)) { + BNerr(BN_F_BN_RAND_RANGE, BN_R_INVALID_RANGE); + return 0; + } - n = BN_num_bits(range); /* n > 0 */ + n = BN_num_bits(range); /* n > 0 */ - /* BN_is_bit_set(range, n - 1) always holds */ + /* BN_is_bit_set(range, n - 1) always holds */ - if (n == 1) - BN_zero(r); - else if (!BN_is_bit_set(range, n - 2) && !BN_is_bit_set(range, n - 3)) - { - /* range = 100..._2, - * so 3*range (= 11..._2) is exactly one bit longer than range */ - do - { - if (!bn_rand(r, n + 1, -1, 0)) return 0; - /* If r < 3*range, use r := r MOD range - * (which is either r, r - range, or r - 2*range). - * Otherwise, iterate once more. - * Since 3*range = 11..._2, each iteration succeeds with - * probability >= .75. */ - if (BN_cmp(r ,range) >= 0) - { - if (!BN_sub(r, r, range)) return 0; - if (BN_cmp(r, range) >= 0) - if (!BN_sub(r, r, range)) return 0; - } + if (n == 1) + BN_zero(r); + else if (!BN_is_bit_set(range, n - 2) && !BN_is_bit_set(range, n - 3)) { + /* + * range = 100..._2, so 3*range (= 11..._2) is exactly one bit longer + * than range + */ + do { + if (!bn_rand(r, n + 1, -1, 0)) + return 0; + /* + * If r < 3*range, use r := r MOD range (which is either r, r - + * range, or r - 2*range). Otherwise, iterate once more. Since + * 3*range = 11..._2, each iteration succeeds with probability >= + * .75. + */ + if (BN_cmp(r, range) >= 0) { + if (!BN_sub(r, r, range)) + return 0; + if (BN_cmp(r, range) >= 0) + if (!BN_sub(r, r, range)) + return 0; + } - if (!--count) - { - BNerr(BN_F_BN_RAND_RANGE, BN_R_TOO_MANY_ITERATIONS); - return 0; - } - - } - while (BN_cmp(r, range) >= 0); - } - else - { - do - { - /* range = 11..._2 or range = 101..._2 */ - if (!bn_rand(r, n, -1, 0)) return 0; + if (!--count) { + BNerr(BN_F_BN_RAND_RANGE, BN_R_TOO_MANY_ITERATIONS); + return 0; + } - if (!--count) - { - BNerr(BN_F_BN_RAND_RANGE, BN_R_TOO_MANY_ITERATIONS); - return 0; - } - } - while (BN_cmp(r, range) >= 0); - } + } + while (BN_cmp(r, range) >= 0); + } else { + do { + /* range = 11..._2 or range = 101..._2 */ + if (!bn_rand(r, n, -1, 0)) + return 0; - bn_check_top(r); - return 1; - } + if (!--count) { + BNerr(BN_F_BN_RAND_RANGE, BN_R_TOO_MANY_ITERATIONS); + return 0; + } + } + while (BN_cmp(r, range) >= 0); + } + bn_check_top(r); + return 1; +} -int BN_rand_range(BIGNUM *r, const BIGNUM *range) - { - return bn_rand_range(0, r, range); - } +int BN_rand_range(BIGNUM *r, const BIGNUM *range) +{ + return bn_rand_range(0, r, range); +} -int BN_pseudo_rand_range(BIGNUM *r, const BIGNUM *range) - { - return bn_rand_range(1, r, range); - } +int BN_pseudo_rand_range(BIGNUM *r, const BIGNUM *range) +{ + return bn_rand_range(1, r, range); +} diff --git a/openssl/crypto/bn/bn_recp.c b/openssl/crypto/bn/bn_recp.c index 2e8efb8da..6826f93b3 100644 --- a/openssl/crypto/bn/bn_recp.c +++ b/openssl/crypto/bn/bn_recp.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -61,174 +61,189 @@ #include "bn_lcl.h" void BN_RECP_CTX_init(BN_RECP_CTX *recp) - { - BN_init(&(recp->N)); - BN_init(&(recp->Nr)); - recp->num_bits=0; - recp->flags=0; - } +{ + BN_init(&(recp->N)); + BN_init(&(recp->Nr)); + recp->num_bits = 0; + recp->flags = 0; +} BN_RECP_CTX *BN_RECP_CTX_new(void) - { - BN_RECP_CTX *ret; +{ + BN_RECP_CTX *ret; - if ((ret=(BN_RECP_CTX *)OPENSSL_malloc(sizeof(BN_RECP_CTX))) == NULL) - return(NULL); + if ((ret = (BN_RECP_CTX *)OPENSSL_malloc(sizeof(BN_RECP_CTX))) == NULL) + return (NULL); - BN_RECP_CTX_init(ret); - ret->flags=BN_FLG_MALLOCED; - return(ret); - } + BN_RECP_CTX_init(ret); + ret->flags = BN_FLG_MALLOCED; + return (ret); +} void BN_RECP_CTX_free(BN_RECP_CTX *recp) - { - if(recp == NULL) - return; +{ + if (recp == NULL) + return; - BN_free(&(recp->N)); - BN_free(&(recp->Nr)); - if (recp->flags & BN_FLG_MALLOCED) - OPENSSL_free(recp); - } + BN_free(&(recp->N)); + BN_free(&(recp->Nr)); + if (recp->flags & BN_FLG_MALLOCED) + OPENSSL_free(recp); +} int BN_RECP_CTX_set(BN_RECP_CTX *recp, const BIGNUM *d, BN_CTX *ctx) - { - if (!BN_copy(&(recp->N),d)) return 0; - BN_zero(&(recp->Nr)); - recp->num_bits=BN_num_bits(d); - recp->shift=0; - return(1); - } +{ + if (!BN_copy(&(recp->N), d)) + return 0; + BN_zero(&(recp->Nr)); + recp->num_bits = BN_num_bits(d); + recp->shift = 0; + return (1); +} int BN_mod_mul_reciprocal(BIGNUM *r, const BIGNUM *x, const BIGNUM *y, - BN_RECP_CTX *recp, BN_CTX *ctx) - { - int ret=0; - BIGNUM *a; - const BIGNUM *ca; - - BN_CTX_start(ctx); - if ((a = BN_CTX_get(ctx)) == NULL) goto err; - if (y != NULL) - { - if (x == y) - { if (!BN_sqr(a,x,ctx)) goto err; } - else - { if (!BN_mul(a,x,y,ctx)) goto err; } - ca = a; - } - else - ca=x; /* Just do the mod */ - - ret = BN_div_recp(NULL,r,ca,recp,ctx); -err: - BN_CTX_end(ctx); - bn_check_top(r); - return(ret); - } + BN_RECP_CTX *recp, BN_CTX *ctx) +{ + int ret = 0; + BIGNUM *a; + const BIGNUM *ca; + + BN_CTX_start(ctx); + if ((a = BN_CTX_get(ctx)) == NULL) + goto err; + if (y != NULL) { + if (x == y) { + if (!BN_sqr(a, x, ctx)) + goto err; + } else { + if (!BN_mul(a, x, y, ctx)) + goto err; + } + ca = a; + } else + ca = x; /* Just do the mod */ + + ret = BN_div_recp(NULL, r, ca, recp, ctx); + err: + BN_CTX_end(ctx); + bn_check_top(r); + return (ret); +} int BN_div_recp(BIGNUM *dv, BIGNUM *rem, const BIGNUM *m, - BN_RECP_CTX *recp, BN_CTX *ctx) - { - int i,j,ret=0; - BIGNUM *a,*b,*d,*r; - - BN_CTX_start(ctx); - a=BN_CTX_get(ctx); - b=BN_CTX_get(ctx); - if (dv != NULL) - d=dv; - else - d=BN_CTX_get(ctx); - if (rem != NULL) - r=rem; - else - r=BN_CTX_get(ctx); - if (a == NULL || b == NULL || d == NULL || r == NULL) goto err; - - if (BN_ucmp(m,&(recp->N)) < 0) - { - BN_zero(d); - if (!BN_copy(r,m)) return 0; - BN_CTX_end(ctx); - return(1); - } - - /* We want the remainder - * Given input of ABCDEF / ab - * we need multiply ABCDEF by 3 digests of the reciprocal of ab - * - */ - - /* i := max(BN_num_bits(m), 2*BN_num_bits(N)) */ - i=BN_num_bits(m); - j=recp->num_bits<<1; - if (j>i) i=j; - - /* Nr := round(2^i / N) */ - if (i != recp->shift) - recp->shift=BN_reciprocal(&(recp->Nr),&(recp->N), - i,ctx); /* BN_reciprocal returns i, or -1 for an error */ - if (recp->shift == -1) goto err; - - /* d := |round(round(m / 2^BN_num_bits(N)) * recp->Nr / 2^(i - BN_num_bits(N)))| - * = |round(round(m / 2^BN_num_bits(N)) * round(2^i / N) / 2^(i - BN_num_bits(N)))| - * <= |(m / 2^BN_num_bits(N)) * (2^i / N) * (2^BN_num_bits(N) / 2^i)| - * = |m/N| - */ - if (!BN_rshift(a,m,recp->num_bits)) goto err; - if (!BN_mul(b,a,&(recp->Nr),ctx)) goto err; - if (!BN_rshift(d,b,i-recp->num_bits)) goto err; - d->neg=0; - - if (!BN_mul(b,&(recp->N),d,ctx)) goto err; - if (!BN_usub(r,m,b)) goto err; - r->neg=0; + BN_RECP_CTX *recp, BN_CTX *ctx) +{ + int i, j, ret = 0; + BIGNUM *a, *b, *d, *r; + + BN_CTX_start(ctx); + a = BN_CTX_get(ctx); + b = BN_CTX_get(ctx); + if (dv != NULL) + d = dv; + else + d = BN_CTX_get(ctx); + if (rem != NULL) + r = rem; + else + r = BN_CTX_get(ctx); + if (a == NULL || b == NULL || d == NULL || r == NULL) + goto err; + + if (BN_ucmp(m, &(recp->N)) < 0) { + BN_zero(d); + if (!BN_copy(r, m)) + return 0; + BN_CTX_end(ctx); + return (1); + } + + /* + * We want the remainder Given input of ABCDEF / ab we need multiply + * ABCDEF by 3 digests of the reciprocal of ab + */ + + /* i := max(BN_num_bits(m), 2*BN_num_bits(N)) */ + i = BN_num_bits(m); + j = recp->num_bits << 1; + if (j > i) + i = j; + + /* Nr := round(2^i / N) */ + if (i != recp->shift) + recp->shift = BN_reciprocal(&(recp->Nr), &(recp->N), i, ctx); + /* BN_reciprocal could have returned -1 for an error */ + if (recp->shift == -1) + goto err; + + /*- + * d := |round(round(m / 2^BN_num_bits(N)) * recp->Nr / 2^(i - BN_num_bits(N)))| + * = |round(round(m / 2^BN_num_bits(N)) * round(2^i / N) / 2^(i - BN_num_bits(N)))| + * <= |(m / 2^BN_num_bits(N)) * (2^i / N) * (2^BN_num_bits(N) / 2^i)| + * = |m/N| + */ + if (!BN_rshift(a, m, recp->num_bits)) + goto err; + if (!BN_mul(b, a, &(recp->Nr), ctx)) + goto err; + if (!BN_rshift(d, b, i - recp->num_bits)) + goto err; + d->neg = 0; + + if (!BN_mul(b, &(recp->N), d, ctx)) + goto err; + if (!BN_usub(r, m, b)) + goto err; + r->neg = 0; #if 1 - j=0; - while (BN_ucmp(r,&(recp->N)) >= 0) - { - if (j++ > 2) - { - BNerr(BN_F_BN_DIV_RECP,BN_R_BAD_RECIPROCAL); - goto err; - } - if (!BN_usub(r,r,&(recp->N))) goto err; - if (!BN_add_word(d,1)) goto err; - } + j = 0; + while (BN_ucmp(r, &(recp->N)) >= 0) { + if (j++ > 2) { + BNerr(BN_F_BN_DIV_RECP, BN_R_BAD_RECIPROCAL); + goto err; + } + if (!BN_usub(r, r, &(recp->N))) + goto err; + if (!BN_add_word(d, 1)) + goto err; + } #endif - r->neg=BN_is_zero(r)?0:m->neg; - d->neg=m->neg^recp->N.neg; - ret=1; -err: - BN_CTX_end(ctx); - bn_check_top(dv); - bn_check_top(rem); - return(ret); - } - -/* len is the expected size of the result - * We actually calculate with an extra word of precision, so - * we can do faster division if the remainder is not required. + r->neg = BN_is_zero(r) ? 0 : m->neg; + d->neg = m->neg ^ recp->N.neg; + ret = 1; + err: + BN_CTX_end(ctx); + bn_check_top(dv); + bn_check_top(rem); + return (ret); +} + +/* + * len is the expected size of the result We actually calculate with an extra + * word of precision, so we can do faster division if the remainder is not + * required. */ /* r := 2^len / m */ int BN_reciprocal(BIGNUM *r, const BIGNUM *m, int len, BN_CTX *ctx) - { - int ret= -1; - BIGNUM *t; - - BN_CTX_start(ctx); - if((t = BN_CTX_get(ctx)) == NULL) goto err; - - if (!BN_set_bit(t,len)) goto err; - - if (!BN_div(r,NULL,t,m,ctx)) goto err; - - ret=len; -err: - bn_check_top(r); - BN_CTX_end(ctx); - return(ret); - } +{ + int ret = -1; + BIGNUM *t; + + BN_CTX_start(ctx); + if ((t = BN_CTX_get(ctx)) == NULL) + goto err; + + if (!BN_set_bit(t, len)) + goto err; + + if (!BN_div(r, NULL, t, m, ctx)) + goto err; + + ret = len; + err: + bn_check_top(r); + BN_CTX_end(ctx); + return (ret); +} diff --git a/openssl/crypto/bn/bn_shift.c b/openssl/crypto/bn/bn_shift.c index a6fca2c42..4f3e8ffed 100644 --- a/openssl/crypto/bn/bn_shift.c +++ b/openssl/crypto/bn/bn_shift.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -61,163 +61,154 @@ #include "bn_lcl.h" int BN_lshift1(BIGNUM *r, const BIGNUM *a) - { - register BN_ULONG *ap,*rp,t,c; - int i; +{ + register BN_ULONG *ap, *rp, t, c; + int i; - bn_check_top(r); - bn_check_top(a); + bn_check_top(r); + bn_check_top(a); - if (r != a) - { - r->neg=a->neg; - if (bn_wexpand(r,a->top+1) == NULL) return(0); - r->top=a->top; - } - else - { - if (bn_wexpand(r,a->top+1) == NULL) return(0); - } - ap=a->d; - rp=r->d; - c=0; - for (i=0; i<a->top; i++) - { - t= *(ap++); - *(rp++)=((t<<1)|c)&BN_MASK2; - c=(t & BN_TBIT)?1:0; - } - if (c) - { - *rp=1; - r->top++; - } - bn_check_top(r); - return(1); - } + if (r != a) { + r->neg = a->neg; + if (bn_wexpand(r, a->top + 1) == NULL) + return (0); + r->top = a->top; + } else { + if (bn_wexpand(r, a->top + 1) == NULL) + return (0); + } + ap = a->d; + rp = r->d; + c = 0; + for (i = 0; i < a->top; i++) { + t = *(ap++); + *(rp++) = ((t << 1) | c) & BN_MASK2; + c = (t & BN_TBIT) ? 1 : 0; + } + if (c) { + *rp = 1; + r->top++; + } + bn_check_top(r); + return (1); +} int BN_rshift1(BIGNUM *r, const BIGNUM *a) - { - BN_ULONG *ap,*rp,t,c; - int i,j; +{ + BN_ULONG *ap, *rp, t, c; + int i, j; - bn_check_top(r); - bn_check_top(a); + bn_check_top(r); + bn_check_top(a); - if (BN_is_zero(a)) - { - BN_zero(r); - return(1); - } - i = a->top; - ap= a->d; - j = i-(ap[i-1]==1); - if (a != r) - { - if (bn_wexpand(r,j) == NULL) return(0); - r->neg=a->neg; - } - rp=r->d; - t=ap[--i]; - c=(t&1)?BN_TBIT:0; - if (t>>=1) rp[i]=t; - while (i>0) - { - t=ap[--i]; - rp[i]=((t>>1)&BN_MASK2)|c; - c=(t&1)?BN_TBIT:0; - } - r->top=j; - bn_check_top(r); - return(1); - } + if (BN_is_zero(a)) { + BN_zero(r); + return (1); + } + i = a->top; + ap = a->d; + j = i - (ap[i - 1] == 1); + if (a != r) { + if (bn_wexpand(r, j) == NULL) + return (0); + r->neg = a->neg; + } + rp = r->d; + t = ap[--i]; + c = (t & 1) ? BN_TBIT : 0; + if (t >>= 1) + rp[i] = t; + while (i > 0) { + t = ap[--i]; + rp[i] = ((t >> 1) & BN_MASK2) | c; + c = (t & 1) ? BN_TBIT : 0; + } + r->top = j; + bn_check_top(r); + return (1); +} int BN_lshift(BIGNUM *r, const BIGNUM *a, int n) - { - int i,nw,lb,rb; - BN_ULONG *t,*f; - BN_ULONG l; +{ + int i, nw, lb, rb; + BN_ULONG *t, *f; + BN_ULONG l; - bn_check_top(r); - bn_check_top(a); + bn_check_top(r); + bn_check_top(a); - r->neg=a->neg; - nw=n/BN_BITS2; - if (bn_wexpand(r,a->top+nw+1) == NULL) return(0); - lb=n%BN_BITS2; - rb=BN_BITS2-lb; - f=a->d; - t=r->d; - t[a->top+nw]=0; - if (lb == 0) - for (i=a->top-1; i>=0; i--) - t[nw+i]=f[i]; - else - for (i=a->top-1; i>=0; i--) - { - l=f[i]; - t[nw+i+1]|=(l>>rb)&BN_MASK2; - t[nw+i]=(l<<lb)&BN_MASK2; - } - memset(t,0,nw*sizeof(t[0])); -/* for (i=0; i<nw; i++) - t[i]=0;*/ - r->top=a->top+nw+1; - bn_correct_top(r); - bn_check_top(r); - return(1); - } + r->neg = a->neg; + nw = n / BN_BITS2; + if (bn_wexpand(r, a->top + nw + 1) == NULL) + return (0); + lb = n % BN_BITS2; + rb = BN_BITS2 - lb; + f = a->d; + t = r->d; + t[a->top + nw] = 0; + if (lb == 0) + for (i = a->top - 1; i >= 0; i--) + t[nw + i] = f[i]; + else + for (i = a->top - 1; i >= 0; i--) { + l = f[i]; + t[nw + i + 1] |= (l >> rb) & BN_MASK2; + t[nw + i] = (l << lb) & BN_MASK2; + } + memset(t, 0, nw * sizeof(t[0])); + /* + * for (i=0; i<nw; i++) t[i]=0; + */ + r->top = a->top + nw + 1; + bn_correct_top(r); + bn_check_top(r); + return (1); +} int BN_rshift(BIGNUM *r, const BIGNUM *a, int n) - { - int i,j,nw,lb,rb; - BN_ULONG *t,*f; - BN_ULONG l,tmp; +{ + int i, j, nw, lb, rb; + BN_ULONG *t, *f; + BN_ULONG l, tmp; - bn_check_top(r); - bn_check_top(a); + bn_check_top(r); + bn_check_top(a); - nw=n/BN_BITS2; - rb=n%BN_BITS2; - lb=BN_BITS2-rb; - if (nw >= a->top || a->top == 0) - { - BN_zero(r); - return(1); - } - i = (BN_num_bits(a)-n+(BN_BITS2-1))/BN_BITS2; - if (r != a) - { - r->neg=a->neg; - if (bn_wexpand(r,i) == NULL) return(0); - } - else - { - if (n == 0) - return 1; /* or the copying loop will go berserk */ - } + nw = n / BN_BITS2; + rb = n % BN_BITS2; + lb = BN_BITS2 - rb; + if (nw >= a->top || a->top == 0) { + BN_zero(r); + return (1); + } + i = (BN_num_bits(a) - n + (BN_BITS2 - 1)) / BN_BITS2; + if (r != a) { + r->neg = a->neg; + if (bn_wexpand(r, i) == NULL) + return (0); + } else { + if (n == 0) + return 1; /* or the copying loop will go berserk */ + } - f= &(a->d[nw]); - t=r->d; - j=a->top-nw; - r->top=i; + f = &(a->d[nw]); + t = r->d; + j = a->top - nw; + r->top = i; - if (rb == 0) - { - for (i=j; i != 0; i--) - *(t++)= *(f++); - } - else - { - l= *(f++); - for (i=j-1; i != 0; i--) - { - tmp =(l>>rb)&BN_MASK2; - l= *(f++); - *(t++) =(tmp|(l<<lb))&BN_MASK2; - } - if ((l = (l>>rb)&BN_MASK2)) *(t) = l; - } - bn_check_top(r); - return(1); - } + if (rb == 0) { + for (i = j; i != 0; i--) + *(t++) = *(f++); + } else { + l = *(f++); + for (i = j - 1; i != 0; i--) { + tmp = (l >> rb) & BN_MASK2; + l = *(f++); + *(t++) = (tmp | (l << lb)) & BN_MASK2; + } + if ((l = (l >> rb) & BN_MASK2)) + *(t) = l; + } + bn_check_top(r); + return (1); +} diff --git a/openssl/crypto/bn/bn_sqr.c b/openssl/crypto/bn/bn_sqr.c index 65bbf165d..3ca69879e 100644 --- a/openssl/crypto/bn/bn_sqr.c +++ b/openssl/crypto/bn/bn_sqr.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -61,140 +61,137 @@ #include "bn_lcl.h" /* r must not be a */ -/* I've just gone over this and it is now %20 faster on x86 - eay - 27 Jun 96 */ +/* + * I've just gone over this and it is now %20 faster on x86 - eay - 27 Jun 96 + */ int BN_sqr(BIGNUM *r, const BIGNUM *a, BN_CTX *ctx) - { - int max,al; - int ret = 0; - BIGNUM *tmp,*rr; +{ + int max, al; + int ret = 0; + BIGNUM *tmp, *rr; #ifdef BN_COUNT - fprintf(stderr,"BN_sqr %d * %d\n",a->top,a->top); + fprintf(stderr, "BN_sqr %d * %d\n", a->top, a->top); #endif - bn_check_top(a); + bn_check_top(a); - al=a->top; - if (al <= 0) - { - r->top=0; - r->neg = 0; - return 1; - } + al = a->top; + if (al <= 0) { + r->top = 0; + r->neg = 0; + return 1; + } - BN_CTX_start(ctx); - rr=(a != r) ? r : BN_CTX_get(ctx); - tmp=BN_CTX_get(ctx); - if (!rr || !tmp) goto err; + BN_CTX_start(ctx); + rr = (a != r) ? r : BN_CTX_get(ctx); + tmp = BN_CTX_get(ctx); + if (!rr || !tmp) + goto err; - max = 2 * al; /* Non-zero (from above) */ - if (bn_wexpand(rr,max) == NULL) goto err; + max = 2 * al; /* Non-zero (from above) */ + if (bn_wexpand(rr, max) == NULL) + goto err; - if (al == 4) - { + if (al == 4) { #ifndef BN_SQR_COMBA - BN_ULONG t[8]; - bn_sqr_normal(rr->d,a->d,4,t); + BN_ULONG t[8]; + bn_sqr_normal(rr->d, a->d, 4, t); #else - bn_sqr_comba4(rr->d,a->d); + bn_sqr_comba4(rr->d, a->d); #endif - } - else if (al == 8) - { + } else if (al == 8) { #ifndef BN_SQR_COMBA - BN_ULONG t[16]; - bn_sqr_normal(rr->d,a->d,8,t); + BN_ULONG t[16]; + bn_sqr_normal(rr->d, a->d, 8, t); #else - bn_sqr_comba8(rr->d,a->d); + bn_sqr_comba8(rr->d, a->d); #endif - } - else - { + } else { #if defined(BN_RECURSION) - if (al < BN_SQR_RECURSIVE_SIZE_NORMAL) - { - BN_ULONG t[BN_SQR_RECURSIVE_SIZE_NORMAL*2]; - bn_sqr_normal(rr->d,a->d,al,t); - } - else - { - int j,k; + if (al < BN_SQR_RECURSIVE_SIZE_NORMAL) { + BN_ULONG t[BN_SQR_RECURSIVE_SIZE_NORMAL * 2]; + bn_sqr_normal(rr->d, a->d, al, t); + } else { + int j, k; - j=BN_num_bits_word((BN_ULONG)al); - j=1<<(j-1); - k=j+j; - if (al == j) - { - if (bn_wexpand(tmp,k*2) == NULL) goto err; - bn_sqr_recursive(rr->d,a->d,al,tmp->d); - } - else - { - if (bn_wexpand(tmp,max) == NULL) goto err; - bn_sqr_normal(rr->d,a->d,al,tmp->d); - } - } + j = BN_num_bits_word((BN_ULONG)al); + j = 1 << (j - 1); + k = j + j; + if (al == j) { + if (bn_wexpand(tmp, k * 2) == NULL) + goto err; + bn_sqr_recursive(rr->d, a->d, al, tmp->d); + } else { + if (bn_wexpand(tmp, max) == NULL) + goto err; + bn_sqr_normal(rr->d, a->d, al, tmp->d); + } + } #else - if (bn_wexpand(tmp,max) == NULL) goto err; - bn_sqr_normal(rr->d,a->d,al,tmp->d); + if (bn_wexpand(tmp, max) == NULL) + goto err; + bn_sqr_normal(rr->d, a->d, al, tmp->d); #endif - } + } - rr->neg=0; - /* If the most-significant half of the top word of 'a' is zero, then - * the square of 'a' will max-1 words. */ - if(a->d[al - 1] == (a->d[al - 1] & BN_MASK2l)) - rr->top = max - 1; - else - rr->top = max; - if (rr != r) BN_copy(r,rr); - ret = 1; + rr->neg = 0; + /* + * If the most-significant half of the top word of 'a' is zero, then the + * square of 'a' will max-1 words. + */ + if (a->d[al - 1] == (a->d[al - 1] & BN_MASK2l)) + rr->top = max - 1; + else + rr->top = max; + if (rr != r) + BN_copy(r, rr); + ret = 1; err: - bn_check_top(rr); - bn_check_top(tmp); - BN_CTX_end(ctx); - return(ret); - } + bn_check_top(rr); + bn_check_top(tmp); + BN_CTX_end(ctx); + return (ret); +} /* tmp must have 2*n words */ void bn_sqr_normal(BN_ULONG *r, const BN_ULONG *a, int n, BN_ULONG *tmp) - { - int i,j,max; - const BN_ULONG *ap; - BN_ULONG *rp; +{ + int i, j, max; + const BN_ULONG *ap; + BN_ULONG *rp; - max=n*2; - ap=a; - rp=r; - rp[0]=rp[max-1]=0; - rp++; - j=n; + max = n * 2; + ap = a; + rp = r; + rp[0] = rp[max - 1] = 0; + rp++; + j = n; - if (--j > 0) - { - ap++; - rp[j]=bn_mul_words(rp,ap,j,ap[-1]); - rp+=2; - } + if (--j > 0) { + ap++; + rp[j] = bn_mul_words(rp, ap, j, ap[-1]); + rp += 2; + } - for (i=n-2; i>0; i--) - { - j--; - ap++; - rp[j]=bn_mul_add_words(rp,ap,j,ap[-1]); - rp+=2; - } + for (i = n - 2; i > 0; i--) { + j--; + ap++; + rp[j] = bn_mul_add_words(rp, ap, j, ap[-1]); + rp += 2; + } - bn_add_words(r,r,r,max); + bn_add_words(r, r, r, max); - /* There will not be a carry */ + /* There will not be a carry */ - bn_sqr_words(tmp,a,n); + bn_sqr_words(tmp, a, n); - bn_add_words(r,r,tmp,max); - } + bn_add_words(r, r, tmp, max); +} #ifdef BN_RECURSION -/* r is 2*n words in size, +/*- + * r is 2*n words in size, * a and b are both n words in size. (There's not actually a 'b' here ...) * n must be a power of 2. * We multiply and return the result. @@ -205,91 +202,89 @@ void bn_sqr_normal(BN_ULONG *r, const BN_ULONG *a, int n, BN_ULONG *tmp) * a[1]*b[1] */ void bn_sqr_recursive(BN_ULONG *r, const BN_ULONG *a, int n2, BN_ULONG *t) - { - int n=n2/2; - int zero,c1; - BN_ULONG ln,lo,*p; +{ + int n = n2 / 2; + int zero, c1; + BN_ULONG ln, lo, *p; -#ifdef BN_COUNT - fprintf(stderr," bn_sqr_recursive %d * %d\n",n2,n2); -#endif - if (n2 == 4) - { -#ifndef BN_SQR_COMBA - bn_sqr_normal(r,a,4,t); -#else - bn_sqr_comba4(r,a); -#endif - return; - } - else if (n2 == 8) - { -#ifndef BN_SQR_COMBA - bn_sqr_normal(r,a,8,t); -#else - bn_sqr_comba8(r,a); -#endif - return; - } - if (n2 < BN_SQR_RECURSIVE_SIZE_NORMAL) - { - bn_sqr_normal(r,a,n2,t); - return; - } - /* r=(a[0]-a[1])*(a[1]-a[0]) */ - c1=bn_cmp_words(a,&(a[n]),n); - zero=0; - if (c1 > 0) - bn_sub_words(t,a,&(a[n]),n); - else if (c1 < 0) - bn_sub_words(t,&(a[n]),a,n); - else - zero=1; +# ifdef BN_COUNT + fprintf(stderr, " bn_sqr_recursive %d * %d\n", n2, n2); +# endif + if (n2 == 4) { +# ifndef BN_SQR_COMBA + bn_sqr_normal(r, a, 4, t); +# else + bn_sqr_comba4(r, a); +# endif + return; + } else if (n2 == 8) { +# ifndef BN_SQR_COMBA + bn_sqr_normal(r, a, 8, t); +# else + bn_sqr_comba8(r, a); +# endif + return; + } + if (n2 < BN_SQR_RECURSIVE_SIZE_NORMAL) { + bn_sqr_normal(r, a, n2, t); + return; + } + /* r=(a[0]-a[1])*(a[1]-a[0]) */ + c1 = bn_cmp_words(a, &(a[n]), n); + zero = 0; + if (c1 > 0) + bn_sub_words(t, a, &(a[n]), n); + else if (c1 < 0) + bn_sub_words(t, &(a[n]), a, n); + else + zero = 1; - /* The result will always be negative unless it is zero */ - p= &(t[n2*2]); + /* The result will always be negative unless it is zero */ + p = &(t[n2 * 2]); - if (!zero) - bn_sqr_recursive(&(t[n2]),t,n,p); - else - memset(&(t[n2]),0,n2*sizeof(BN_ULONG)); - bn_sqr_recursive(r,a,n,p); - bn_sqr_recursive(&(r[n2]),&(a[n]),n,p); + if (!zero) + bn_sqr_recursive(&(t[n2]), t, n, p); + else + memset(&(t[n2]), 0, n2 * sizeof(BN_ULONG)); + bn_sqr_recursive(r, a, n, p); + bn_sqr_recursive(&(r[n2]), &(a[n]), n, p); - /* t[32] holds (a[0]-a[1])*(a[1]-a[0]), it is negative or zero - * r[10] holds (a[0]*b[0]) - * r[32] holds (b[1]*b[1]) - */ + /*- + * t[32] holds (a[0]-a[1])*(a[1]-a[0]), it is negative or zero + * r[10] holds (a[0]*b[0]) + * r[32] holds (b[1]*b[1]) + */ - c1=(int)(bn_add_words(t,r,&(r[n2]),n2)); + c1 = (int)(bn_add_words(t, r, &(r[n2]), n2)); - /* t[32] is negative */ - c1-=(int)(bn_sub_words(&(t[n2]),t,&(t[n2]),n2)); + /* t[32] is negative */ + c1 -= (int)(bn_sub_words(&(t[n2]), t, &(t[n2]), n2)); - /* t[32] holds (a[0]-a[1])*(a[1]-a[0])+(a[0]*a[0])+(a[1]*a[1]) - * r[10] holds (a[0]*a[0]) - * r[32] holds (a[1]*a[1]) - * c1 holds the carry bits - */ - c1+=(int)(bn_add_words(&(r[n]),&(r[n]),&(t[n2]),n2)); - if (c1) - { - p= &(r[n+n2]); - lo= *p; - ln=(lo+c1)&BN_MASK2; - *p=ln; + /*- + * t[32] holds (a[0]-a[1])*(a[1]-a[0])+(a[0]*a[0])+(a[1]*a[1]) + * r[10] holds (a[0]*a[0]) + * r[32] holds (a[1]*a[1]) + * c1 holds the carry bits + */ + c1 += (int)(bn_add_words(&(r[n]), &(r[n]), &(t[n2]), n2)); + if (c1) { + p = &(r[n + n2]); + lo = *p; + ln = (lo + c1) & BN_MASK2; + *p = ln; - /* The overflow will stop before we over write - * words we should not overwrite */ - if (ln < (BN_ULONG)c1) - { - do { - p++; - lo= *p; - ln=(lo+1)&BN_MASK2; - *p=ln; - } while (ln == 0); - } - } - } + /* + * The overflow will stop before we over write words we should not + * overwrite + */ + if (ln < (BN_ULONG)c1) { + do { + p++; + lo = *p; + ln = (lo + 1) & BN_MASK2; + *p = ln; + } while (ln == 0); + } + } +} #endif diff --git a/openssl/crypto/bn/bn_sqrt.c b/openssl/crypto/bn/bn_sqrt.c index 6beaf9e5e..232af99a2 100644 --- a/openssl/crypto/bn/bn_sqrt.c +++ b/openssl/crypto/bn/bn_sqrt.c @@ -1,6 +1,8 @@ /* crypto/bn/bn_sqrt.c */ -/* Written by Lenka Fibikova <fibikova@exp-math.uni-essen.de> - * and Bodo Moeller for the OpenSSL project. */ +/* + * Written by Lenka Fibikova <fibikova@exp-math.uni-essen.de> and Bodo + * Moeller for the OpenSSL project. + */ /* ==================================================================== * Copyright (c) 1998-2000 The OpenSSL Project. All rights reserved. * @@ -9,7 +11,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -58,336 +60,350 @@ #include "cryptlib.h" #include "bn_lcl.h" - -BIGNUM *BN_mod_sqrt(BIGNUM *in, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) -/* Returns 'ret' such that - * ret^2 == a (mod p), - * using the Tonelli/Shanks algorithm (cf. Henri Cohen, "A Course - * in Algebraic Computational Number Theory", algorithm 1.5.1). - * 'p' must be prime! +BIGNUM *BN_mod_sqrt(BIGNUM *in, const BIGNUM *a, const BIGNUM *p, BN_CTX *ctx) +/* + * Returns 'ret' such that ret^2 == a (mod p), using the Tonelli/Shanks + * algorithm (cf. Henri Cohen, "A Course in Algebraic Computational Number + * Theory", algorithm 1.5.1). 'p' must be prime! */ - { - BIGNUM *ret = in; - int err = 1; - int r; - BIGNUM *A, *b, *q, *t, *x, *y; - int e, i, j; - - if (!BN_is_odd(p) || BN_abs_is_word(p, 1)) - { - if (BN_abs_is_word(p, 2)) - { - if (ret == NULL) - ret = BN_new(); - if (ret == NULL) - goto end; - if (!BN_set_word(ret, BN_is_bit_set(a, 0))) - { - if (ret != in) - BN_free(ret); - return NULL; - } - bn_check_top(ret); - return ret; - } - - BNerr(BN_F_BN_MOD_SQRT, BN_R_P_IS_NOT_PRIME); - return(NULL); - } - - if (BN_is_zero(a) || BN_is_one(a)) - { - if (ret == NULL) - ret = BN_new(); - if (ret == NULL) - goto end; - if (!BN_set_word(ret, BN_is_one(a))) - { - if (ret != in) - BN_free(ret); - return NULL; - } - bn_check_top(ret); - return ret; - } - - BN_CTX_start(ctx); - A = BN_CTX_get(ctx); - b = BN_CTX_get(ctx); - q = BN_CTX_get(ctx); - t = BN_CTX_get(ctx); - x = BN_CTX_get(ctx); - y = BN_CTX_get(ctx); - if (y == NULL) goto end; - - if (ret == NULL) - ret = BN_new(); - if (ret == NULL) goto end; - - /* A = a mod p */ - if (!BN_nnmod(A, a, p, ctx)) goto end; - - /* now write |p| - 1 as 2^e*q where q is odd */ - e = 1; - while (!BN_is_bit_set(p, e)) - e++; - /* we'll set q later (if needed) */ - - if (e == 1) - { - /* The easy case: (|p|-1)/2 is odd, so 2 has an inverse - * modulo (|p|-1)/2, and square roots can be computed - * directly by modular exponentiation. - * We have - * 2 * (|p|+1)/4 == 1 (mod (|p|-1)/2), - * so we can use exponent (|p|+1)/4, i.e. (|p|-3)/4 + 1. - */ - if (!BN_rshift(q, p, 2)) goto end; - q->neg = 0; - if (!BN_add_word(q, 1)) goto end; - if (!BN_mod_exp(ret, A, q, p, ctx)) goto end; - err = 0; - goto vrfy; - } - - if (e == 2) - { - /* |p| == 5 (mod 8) - * - * In this case 2 is always a non-square since - * Legendre(2,p) = (-1)^((p^2-1)/8) for any odd prime. - * So if a really is a square, then 2*a is a non-square. - * Thus for - * b := (2*a)^((|p|-5)/8), - * i := (2*a)*b^2 - * we have - * i^2 = (2*a)^((1 + (|p|-5)/4)*2) - * = (2*a)^((p-1)/2) - * = -1; - * so if we set - * x := a*b*(i-1), - * then - * x^2 = a^2 * b^2 * (i^2 - 2*i + 1) - * = a^2 * b^2 * (-2*i) - * = a*(-i)*(2*a*b^2) - * = a*(-i)*i - * = a. - * - * (This is due to A.O.L. Atkin, - * <URL: http://listserv.nodak.edu/scripts/wa.exe?A2=ind9211&L=nmbrthry&O=T&P=562>, - * November 1992.) - */ - - /* t := 2*a */ - if (!BN_mod_lshift1_quick(t, A, p)) goto end; - - /* b := (2*a)^((|p|-5)/8) */ - if (!BN_rshift(q, p, 3)) goto end; - q->neg = 0; - if (!BN_mod_exp(b, t, q, p, ctx)) goto end; - - /* y := b^2 */ - if (!BN_mod_sqr(y, b, p, ctx)) goto end; - - /* t := (2*a)*b^2 - 1*/ - if (!BN_mod_mul(t, t, y, p, ctx)) goto end; - if (!BN_sub_word(t, 1)) goto end; - - /* x = a*b*t */ - if (!BN_mod_mul(x, A, b, p, ctx)) goto end; - if (!BN_mod_mul(x, x, t, p, ctx)) goto end; - - if (!BN_copy(ret, x)) goto end; - err = 0; - goto vrfy; - } - - /* e > 2, so we really have to use the Tonelli/Shanks algorithm. - * First, find some y that is not a square. */ - if (!BN_copy(q, p)) goto end; /* use 'q' as temp */ - q->neg = 0; - i = 2; - do - { - /* For efficiency, try small numbers first; - * if this fails, try random numbers. - */ - if (i < 22) - { - if (!BN_set_word(y, i)) goto end; - } - else - { - if (!BN_pseudo_rand(y, BN_num_bits(p), 0, 0)) goto end; - if (BN_ucmp(y, p) >= 0) - { - if (!(p->neg ? BN_add : BN_sub)(y, y, p)) goto end; - } - /* now 0 <= y < |p| */ - if (BN_is_zero(y)) - if (!BN_set_word(y, i)) goto end; - } - - r = BN_kronecker(y, q, ctx); /* here 'q' is |p| */ - if (r < -1) goto end; - if (r == 0) - { - /* m divides p */ - BNerr(BN_F_BN_MOD_SQRT, BN_R_P_IS_NOT_PRIME); - goto end; - } - } - while (r == 1 && ++i < 82); - - if (r != -1) - { - /* Many rounds and still no non-square -- this is more likely - * a bug than just bad luck. - * Even if p is not prime, we should have found some y - * such that r == -1. - */ - BNerr(BN_F_BN_MOD_SQRT, BN_R_TOO_MANY_ITERATIONS); - goto end; - } - - /* Here's our actual 'q': */ - if (!BN_rshift(q, q, e)) goto end; - - /* Now that we have some non-square, we can find an element - * of order 2^e by computing its q'th power. */ - if (!BN_mod_exp(y, y, q, p, ctx)) goto end; - if (BN_is_one(y)) - { - BNerr(BN_F_BN_MOD_SQRT, BN_R_P_IS_NOT_PRIME); - goto end; - } - - /* Now we know that (if p is indeed prime) there is an integer - * k, 0 <= k < 2^e, such that - * - * a^q * y^k == 1 (mod p). - * - * As a^q is a square and y is not, k must be even. - * q+1 is even, too, so there is an element - * - * X := a^((q+1)/2) * y^(k/2), - * - * and it satisfies - * - * X^2 = a^q * a * y^k - * = a, - * - * so it is the square root that we are looking for. - */ - - /* t := (q-1)/2 (note that q is odd) */ - if (!BN_rshift1(t, q)) goto end; - - /* x := a^((q-1)/2) */ - if (BN_is_zero(t)) /* special case: p = 2^e + 1 */ - { - if (!BN_nnmod(t, A, p, ctx)) goto end; - if (BN_is_zero(t)) - { - /* special case: a == 0 (mod p) */ - BN_zero(ret); - err = 0; - goto end; - } - else - if (!BN_one(x)) goto end; - } - else - { - if (!BN_mod_exp(x, A, t, p, ctx)) goto end; - if (BN_is_zero(x)) - { - /* special case: a == 0 (mod p) */ - BN_zero(ret); - err = 0; - goto end; - } - } - - /* b := a*x^2 (= a^q) */ - if (!BN_mod_sqr(b, x, p, ctx)) goto end; - if (!BN_mod_mul(b, b, A, p, ctx)) goto end; - - /* x := a*x (= a^((q+1)/2)) */ - if (!BN_mod_mul(x, x, A, p, ctx)) goto end; - - while (1) - { - /* Now b is a^q * y^k for some even k (0 <= k < 2^E - * where E refers to the original value of e, which we - * don't keep in a variable), and x is a^((q+1)/2) * y^(k/2). - * - * We have a*b = x^2, - * y^2^(e-1) = -1, - * b^2^(e-1) = 1. - */ - - if (BN_is_one(b)) - { - if (!BN_copy(ret, x)) goto end; - err = 0; - goto vrfy; - } - - - /* find smallest i such that b^(2^i) = 1 */ - i = 1; - if (!BN_mod_sqr(t, b, p, ctx)) goto end; - while (!BN_is_one(t)) - { - i++; - if (i == e) - { - BNerr(BN_F_BN_MOD_SQRT, BN_R_NOT_A_SQUARE); - goto end; - } - if (!BN_mod_mul(t, t, t, p, ctx)) goto end; - } - - - /* t := y^2^(e - i - 1) */ - if (!BN_copy(t, y)) goto end; - for (j = e - i - 1; j > 0; j--) - { - if (!BN_mod_sqr(t, t, p, ctx)) goto end; - } - if (!BN_mod_mul(y, t, t, p, ctx)) goto end; - if (!BN_mod_mul(x, x, t, p, ctx)) goto end; - if (!BN_mod_mul(b, b, y, p, ctx)) goto end; - e = i; - } +{ + BIGNUM *ret = in; + int err = 1; + int r; + BIGNUM *A, *b, *q, *t, *x, *y; + int e, i, j; + + if (!BN_is_odd(p) || BN_abs_is_word(p, 1)) { + if (BN_abs_is_word(p, 2)) { + if (ret == NULL) + ret = BN_new(); + if (ret == NULL) + goto end; + if (!BN_set_word(ret, BN_is_bit_set(a, 0))) { + if (ret != in) + BN_free(ret); + return NULL; + } + bn_check_top(ret); + return ret; + } + + BNerr(BN_F_BN_MOD_SQRT, BN_R_P_IS_NOT_PRIME); + return (NULL); + } + + if (BN_is_zero(a) || BN_is_one(a)) { + if (ret == NULL) + ret = BN_new(); + if (ret == NULL) + goto end; + if (!BN_set_word(ret, BN_is_one(a))) { + if (ret != in) + BN_free(ret); + return NULL; + } + bn_check_top(ret); + return ret; + } + + BN_CTX_start(ctx); + A = BN_CTX_get(ctx); + b = BN_CTX_get(ctx); + q = BN_CTX_get(ctx); + t = BN_CTX_get(ctx); + x = BN_CTX_get(ctx); + y = BN_CTX_get(ctx); + if (y == NULL) + goto end; + + if (ret == NULL) + ret = BN_new(); + if (ret == NULL) + goto end; + + /* A = a mod p */ + if (!BN_nnmod(A, a, p, ctx)) + goto end; + + /* now write |p| - 1 as 2^e*q where q is odd */ + e = 1; + while (!BN_is_bit_set(p, e)) + e++; + /* we'll set q later (if needed) */ + + if (e == 1) { + /*- + * The easy case: (|p|-1)/2 is odd, so 2 has an inverse + * modulo (|p|-1)/2, and square roots can be computed + * directly by modular exponentiation. + * We have + * 2 * (|p|+1)/4 == 1 (mod (|p|-1)/2), + * so we can use exponent (|p|+1)/4, i.e. (|p|-3)/4 + 1. + */ + if (!BN_rshift(q, p, 2)) + goto end; + q->neg = 0; + if (!BN_add_word(q, 1)) + goto end; + if (!BN_mod_exp(ret, A, q, p, ctx)) + goto end; + err = 0; + goto vrfy; + } + + if (e == 2) { + /*- + * |p| == 5 (mod 8) + * + * In this case 2 is always a non-square since + * Legendre(2,p) = (-1)^((p^2-1)/8) for any odd prime. + * So if a really is a square, then 2*a is a non-square. + * Thus for + * b := (2*a)^((|p|-5)/8), + * i := (2*a)*b^2 + * we have + * i^2 = (2*a)^((1 + (|p|-5)/4)*2) + * = (2*a)^((p-1)/2) + * = -1; + * so if we set + * x := a*b*(i-1), + * then + * x^2 = a^2 * b^2 * (i^2 - 2*i + 1) + * = a^2 * b^2 * (-2*i) + * = a*(-i)*(2*a*b^2) + * = a*(-i)*i + * = a. + * + * (This is due to A.O.L. Atkin, + * <URL: http://listserv.nodak.edu/scripts/wa.exe?A2=ind9211&L=nmbrthry&O=T&P=562>, + * November 1992.) + */ + + /* t := 2*a */ + if (!BN_mod_lshift1_quick(t, A, p)) + goto end; + + /* b := (2*a)^((|p|-5)/8) */ + if (!BN_rshift(q, p, 3)) + goto end; + q->neg = 0; + if (!BN_mod_exp(b, t, q, p, ctx)) + goto end; + + /* y := b^2 */ + if (!BN_mod_sqr(y, b, p, ctx)) + goto end; + + /* t := (2*a)*b^2 - 1 */ + if (!BN_mod_mul(t, t, y, p, ctx)) + goto end; + if (!BN_sub_word(t, 1)) + goto end; + + /* x = a*b*t */ + if (!BN_mod_mul(x, A, b, p, ctx)) + goto end; + if (!BN_mod_mul(x, x, t, p, ctx)) + goto end; + + if (!BN_copy(ret, x)) + goto end; + err = 0; + goto vrfy; + } + + /* + * e > 2, so we really have to use the Tonelli/Shanks algorithm. First, + * find some y that is not a square. + */ + if (!BN_copy(q, p)) + goto end; /* use 'q' as temp */ + q->neg = 0; + i = 2; + do { + /* + * For efficiency, try small numbers first; if this fails, try random + * numbers. + */ + if (i < 22) { + if (!BN_set_word(y, i)) + goto end; + } else { + if (!BN_pseudo_rand(y, BN_num_bits(p), 0, 0)) + goto end; + if (BN_ucmp(y, p) >= 0) { + if (!(p->neg ? BN_add : BN_sub) (y, y, p)) + goto end; + } + /* now 0 <= y < |p| */ + if (BN_is_zero(y)) + if (!BN_set_word(y, i)) + goto end; + } + + r = BN_kronecker(y, q, ctx); /* here 'q' is |p| */ + if (r < -1) + goto end; + if (r == 0) { + /* m divides p */ + BNerr(BN_F_BN_MOD_SQRT, BN_R_P_IS_NOT_PRIME); + goto end; + } + } + while (r == 1 && ++i < 82); + + if (r != -1) { + /* + * Many rounds and still no non-square -- this is more likely a bug + * than just bad luck. Even if p is not prime, we should have found + * some y such that r == -1. + */ + BNerr(BN_F_BN_MOD_SQRT, BN_R_TOO_MANY_ITERATIONS); + goto end; + } + + /* Here's our actual 'q': */ + if (!BN_rshift(q, q, e)) + goto end; + + /* + * Now that we have some non-square, we can find an element of order 2^e + * by computing its q'th power. + */ + if (!BN_mod_exp(y, y, q, p, ctx)) + goto end; + if (BN_is_one(y)) { + BNerr(BN_F_BN_MOD_SQRT, BN_R_P_IS_NOT_PRIME); + goto end; + } + + /*- + * Now we know that (if p is indeed prime) there is an integer + * k, 0 <= k < 2^e, such that + * + * a^q * y^k == 1 (mod p). + * + * As a^q is a square and y is not, k must be even. + * q+1 is even, too, so there is an element + * + * X := a^((q+1)/2) * y^(k/2), + * + * and it satisfies + * + * X^2 = a^q * a * y^k + * = a, + * + * so it is the square root that we are looking for. + */ + + /* t := (q-1)/2 (note that q is odd) */ + if (!BN_rshift1(t, q)) + goto end; + + /* x := a^((q-1)/2) */ + if (BN_is_zero(t)) { /* special case: p = 2^e + 1 */ + if (!BN_nnmod(t, A, p, ctx)) + goto end; + if (BN_is_zero(t)) { + /* special case: a == 0 (mod p) */ + BN_zero(ret); + err = 0; + goto end; + } else if (!BN_one(x)) + goto end; + } else { + if (!BN_mod_exp(x, A, t, p, ctx)) + goto end; + if (BN_is_zero(x)) { + /* special case: a == 0 (mod p) */ + BN_zero(ret); + err = 0; + goto end; + } + } + + /* b := a*x^2 (= a^q) */ + if (!BN_mod_sqr(b, x, p, ctx)) + goto end; + if (!BN_mod_mul(b, b, A, p, ctx)) + goto end; + + /* x := a*x (= a^((q+1)/2)) */ + if (!BN_mod_mul(x, x, A, p, ctx)) + goto end; + + while (1) { + /*- + * Now b is a^q * y^k for some even k (0 <= k < 2^E + * where E refers to the original value of e, which we + * don't keep in a variable), and x is a^((q+1)/2) * y^(k/2). + * + * We have a*b = x^2, + * y^2^(e-1) = -1, + * b^2^(e-1) = 1. + */ + + if (BN_is_one(b)) { + if (!BN_copy(ret, x)) + goto end; + err = 0; + goto vrfy; + } + + /* find smallest i such that b^(2^i) = 1 */ + i = 1; + if (!BN_mod_sqr(t, b, p, ctx)) + goto end; + while (!BN_is_one(t)) { + i++; + if (i == e) { + BNerr(BN_F_BN_MOD_SQRT, BN_R_NOT_A_SQUARE); + goto end; + } + if (!BN_mod_mul(t, t, t, p, ctx)) + goto end; + } + + /* t := y^2^(e - i - 1) */ + if (!BN_copy(t, y)) + goto end; + for (j = e - i - 1; j > 0; j--) { + if (!BN_mod_sqr(t, t, p, ctx)) + goto end; + } + if (!BN_mod_mul(y, t, t, p, ctx)) + goto end; + if (!BN_mod_mul(x, x, t, p, ctx)) + goto end; + if (!BN_mod_mul(b, b, y, p, ctx)) + goto end; + e = i; + } vrfy: - if (!err) - { - /* verify the result -- the input might have been not a square - * (test added in 0.9.8) */ - - if (!BN_mod_sqr(x, ret, p, ctx)) - err = 1; - - if (!err && 0 != BN_cmp(x, A)) - { - BNerr(BN_F_BN_MOD_SQRT, BN_R_NOT_A_SQUARE); - err = 1; - } - } + if (!err) { + /* + * verify the result -- the input might have been not a square (test + * added in 0.9.8) + */ + + if (!BN_mod_sqr(x, ret, p, ctx)) + err = 1; + + if (!err && 0 != BN_cmp(x, A)) { + BNerr(BN_F_BN_MOD_SQRT, BN_R_NOT_A_SQUARE); + err = 1; + } + } end: - if (err) - { - if (ret != NULL && ret != in) - { - BN_clear_free(ret); - } - ret = NULL; - } - BN_CTX_end(ctx); - bn_check_top(ret); - return ret; - } + if (err) { + if (ret != NULL && ret != in) { + BN_clear_free(ret); + } + ret = NULL; + } + BN_CTX_end(ctx); + bn_check_top(ret); + return ret; +} diff --git a/openssl/crypto/bn/bn_word.c b/openssl/crypto/bn/bn_word.c index de83a15b9..b031a60b5 100644 --- a/openssl/crypto/bn/bn_word.c +++ b/openssl/crypto/bn/bn_word.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -61,178 +61,167 @@ #include "bn_lcl.h" BN_ULONG BN_mod_word(const BIGNUM *a, BN_ULONG w) - { +{ #ifndef BN_LLONG - BN_ULONG ret=0; + BN_ULONG ret = 0; #else - BN_ULLONG ret=0; + BN_ULLONG ret = 0; #endif - int i; + int i; - if (w == 0) - return (BN_ULONG)-1; + if (w == 0) + return (BN_ULONG)-1; - bn_check_top(a); - w&=BN_MASK2; - for (i=a->top-1; i>=0; i--) - { + bn_check_top(a); + w &= BN_MASK2; + for (i = a->top - 1; i >= 0; i--) { #ifndef BN_LLONG - ret=((ret<<BN_BITS4)|((a->d[i]>>BN_BITS4)&BN_MASK2l))%w; - ret=((ret<<BN_BITS4)|(a->d[i]&BN_MASK2l))%w; + ret = ((ret << BN_BITS4) | ((a->d[i] >> BN_BITS4) & BN_MASK2l)) % w; + ret = ((ret << BN_BITS4) | (a->d[i] & BN_MASK2l)) % w; #else - ret=(BN_ULLONG)(((ret<<(BN_ULLONG)BN_BITS2)|a->d[i])% - (BN_ULLONG)w); + ret = (BN_ULLONG) (((ret << (BN_ULLONG) BN_BITS2) | a->d[i]) % + (BN_ULLONG) w); #endif - } - return((BN_ULONG)ret); - } + } + return ((BN_ULONG)ret); +} BN_ULONG BN_div_word(BIGNUM *a, BN_ULONG w) - { - BN_ULONG ret = 0; - int i, j; - - bn_check_top(a); - w &= BN_MASK2; - - if (!w) - /* actually this an error (division by zero) */ - return (BN_ULONG)-1; - if (a->top == 0) - return 0; - - /* normalize input (so bn_div_words doesn't complain) */ - j = BN_BITS2 - BN_num_bits_word(w); - w <<= j; - if (!BN_lshift(a, a, j)) - return (BN_ULONG)-1; - - for (i=a->top-1; i>=0; i--) - { - BN_ULONG l,d; - - l=a->d[i]; - d=bn_div_words(ret,l,w); - ret=(l-((d*w)&BN_MASK2))&BN_MASK2; - a->d[i]=d; - } - if ((a->top > 0) && (a->d[a->top-1] == 0)) - a->top--; - ret >>= j; - bn_check_top(a); - return(ret); - } +{ + BN_ULONG ret = 0; + int i, j; + + bn_check_top(a); + w &= BN_MASK2; + + if (!w) + /* actually this an error (division by zero) */ + return (BN_ULONG)-1; + if (a->top == 0) + return 0; + + /* normalize input (so bn_div_words doesn't complain) */ + j = BN_BITS2 - BN_num_bits_word(w); + w <<= j; + if (!BN_lshift(a, a, j)) + return (BN_ULONG)-1; + + for (i = a->top - 1; i >= 0; i--) { + BN_ULONG l, d; + + l = a->d[i]; + d = bn_div_words(ret, l, w); + ret = (l - ((d * w) & BN_MASK2)) & BN_MASK2; + a->d[i] = d; + } + if ((a->top > 0) && (a->d[a->top - 1] == 0)) + a->top--; + ret >>= j; + bn_check_top(a); + return (ret); +} int BN_add_word(BIGNUM *a, BN_ULONG w) - { - BN_ULONG l; - int i; - - bn_check_top(a); - w &= BN_MASK2; - - /* degenerate case: w is zero */ - if (!w) return 1; - /* degenerate case: a is zero */ - if(BN_is_zero(a)) return BN_set_word(a, w); - /* handle 'a' when negative */ - if (a->neg) - { - a->neg=0; - i=BN_sub_word(a,w); - if (!BN_is_zero(a)) - a->neg=!(a->neg); - return(i); - } - for (i=0;w!=0 && i<a->top;i++) - { - a->d[i] = l = (a->d[i]+w)&BN_MASK2; - w = (w>l)?1:0; - } - if (w && i==a->top) - { - if (bn_wexpand(a,a->top+1) == NULL) return 0; - a->top++; - a->d[i]=w; - } - bn_check_top(a); - return(1); - } +{ + BN_ULONG l; + int i; + + bn_check_top(a); + w &= BN_MASK2; + + /* degenerate case: w is zero */ + if (!w) + return 1; + /* degenerate case: a is zero */ + if (BN_is_zero(a)) + return BN_set_word(a, w); + /* handle 'a' when negative */ + if (a->neg) { + a->neg = 0; + i = BN_sub_word(a, w); + if (!BN_is_zero(a)) + a->neg = !(a->neg); + return (i); + } + for (i = 0; w != 0 && i < a->top; i++) { + a->d[i] = l = (a->d[i] + w) & BN_MASK2; + w = (w > l) ? 1 : 0; + } + if (w && i == a->top) { + if (bn_wexpand(a, a->top + 1) == NULL) + return 0; + a->top++; + a->d[i] = w; + } + bn_check_top(a); + return (1); +} int BN_sub_word(BIGNUM *a, BN_ULONG w) - { - int i; - - bn_check_top(a); - w &= BN_MASK2; - - /* degenerate case: w is zero */ - if (!w) return 1; - /* degenerate case: a is zero */ - if(BN_is_zero(a)) - { - i = BN_set_word(a,w); - if (i != 0) - BN_set_negative(a, 1); - return i; - } - /* handle 'a' when negative */ - if (a->neg) - { - a->neg=0; - i=BN_add_word(a,w); - a->neg=1; - return(i); - } - - if ((a->top == 1) && (a->d[0] < w)) - { - a->d[0]=w-a->d[0]; - a->neg=1; - return(1); - } - i=0; - for (;;) - { - if (a->d[i] >= w) - { - a->d[i]-=w; - break; - } - else - { - a->d[i]=(a->d[i]-w)&BN_MASK2; - i++; - w=1; - } - } - if ((a->d[i] == 0) && (i == (a->top-1))) - a->top--; - bn_check_top(a); - return(1); - } +{ + int i; + + bn_check_top(a); + w &= BN_MASK2; + + /* degenerate case: w is zero */ + if (!w) + return 1; + /* degenerate case: a is zero */ + if (BN_is_zero(a)) { + i = BN_set_word(a, w); + if (i != 0) + BN_set_negative(a, 1); + return i; + } + /* handle 'a' when negative */ + if (a->neg) { + a->neg = 0; + i = BN_add_word(a, w); + a->neg = 1; + return (i); + } + + if ((a->top == 1) && (a->d[0] < w)) { + a->d[0] = w - a->d[0]; + a->neg = 1; + return (1); + } + i = 0; + for (;;) { + if (a->d[i] >= w) { + a->d[i] -= w; + break; + } else { + a->d[i] = (a->d[i] - w) & BN_MASK2; + i++; + w = 1; + } + } + if ((a->d[i] == 0) && (i == (a->top - 1))) + a->top--; + bn_check_top(a); + return (1); +} int BN_mul_word(BIGNUM *a, BN_ULONG w) - { - BN_ULONG ll; - - bn_check_top(a); - w&=BN_MASK2; - if (a->top) - { - if (w == 0) - BN_zero(a); - else - { - ll=bn_mul_words(a->d,a->d,a->top,w); - if (ll) - { - if (bn_wexpand(a,a->top+1) == NULL) return(0); - a->d[a->top++]=ll; - } - } - } - bn_check_top(a); - return(1); - } - +{ + BN_ULONG ll; + + bn_check_top(a); + w &= BN_MASK2; + if (a->top) { + if (w == 0) + BN_zero(a); + else { + ll = bn_mul_words(a->d, a->d, a->top, w); + if (ll) { + if (bn_wexpand(a, a->top + 1) == NULL) + return (0); + a->d[a->top++] = ll; + } + } + } + bn_check_top(a); + return (1); +} diff --git a/openssl/crypto/bn/bn_x931p.c b/openssl/crypto/bn/bn_x931p.c index 04c5c874e..6d76b1284 100644 --- a/openssl/crypto/bn/bn_x931p.c +++ b/openssl/crypto/bn/bn_x931p.c @@ -1,6 +1,7 @@ /* bn_x931p.c */ -/* Written by Dr Stephen N Henson (steve@openssl.org) for the OpenSSL - * project 2005. +/* + * Written by Dr Stephen N Henson (steve@openssl.org) for the OpenSSL project + * 2005. */ /* ==================================================================== * Copyright (c) 2005 The OpenSSL Project. All rights reserved. @@ -10,7 +11,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -61,212 +62,213 @@ /* X9.31 routines for prime derivation */ -/* X9.31 prime derivation. This is used to generate the primes pi - * (p1, p2, q1, q2) from a parameter Xpi by checking successive odd - * integers. +/* + * X9.31 prime derivation. This is used to generate the primes pi (p1, p2, + * q1, q2) from a parameter Xpi by checking successive odd integers. */ static int bn_x931_derive_pi(BIGNUM *pi, const BIGNUM *Xpi, BN_CTX *ctx, - BN_GENCB *cb) - { - int i = 0; - if (!BN_copy(pi, Xpi)) - return 0; - if (!BN_is_odd(pi) && !BN_add_word(pi, 1)) - return 0; - for(;;) - { - i++; - BN_GENCB_call(cb, 0, i); - /* NB 27 MR is specificed in X9.31 */ - if (BN_is_prime_fasttest_ex(pi, 27, ctx, 1, cb)) - break; - if (!BN_add_word(pi, 2)) - return 0; - } - BN_GENCB_call(cb, 2, i); - return 1; - } - -/* This is the main X9.31 prime derivation function. From parameters - * Xp1, Xp2 and Xp derive the prime p. If the parameters p1 or p2 are - * not NULL they will be returned too: this is needed for testing. + BN_GENCB *cb) +{ + int i = 0; + if (!BN_copy(pi, Xpi)) + return 0; + if (!BN_is_odd(pi) && !BN_add_word(pi, 1)) + return 0; + for (;;) { + i++; + BN_GENCB_call(cb, 0, i); + /* NB 27 MR is specificed in X9.31 */ + if (BN_is_prime_fasttest_ex(pi, 27, ctx, 1, cb)) + break; + if (!BN_add_word(pi, 2)) + return 0; + } + BN_GENCB_call(cb, 2, i); + return 1; +} + +/* + * This is the main X9.31 prime derivation function. From parameters Xp1, Xp2 + * and Xp derive the prime p. If the parameters p1 or p2 are not NULL they + * will be returned too: this is needed for testing. */ int BN_X931_derive_prime_ex(BIGNUM *p, BIGNUM *p1, BIGNUM *p2, - const BIGNUM *Xp, const BIGNUM *Xp1, const BIGNUM *Xp2, - const BIGNUM *e, BN_CTX *ctx, BN_GENCB *cb) - { - int ret = 0; + const BIGNUM *Xp, const BIGNUM *Xp1, + const BIGNUM *Xp2, const BIGNUM *e, BN_CTX *ctx, + BN_GENCB *cb) +{ + int ret = 0; - BIGNUM *t, *p1p2, *pm1; + BIGNUM *t, *p1p2, *pm1; - /* Only even e supported */ - if (!BN_is_odd(e)) - return 0; + /* Only even e supported */ + if (!BN_is_odd(e)) + return 0; - BN_CTX_start(ctx); - if (!p1) - p1 = BN_CTX_get(ctx); + BN_CTX_start(ctx); + if (!p1) + p1 = BN_CTX_get(ctx); - if (!p2) - p2 = BN_CTX_get(ctx); + if (!p2) + p2 = BN_CTX_get(ctx); - t = BN_CTX_get(ctx); + t = BN_CTX_get(ctx); - p1p2 = BN_CTX_get(ctx); + p1p2 = BN_CTX_get(ctx); - pm1 = BN_CTX_get(ctx); + pm1 = BN_CTX_get(ctx); - if (!bn_x931_derive_pi(p1, Xp1, ctx, cb)) - goto err; + if (!bn_x931_derive_pi(p1, Xp1, ctx, cb)) + goto err; - if (!bn_x931_derive_pi(p2, Xp2, ctx, cb)) - goto err; + if (!bn_x931_derive_pi(p2, Xp2, ctx, cb)) + goto err; - if (!BN_mul(p1p2, p1, p2, ctx)) - goto err; + if (!BN_mul(p1p2, p1, p2, ctx)) + goto err; - /* First set p to value of Rp */ + /* First set p to value of Rp */ - if (!BN_mod_inverse(p, p2, p1, ctx)) - goto err; + if (!BN_mod_inverse(p, p2, p1, ctx)) + goto err; - if (!BN_mul(p, p, p2, ctx)) - goto err; + if (!BN_mul(p, p, p2, ctx)) + goto err; - if (!BN_mod_inverse(t, p1, p2, ctx)) - goto err; + if (!BN_mod_inverse(t, p1, p2, ctx)) + goto err; - if (!BN_mul(t, t, p1, ctx)) - goto err; + if (!BN_mul(t, t, p1, ctx)) + goto err; - if (!BN_sub(p, p, t)) - goto err; + if (!BN_sub(p, p, t)) + goto err; - if (p->neg && !BN_add(p, p, p1p2)) - goto err; + if (p->neg && !BN_add(p, p, p1p2)) + goto err; - /* p now equals Rp */ + /* p now equals Rp */ - if (!BN_mod_sub(p, p, Xp, p1p2, ctx)) - goto err; + if (!BN_mod_sub(p, p, Xp, p1p2, ctx)) + goto err; - if (!BN_add(p, p, Xp)) - goto err; + if (!BN_add(p, p, Xp)) + goto err; - /* p now equals Yp0 */ + /* p now equals Yp0 */ - for (;;) - { - int i = 1; - BN_GENCB_call(cb, 0, i++); - if (!BN_copy(pm1, p)) - goto err; - if (!BN_sub_word(pm1, 1)) - goto err; - if (!BN_gcd(t, pm1, e, ctx)) - goto err; - if (BN_is_one(t) - /* X9.31 specifies 8 MR and 1 Lucas test or any prime test - * offering similar or better guarantees 50 MR is considerably - * better. - */ - && BN_is_prime_fasttest_ex(p, 50, ctx, 1, cb)) - break; - if (!BN_add(p, p, p1p2)) - goto err; - } + for (;;) { + int i = 1; + BN_GENCB_call(cb, 0, i++); + if (!BN_copy(pm1, p)) + goto err; + if (!BN_sub_word(pm1, 1)) + goto err; + if (!BN_gcd(t, pm1, e, ctx)) + goto err; + if (BN_is_one(t) + /* + * X9.31 specifies 8 MR and 1 Lucas test or any prime test + * offering similar or better guarantees 50 MR is considerably + * better. + */ + && BN_is_prime_fasttest_ex(p, 50, ctx, 1, cb)) + break; + if (!BN_add(p, p, p1p2)) + goto err; + } - BN_GENCB_call(cb, 3, 0); + BN_GENCB_call(cb, 3, 0); - ret = 1; + ret = 1; - err: + err: - BN_CTX_end(ctx); + BN_CTX_end(ctx); - return ret; - } + return ret; +} -/* Generate pair of paramters Xp, Xq for X9.31 prime generation. - * Note: nbits paramter is sum of number of bits in both. +/* + * Generate pair of paramters Xp, Xq for X9.31 prime generation. Note: nbits + * paramter is sum of number of bits in both. */ int BN_X931_generate_Xpq(BIGNUM *Xp, BIGNUM *Xq, int nbits, BN_CTX *ctx) - { - BIGNUM *t; - int i; - /* Number of bits for each prime is of the form - * 512+128s for s = 0, 1, ... - */ - if ((nbits < 1024) || (nbits & 0xff)) - return 0; - nbits >>= 1; - /* The random value Xp must be between sqrt(2) * 2^(nbits-1) and - * 2^nbits - 1. By setting the top two bits we ensure that the lower - * bound is exceeded. - */ - if (!BN_rand(Xp, nbits, 1, 0)) - return 0; - - BN_CTX_start(ctx); - t = BN_CTX_get(ctx); - - for (i = 0; i < 1000; i++) - { - if (!BN_rand(Xq, nbits, 1, 0)) - return 0; - /* Check that |Xp - Xq| > 2^(nbits - 100) */ - BN_sub(t, Xp, Xq); - if (BN_num_bits(t) > (nbits - 100)) - break; - } - - BN_CTX_end(ctx); - - if (i < 1000) - return 1; - - return 0; - - } - -/* Generate primes using X9.31 algorithm. Of the values p, p1, p2, Xp1 - * and Xp2 only 'p' needs to be non-NULL. If any of the others are not NULL - * the relevant parameter will be stored in it. - * - * Due to the fact that |Xp - Xq| > 2^(nbits - 100) must be satisfied Xp and Xq - * are generated using the previous function and supplied as input. +{ + BIGNUM *t; + int i; + /* + * Number of bits for each prime is of the form 512+128s for s = 0, 1, + * ... + */ + if ((nbits < 1024) || (nbits & 0xff)) + return 0; + nbits >>= 1; + /* + * The random value Xp must be between sqrt(2) * 2^(nbits-1) and 2^nbits + * - 1. By setting the top two bits we ensure that the lower bound is + * exceeded. + */ + if (!BN_rand(Xp, nbits, 1, 0)) + return 0; + + BN_CTX_start(ctx); + t = BN_CTX_get(ctx); + + for (i = 0; i < 1000; i++) { + if (!BN_rand(Xq, nbits, 1, 0)) + return 0; + /* Check that |Xp - Xq| > 2^(nbits - 100) */ + BN_sub(t, Xp, Xq); + if (BN_num_bits(t) > (nbits - 100)) + break; + } + + BN_CTX_end(ctx); + + if (i < 1000) + return 1; + + return 0; + +} + +/* + * Generate primes using X9.31 algorithm. Of the values p, p1, p2, Xp1 and + * Xp2 only 'p' needs to be non-NULL. If any of the others are not NULL the + * relevant parameter will be stored in it. Due to the fact that |Xp - Xq| > + * 2^(nbits - 100) must be satisfied Xp and Xq are generated using the + * previous function and supplied as input. */ int BN_X931_generate_prime_ex(BIGNUM *p, BIGNUM *p1, BIGNUM *p2, - BIGNUM *Xp1, BIGNUM *Xp2, - const BIGNUM *Xp, - const BIGNUM *e, BN_CTX *ctx, - BN_GENCB *cb) - { - int ret = 0; - - BN_CTX_start(ctx); - if (!Xp1) - Xp1 = BN_CTX_get(ctx); - if (!Xp2) - Xp2 = BN_CTX_get(ctx); + BIGNUM *Xp1, BIGNUM *Xp2, + const BIGNUM *Xp, + const BIGNUM *e, BN_CTX *ctx, BN_GENCB *cb) +{ + int ret = 0; - if (!BN_rand(Xp1, 101, 0, 0)) - goto error; - if (!BN_rand(Xp2, 101, 0, 0)) - goto error; - if (!BN_X931_derive_prime_ex(p, p1, p2, Xp, Xp1, Xp2, e, ctx, cb)) - goto error; + BN_CTX_start(ctx); + if (!Xp1) + Xp1 = BN_CTX_get(ctx); + if (!Xp2) + Xp2 = BN_CTX_get(ctx); - ret = 1; + if (!BN_rand(Xp1, 101, 0, 0)) + goto error; + if (!BN_rand(Xp2, 101, 0, 0)) + goto error; + if (!BN_X931_derive_prime_ex(p, p1, p2, Xp, Xp1, Xp2, e, ctx, cb)) + goto error; - error: - BN_CTX_end(ctx); + ret = 1; - return ret; + error: + BN_CTX_end(ctx); - } + return ret; +} diff --git a/openssl/crypto/bn/bnspeed.c b/openssl/crypto/bn/bnspeed.c index b554ac8cf..e387fdfbc 100644 --- a/openssl/crypto/bn/bnspeed.c +++ b/openssl/crypto/bn/bnspeed.c @@ -7,21 +7,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -36,10 +36,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -51,7 +51,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -60,7 +60,7 @@ /* most of this code has been pilfered from my libdes speed.c program */ -#define BASENUM 1000000 +#define BASENUM 1000000 #undef PROG #define PROG bnspeed_main @@ -72,33 +72,35 @@ #include <openssl/err.h> #if !defined(OPENSSL_SYS_MSDOS) && (!defined(OPENSSL_SYS_VMS) || defined(__DECC)) && !defined(OPENSSL_SYS_MACOSX) -#define TIMES +# define TIMES #endif #ifndef _IRIX -#include <time.h> +# include <time.h> #endif #ifdef TIMES -#include <sys/types.h> -#include <sys/times.h> +# include <sys/types.h> +# include <sys/times.h> #endif -/* Depending on the VMS version, the tms structure is perhaps defined. - The __TMS macro will show if it was. If it wasn't defined, we should - undefine TIMES, since that tells the rest of the program how things - should be handled. -- Richard Levitte */ +/* + * Depending on the VMS version, the tms structure is perhaps defined. The + * __TMS macro will show if it was. If it wasn't defined, we should undefine + * TIMES, since that tells the rest of the program how things should be + * handled. -- Richard Levitte + */ #if defined(OPENSSL_SYS_VMS_DECC) && !defined(__TMS) -#undef TIMES +# undef TIMES #endif #ifndef TIMES -#include <sys/timeb.h> +# include <sys/timeb.h> #endif #if defined(sun) || defined(__ultrix) -#define _POSIX_SOURCE -#include <limits.h> -#include <sys/param.h> +# define _POSIX_SOURCE +# include <limits.h> +# include <sys/param.h> #endif #include <openssl/bn.h> @@ -107,127 +109,124 @@ /* The following if from times(3) man page. It may need to be changed */ #ifndef HZ # ifndef CLK_TCK -# ifndef _BSD_CLK_TCK_ /* FreeBSD hack */ -# define HZ 100.0 -# else /* _BSD_CLK_TCK_ */ +# ifndef _BSD_CLK_TCK_ /* FreeBSD hack */ +# define HZ 100.0 +# else /* _BSD_CLK_TCK_ */ # define HZ ((double)_BSD_CLK_TCK_) # endif -# else /* CLK_TCK */ +# else /* CLK_TCK */ # define HZ ((double)CLK_TCK) # endif #endif #undef BUFSIZE -#define BUFSIZE ((long)1024*8) -int run=0; +#define BUFSIZE ((long)1024*8) +int run = 0; static double Time_F(int s); -#define START 0 -#define STOP 1 +#define START 0 +#define STOP 1 static double Time_F(int s) - { - double ret; +{ + double ret; #ifdef TIMES - static struct tms tstart,tend; - - if (s == START) - { - times(&tstart); - return(0); - } - else - { - times(&tend); - ret=((double)(tend.tms_utime-tstart.tms_utime))/HZ; - return((ret < 1e-3)?1e-3:ret); - } -#else /* !times() */ - static struct timeb tstart,tend; - long i; - - if (s == START) - { - ftime(&tstart); - return(0); - } - else - { - ftime(&tend); - i=(long)tend.millitm-(long)tstart.millitm; - ret=((double)(tend.time-tstart.time))+((double)i)/1000.0; - return((ret < 0.001)?0.001:ret); - } + static struct tms tstart, tend; + + if (s == START) { + times(&tstart); + return (0); + } else { + times(&tend); + ret = ((double)(tend.tms_utime - tstart.tms_utime)) / HZ; + return ((ret < 1e-3) ? 1e-3 : ret); + } +#else /* !times() */ + static struct timeb tstart, tend; + long i; + + if (s == START) { + ftime(&tstart); + return (0); + } else { + ftime(&tend); + i = (long)tend.millitm - (long)tstart.millitm; + ret = ((double)(tend.time - tstart.time)) + ((double)i) / 1000.0; + return ((ret < 0.001) ? 0.001 : ret); + } #endif - } +} + +#define NUM_SIZES 5 +static int sizes[NUM_SIZES] = { 128, 256, 512, 1024, 2048 }; -#define NUM_SIZES 5 -static int sizes[NUM_SIZES]={128,256,512,1024,2048}; -/*static int sizes[NUM_SIZES]={59,179,299,419,539}; */ +/* + * static int sizes[NUM_SIZES]={59,179,299,419,539}; + */ -void do_mul(BIGNUM *r,BIGNUM *a,BIGNUM *b,BN_CTX *ctx); +void do_mul(BIGNUM *r, BIGNUM *a, BIGNUM *b, BN_CTX *ctx); int main(int argc, char **argv) - { - BN_CTX *ctx; - BIGNUM a,b,c; +{ + BN_CTX *ctx; + BIGNUM a, b, c; - ctx=BN_CTX_new(); - BN_init(&a); - BN_init(&b); - BN_init(&c); + ctx = BN_CTX_new(); + BN_init(&a); + BN_init(&b); + BN_init(&c); - do_mul(&a,&b,&c,ctx); - } + do_mul(&a, &b, &c, ctx); +} void do_mul(BIGNUM *r, BIGNUM *a, BIGNUM *b, BN_CTX *ctx) - { - int i,j,k; - double tm; - long num; - - for (i=0; i<NUM_SIZES; i++) - { - num=BASENUM; - if (i) num/=(i*3); - BN_rand(a,sizes[i],1,0); - for (j=i; j<NUM_SIZES; j++) - { - BN_rand(b,sizes[j],1,0); - Time_F(START); - for (k=0; k<num; k++) - BN_mul(r,b,a,ctx); - tm=Time_F(STOP); - printf("mul %4d x %4d -> %8.3fms\n",sizes[i],sizes[j],tm*1000.0/num); - } - } - - for (i=0; i<NUM_SIZES; i++) - { - num=BASENUM; - if (i) num/=(i*3); - BN_rand(a,sizes[i],1,0); - Time_F(START); - for (k=0; k<num; k++) - BN_sqr(r,a,ctx); - tm=Time_F(STOP); - printf("sqr %4d x %4d -> %8.3fms\n",sizes[i],sizes[i],tm*1000.0/num); - } - - for (i=0; i<NUM_SIZES; i++) - { - num=BASENUM/10; - if (i) num/=(i*3); - BN_rand(a,sizes[i]-1,1,0); - for (j=i; j<NUM_SIZES; j++) - { - BN_rand(b,sizes[j],1,0); - Time_F(START); - for (k=0; k<100000; k++) - BN_div(r, NULL, b, a,ctx); - tm=Time_F(STOP); - printf("div %4d / %4d -> %8.3fms\n",sizes[j],sizes[i]-1,tm*1000.0/num); - } - } - } - +{ + int i, j, k; + double tm; + long num; + + for (i = 0; i < NUM_SIZES; i++) { + num = BASENUM; + if (i) + num /= (i * 3); + BN_rand(a, sizes[i], 1, 0); + for (j = i; j < NUM_SIZES; j++) { + BN_rand(b, sizes[j], 1, 0); + Time_F(START); + for (k = 0; k < num; k++) + BN_mul(r, b, a, ctx); + tm = Time_F(STOP); + printf("mul %4d x %4d -> %8.3fms\n", sizes[i], sizes[j], + tm * 1000.0 / num); + } + } + + for (i = 0; i < NUM_SIZES; i++) { + num = BASENUM; + if (i) + num /= (i * 3); + BN_rand(a, sizes[i], 1, 0); + Time_F(START); + for (k = 0; k < num; k++) + BN_sqr(r, a, ctx); + tm = Time_F(STOP); + printf("sqr %4d x %4d -> %8.3fms\n", sizes[i], sizes[i], + tm * 1000.0 / num); + } + + for (i = 0; i < NUM_SIZES; i++) { + num = BASENUM / 10; + if (i) + num /= (i * 3); + BN_rand(a, sizes[i] - 1, 1, 0); + for (j = i; j < NUM_SIZES; j++) { + BN_rand(b, sizes[j], 1, 0); + Time_F(START); + for (k = 0; k < 100000; k++) + BN_div(r, NULL, b, a, ctx); + tm = Time_F(STOP); + printf("div %4d / %4d -> %8.3fms\n", sizes[j], sizes[i] - 1, + tm * 1000.0 / num); + } + } +} diff --git a/openssl/crypto/bn/bntest.c b/openssl/crypto/bn/bntest.c index 06f5954ac..06662c58b 100644 --- a/openssl/crypto/bn/bntest.c +++ b/openssl/crypto/bn/bntest.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -58,21 +58,23 @@ /* ==================================================================== * Copyright 2002 Sun Microsystems, Inc. ALL RIGHTS RESERVED. * - * Portions of the attached software ("Contribution") are developed by + * Portions of the attached software ("Contribution") are developed by * SUN MICROSYSTEMS, INC., and are contributed to the OpenSSL project. * * The Contribution is licensed pursuant to the Eric Young open source * license provided above. * - * The binary polynomial arithmetic software is originally written by + * The binary polynomial arithmetic software is originally written by * Sheueling Chang Shantz and Douglas Stebila of Sun Microsystems Laboratories. * */ -/* Until the key-gen callbacks are modified to use newer prototypes, we allow - * deprecated functions for openssl-internal code */ +/* + * Until the key-gen callbacks are modified to use newer prototypes, we allow + * deprecated functions for openssl-internal code + */ #ifdef OPENSSL_NO_DEPRECATED -#undef OPENSSL_NO_DEPRECATED +# undef OPENSSL_NO_DEPRECATED #endif #include <stdio.h> @@ -87,1927 +89,1997 @@ #include <openssl/x509.h> #include <openssl/err.h> -const int num0 = 100; /* number of tests */ -const int num1 = 50; /* additional tests for some functions */ -const int num2 = 5; /* number of tests for slow functions */ +const int num0 = 100; /* number of tests */ +const int num1 = 50; /* additional tests for some functions */ +const int num2 = 5; /* number of tests for slow functions */ int test_add(BIO *bp); int test_sub(BIO *bp); int test_lshift1(BIO *bp); -int test_lshift(BIO *bp,BN_CTX *ctx,BIGNUM *a_); +int test_lshift(BIO *bp, BN_CTX *ctx, BIGNUM *a_); int test_rshift1(BIO *bp); -int test_rshift(BIO *bp,BN_CTX *ctx); -int test_div(BIO *bp,BN_CTX *ctx); +int test_rshift(BIO *bp, BN_CTX *ctx); +int test_div(BIO *bp, BN_CTX *ctx); int test_div_word(BIO *bp); -int test_div_recp(BIO *bp,BN_CTX *ctx); +int test_div_recp(BIO *bp, BN_CTX *ctx); int test_mul(BIO *bp); -int test_sqr(BIO *bp,BN_CTX *ctx); -int test_mont(BIO *bp,BN_CTX *ctx); -int test_mod(BIO *bp,BN_CTX *ctx); -int test_mod_mul(BIO *bp,BN_CTX *ctx); -int test_mod_exp(BIO *bp,BN_CTX *ctx); -int test_mod_exp_mont_consttime(BIO *bp,BN_CTX *ctx); -int test_exp(BIO *bp,BN_CTX *ctx); +int test_sqr(BIO *bp, BN_CTX *ctx); +int test_mont(BIO *bp, BN_CTX *ctx); +int test_mod(BIO *bp, BN_CTX *ctx); +int test_mod_mul(BIO *bp, BN_CTX *ctx); +int test_mod_exp(BIO *bp, BN_CTX *ctx); +int test_mod_exp_mont_consttime(BIO *bp, BN_CTX *ctx); +int test_mod_exp_mont5(BIO *bp, BN_CTX *ctx); +int test_exp(BIO *bp, BN_CTX *ctx); int test_gf2m_add(BIO *bp); int test_gf2m_mod(BIO *bp); -int test_gf2m_mod_mul(BIO *bp,BN_CTX *ctx); -int test_gf2m_mod_sqr(BIO *bp,BN_CTX *ctx); -int test_gf2m_mod_inv(BIO *bp,BN_CTX *ctx); -int test_gf2m_mod_div(BIO *bp,BN_CTX *ctx); -int test_gf2m_mod_exp(BIO *bp,BN_CTX *ctx); -int test_gf2m_mod_sqrt(BIO *bp,BN_CTX *ctx); -int test_gf2m_mod_solve_quad(BIO *bp,BN_CTX *ctx); -int test_kron(BIO *bp,BN_CTX *ctx); -int test_sqrt(BIO *bp,BN_CTX *ctx); +int test_gf2m_mod_mul(BIO *bp, BN_CTX *ctx); +int test_gf2m_mod_sqr(BIO *bp, BN_CTX *ctx); +int test_gf2m_mod_inv(BIO *bp, BN_CTX *ctx); +int test_gf2m_mod_div(BIO *bp, BN_CTX *ctx); +int test_gf2m_mod_exp(BIO *bp, BN_CTX *ctx); +int test_gf2m_mod_sqrt(BIO *bp, BN_CTX *ctx); +int test_gf2m_mod_solve_quad(BIO *bp, BN_CTX *ctx); +int test_kron(BIO *bp, BN_CTX *ctx); +int test_sqrt(BIO *bp, BN_CTX *ctx); int rand_neg(void); -static int results=0; +static int results = 0; -static unsigned char lst[]="\xC6\x4F\x43\x04\x2A\xEA\xCA\x6E\x58\x36\x80\x5B\xE8\xC9" -"\x9B\x04\x5D\x48\x36\xC2\xFD\x16\xC9\x64\xF0"; +static unsigned char lst[] = + "\xC6\x4F\x43\x04\x2A\xEA\xCA\x6E\x58\x36\x80\x5B\xE8\xC9" + "\x9B\x04\x5D\x48\x36\xC2\xFD\x16\xC9\x64\xF0"; -static const char rnd_seed[] = "string to make the random number generator think it has entropy"; +static const char rnd_seed[] = + "string to make the random number generator think it has entropy"; static void message(BIO *out, char *m) - { - fprintf(stderr, "test %s\n", m); - BIO_puts(out, "print \"test "); - BIO_puts(out, m); - BIO_puts(out, "\\n\"\n"); - } +{ + fprintf(stderr, "test %s\n", m); + BIO_puts(out, "print \"test "); + BIO_puts(out, m); + BIO_puts(out, "\\n\"\n"); +} int main(int argc, char *argv[]) - { - BN_CTX *ctx; - BIO *out; - char *outfile=NULL; - - results = 0; - - RAND_seed(rnd_seed, sizeof rnd_seed); /* or BN_generate_prime may fail */ - - argc--; - argv++; - while (argc >= 1) - { - if (strcmp(*argv,"-results") == 0) - results=1; - else if (strcmp(*argv,"-out") == 0) - { - if (--argc < 1) break; - outfile= *(++argv); - } - argc--; - argv++; - } - - - ctx=BN_CTX_new(); - if (ctx == NULL) EXIT(1); - - out=BIO_new(BIO_s_file()); - if (out == NULL) EXIT(1); - if (outfile == NULL) - { - BIO_set_fp(out,stdout,BIO_NOCLOSE); - } - else - { - if (!BIO_write_filename(out,outfile)) - { - perror(outfile); - EXIT(1); - } - } - - if (!results) - BIO_puts(out,"obase=16\nibase=16\n"); - - message(out,"BN_add"); - if (!test_add(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_sub"); - if (!test_sub(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_lshift1"); - if (!test_lshift1(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_lshift (fixed)"); - if (!test_lshift(out,ctx,BN_bin2bn(lst,sizeof(lst)-1,NULL))) - goto err; - (void)BIO_flush(out); - - message(out,"BN_lshift"); - if (!test_lshift(out,ctx,NULL)) goto err; - (void)BIO_flush(out); - - message(out,"BN_rshift1"); - if (!test_rshift1(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_rshift"); - if (!test_rshift(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_sqr"); - if (!test_sqr(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mul"); - if (!test_mul(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_div"); - if (!test_div(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_div_word"); - if (!test_div_word(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_div_recp"); - if (!test_div_recp(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mod"); - if (!test_mod(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mod_mul"); - if (!test_mod_mul(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mont"); - if (!test_mont(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mod_exp"); - if (!test_mod_exp(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mod_exp_mont_consttime"); - if (!test_mod_exp_mont_consttime(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_exp"); - if (!test_exp(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_kronecker"); - if (!test_kron(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_mod_sqrt"); - if (!test_sqrt(out,ctx)) goto err; - (void)BIO_flush(out); +{ + BN_CTX *ctx; + BIO *out; + char *outfile = NULL; + + results = 0; + + RAND_seed(rnd_seed, sizeof rnd_seed); /* or BN_generate_prime may fail */ + + argc--; + argv++; + while (argc >= 1) { + if (strcmp(*argv, "-results") == 0) + results = 1; + else if (strcmp(*argv, "-out") == 0) { + if (--argc < 1) + break; + outfile = *(++argv); + } + argc--; + argv++; + } + + ctx = BN_CTX_new(); + if (ctx == NULL) + EXIT(1); + + out = BIO_new(BIO_s_file()); + if (out == NULL) + EXIT(1); + if (outfile == NULL) { + BIO_set_fp(out, stdout, BIO_NOCLOSE); + } else { + if (!BIO_write_filename(out, outfile)) { + perror(outfile); + EXIT(1); + } + } + + if (!results) + BIO_puts(out, "obase=16\nibase=16\n"); + + message(out, "BN_add"); + if (!test_add(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_sub"); + if (!test_sub(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_lshift1"); + if (!test_lshift1(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_lshift (fixed)"); + if (!test_lshift(out, ctx, BN_bin2bn(lst, sizeof(lst) - 1, NULL))) + goto err; + (void)BIO_flush(out); + + message(out, "BN_lshift"); + if (!test_lshift(out, ctx, NULL)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_rshift1"); + if (!test_rshift1(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_rshift"); + if (!test_rshift(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_sqr"); + if (!test_sqr(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mul"); + if (!test_mul(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_div"); + if (!test_div(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_div_word"); + if (!test_div_word(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_div_recp"); + if (!test_div_recp(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mod"); + if (!test_mod(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mod_mul"); + if (!test_mod_mul(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mont"); + if (!test_mont(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mod_exp"); + if (!test_mod_exp(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mod_exp_mont_consttime"); + if (!test_mod_exp_mont_consttime(out, ctx)) + goto err; + if (!test_mod_exp_mont5(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_exp"); + if (!test_exp(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_kronecker"); + if (!test_kron(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_mod_sqrt"); + if (!test_sqrt(out, ctx)) + goto err; + (void)BIO_flush(out); #ifndef OPENSSL_NO_EC2M - message(out,"BN_GF2m_add"); - if (!test_gf2m_add(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod"); - if (!test_gf2m_mod(out)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_mul"); - if (!test_gf2m_mod_mul(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_sqr"); - if (!test_gf2m_mod_sqr(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_inv"); - if (!test_gf2m_mod_inv(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_div"); - if (!test_gf2m_mod_div(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_exp"); - if (!test_gf2m_mod_exp(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_sqrt"); - if (!test_gf2m_mod_sqrt(out,ctx)) goto err; - (void)BIO_flush(out); - - message(out,"BN_GF2m_mod_solve_quad"); - if (!test_gf2m_mod_solve_quad(out,ctx)) goto err; - (void)BIO_flush(out); + message(out, "BN_GF2m_add"); + if (!test_gf2m_add(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod"); + if (!test_gf2m_mod(out)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_mul"); + if (!test_gf2m_mod_mul(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_sqr"); + if (!test_gf2m_mod_sqr(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_inv"); + if (!test_gf2m_mod_inv(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_div"); + if (!test_gf2m_mod_div(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_exp"); + if (!test_gf2m_mod_exp(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_sqrt"); + if (!test_gf2m_mod_sqrt(out, ctx)) + goto err; + (void)BIO_flush(out); + + message(out, "BN_GF2m_mod_solve_quad"); + if (!test_gf2m_mod_solve_quad(out, ctx)) + goto err; + (void)BIO_flush(out); #endif - BN_CTX_free(ctx); - BIO_free(out); - -/**/ - EXIT(0); -err: - BIO_puts(out,"1\n"); /* make sure the Perl script fed by bc notices - * the failure, see test_bn in test/Makefile.ssl*/ - (void)BIO_flush(out); - ERR_load_crypto_strings(); - ERR_print_errors_fp(stderr); - EXIT(1); - return(1); - } + BN_CTX_free(ctx); + BIO_free(out); + + EXIT(0); + err: + BIO_puts(out, "1\n"); /* make sure the Perl script fed by bc + * notices the failure, see test_bn in + * test/Makefile.ssl */ + (void)BIO_flush(out); + ERR_load_crypto_strings(); + ERR_print_errors_fp(stderr); + EXIT(1); + return (1); +} int test_add(BIO *bp) - { - BIGNUM a,b,c; - int i; - - BN_init(&a); - BN_init(&b); - BN_init(&c); - - BN_bntest_rand(&a,512,0,0); - for (i=0; i<num0; i++) - { - BN_bntest_rand(&b,450+i,0,0); - a.neg=rand_neg(); - b.neg=rand_neg(); - BN_add(&c,&a,&b); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," + "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } - a.neg=!a.neg; - b.neg=!b.neg; - BN_add(&c,&c,&b); - BN_add(&c,&c,&a); - if(!BN_is_zero(&c)) - { - fprintf(stderr,"Add test failed!\n"); - return 0; - } - } - BN_free(&a); - BN_free(&b); - BN_free(&c); - return(1); - } +{ + BIGNUM a, b, c; + int i; + + BN_init(&a); + BN_init(&b); + BN_init(&c); + + BN_bntest_rand(&a, 512, 0, 0); + for (i = 0; i < num0; i++) { + BN_bntest_rand(&b, 450 + i, 0, 0); + a.neg = rand_neg(); + b.neg = rand_neg(); + BN_add(&c, &a, &b); + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " + "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &c); + BIO_puts(bp, "\n"); + } + a.neg = !a.neg; + b.neg = !b.neg; + BN_add(&c, &c, &b); + BN_add(&c, &c, &a); + if (!BN_is_zero(&c)) { + fprintf(stderr, "Add test failed!\n"); + return 0; + } + } + BN_free(&a); + BN_free(&b); + BN_free(&c); + return (1); +} int test_sub(BIO *bp) - { - BIGNUM a,b,c; - int i; - - BN_init(&a); - BN_init(&b); - BN_init(&c); - - for (i=0; i<num0+num1; i++) - { - if (i < num1) - { - BN_bntest_rand(&a,512,0,0); - BN_copy(&b,&a); - if (BN_set_bit(&a,i)==0) return(0); - BN_add_word(&b,i); - } - else - { - BN_bntest_rand(&b,400+i-num1,0,0); - a.neg=rand_neg(); - b.neg=rand_neg(); - } - BN_sub(&c,&a,&b); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," - "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } - BN_add(&c,&c,&b); - BN_sub(&c,&c,&a); - if(!BN_is_zero(&c)) - { - fprintf(stderr,"Subtract test failed!\n"); - return 0; - } - } - BN_free(&a); - BN_free(&b); - BN_free(&c); - return(1); - } +{ + BIGNUM a, b, c; + int i; + + BN_init(&a); + BN_init(&b); + BN_init(&c); + + for (i = 0; i < num0 + num1; i++) { + if (i < num1) { + BN_bntest_rand(&a, 512, 0, 0); + BN_copy(&b, &a); + if (BN_set_bit(&a, i) == 0) + return (0); + BN_add_word(&b, i); + } else { + BN_bntest_rand(&b, 400 + i - num1, 0, 0); + a.neg = rand_neg(); + b.neg = rand_neg(); + } + BN_sub(&c, &a, &b); + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " - "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &c); + BIO_puts(bp, "\n"); + } + BN_add(&c, &c, &b); + BN_sub(&c, &c, &a); + if (!BN_is_zero(&c)) { + fprintf(stderr, "Subtract test failed!\n"); + return 0; + } + } + BN_free(&a); + BN_free(&b); + BN_free(&c); + return (1); +} int test_div(BIO *bp, BN_CTX *ctx) - { - BIGNUM a,b,c,d,e; - int i; - - BN_init(&a); - BN_init(&b); - BN_init(&c); - BN_init(&d); - BN_init(&e); - - for (i=0; i<num0+num1; i++) - { - if (i < num1) - { - BN_bntest_rand(&a,400,0,0); - BN_copy(&b,&a); - BN_lshift(&a,&a,i); - BN_add_word(&a,i); - } - else - BN_bntest_rand(&b,50+3*(i-num1),0,0); - a.neg=rand_neg(); - b.neg=rand_neg(); - BN_div(&d,&c,&a,&b,ctx); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," / "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&d); - BIO_puts(bp,"\n"); - - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," % "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } - BN_mul(&e,&d,&b,ctx); - BN_add(&d,&e,&c); - BN_sub(&d,&d,&a); - if(!BN_is_zero(&d)) - { - fprintf(stderr,"Division test failed!\n"); - return 0; - } - } - BN_free(&a); - BN_free(&b); - BN_free(&c); - BN_free(&d); - BN_free(&e); - return(1); - } - -static void print_word(BIO *bp,BN_ULONG w) - { +{ + BIGNUM a, b, c, d, e; + int i; + + BN_init(&a); + BN_init(&b); + BN_init(&c); + BN_init(&d); + BN_init(&e); + + for (i = 0; i < num0 + num1; i++) { + if (i < num1) { + BN_bntest_rand(&a, 400, 0, 0); + BN_copy(&b, &a); + BN_lshift(&a, &a, i); + BN_add_word(&a, i); + } else + BN_bntest_rand(&b, 50 + 3 * (i - num1), 0, 0); + a.neg = rand_neg(); + b.neg = rand_neg(); + BN_div(&d, &c, &a, &b, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " / "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &d); + BIO_puts(bp, "\n"); + + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " % "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &c); + BIO_puts(bp, "\n"); + } + BN_mul(&e, &d, &b, ctx); + BN_add(&d, &e, &c); + BN_sub(&d, &d, &a); + if (!BN_is_zero(&d)) { + fprintf(stderr, "Division test failed!\n"); + return 0; + } + } + BN_free(&a); + BN_free(&b); + BN_free(&c); + BN_free(&d); + BN_free(&e); + return (1); +} + +static void print_word(BIO *bp, BN_ULONG w) +{ #ifdef SIXTY_FOUR_BIT - if (sizeof(w) > sizeof(unsigned long)) - { - unsigned long h=(unsigned long)(w>>32), - l=(unsigned long)(w); - - if (h) BIO_printf(bp,"%lX%08lX",h,l); - else BIO_printf(bp,"%lX",l); - return; - } + if (sizeof(w) > sizeof(unsigned long)) { + unsigned long h = (unsigned long)(w >> 32), l = (unsigned long)(w); + + if (h) + BIO_printf(bp, "%lX%08lX", h, l); + else + BIO_printf(bp, "%lX", l); + return; + } #endif - BIO_printf(bp,BN_HEX_FMT1,w); - } + BIO_printf(bp, BN_HEX_FMT1, w); +} int test_div_word(BIO *bp) - { - BIGNUM a,b; - BN_ULONG r,s; - int i; - - BN_init(&a); - BN_init(&b); - - for (i=0; i<num0; i++) - { - do { - BN_bntest_rand(&a,512,-1,0); - BN_bntest_rand(&b,BN_BITS2,-1,0); - s = b.d[0]; - } while (!s); - - BN_copy(&b, &a); - r = BN_div_word(&b, s); - - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," / "); - print_word(bp,s); - BIO_puts(bp," - "); - } - BN_print(bp,&b); - BIO_puts(bp,"\n"); - - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," % "); - print_word(bp,s); - BIO_puts(bp," - "); - } - print_word(bp,r); - BIO_puts(bp,"\n"); - } - BN_mul_word(&b,s); - BN_add_word(&b,r); - BN_sub(&b,&a,&b); - if(!BN_is_zero(&b)) - { - fprintf(stderr,"Division (word) test failed!\n"); - return 0; - } - } - BN_free(&a); - BN_free(&b); - return(1); - } +{ + BIGNUM a, b; + BN_ULONG r, s; + int i; + + BN_init(&a); + BN_init(&b); + + for (i = 0; i < num0; i++) { + do { + BN_bntest_rand(&a, 512, -1, 0); + BN_bntest_rand(&b, BN_BITS2, -1, 0); + s = b.d[0]; + } while (!s); + + BN_copy(&b, &a); + r = BN_div_word(&b, s); + + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " / "); + print_word(bp, s); + BIO_puts(bp, " - "); + } + BN_print(bp, &b); + BIO_puts(bp, "\n"); + + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " % "); + print_word(bp, s); + BIO_puts(bp, " - "); + } + print_word(bp, r); + BIO_puts(bp, "\n"); + } + BN_mul_word(&b, s); + BN_add_word(&b, r); + BN_sub(&b, &a, &b); + if (!BN_is_zero(&b)) { + fprintf(stderr, "Division (word) test failed!\n"); + return 0; + } + } + BN_free(&a); + BN_free(&b); + return (1); +} int test_div_recp(BIO *bp, BN_CTX *ctx) - { - BIGNUM a,b,c,d,e; - BN_RECP_CTX recp; - int i; - - BN_RECP_CTX_init(&recp); - BN_init(&a); - BN_init(&b); - BN_init(&c); - BN_init(&d); - BN_init(&e); - - for (i=0; i<num0+num1; i++) - { - if (i < num1) - { - BN_bntest_rand(&a,400,0,0); - BN_copy(&b,&a); - BN_lshift(&a,&a,i); - BN_add_word(&a,i); - } - else - BN_bntest_rand(&b,50+3*(i-num1),0,0); - a.neg=rand_neg(); - b.neg=rand_neg(); - BN_RECP_CTX_set(&recp,&b,ctx); - BN_div_recp(&d,&c,&a,&recp,ctx); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," / "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&d); - BIO_puts(bp,"\n"); - - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," % "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } - BN_mul(&e,&d,&b,ctx); - BN_add(&d,&e,&c); - BN_sub(&d,&d,&a); - if(!BN_is_zero(&d)) - { - fprintf(stderr,"Reciprocal division test failed!\n"); - fprintf(stderr,"a="); - BN_print_fp(stderr,&a); - fprintf(stderr,"\nb="); - BN_print_fp(stderr,&b); - fprintf(stderr,"\n"); - return 0; - } - } - BN_free(&a); - BN_free(&b); - BN_free(&c); - BN_free(&d); - BN_free(&e); - BN_RECP_CTX_free(&recp); - return(1); - } +{ + BIGNUM a, b, c, d, e; + BN_RECP_CTX recp; + int i; + + BN_RECP_CTX_init(&recp); + BN_init(&a); + BN_init(&b); + BN_init(&c); + BN_init(&d); + BN_init(&e); + + for (i = 0; i < num0 + num1; i++) { + if (i < num1) { + BN_bntest_rand(&a, 400, 0, 0); + BN_copy(&b, &a); + BN_lshift(&a, &a, i); + BN_add_word(&a, i); + } else + BN_bntest_rand(&b, 50 + 3 * (i - num1), 0, 0); + a.neg = rand_neg(); + b.neg = rand_neg(); + BN_RECP_CTX_set(&recp, &b, ctx); + BN_div_recp(&d, &c, &a, &recp, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " / "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &d); + BIO_puts(bp, "\n"); + + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " % "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &c); + BIO_puts(bp, "\n"); + } + BN_mul(&e, &d, &b, ctx); + BN_add(&d, &e, &c); + BN_sub(&d, &d, &a); + if (!BN_is_zero(&d)) { + fprintf(stderr, "Reciprocal division test failed!\n"); + fprintf(stderr, "a="); + BN_print_fp(stderr, &a); + fprintf(stderr, "\nb="); + BN_print_fp(stderr, &b); + fprintf(stderr, "\n"); + return 0; + } + } + BN_free(&a); + BN_free(&b); + BN_free(&c); + BN_free(&d); + BN_free(&e); + BN_RECP_CTX_free(&recp); + return (1); +} int test_mul(BIO *bp) - { - BIGNUM a,b,c,d,e; - int i; - BN_CTX *ctx; - - ctx = BN_CTX_new(); - if (ctx == NULL) EXIT(1); - - BN_init(&a); - BN_init(&b); - BN_init(&c); - BN_init(&d); - BN_init(&e); - - for (i=0; i<num0+num1; i++) - { - if (i <= num1) - { - BN_bntest_rand(&a,100,0,0); - BN_bntest_rand(&b,100,0,0); - } - else - BN_bntest_rand(&b,i-num1,0,0); - a.neg=rand_neg(); - b.neg=rand_neg(); - BN_mul(&c,&a,&b,ctx); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," * "); - BN_print(bp,&b); - BIO_puts(bp," - "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } - BN_div(&d,&e,&c,&a,ctx); - BN_sub(&d,&d,&b); - if(!BN_is_zero(&d) || !BN_is_zero(&e)) - { - fprintf(stderr,"Multiplication test failed!\n"); - return 0; - } - } - BN_free(&a); - BN_free(&b); - BN_free(&c); - BN_free(&d); - BN_free(&e); - BN_CTX_free(ctx); - return(1); - } +{ + BIGNUM a, b, c, d, e; + int i; + BN_CTX *ctx; + + ctx = BN_CTX_new(); + if (ctx == NULL) + EXIT(1); + + BN_init(&a); + BN_init(&b); + BN_init(&c); + BN_init(&d); + BN_init(&e); + + for (i = 0; i < num0 + num1; i++) { + if (i <= num1) { + BN_bntest_rand(&a, 100, 0, 0); + BN_bntest_rand(&b, 100, 0, 0); + } else + BN_bntest_rand(&b, i - num1, 0, 0); + a.neg = rand_neg(); + b.neg = rand_neg(); + BN_mul(&c, &a, &b, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " * "); + BN_print(bp, &b); + BIO_puts(bp, " - "); + } + BN_print(bp, &c); + BIO_puts(bp, "\n"); + } + BN_div(&d, &e, &c, &a, ctx); + BN_sub(&d, &d, &b); + if (!BN_is_zero(&d) || !BN_is_zero(&e)) { + fprintf(stderr, "Multiplication test failed!\n"); + return 0; + } + } + BN_free(&a); + BN_free(&b); + BN_free(&c); + BN_free(&d); + BN_free(&e); + BN_CTX_free(ctx); + return (1); +} int test_sqr(BIO *bp, BN_CTX *ctx) - { - BIGNUM a,c,d,e; - int i; - - BN_init(&a); - BN_init(&c); - BN_init(&d); - BN_init(&e); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(&a,40+i*10,0,0); - a.neg=rand_neg(); - BN_sqr(&c,&a,ctx); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," * "); - BN_print(bp,&a); - BIO_puts(bp," - "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } - BN_div(&d,&e,&c,&a,ctx); - BN_sub(&d,&d,&a); - if(!BN_is_zero(&d) || !BN_is_zero(&e)) - { - fprintf(stderr,"Square test failed!\n"); - return 0; - } - } - BN_free(&a); - BN_free(&c); - BN_free(&d); - BN_free(&e); - return(1); - } +{ + BIGNUM *a, *c, *d, *e; + int i, ret = 0; + + a = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + if (a == NULL || c == NULL || d == NULL || e == NULL) { + goto err; + } + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 40 + i * 10, 0, 0); + a->neg = rand_neg(); + BN_sqr(c, a, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, a); + BIO_puts(bp, " - "); + } + BN_print(bp, c); + BIO_puts(bp, "\n"); + } + BN_div(d, e, c, a, ctx); + BN_sub(d, d, a); + if (!BN_is_zero(d) || !BN_is_zero(e)) { + fprintf(stderr, "Square test failed!\n"); + goto err; + } + } + + /* Regression test for a BN_sqr overflow bug. */ + BN_hex2bn(&a, + "80000000000000008000000000000001" + "FFFFFFFFFFFFFFFE0000000000000000"); + BN_sqr(c, a, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, a); + BIO_puts(bp, " - "); + } + BN_print(bp, c); + BIO_puts(bp, "\n"); + } + BN_mul(d, a, a, ctx); + if (BN_cmp(c, d)) { + fprintf(stderr, "Square test failed: BN_sqr and BN_mul produce " + "different results!\n"); + goto err; + } + + /* Regression test for a BN_sqr overflow bug. */ + BN_hex2bn(&a, + "80000000000000000000000080000001" + "FFFFFFFE000000000000000000000000"); + BN_sqr(c, a, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, a); + BIO_puts(bp, " - "); + } + BN_print(bp, c); + BIO_puts(bp, "\n"); + } + BN_mul(d, a, a, ctx); + if (BN_cmp(c, d)) { + fprintf(stderr, "Square test failed: BN_sqr and BN_mul produce " + "different results!\n"); + goto err; + } + ret = 1; + err: + if (a != NULL) + BN_free(a); + if (c != NULL) + BN_free(c); + if (d != NULL) + BN_free(d); + if (e != NULL) + BN_free(e); + return ret; +} int test_mont(BIO *bp, BN_CTX *ctx) - { - BIGNUM a,b,c,d,A,B; - BIGNUM n; - int i; - BN_MONT_CTX *mont; - - BN_init(&a); - BN_init(&b); - BN_init(&c); - BN_init(&d); - BN_init(&A); - BN_init(&B); - BN_init(&n); - - mont=BN_MONT_CTX_new(); - if (mont == NULL) - return 0; - - BN_bntest_rand(&a,100,0,0); /**/ - BN_bntest_rand(&b,100,0,0); /**/ - for (i=0; i<num2; i++) - { - int bits = (200*(i+1))/num2; - - if (bits == 0) - continue; - BN_bntest_rand(&n,bits,0,1); - BN_MONT_CTX_set(mont,&n,ctx); - - BN_nnmod(&a,&a,&n,ctx); - BN_nnmod(&b,&b,&n,ctx); - - BN_to_montgomery(&A,&a,mont,ctx); - BN_to_montgomery(&B,&b,mont,ctx); - - BN_mod_mul_montgomery(&c,&A,&B,mont,ctx);/**/ - BN_from_montgomery(&A,&c,mont,ctx);/**/ - if (bp != NULL) - { - if (!results) - { +{ + BIGNUM a, b, c, d, A, B; + BIGNUM n; + int i; + BN_MONT_CTX *mont; + + BN_init(&a); + BN_init(&b); + BN_init(&c); + BN_init(&d); + BN_init(&A); + BN_init(&B); + BN_init(&n); + + mont = BN_MONT_CTX_new(); + if (mont == NULL) + return 0; + + BN_bntest_rand(&a, 100, 0, 0); + BN_bntest_rand(&b, 100, 0, 0); + for (i = 0; i < num2; i++) { + int bits = (200 * (i + 1)) / num2; + + if (bits == 0) + continue; + BN_bntest_rand(&n, bits, 0, 1); + BN_MONT_CTX_set(mont, &n, ctx); + + BN_nnmod(&a, &a, &n, ctx); + BN_nnmod(&b, &b, &n, ctx); + + BN_to_montgomery(&A, &a, mont, ctx); + BN_to_montgomery(&B, &b, mont, ctx); + + BN_mod_mul_montgomery(&c, &A, &B, mont, ctx); + BN_from_montgomery(&A, &c, mont, ctx); + if (bp != NULL) { + if (!results) { #ifdef undef -fprintf(stderr,"%d * %d %% %d\n", -BN_num_bits(&a), -BN_num_bits(&b), -BN_num_bits(mont->N)); + fprintf(stderr, "%d * %d %% %d\n", + BN_num_bits(&a), + BN_num_bits(&b), BN_num_bits(mont->N)); #endif - BN_print(bp,&a); - BIO_puts(bp," * "); - BN_print(bp,&b); - BIO_puts(bp," % "); - BN_print(bp,&(mont->N)); - BIO_puts(bp," - "); - } - BN_print(bp,&A); - BIO_puts(bp,"\n"); - } - BN_mod_mul(&d,&a,&b,&n,ctx); - BN_sub(&d,&d,&A); - if(!BN_is_zero(&d)) - { - fprintf(stderr,"Montgomery multiplication test failed!\n"); - return 0; - } - } - BN_MONT_CTX_free(mont); - BN_free(&a); - BN_free(&b); - BN_free(&c); - BN_free(&d); - BN_free(&A); - BN_free(&B); - BN_free(&n); - return(1); - } + BN_print(bp, &a); + BIO_puts(bp, " * "); + BN_print(bp, &b); + BIO_puts(bp, " % "); + BN_print(bp, &(mont->N)); + BIO_puts(bp, " - "); + } + BN_print(bp, &A); + BIO_puts(bp, "\n"); + } + BN_mod_mul(&d, &a, &b, &n, ctx); + BN_sub(&d, &d, &A); + if (!BN_is_zero(&d)) { + fprintf(stderr, "Montgomery multiplication test failed!\n"); + return 0; + } + } + BN_MONT_CTX_free(mont); + BN_free(&a); + BN_free(&b); + BN_free(&c); + BN_free(&d); + BN_free(&A); + BN_free(&B); + BN_free(&n); + return (1); +} int test_mod(BIO *bp, BN_CTX *ctx) - { - BIGNUM *a,*b,*c,*d,*e; - int i; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - - BN_bntest_rand(a,1024,0,0); /**/ - for (i=0; i<num0; i++) - { - BN_bntest_rand(b,450+i*10,0,0); /**/ - a->neg=rand_neg(); - b->neg=rand_neg(); - BN_mod(c,a,b,ctx);/**/ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," % "); - BN_print(bp,b); - BIO_puts(bp," - "); - } - BN_print(bp,c); - BIO_puts(bp,"\n"); - } - BN_div(d,e,a,b,ctx); - BN_sub(e,e,c); - if(!BN_is_zero(e)) - { - fprintf(stderr,"Modulo test failed!\n"); - return 0; - } - } - BN_free(a); - BN_free(b); - BN_free(c); - BN_free(d); - BN_free(e); - return(1); - } +{ + BIGNUM *a, *b, *c, *d, *e; + int i; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + + BN_bntest_rand(a, 1024, 0, 0); + for (i = 0; i < num0; i++) { + BN_bntest_rand(b, 450 + i * 10, 0, 0); + a->neg = rand_neg(); + b->neg = rand_neg(); + BN_mod(c, a, b, ctx); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " % "); + BN_print(bp, b); + BIO_puts(bp, " - "); + } + BN_print(bp, c); + BIO_puts(bp, "\n"); + } + BN_div(d, e, a, b, ctx); + BN_sub(e, e, c); + if (!BN_is_zero(e)) { + fprintf(stderr, "Modulo test failed!\n"); + return 0; + } + } + BN_free(a); + BN_free(b); + BN_free(c); + BN_free(d); + BN_free(e); + return (1); +} int test_mod_mul(BIO *bp, BN_CTX *ctx) - { - BIGNUM *a,*b,*c,*d,*e; - int i,j; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - - for (j=0; j<3; j++) { - BN_bntest_rand(c,1024,0,0); /**/ - for (i=0; i<num0; i++) - { - BN_bntest_rand(a,475+i*10,0,0); /**/ - BN_bntest_rand(b,425+i*11,0,0); /**/ - a->neg=rand_neg(); - b->neg=rand_neg(); - if (!BN_mod_mul(e,a,b,c,ctx)) - { - unsigned long l; - - while ((l=ERR_get_error())) - fprintf(stderr,"ERROR:%s\n", - ERR_error_string(l,NULL)); - EXIT(1); - } - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," * "); - BN_print(bp,b); - BIO_puts(bp," % "); - BN_print(bp,c); - if ((a->neg ^ b->neg) && !BN_is_zero(e)) - { - /* If (a*b) % c is negative, c must be added - * in order to obtain the normalized remainder - * (new with OpenSSL 0.9.7, previous versions of - * BN_mod_mul could generate negative results) - */ - BIO_puts(bp," + "); - BN_print(bp,c); - } - BIO_puts(bp," - "); - } - BN_print(bp,e); - BIO_puts(bp,"\n"); - } - BN_mul(d,a,b,ctx); - BN_sub(d,d,e); - BN_div(a,b,d,c,ctx); - if(!BN_is_zero(b)) - { - fprintf(stderr,"Modulo multiply test failed!\n"); - ERR_print_errors_fp(stderr); - return 0; - } - } - } - BN_free(a); - BN_free(b); - BN_free(c); - BN_free(d); - BN_free(e); - return(1); - } +{ + BIGNUM *a, *b, *c, *d, *e; + int i, j; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + + for (j = 0; j < 3; j++) { + BN_bntest_rand(c, 1024, 0, 0); + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 475 + i * 10, 0, 0); + BN_bntest_rand(b, 425 + i * 11, 0, 0); + a->neg = rand_neg(); + b->neg = rand_neg(); + if (!BN_mod_mul(e, a, b, c, ctx)) { + unsigned long l; + + while ((l = ERR_get_error())) + fprintf(stderr, "ERROR:%s\n", ERR_error_string(l, NULL)); + EXIT(1); + } + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, b); + BIO_puts(bp, " % "); + BN_print(bp, c); + if ((a->neg ^ b->neg) && !BN_is_zero(e)) { + /* + * If (a*b) % c is negative, c must be added in order + * to obtain the normalized remainder (new with + * OpenSSL 0.9.7, previous versions of BN_mod_mul + * could generate negative results) + */ + BIO_puts(bp, " + "); + BN_print(bp, c); + } + BIO_puts(bp, " - "); + } + BN_print(bp, e); + BIO_puts(bp, "\n"); + } + BN_mul(d, a, b, ctx); + BN_sub(d, d, e); + BN_div(a, b, d, c, ctx); + if (!BN_is_zero(b)) { + fprintf(stderr, "Modulo multiply test failed!\n"); + ERR_print_errors_fp(stderr); + return 0; + } + } + } + BN_free(a); + BN_free(b); + BN_free(c); + BN_free(d); + BN_free(e); + return (1); +} int test_mod_exp(BIO *bp, BN_CTX *ctx) - { - BIGNUM *a,*b,*c,*d,*e; - int i; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - - BN_bntest_rand(c,30,0,1); /* must be odd for montgomery */ - for (i=0; i<num2; i++) - { - BN_bntest_rand(a,20+i*5,0,0); /**/ - BN_bntest_rand(b,2+i,0,0); /**/ - - if (!BN_mod_exp(d,a,b,c,ctx)) - return(0); - - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," ^ "); - BN_print(bp,b); - BIO_puts(bp," % "); - BN_print(bp,c); - BIO_puts(bp," - "); - } - BN_print(bp,d); - BIO_puts(bp,"\n"); - } - BN_exp(e,a,b,ctx); - BN_sub(e,e,d); - BN_div(a,b,e,c,ctx); - if(!BN_is_zero(b)) - { - fprintf(stderr,"Modulo exponentiation test failed!\n"); - return 0; - } - } - BN_free(a); - BN_free(b); - BN_free(c); - BN_free(d); - BN_free(e); - return(1); - } +{ + BIGNUM *a, *b, *c, *d, *e; + int i; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + + BN_bntest_rand(c, 30, 0, 1); /* must be odd for montgomery */ + for (i = 0; i < num2; i++) { + BN_bntest_rand(a, 20 + i * 5, 0, 0); + BN_bntest_rand(b, 2 + i, 0, 0); + + if (!BN_mod_exp(d, a, b, c, ctx)) + return (0); + + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " ^ "); + BN_print(bp, b); + BIO_puts(bp, " % "); + BN_print(bp, c); + BIO_puts(bp, " - "); + } + BN_print(bp, d); + BIO_puts(bp, "\n"); + } + BN_exp(e, a, b, ctx); + BN_sub(e, e, d); + BN_div(a, b, e, c, ctx); + if (!BN_is_zero(b)) { + fprintf(stderr, "Modulo exponentiation test failed!\n"); + return 0; + } + } + BN_free(a); + BN_free(b); + BN_free(c); + BN_free(d); + BN_free(e); + return (1); +} int test_mod_exp_mont_consttime(BIO *bp, BN_CTX *ctx) - { - BIGNUM *a,*b,*c,*d,*e; - int i; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - - BN_bntest_rand(c,30,0,1); /* must be odd for montgomery */ - for (i=0; i<num2; i++) - { - BN_bntest_rand(a,20+i*5,0,0); /**/ - BN_bntest_rand(b,2+i,0,0); /**/ - - if (!BN_mod_exp_mont_consttime(d,a,b,c,ctx,NULL)) - return(00); - - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," ^ "); - BN_print(bp,b); - BIO_puts(bp," % "); - BN_print(bp,c); - BIO_puts(bp," - "); - } - BN_print(bp,d); - BIO_puts(bp,"\n"); - } - BN_exp(e,a,b,ctx); - BN_sub(e,e,d); - BN_div(a,b,e,c,ctx); - if(!BN_is_zero(b)) - { - fprintf(stderr,"Modulo exponentiation test failed!\n"); - return 0; - } - } - BN_free(a); - BN_free(b); - BN_free(c); - BN_free(d); - BN_free(e); - return(1); - } +{ + BIGNUM *a, *b, *c, *d, *e; + int i; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + + BN_bntest_rand(c, 30, 0, 1); /* must be odd for montgomery */ + for (i = 0; i < num2; i++) { + BN_bntest_rand(a, 20 + i * 5, 0, 0); + BN_bntest_rand(b, 2 + i, 0, 0); + + if (!BN_mod_exp_mont_consttime(d, a, b, c, ctx, NULL)) + return (00); + + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " ^ "); + BN_print(bp, b); + BIO_puts(bp, " % "); + BN_print(bp, c); + BIO_puts(bp, " - "); + } + BN_print(bp, d); + BIO_puts(bp, "\n"); + } + BN_exp(e, a, b, ctx); + BN_sub(e, e, d); + BN_div(a, b, e, c, ctx); + if (!BN_is_zero(b)) { + fprintf(stderr, "Modulo exponentiation test failed!\n"); + return 0; + } + } + BN_free(a); + BN_free(b); + BN_free(c); + BN_free(d); + BN_free(e); + return (1); +} + +/* + * Test constant-time modular exponentiation with 1024-bit inputs, which on + * x86_64 cause a different code branch to be taken. + */ +int test_mod_exp_mont5(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *p, *m, *d, *e; + + BN_MONT_CTX *mont; + + a = BN_new(); + p = BN_new(); + m = BN_new(); + d = BN_new(); + e = BN_new(); + + mont = BN_MONT_CTX_new(); + + BN_bntest_rand(m, 1024, 0, 1); /* must be odd for montgomery */ + /* Zero exponent */ + BN_bntest_rand(a, 1024, 0, 0); + BN_zero(p); + if (!BN_mod_exp_mont_consttime(d, a, p, m, ctx, NULL)) + return 0; + if (!BN_is_one(d)) { + fprintf(stderr, "Modular exponentiation test failed!\n"); + return 0; + } + /* Zero input */ + BN_bntest_rand(p, 1024, 0, 0); + BN_zero(a); + if (!BN_mod_exp_mont_consttime(d, a, p, m, ctx, NULL)) + return 0; + if (!BN_is_zero(d)) { + fprintf(stderr, "Modular exponentiation test failed!\n"); + return 0; + } + /* + * Craft an input whose Montgomery representation is 1, i.e., shorter + * than the modulus m, in order to test the const time precomputation + * scattering/gathering. + */ + BN_one(a); + BN_MONT_CTX_set(mont, m, ctx); + if (!BN_from_montgomery(e, a, mont, ctx)) + return 0; + if (!BN_mod_exp_mont_consttime(d, e, p, m, ctx, NULL)) + return 0; + if (!BN_mod_exp_simple(a, e, p, m, ctx)) + return 0; + if (BN_cmp(a, d) != 0) { + fprintf(stderr, "Modular exponentiation test failed!\n"); + return 0; + } + /* Finally, some regular test vectors. */ + BN_bntest_rand(e, 1024, 0, 0); + if (!BN_mod_exp_mont_consttime(d, e, p, m, ctx, NULL)) + return 0; + if (!BN_mod_exp_simple(a, e, p, m, ctx)) + return 0; + if (BN_cmp(a, d) != 0) { + fprintf(stderr, "Modular exponentiation test failed!\n"); + return 0; + } + BN_free(a); + BN_free(p); + BN_free(m); + BN_free(d); + BN_free(e); + return (1); +} int test_exp(BIO *bp, BN_CTX *ctx) - { - BIGNUM *a,*b,*d,*e,*one; - int i; - - a=BN_new(); - b=BN_new(); - d=BN_new(); - e=BN_new(); - one=BN_new(); - BN_one(one); - - for (i=0; i<num2; i++) - { - BN_bntest_rand(a,20+i*5,0,0); /**/ - BN_bntest_rand(b,2+i,0,0); /**/ - - if (BN_exp(d,a,b,ctx) <= 0) - return(0); - - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," ^ "); - BN_print(bp,b); - BIO_puts(bp," - "); - } - BN_print(bp,d); - BIO_puts(bp,"\n"); - } - BN_one(e); - for( ; !BN_is_zero(b) ; BN_sub(b,b,one)) - BN_mul(e,e,a,ctx); - BN_sub(e,e,d); - if(!BN_is_zero(e)) - { - fprintf(stderr,"Exponentiation test failed!\n"); - return 0; - } - } - BN_free(a); - BN_free(b); - BN_free(d); - BN_free(e); - BN_free(one); - return(1); - } +{ + BIGNUM *a, *b, *d, *e, *one; + int i; + + a = BN_new(); + b = BN_new(); + d = BN_new(); + e = BN_new(); + one = BN_new(); + BN_one(one); + + for (i = 0; i < num2; i++) { + BN_bntest_rand(a, 20 + i * 5, 0, 0); + BN_bntest_rand(b, 2 + i, 0, 0); + + if (BN_exp(d, a, b, ctx) <= 0) + return (0); + + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " ^ "); + BN_print(bp, b); + BIO_puts(bp, " - "); + } + BN_print(bp, d); + BIO_puts(bp, "\n"); + } + BN_one(e); + for (; !BN_is_zero(b); BN_sub(b, b, one)) + BN_mul(e, e, a, ctx); + BN_sub(e, e, d); + if (!BN_is_zero(e)) { + fprintf(stderr, "Exponentiation test failed!\n"); + return 0; + } + } + BN_free(a); + BN_free(b); + BN_free(d); + BN_free(e); + BN_free(one); + return (1); +} + #ifndef OPENSSL_NO_EC2M int test_gf2m_add(BIO *bp) - { - BIGNUM a,b,c; - int i, ret = 0; - - BN_init(&a); - BN_init(&b); - BN_init(&c); - - for (i=0; i<num0; i++) - { - BN_rand(&a,512,0,0); - BN_copy(&b, BN_value_one()); - a.neg=rand_neg(); - b.neg=rand_neg(); - BN_GF2m_add(&c,&a,&b); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,&a); - BIO_puts(bp," ^ "); - BN_print(bp,&b); - BIO_puts(bp," = "); - } - BN_print(bp,&c); - BIO_puts(bp,"\n"); - } -#endif - /* Test that two added values have the correct parity. */ - if((BN_is_odd(&a) && BN_is_odd(&c)) || (!BN_is_odd(&a) && !BN_is_odd(&c))) - { - fprintf(stderr,"GF(2^m) addition test (a) failed!\n"); - goto err; - } - BN_GF2m_add(&c,&c,&c); - /* Test that c + c = 0. */ - if(!BN_is_zero(&c)) - { - fprintf(stderr,"GF(2^m) addition test (b) failed!\n"); - goto err; - } - } - ret = 1; - err: - BN_free(&a); - BN_free(&b); - BN_free(&c); - return ret; - } +{ + BIGNUM a, b, c; + int i, ret = 0; + + BN_init(&a); + BN_init(&b); + BN_init(&c); + + for (i = 0; i < num0; i++) { + BN_rand(&a, 512, 0, 0); + BN_copy(&b, BN_value_one()); + a.neg = rand_neg(); + b.neg = rand_neg(); + BN_GF2m_add(&c, &a, &b); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, &a); + BIO_puts(bp, " ^ "); + BN_print(bp, &b); + BIO_puts(bp, " = "); + } + BN_print(bp, &c); + BIO_puts(bp, "\n"); + } +# endif + /* Test that two added values have the correct parity. */ + if ((BN_is_odd(&a) && BN_is_odd(&c)) + || (!BN_is_odd(&a) && !BN_is_odd(&c))) { + fprintf(stderr, "GF(2^m) addition test (a) failed!\n"); + goto err; + } + BN_GF2m_add(&c, &c, &c); + /* Test that c + c = 0. */ + if (!BN_is_zero(&c)) { + fprintf(stderr, "GF(2^m) addition test (b) failed!\n"); + goto err; + } + } + ret = 1; + err: + BN_free(&a); + BN_free(&b); + BN_free(&c); + return ret; +} int test_gf2m_mod(BIO *bp) - { - BIGNUM *a,*b[2],*c,*d,*e; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 1024, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod(c, a, b[j]); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," % "); - BN_print(bp,b[j]); - BIO_puts(bp," - "); - BN_print(bp,c); - BIO_puts(bp,"\n"); - } - } -#endif - BN_GF2m_add(d, a, c); - BN_GF2m_mod(e, d, b[j]); - /* Test that a + (a mod p) mod p == 0. */ - if(!BN_is_zero(e)) - { - fprintf(stderr,"GF(2^m) modulo test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - BN_free(e); - return ret; - } - -int test_gf2m_mod_mul(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d,*e,*f,*g,*h; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - f=BN_new(); - g=BN_new(); - h=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 1024, 0, 0); - BN_bntest_rand(c, 1024, 0, 0); - BN_bntest_rand(d, 1024, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod_mul(e, a, c, b[j], ctx); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," * "); - BN_print(bp,c); - BIO_puts(bp," % "); - BN_print(bp,b[j]); - BIO_puts(bp," - "); - BN_print(bp,e); - BIO_puts(bp,"\n"); - } - } -#endif - BN_GF2m_add(f, a, d); - BN_GF2m_mod_mul(g, f, c, b[j], ctx); - BN_GF2m_mod_mul(h, d, c, b[j], ctx); - BN_GF2m_add(f, e, g); - BN_GF2m_add(f, f, h); - /* Test that (a+d)*c = a*c + d*c. */ - if(!BN_is_zero(f)) - { - fprintf(stderr,"GF(2^m) modular multiplication test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - BN_free(e); - BN_free(f); - BN_free(g); - BN_free(h); - return ret; - } - -int test_gf2m_mod_sqr(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 1024, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod_sqr(c, a, b[j], ctx); - BN_copy(d, a); - BN_GF2m_mod_mul(d, a, d, b[j], ctx); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," ^ 2 % "); - BN_print(bp,b[j]); - BIO_puts(bp, " = "); - BN_print(bp,c); - BIO_puts(bp,"; a * a = "); - BN_print(bp,d); - BIO_puts(bp,"\n"); - } - } -#endif - BN_GF2m_add(d, c, d); - /* Test that a*a = a^2. */ - if(!BN_is_zero(d)) - { - fprintf(stderr,"GF(2^m) modular squaring test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - return ret; - } - -int test_gf2m_mod_inv(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 512, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod_inv(c, a, b[j], ctx); - BN_GF2m_mod_mul(d, a, c, b[j], ctx); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp, " * "); - BN_print(bp,c); - BIO_puts(bp," - 1 % "); - BN_print(bp,b[j]); - BIO_puts(bp,"\n"); - } - } -#endif - /* Test that ((1/a)*a) = 1. */ - if(!BN_is_one(d)) - { - fprintf(stderr,"GF(2^m) modular inversion test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - return ret; - } - -int test_gf2m_mod_div(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d,*e,*f; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - f=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 512, 0, 0); - BN_bntest_rand(c, 512, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod_div(d, a, c, b[j], ctx); - BN_GF2m_mod_mul(e, d, c, b[j], ctx); - BN_GF2m_mod_div(f, a, e, b[j], ctx); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp, " = "); - BN_print(bp,c); - BIO_puts(bp," * "); - BN_print(bp,d); - BIO_puts(bp, " % "); - BN_print(bp,b[j]); - BIO_puts(bp,"\n"); - } - } -#endif - /* Test that ((a/c)*c)/a = 1. */ - if(!BN_is_one(f)) - { - fprintf(stderr,"GF(2^m) modular division test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - BN_free(e); - BN_free(f); - return ret; - } - -int test_gf2m_mod_exp(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d,*e,*f; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - f=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 512, 0, 0); - BN_bntest_rand(c, 512, 0, 0); - BN_bntest_rand(d, 512, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod_exp(e, a, c, b[j], ctx); - BN_GF2m_mod_exp(f, a, d, b[j], ctx); - BN_GF2m_mod_mul(e, e, f, b[j], ctx); - BN_add(f, c, d); - BN_GF2m_mod_exp(f, a, f, b[j], ctx); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp, " ^ ("); - BN_print(bp,c); - BIO_puts(bp," + "); - BN_print(bp,d); - BIO_puts(bp, ") = "); - BN_print(bp,e); - BIO_puts(bp, "; - "); - BN_print(bp,f); - BIO_puts(bp, " % "); - BN_print(bp,b[j]); - BIO_puts(bp,"\n"); - } - } -#endif - BN_GF2m_add(f, e, f); - /* Test that a^(c+d)=a^c*a^d. */ - if(!BN_is_zero(f)) - { - fprintf(stderr,"GF(2^m) modular exponentiation test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - BN_free(e); - BN_free(f); - return ret; - } - -int test_gf2m_mod_sqrt(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d,*e,*f; - int i, j, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - f=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 512, 0, 0); - for (j=0; j < 2; j++) - { - BN_GF2m_mod(c, a, b[j]); - BN_GF2m_mod_sqrt(d, a, b[j], ctx); - BN_GF2m_mod_sqr(e, d, b[j], ctx); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,d); - BIO_puts(bp, " ^ 2 - "); - BN_print(bp,a); - BIO_puts(bp,"\n"); - } - } -#endif - BN_GF2m_add(f, c, e); - /* Test that d^2 = a, where d = sqrt(a). */ - if(!BN_is_zero(f)) - { - fprintf(stderr,"GF(2^m) modular square root test failed!\n"); - goto err; - } - } - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - BN_free(e); - BN_free(f); - return ret; - } - -int test_gf2m_mod_solve_quad(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b[2],*c,*d,*e; - int i, j, s = 0, t, ret = 0; - int p0[] = {163,7,6,3,0,-1}; - int p1[] = {193,15,0,-1}; - - a=BN_new(); - b[0]=BN_new(); - b[1]=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - - BN_GF2m_arr2poly(p0, b[0]); - BN_GF2m_arr2poly(p1, b[1]); - - for (i=0; i<num0; i++) - { - BN_bntest_rand(a, 512, 0, 0); - for (j=0; j < 2; j++) - { - t = BN_GF2m_mod_solve_quad(c, a, b[j], ctx); - if (t) - { - s++; - BN_GF2m_mod_sqr(d, c, b[j], ctx); - BN_GF2m_add(d, c, d); - BN_GF2m_mod(e, a, b[j]); -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BN_print(bp,c); - BIO_puts(bp, " is root of z^2 + z = "); - BN_print(bp,a); - BIO_puts(bp, " % "); - BN_print(bp,b[j]); - BIO_puts(bp, "\n"); - } - } -#endif - BN_GF2m_add(e, e, d); - /* Test that solution of quadratic c satisfies c^2 + c = a. */ - if(!BN_is_zero(e)) - { - fprintf(stderr,"GF(2^m) modular solve quadratic test failed!\n"); - goto err; - } - - } - else - { -#if 0 /* make test uses ouput in bc but bc can't handle GF(2^m) arithmetic */ - if (bp != NULL) - { - if (!results) - { - BIO_puts(bp, "There are no roots of z^2 + z = "); - BN_print(bp,a); - BIO_puts(bp, " % "); - BN_print(bp,b[j]); - BIO_puts(bp, "\n"); - } - } -#endif - } - } - } - if (s == 0) - { - fprintf(stderr,"All %i tests of GF(2^m) modular solve quadratic resulted in no roots;\n", num0); - fprintf(stderr,"this is very unlikely and probably indicates an error.\n"); - goto err; - } - ret = 1; - err: - BN_free(a); - BN_free(b[0]); - BN_free(b[1]); - BN_free(c); - BN_free(d); - BN_free(e); - return ret; - } +{ + BIGNUM *a, *b[2], *c, *d, *e; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 1024, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod(c, a, b[j]); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " % "); + BN_print(bp, b[j]); + BIO_puts(bp, " - "); + BN_print(bp, c); + BIO_puts(bp, "\n"); + } + } +# endif + BN_GF2m_add(d, a, c); + BN_GF2m_mod(e, d, b[j]); + /* Test that a + (a mod p) mod p == 0. */ + if (!BN_is_zero(e)) { + fprintf(stderr, "GF(2^m) modulo test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + BN_free(e); + return ret; +} + +int test_gf2m_mod_mul(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d, *e, *f, *g, *h; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + f = BN_new(); + g = BN_new(); + h = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 1024, 0, 0); + BN_bntest_rand(c, 1024, 0, 0); + BN_bntest_rand(d, 1024, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod_mul(e, a, c, b[j], ctx); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, c); + BIO_puts(bp, " % "); + BN_print(bp, b[j]); + BIO_puts(bp, " - "); + BN_print(bp, e); + BIO_puts(bp, "\n"); + } + } +# endif + BN_GF2m_add(f, a, d); + BN_GF2m_mod_mul(g, f, c, b[j], ctx); + BN_GF2m_mod_mul(h, d, c, b[j], ctx); + BN_GF2m_add(f, e, g); + BN_GF2m_add(f, f, h); + /* Test that (a+d)*c = a*c + d*c. */ + if (!BN_is_zero(f)) { + fprintf(stderr, + "GF(2^m) modular multiplication test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + BN_free(e); + BN_free(f); + BN_free(g); + BN_free(h); + return ret; +} + +int test_gf2m_mod_sqr(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 1024, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod_sqr(c, a, b[j], ctx); + BN_copy(d, a); + BN_GF2m_mod_mul(d, a, d, b[j], ctx); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " ^ 2 % "); + BN_print(bp, b[j]); + BIO_puts(bp, " = "); + BN_print(bp, c); + BIO_puts(bp, "; a * a = "); + BN_print(bp, d); + BIO_puts(bp, "\n"); + } + } +# endif + BN_GF2m_add(d, c, d); + /* Test that a*a = a^2. */ + if (!BN_is_zero(d)) { + fprintf(stderr, "GF(2^m) modular squaring test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + return ret; +} + +int test_gf2m_mod_inv(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 512, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod_inv(c, a, b[j], ctx); + BN_GF2m_mod_mul(d, a, c, b[j], ctx); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, c); + BIO_puts(bp, " - 1 % "); + BN_print(bp, b[j]); + BIO_puts(bp, "\n"); + } + } +# endif + /* Test that ((1/a)*a) = 1. */ + if (!BN_is_one(d)) { + fprintf(stderr, "GF(2^m) modular inversion test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + return ret; +} + +int test_gf2m_mod_div(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d, *e, *f; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + f = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 512, 0, 0); + BN_bntest_rand(c, 512, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod_div(d, a, c, b[j], ctx); + BN_GF2m_mod_mul(e, d, c, b[j], ctx); + BN_GF2m_mod_div(f, a, e, b[j], ctx); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " = "); + BN_print(bp, c); + BIO_puts(bp, " * "); + BN_print(bp, d); + BIO_puts(bp, " % "); + BN_print(bp, b[j]); + BIO_puts(bp, "\n"); + } + } +# endif + /* Test that ((a/c)*c)/a = 1. */ + if (!BN_is_one(f)) { + fprintf(stderr, "GF(2^m) modular division test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + BN_free(e); + BN_free(f); + return ret; +} + +int test_gf2m_mod_exp(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d, *e, *f; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + f = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 512, 0, 0); + BN_bntest_rand(c, 512, 0, 0); + BN_bntest_rand(d, 512, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod_exp(e, a, c, b[j], ctx); + BN_GF2m_mod_exp(f, a, d, b[j], ctx); + BN_GF2m_mod_mul(e, e, f, b[j], ctx); + BN_add(f, c, d); + BN_GF2m_mod_exp(f, a, f, b[j], ctx); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " ^ ("); + BN_print(bp, c); + BIO_puts(bp, " + "); + BN_print(bp, d); + BIO_puts(bp, ") = "); + BN_print(bp, e); + BIO_puts(bp, "; - "); + BN_print(bp, f); + BIO_puts(bp, " % "); + BN_print(bp, b[j]); + BIO_puts(bp, "\n"); + } + } +# endif + BN_GF2m_add(f, e, f); + /* Test that a^(c+d)=a^c*a^d. */ + if (!BN_is_zero(f)) { + fprintf(stderr, + "GF(2^m) modular exponentiation test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + BN_free(e); + BN_free(f); + return ret; +} + +int test_gf2m_mod_sqrt(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d, *e, *f; + int i, j, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + f = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 512, 0, 0); + for (j = 0; j < 2; j++) { + BN_GF2m_mod(c, a, b[j]); + BN_GF2m_mod_sqrt(d, a, b[j], ctx); + BN_GF2m_mod_sqr(e, d, b[j], ctx); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, d); + BIO_puts(bp, " ^ 2 - "); + BN_print(bp, a); + BIO_puts(bp, "\n"); + } + } +# endif + BN_GF2m_add(f, c, e); + /* Test that d^2 = a, where d = sqrt(a). */ + if (!BN_is_zero(f)) { + fprintf(stderr, "GF(2^m) modular square root test failed!\n"); + goto err; + } + } + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + BN_free(e); + BN_free(f); + return ret; +} + +int test_gf2m_mod_solve_quad(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b[2], *c, *d, *e; + int i, j, s = 0, t, ret = 0; + int p0[] = { 163, 7, 6, 3, 0, -1 }; + int p1[] = { 193, 15, 0, -1 }; + + a = BN_new(); + b[0] = BN_new(); + b[1] = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + + BN_GF2m_arr2poly(p0, b[0]); + BN_GF2m_arr2poly(p1, b[1]); + + for (i = 0; i < num0; i++) { + BN_bntest_rand(a, 512, 0, 0); + for (j = 0; j < 2; j++) { + t = BN_GF2m_mod_solve_quad(c, a, b[j], ctx); + if (t) { + s++; + BN_GF2m_mod_sqr(d, c, b[j], ctx); + BN_GF2m_add(d, c, d); + BN_GF2m_mod(e, a, b[j]); +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BN_print(bp, c); + BIO_puts(bp, " is root of z^2 + z = "); + BN_print(bp, a); + BIO_puts(bp, " % "); + BN_print(bp, b[j]); + BIO_puts(bp, "\n"); + } + } +# endif + BN_GF2m_add(e, e, d); + /* + * Test that solution of quadratic c satisfies c^2 + c = a. + */ + if (!BN_is_zero(e)) { + fprintf(stderr, + "GF(2^m) modular solve quadratic test failed!\n"); + goto err; + } + + } else { +# if 0 /* make test uses ouput in bc but bc can't + * handle GF(2^m) arithmetic */ + if (bp != NULL) { + if (!results) { + BIO_puts(bp, "There are no roots of z^2 + z = "); + BN_print(bp, a); + BIO_puts(bp, " % "); + BN_print(bp, b[j]); + BIO_puts(bp, "\n"); + } + } +# endif + } + } + } + if (s == 0) { + fprintf(stderr, + "All %i tests of GF(2^m) modular solve quadratic resulted in no roots;\n", + num0); + fprintf(stderr, + "this is very unlikely and probably indicates an error.\n"); + goto err; + } + ret = 1; + err: + BN_free(a); + BN_free(b[0]); + BN_free(b[1]); + BN_free(c); + BN_free(d); + BN_free(e); + return ret; +} #endif static int genprime_cb(int p, int n, BN_GENCB *arg) - { - char c='*'; - - if (p == 0) c='.'; - if (p == 1) c='+'; - if (p == 2) c='*'; - if (p == 3) c='\n'; - putc(c, stderr); - fflush(stderr); - return 1; - } +{ + char c = '*'; + + if (p == 0) + c = '.'; + if (p == 1) + c = '+'; + if (p == 2) + c = '*'; + if (p == 3) + c = '\n'; + putc(c, stderr); + fflush(stderr); + return 1; +} int test_kron(BIO *bp, BN_CTX *ctx) - { - BN_GENCB cb; - BIGNUM *a,*b,*r,*t; - int i; - int legendre, kronecker; - int ret = 0; - - a = BN_new(); - b = BN_new(); - r = BN_new(); - t = BN_new(); - if (a == NULL || b == NULL || r == NULL || t == NULL) goto err; - - BN_GENCB_set(&cb, genprime_cb, NULL); - - /* We test BN_kronecker(a, b, ctx) just for b odd (Jacobi symbol). - * In this case we know that if b is prime, then BN_kronecker(a, b, ctx) - * is congruent to $a^{(b-1)/2}$, modulo $b$ (Legendre symbol). - * So we generate a random prime b and compare these values - * for a number of random a's. (That is, we run the Solovay-Strassen - * primality test to confirm that b is prime, except that we - * don't want to test whether b is prime but whether BN_kronecker - * works.) */ - - if (!BN_generate_prime_ex(b, 512, 0, NULL, NULL, &cb)) goto err; - b->neg = rand_neg(); - putc('\n', stderr); - - for (i = 0; i < num0; i++) - { - if (!BN_bntest_rand(a, 512, 0, 0)) goto err; - a->neg = rand_neg(); - - /* t := (|b|-1)/2 (note that b is odd) */ - if (!BN_copy(t, b)) goto err; - t->neg = 0; - if (!BN_sub_word(t, 1)) goto err; - if (!BN_rshift1(t, t)) goto err; - /* r := a^t mod b */ - b->neg=0; - - if (!BN_mod_exp_recp(r, a, t, b, ctx)) goto err; - b->neg=1; - - if (BN_is_word(r, 1)) - legendre = 1; - else if (BN_is_zero(r)) - legendre = 0; - else - { - if (!BN_add_word(r, 1)) goto err; - if (0 != BN_ucmp(r, b)) - { - fprintf(stderr, "Legendre symbol computation failed\n"); - goto err; - } - legendre = -1; - } - - kronecker = BN_kronecker(a, b, ctx); - if (kronecker < -1) goto err; - /* we actually need BN_kronecker(a, |b|) */ - if (a->neg && b->neg) - kronecker = -kronecker; - - if (legendre != kronecker) - { - fprintf(stderr, "legendre != kronecker; a = "); - BN_print_fp(stderr, a); - fprintf(stderr, ", b = "); - BN_print_fp(stderr, b); - fprintf(stderr, "\n"); - goto err; - } - - putc('.', stderr); - fflush(stderr); - } - - putc('\n', stderr); - fflush(stderr); - ret = 1; +{ + BN_GENCB cb; + BIGNUM *a, *b, *r, *t; + int i; + int legendre, kronecker; + int ret = 0; + + a = BN_new(); + b = BN_new(); + r = BN_new(); + t = BN_new(); + if (a == NULL || b == NULL || r == NULL || t == NULL) + goto err; + + BN_GENCB_set(&cb, genprime_cb, NULL); + + /* + * We test BN_kronecker(a, b, ctx) just for b odd (Jacobi symbol). In + * this case we know that if b is prime, then BN_kronecker(a, b, ctx) is + * congruent to $a^{(b-1)/2}$, modulo $b$ (Legendre symbol). So we + * generate a random prime b and compare these values for a number of + * random a's. (That is, we run the Solovay-Strassen primality test to + * confirm that b is prime, except that we don't want to test whether b + * is prime but whether BN_kronecker works.) + */ + + if (!BN_generate_prime_ex(b, 512, 0, NULL, NULL, &cb)) + goto err; + b->neg = rand_neg(); + putc('\n', stderr); + + for (i = 0; i < num0; i++) { + if (!BN_bntest_rand(a, 512, 0, 0)) + goto err; + a->neg = rand_neg(); + + /* t := (|b|-1)/2 (note that b is odd) */ + if (!BN_copy(t, b)) + goto err; + t->neg = 0; + if (!BN_sub_word(t, 1)) + goto err; + if (!BN_rshift1(t, t)) + goto err; + /* r := a^t mod b */ + b->neg = 0; + + if (!BN_mod_exp_recp(r, a, t, b, ctx)) + goto err; + b->neg = 1; + + if (BN_is_word(r, 1)) + legendre = 1; + else if (BN_is_zero(r)) + legendre = 0; + else { + if (!BN_add_word(r, 1)) + goto err; + if (0 != BN_ucmp(r, b)) { + fprintf(stderr, "Legendre symbol computation failed\n"); + goto err; + } + legendre = -1; + } + + kronecker = BN_kronecker(a, b, ctx); + if (kronecker < -1) + goto err; + /* we actually need BN_kronecker(a, |b|) */ + if (a->neg && b->neg) + kronecker = -kronecker; + + if (legendre != kronecker) { + fprintf(stderr, "legendre != kronecker; a = "); + BN_print_fp(stderr, a); + fprintf(stderr, ", b = "); + BN_print_fp(stderr, b); + fprintf(stderr, "\n"); + goto err; + } + + putc('.', stderr); + fflush(stderr); + } + + putc('\n', stderr); + fflush(stderr); + ret = 1; err: - if (a != NULL) BN_free(a); - if (b != NULL) BN_free(b); - if (r != NULL) BN_free(r); - if (t != NULL) BN_free(t); - return ret; - } + if (a != NULL) + BN_free(a); + if (b != NULL) + BN_free(b); + if (r != NULL) + BN_free(r); + if (t != NULL) + BN_free(t); + return ret; +} int test_sqrt(BIO *bp, BN_CTX *ctx) - { - BN_GENCB cb; - BIGNUM *a,*p,*r; - int i, j; - int ret = 0; - - a = BN_new(); - p = BN_new(); - r = BN_new(); - if (a == NULL || p == NULL || r == NULL) goto err; - - BN_GENCB_set(&cb, genprime_cb, NULL); - - for (i = 0; i < 16; i++) - { - if (i < 8) - { - unsigned primes[8] = { 2, 3, 5, 7, 11, 13, 17, 19 }; - - if (!BN_set_word(p, primes[i])) goto err; - } - else - { - if (!BN_set_word(a, 32)) goto err; - if (!BN_set_word(r, 2*i + 1)) goto err; - - if (!BN_generate_prime_ex(p, 256, 0, a, r, &cb)) goto err; - putc('\n', stderr); - } - p->neg = rand_neg(); - - for (j = 0; j < num2; j++) - { - /* construct 'a' such that it is a square modulo p, - * but in general not a proper square and not reduced modulo p */ - if (!BN_bntest_rand(r, 256, 0, 3)) goto err; - if (!BN_nnmod(r, r, p, ctx)) goto err; - if (!BN_mod_sqr(r, r, p, ctx)) goto err; - if (!BN_bntest_rand(a, 256, 0, 3)) goto err; - if (!BN_nnmod(a, a, p, ctx)) goto err; - if (!BN_mod_sqr(a, a, p, ctx)) goto err; - if (!BN_mul(a, a, r, ctx)) goto err; - if (rand_neg()) - if (!BN_sub(a, a, p)) goto err; - - if (!BN_mod_sqrt(r, a, p, ctx)) goto err; - if (!BN_mod_sqr(r, r, p, ctx)) goto err; - - if (!BN_nnmod(a, a, p, ctx)) goto err; - - if (BN_cmp(a, r) != 0) - { - fprintf(stderr, "BN_mod_sqrt failed: a = "); - BN_print_fp(stderr, a); - fprintf(stderr, ", r = "); - BN_print_fp(stderr, r); - fprintf(stderr, ", p = "); - BN_print_fp(stderr, p); - fprintf(stderr, "\n"); - goto err; - } - - putc('.', stderr); - fflush(stderr); - } - - putc('\n', stderr); - fflush(stderr); - } - ret = 1; +{ + BN_GENCB cb; + BIGNUM *a, *p, *r; + int i, j; + int ret = 0; + + a = BN_new(); + p = BN_new(); + r = BN_new(); + if (a == NULL || p == NULL || r == NULL) + goto err; + + BN_GENCB_set(&cb, genprime_cb, NULL); + + for (i = 0; i < 16; i++) { + if (i < 8) { + unsigned primes[8] = { 2, 3, 5, 7, 11, 13, 17, 19 }; + + if (!BN_set_word(p, primes[i])) + goto err; + } else { + if (!BN_set_word(a, 32)) + goto err; + if (!BN_set_word(r, 2 * i + 1)) + goto err; + + if (!BN_generate_prime_ex(p, 256, 0, a, r, &cb)) + goto err; + putc('\n', stderr); + } + p->neg = rand_neg(); + + for (j = 0; j < num2; j++) { + /* + * construct 'a' such that it is a square modulo p, but in + * general not a proper square and not reduced modulo p + */ + if (!BN_bntest_rand(r, 256, 0, 3)) + goto err; + if (!BN_nnmod(r, r, p, ctx)) + goto err; + if (!BN_mod_sqr(r, r, p, ctx)) + goto err; + if (!BN_bntest_rand(a, 256, 0, 3)) + goto err; + if (!BN_nnmod(a, a, p, ctx)) + goto err; + if (!BN_mod_sqr(a, a, p, ctx)) + goto err; + if (!BN_mul(a, a, r, ctx)) + goto err; + if (rand_neg()) + if (!BN_sub(a, a, p)) + goto err; + + if (!BN_mod_sqrt(r, a, p, ctx)) + goto err; + if (!BN_mod_sqr(r, r, p, ctx)) + goto err; + + if (!BN_nnmod(a, a, p, ctx)) + goto err; + + if (BN_cmp(a, r) != 0) { + fprintf(stderr, "BN_mod_sqrt failed: a = "); + BN_print_fp(stderr, a); + fprintf(stderr, ", r = "); + BN_print_fp(stderr, r); + fprintf(stderr, ", p = "); + BN_print_fp(stderr, p); + fprintf(stderr, "\n"); + goto err; + } + + putc('.', stderr); + fflush(stderr); + } + + putc('\n', stderr); + fflush(stderr); + } + ret = 1; err: - if (a != NULL) BN_free(a); - if (p != NULL) BN_free(p); - if (r != NULL) BN_free(r); - return ret; - } - -int test_lshift(BIO *bp,BN_CTX *ctx,BIGNUM *a_) - { - BIGNUM *a,*b,*c,*d; - int i; - - b=BN_new(); - c=BN_new(); - d=BN_new(); - BN_one(c); - - if(a_) - a=a_; - else - { - a=BN_new(); - BN_bntest_rand(a,200,0,0); /**/ - a->neg=rand_neg(); - } - for (i=0; i<num0; i++) - { - BN_lshift(b,a,i+1); - BN_add(c,c,c); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," * "); - BN_print(bp,c); - BIO_puts(bp," - "); - } - BN_print(bp,b); - BIO_puts(bp,"\n"); - } - BN_mul(d,a,c,ctx); - BN_sub(d,d,b); - if(!BN_is_zero(d)) - { - fprintf(stderr,"Left shift test failed!\n"); - fprintf(stderr,"a="); - BN_print_fp(stderr,a); - fprintf(stderr,"\nb="); - BN_print_fp(stderr,b); - fprintf(stderr,"\nc="); - BN_print_fp(stderr,c); - fprintf(stderr,"\nd="); - BN_print_fp(stderr,d); - fprintf(stderr,"\n"); - return 0; - } - } - BN_free(a); - BN_free(b); - BN_free(c); - BN_free(d); - return(1); - } + if (a != NULL) + BN_free(a); + if (p != NULL) + BN_free(p); + if (r != NULL) + BN_free(r); + return ret; +} + +int test_lshift(BIO *bp, BN_CTX *ctx, BIGNUM *a_) +{ + BIGNUM *a, *b, *c, *d; + int i; + + b = BN_new(); + c = BN_new(); + d = BN_new(); + BN_one(c); + + if (a_) + a = a_; + else { + a = BN_new(); + BN_bntest_rand(a, 200, 0, 0); + a->neg = rand_neg(); + } + for (i = 0; i < num0; i++) { + BN_lshift(b, a, i + 1); + BN_add(c, c, c); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * "); + BN_print(bp, c); + BIO_puts(bp, " - "); + } + BN_print(bp, b); + BIO_puts(bp, "\n"); + } + BN_mul(d, a, c, ctx); + BN_sub(d, d, b); + if (!BN_is_zero(d)) { + fprintf(stderr, "Left shift test failed!\n"); + fprintf(stderr, "a="); + BN_print_fp(stderr, a); + fprintf(stderr, "\nb="); + BN_print_fp(stderr, b); + fprintf(stderr, "\nc="); + BN_print_fp(stderr, c); + fprintf(stderr, "\nd="); + BN_print_fp(stderr, d); + fprintf(stderr, "\n"); + return 0; + } + } + BN_free(a); + BN_free(b); + BN_free(c); + BN_free(d); + return (1); +} int test_lshift1(BIO *bp) - { - BIGNUM *a,*b,*c; - int i; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - - BN_bntest_rand(a,200,0,0); /**/ - a->neg=rand_neg(); - for (i=0; i<num0; i++) - { - BN_lshift1(b,a); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," * 2"); - BIO_puts(bp," - "); - } - BN_print(bp,b); - BIO_puts(bp,"\n"); - } - BN_add(c,a,a); - BN_sub(a,b,c); - if(!BN_is_zero(a)) - { - fprintf(stderr,"Left shift one test failed!\n"); - return 0; - } - - BN_copy(a,b); - } - BN_free(a); - BN_free(b); - BN_free(c); - return(1); - } - -int test_rshift(BIO *bp,BN_CTX *ctx) - { - BIGNUM *a,*b,*c,*d,*e; - int i; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - d=BN_new(); - e=BN_new(); - BN_one(c); - - BN_bntest_rand(a,200,0,0); /**/ - a->neg=rand_neg(); - for (i=0; i<num0; i++) - { - BN_rshift(b,a,i+1); - BN_add(c,c,c); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," / "); - BN_print(bp,c); - BIO_puts(bp," - "); - } - BN_print(bp,b); - BIO_puts(bp,"\n"); - } - BN_div(d,e,a,c,ctx); - BN_sub(d,d,b); - if(!BN_is_zero(d)) - { - fprintf(stderr,"Right shift test failed!\n"); - return 0; - } - } - BN_free(a); - BN_free(b); - BN_free(c); - BN_free(d); - BN_free(e); - return(1); - } +{ + BIGNUM *a, *b, *c; + int i; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + + BN_bntest_rand(a, 200, 0, 0); + a->neg = rand_neg(); + for (i = 0; i < num0; i++) { + BN_lshift1(b, a); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " * 2"); + BIO_puts(bp, " - "); + } + BN_print(bp, b); + BIO_puts(bp, "\n"); + } + BN_add(c, a, a); + BN_sub(a, b, c); + if (!BN_is_zero(a)) { + fprintf(stderr, "Left shift one test failed!\n"); + return 0; + } + + BN_copy(a, b); + } + BN_free(a); + BN_free(b); + BN_free(c); + return (1); +} + +int test_rshift(BIO *bp, BN_CTX *ctx) +{ + BIGNUM *a, *b, *c, *d, *e; + int i; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + d = BN_new(); + e = BN_new(); + BN_one(c); + + BN_bntest_rand(a, 200, 0, 0); + a->neg = rand_neg(); + for (i = 0; i < num0; i++) { + BN_rshift(b, a, i + 1); + BN_add(c, c, c); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " / "); + BN_print(bp, c); + BIO_puts(bp, " - "); + } + BN_print(bp, b); + BIO_puts(bp, "\n"); + } + BN_div(d, e, a, c, ctx); + BN_sub(d, d, b); + if (!BN_is_zero(d)) { + fprintf(stderr, "Right shift test failed!\n"); + return 0; + } + } + BN_free(a); + BN_free(b); + BN_free(c); + BN_free(d); + BN_free(e); + return (1); +} int test_rshift1(BIO *bp) - { - BIGNUM *a,*b,*c; - int i; - - a=BN_new(); - b=BN_new(); - c=BN_new(); - - BN_bntest_rand(a,200,0,0); /**/ - a->neg=rand_neg(); - for (i=0; i<num0; i++) - { - BN_rshift1(b,a); - if (bp != NULL) - { - if (!results) - { - BN_print(bp,a); - BIO_puts(bp," / 2"); - BIO_puts(bp," - "); - } - BN_print(bp,b); - BIO_puts(bp,"\n"); - } - BN_sub(c,a,b); - BN_sub(c,c,b); - if(!BN_is_zero(c) && !BN_abs_is_word(c, 1)) - { - fprintf(stderr,"Right shift one test failed!\n"); - return 0; - } - BN_copy(a,b); - } - BN_free(a); - BN_free(b); - BN_free(c); - return(1); - } +{ + BIGNUM *a, *b, *c; + int i; + + a = BN_new(); + b = BN_new(); + c = BN_new(); + + BN_bntest_rand(a, 200, 0, 0); + a->neg = rand_neg(); + for (i = 0; i < num0; i++) { + BN_rshift1(b, a); + if (bp != NULL) { + if (!results) { + BN_print(bp, a); + BIO_puts(bp, " / 2"); + BIO_puts(bp, " - "); + } + BN_print(bp, b); + BIO_puts(bp, "\n"); + } + BN_sub(c, a, b); + BN_sub(c, c, b); + if (!BN_is_zero(c) && !BN_abs_is_word(c, 1)) { + fprintf(stderr, "Right shift one test failed!\n"); + return 0; + } + BN_copy(a, b); + } + BN_free(a); + BN_free(b); + BN_free(c); + return (1); +} int rand_neg(void) - { - static unsigned int neg=0; - static int sign[8]={0,0,0,1,1,0,1,1}; +{ + static unsigned int neg = 0; + static int sign[8] = { 0, 0, 0, 1, 1, 0, 1, 1 }; - return(sign[(neg++)%8]); - } + return (sign[(neg++) % 8]); +} diff --git a/openssl/crypto/bn/divtest.c b/openssl/crypto/bn/divtest.c index d3fc688f3..2590b4581 100644 --- a/openssl/crypto/bn/divtest.c +++ b/openssl/crypto/bn/divtest.c @@ -4,13 +4,13 @@ static int Rand(n) { unsigned char x[2]; - RAND_pseudo_bytes(x,2); - return (x[0] + 2*x[1]); + RAND_pseudo_bytes(x, 2); + return (x[0] + 2 * x[1]); } static void bug(char *m, BIGNUM *a, BIGNUM *b) { - printf("%s!\na=",m); + printf("%s!\na=", m); BN_print_fp(stdout, a); printf("\nb="); BN_print_fp(stdout, b); @@ -20,22 +20,23 @@ static void bug(char *m, BIGNUM *a, BIGNUM *b) main() { - BIGNUM *a=BN_new(), *b=BN_new(), *c=BN_new(), *d=BN_new(), - *C=BN_new(), *D=BN_new(); - BN_RECP_CTX *recp=BN_RECP_CTX_new(); - BN_CTX *ctx=BN_CTX_new(); + BIGNUM *a = BN_new(), *b = BN_new(), *c = BN_new(), *d = BN_new(), + *C = BN_new(), *D = BN_new(); + BN_RECP_CTX *recp = BN_RECP_CTX_new(); + BN_CTX *ctx = BN_CTX_new(); - for(;;) { - BN_pseudo_rand(a,Rand(),0,0); - BN_pseudo_rand(b,Rand(),0,0); - if (BN_is_zero(b)) continue; + for (;;) { + BN_pseudo_rand(a, Rand(), 0, 0); + BN_pseudo_rand(b, Rand(), 0, 0); + if (BN_is_zero(b)) + continue; - BN_RECP_CTX_set(recp,b,ctx); - if (BN_div(C,D,a,b,ctx) != 1) - bug("BN_div failed",a,b); - if (BN_div_recp(c,d,a,recp,ctx) != 1) - bug("BN_div_recp failed",a,b); - else if (BN_cmp(c,C) != 0 || BN_cmp(c,C) != 0) - bug("mismatch",a,b); + BN_RECP_CTX_set(recp, b, ctx); + if (BN_div(C, D, a, b, ctx) != 1) + bug("BN_div failed", a, b); + if (BN_div_recp(c, d, a, recp, ctx) != 1) + bug("BN_div_recp failed", a, b); + else if (BN_cmp(c, C) != 0 || BN_cmp(c, C) != 0) + bug("mismatch", a, b); } } diff --git a/openssl/crypto/bn/exp.c b/openssl/crypto/bn/exp.c index 4865b0ef7..fbce28c5b 100644 --- a/openssl/crypto/bn/exp.c +++ b/openssl/crypto/bn/exp.c @@ -4,59 +4,58 @@ #include <openssl/tmdiff.h> #include "bn_lcl.h" -#define SIZE 256 -#define NUM (8*8*8) -#define MOD (8*8*8*8*8) +#define SIZE 256 +#define NUM (8*8*8) +#define MOD (8*8*8*8*8) -main(argc,argv) +main(argc, argv) int argc; char *argv[]; - { - BN_CTX ctx; - BIGNUM a,b,c,r,rr,t,l; - int j,i,size=SIZE,num=NUM,mod=MOD; - char *start,*end; - BN_MONT_CTX mont; - double d,md; - - BN_MONT_CTX_init(&mont); - BN_CTX_init(&ctx); - BN_init(&a); - BN_init(&b); - BN_init(&c); - BN_init(&r); - - start=ms_time_new(); - end=ms_time_new(); - while (size <= 1024*8) - { - BN_rand(&a,size,0,0); - BN_rand(&b,size,1,0); - BN_rand(&c,size,0,1); - - BN_mod(&a,&a,&c,&ctx); - - ms_time_get(start); - for (i=0; i<10; i++) - BN_MONT_CTX_set(&mont,&c,&ctx); - ms_time_get(end); - md=ms_time_diff(start,end); - - ms_time_get(start); - for (i=0; i<num; i++) - { - /* bn_mull(&r,&a,&b,&ctx); */ - /* BN_sqr(&r,&a,&ctx); */ - BN_mod_exp_mont(&r,&a,&b,&c,&ctx,&mont); - } - ms_time_get(end); - d=ms_time_diff(start,end)/* *50/33 */; - printf("%5d bit:%6.2f %6d %6.4f %4d m_set(%5.4f)\n",size, - d,num,d/num,(int)((d/num)*mod),md/10.0); - num/=8; - mod/=8; - if (num <= 0) num=1; - size*=2; - } - - } +{ + BN_CTX ctx; + BIGNUM a, b, c, r, rr, t, l; + int j, i, size = SIZE, num = NUM, mod = MOD; + char *start, *end; + BN_MONT_CTX mont; + double d, md; + + BN_MONT_CTX_init(&mont); + BN_CTX_init(&ctx); + BN_init(&a); + BN_init(&b); + BN_init(&c); + BN_init(&r); + + start = ms_time_new(); + end = ms_time_new(); + while (size <= 1024 * 8) { + BN_rand(&a, size, 0, 0); + BN_rand(&b, size, 1, 0); + BN_rand(&c, size, 0, 1); + + BN_mod(&a, &a, &c, &ctx); + + ms_time_get(start); + for (i = 0; i < 10; i++) + BN_MONT_CTX_set(&mont, &c, &ctx); + ms_time_get(end); + md = ms_time_diff(start, end); + + ms_time_get(start); + for (i = 0; i < num; i++) { + /* bn_mull(&r,&a,&b,&ctx); */ + /* BN_sqr(&r,&a,&ctx); */ + BN_mod_exp_mont(&r, &a, &b, &c, &ctx, &mont); + } + ms_time_get(end); + d = ms_time_diff(start, end) /* *50/33 */ ; + printf("%5d bit:%6.2f %6d %6.4f %4d m_set(%5.4f)\n", size, + d, num, d / num, (int)((d / num) * mod), md / 10.0); + num /= 8; + mod /= 8; + if (num <= 0) + num = 1; + size *= 2; + } + +} diff --git a/openssl/crypto/bn/expspeed.c b/openssl/crypto/bn/expspeed.c index 4d5f221f3..513a568a4 100644 --- a/openssl/crypto/bn/expspeed.c +++ b/openssl/crypto/bn/expspeed.c @@ -7,21 +7,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -36,10 +36,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -51,7 +51,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -60,12 +60,13 @@ /* most of this code has been pilfered from my libdes speed.c program */ -#define BASENUM 5000 +#define BASENUM 5000 #define NUM_START 0 - -/* determine timings for modexp, modmul, modsqr, gcd, Kronecker symbol, - * modular inverse, or modular square roots */ +/* + * determine timings for modexp, modmul, modsqr, gcd, Kronecker symbol, + * modular inverse, or modular square roots + */ #define TEST_EXP #undef TEST_MUL #undef TEST_SQR @@ -73,19 +74,18 @@ #undef TEST_KRON #undef TEST_INV #undef TEST_SQRT -#define P_MOD_64 9 /* least significant 6 bits for prime to be used for BN_sqrt timings */ +#define P_MOD_64 9 /* least significant 6 bits for prime to be + * used for BN_sqrt timings */ #if defined(TEST_EXP) + defined(TEST_MUL) + defined(TEST_SQR) + defined(TEST_GCD) + defined(TEST_KRON) + defined(TEST_INV) +defined(TEST_SQRT) != 1 -# error "choose one test" +# error "choose one test" #endif #if defined(TEST_INV) || defined(TEST_SQRT) -# define C_PRIME +# define C_PRIME static void genprime_cb(int p, int n, void *arg); #endif - - #undef PROG #define PROG bnspeed_main @@ -98,33 +98,35 @@ static void genprime_cb(int p, int n, void *arg); #include <openssl/rand.h> #if !defined(OPENSSL_SYS_MSDOS) && (!defined(OPENSSL_SYS_VMS) || defined(__DECC)) && !defined(OPENSSL_SYS_MACOSX) -#define TIMES +# define TIMES #endif #ifndef _IRIX -#include <time.h> +# include <time.h> #endif #ifdef TIMES -#include <sys/types.h> -#include <sys/times.h> +# include <sys/types.h> +# include <sys/times.h> #endif -/* Depending on the VMS version, the tms structure is perhaps defined. - The __TMS macro will show if it was. If it wasn't defined, we should - undefine TIMES, since that tells the rest of the program how things - should be handled. -- Richard Levitte */ +/* + * Depending on the VMS version, the tms structure is perhaps defined. The + * __TMS macro will show if it was. If it wasn't defined, we should undefine + * TIMES, since that tells the rest of the program how things should be + * handled. -- Richard Levitte + */ #if defined(OPENSSL_SYS_VMS_DECC) && !defined(__TMS) -#undef TIMES +# undef TIMES #endif #ifndef TIMES -#include <sys/timeb.h> +# include <sys/timeb.h> #endif #if defined(sun) || defined(__ultrix) -#define _POSIX_SOURCE -#include <limits.h> -#include <sys/param.h> +# define _POSIX_SOURCE +# include <limits.h> +# include <sys/param.h> #endif #include <openssl/bn.h> @@ -133,221 +135,247 @@ static void genprime_cb(int p, int n, void *arg); /* The following if from times(3) man page. It may need to be changed */ #ifndef HZ # ifndef CLK_TCK -# ifndef _BSD_CLK_TCK_ /* FreeBSD hack */ -# define HZ 100.0 -# else /* _BSD_CLK_TCK_ */ +# ifndef _BSD_CLK_TCK_ /* FreeBSD hack */ +# define HZ 100.0 +# else /* _BSD_CLK_TCK_ */ # define HZ ((double)_BSD_CLK_TCK_) # endif -# else /* CLK_TCK */ +# else /* CLK_TCK */ # define HZ ((double)CLK_TCK) # endif #endif #undef BUFSIZE -#define BUFSIZE ((long)1024*8) -int run=0; +#define BUFSIZE ((long)1024*8) +int run = 0; static double Time_F(int s); -#define START 0 -#define STOP 1 +#define START 0 +#define STOP 1 static double Time_F(int s) - { - double ret; +{ + double ret; #ifdef TIMES - static struct tms tstart,tend; - - if (s == START) - { - times(&tstart); - return(0); - } - else - { - times(&tend); - ret=((double)(tend.tms_utime-tstart.tms_utime))/HZ; - return((ret < 1e-3)?1e-3:ret); - } -#else /* !times() */ - static struct timeb tstart,tend; - long i; - - if (s == START) - { - ftime(&tstart); - return(0); - } - else - { - ftime(&tend); - i=(long)tend.millitm-(long)tstart.millitm; - ret=((double)(tend.time-tstart.time))+((double)i)/1000.0; - return((ret < 0.001)?0.001:ret); - } + static struct tms tstart, tend; + + if (s == START) { + times(&tstart); + return (0); + } else { + times(&tend); + ret = ((double)(tend.tms_utime - tstart.tms_utime)) / HZ; + return ((ret < 1e-3) ? 1e-3 : ret); + } +#else /* !times() */ + static struct timeb tstart, tend; + long i; + + if (s == START) { + ftime(&tstart); + return (0); + } else { + ftime(&tend); + i = (long)tend.millitm - (long)tstart.millitm; + ret = ((double)(tend.time - tstart.time)) + ((double)i) / 1000.0; + return ((ret < 0.001) ? 0.001 : ret); + } #endif - } +} -#define NUM_SIZES 7 +#define NUM_SIZES 7 #if NUM_START > NUM_SIZES -# error "NUM_START > NUM_SIZES" +# error "NUM_START > NUM_SIZES" #endif -static int sizes[NUM_SIZES]={128,256,512,1024,2048,4096,8192}; -static int mul_c[NUM_SIZES]={8*8*8*8*8*8,8*8*8*8*8,8*8*8*8,8*8*8,8*8,8,1}; -/*static int sizes[NUM_SIZES]={59,179,299,419,539}; */ +static int sizes[NUM_SIZES] = { 128, 256, 512, 1024, 2048, 4096, 8192 }; + +static int mul_c[NUM_SIZES] = + { 8 * 8 * 8 * 8 * 8 * 8, 8 * 8 * 8 * 8 * 8, 8 * 8 * 8 * 8, 8 * 8 * 8, + 8 * 8, 8, 1 +}; + +/* + * static int sizes[NUM_SIZES]={59,179,299,419,539}; + */ #define RAND_SEED(string) { const char str[] = string; RAND_seed(string, sizeof str); } -void do_mul_exp(BIGNUM *r,BIGNUM *a,BIGNUM *b,BIGNUM *c,BN_CTX *ctx); +void do_mul_exp(BIGNUM *r, BIGNUM *a, BIGNUM *b, BIGNUM *c, BN_CTX *ctx); int main(int argc, char **argv) - { - BN_CTX *ctx; - BIGNUM *a,*b,*c,*r; +{ + BN_CTX *ctx; + BIGNUM *a, *b, *c, *r; #if 1 - if (!CRYPTO_set_mem_debug_functions(0,0,0,0,0)) - abort(); + if (!CRYPTO_set_mem_debug_functions(0, 0, 0, 0, 0)) + abort(); #endif - ctx=BN_CTX_new(); - a=BN_new(); - b=BN_new(); - c=BN_new(); - r=BN_new(); + ctx = BN_CTX_new(); + a = BN_new(); + b = BN_new(); + c = BN_new(); + r = BN_new(); - while (!RAND_status()) - /* not enough bits */ - RAND_SEED("I demand a manual recount!"); + while (!RAND_status()) + /* not enough bits */ + RAND_SEED("I demand a manual recount!"); - do_mul_exp(r,a,b,c,ctx); - return 0; - } + do_mul_exp(r, a, b, c, ctx); + return 0; +} void do_mul_exp(BIGNUM *r, BIGNUM *a, BIGNUM *b, BIGNUM *c, BN_CTX *ctx) - { - int i,k; - double tm; - long num; - - num=BASENUM; - for (i=NUM_START; i<NUM_SIZES; i++) - { +{ + int i, k; + double tm; + long num; + + num = BASENUM; + for (i = NUM_START; i < NUM_SIZES; i++) { #ifdef C_PRIME -# ifdef TEST_SQRT - if (!BN_set_word(a, 64)) goto err; - if (!BN_set_word(b, P_MOD_64)) goto err; -# define ADD a -# define REM b -# else -# define ADD NULL -# define REM NULL -# endif - if (!BN_generate_prime(c,sizes[i],0,ADD,REM,genprime_cb,NULL)) goto err; - putc('\n', stderr); - fflush(stderr); +# ifdef TEST_SQRT + if (!BN_set_word(a, 64)) + goto err; + if (!BN_set_word(b, P_MOD_64)) + goto err; +# define ADD a +# define REM b +# else +# define ADD NULL +# define REM NULL +# endif + if (!BN_generate_prime(c, sizes[i], 0, ADD, REM, genprime_cb, NULL)) + goto err; + putc('\n', stderr); + fflush(stderr); #endif - for (k=0; k<num; k++) - { - if (k%50 == 0) /* Average over num/50 different choices of random numbers. */ - { - if (!BN_pseudo_rand(a,sizes[i],1,0)) goto err; + for (k = 0; k < num; k++) { + if (k % 50 == 0) { /* Average over num/50 different choices of + * random numbers. */ + if (!BN_pseudo_rand(a, sizes[i], 1, 0)) + goto err; - if (!BN_pseudo_rand(b,sizes[i],1,0)) goto err; + if (!BN_pseudo_rand(b, sizes[i], 1, 0)) + goto err; #ifndef C_PRIME - if (!BN_pseudo_rand(c,sizes[i],1,1)) goto err; + if (!BN_pseudo_rand(c, sizes[i], 1, 1)) + goto err; #endif -#ifdef TEST_SQRT - if (!BN_mod_sqr(a,a,c,ctx)) goto err; - if (!BN_mod_sqr(b,b,c,ctx)) goto err; +#ifdef TEST_SQRT + if (!BN_mod_sqr(a, a, c, ctx)) + goto err; + if (!BN_mod_sqr(b, b, c, ctx)) + goto err; #else - if (!BN_nnmod(a,a,c,ctx)) goto err; - if (!BN_nnmod(b,b,c,ctx)) goto err; + if (!BN_nnmod(a, a, c, ctx)) + goto err; + if (!BN_nnmod(b, b, c, ctx)) + goto err; #endif - if (k == 0) - Time_F(START); - } - + if (k == 0) + Time_F(START); + } #if defined(TEST_EXP) - if (!BN_mod_exp(r,a,b,c,ctx)) goto err; + if (!BN_mod_exp(r, a, b, c, ctx)) + goto err; #elif defined(TEST_MUL) - { - int i = 0; - for (i = 0; i < 50; i++) - if (!BN_mod_mul(r,a,b,c,ctx)) goto err; - } + { + int i = 0; + for (i = 0; i < 50; i++) + if (!BN_mod_mul(r, a, b, c, ctx)) + goto err; + } #elif defined(TEST_SQR) - { - int i = 0; - for (i = 0; i < 50; i++) - { - if (!BN_mod_sqr(r,a,c,ctx)) goto err; - if (!BN_mod_sqr(r,b,c,ctx)) goto err; - } - } + { + int i = 0; + for (i = 0; i < 50; i++) { + if (!BN_mod_sqr(r, a, c, ctx)) + goto err; + if (!BN_mod_sqr(r, b, c, ctx)) + goto err; + } + } #elif defined(TEST_GCD) - if (!BN_gcd(r,a,b,ctx)) goto err; - if (!BN_gcd(r,b,c,ctx)) goto err; - if (!BN_gcd(r,c,a,ctx)) goto err; + if (!BN_gcd(r, a, b, ctx)) + goto err; + if (!BN_gcd(r, b, c, ctx)) + goto err; + if (!BN_gcd(r, c, a, ctx)) + goto err; #elif defined(TEST_KRON) - if (-2 == BN_kronecker(a,b,ctx)) goto err; - if (-2 == BN_kronecker(b,c,ctx)) goto err; - if (-2 == BN_kronecker(c,a,ctx)) goto err; + if (-2 == BN_kronecker(a, b, ctx)) + goto err; + if (-2 == BN_kronecker(b, c, ctx)) + goto err; + if (-2 == BN_kronecker(c, a, ctx)) + goto err; #elif defined(TEST_INV) - if (!BN_mod_inverse(r,a,c,ctx)) goto err; - if (!BN_mod_inverse(r,b,c,ctx)) goto err; -#else /* TEST_SQRT */ - if (!BN_mod_sqrt(r,a,c,ctx)) goto err; - if (!BN_mod_sqrt(r,b,c,ctx)) goto err; + if (!BN_mod_inverse(r, a, c, ctx)) + goto err; + if (!BN_mod_inverse(r, b, c, ctx)) + goto err; +#else /* TEST_SQRT */ + if (!BN_mod_sqrt(r, a, c, ctx)) + goto err; + if (!BN_mod_sqrt(r, b, c, ctx)) + goto err; #endif - } - tm=Time_F(STOP); - printf( + } + tm = Time_F(STOP); + printf( #if defined(TEST_EXP) - "modexp %4d ^ %4d %% %4d" + "modexp %4d ^ %4d %% %4d" #elif defined(TEST_MUL) - "50*modmul %4d %4d %4d" + "50*modmul %4d %4d %4d" #elif defined(TEST_SQR) - "100*modsqr %4d %4d %4d" + "100*modsqr %4d %4d %4d" #elif defined(TEST_GCD) - "3*gcd %4d %4d %4d" + "3*gcd %4d %4d %4d" #elif defined(TEST_KRON) - "3*kronecker %4d %4d %4d" + "3*kronecker %4d %4d %4d" #elif defined(TEST_INV) - "2*inv %4d %4d mod %4d" -#else /* TEST_SQRT */ - "2*sqrt [prime == %d (mod 64)] %4d %4d mod %4d" + "2*inv %4d %4d mod %4d" +#else /* TEST_SQRT */ + "2*sqrt [prime == %d (mod 64)] %4d %4d mod %4d" #endif - " -> %8.6fms %5.1f (%ld)\n", + " -> %8.6fms %5.1f (%ld)\n", #ifdef TEST_SQRT - P_MOD_64, + P_MOD_64, #endif - sizes[i],sizes[i],sizes[i],tm*1000.0/num,tm*mul_c[i]/num, num); - num/=7; - if (num <= 0) num=1; - } - return; + sizes[i], sizes[i], sizes[i], tm * 1000.0 / num, + tm * mul_c[i] / num, num); + num /= 7; + if (num <= 0) + num = 1; + } + return; err: - ERR_print_errors_fp(stderr); - } - + ERR_print_errors_fp(stderr); +} #ifdef C_PRIME static void genprime_cb(int p, int n, void *arg) - { - char c='*'; - - if (p == 0) c='.'; - if (p == 1) c='+'; - if (p == 2) c='*'; - if (p == 3) c='\n'; - putc(c, stderr); - fflush(stderr); - (void)n; - (void)arg; - } +{ + char c = '*'; + + if (p == 0) + c = '.'; + if (p == 1) + c = '+'; + if (p == 2) + c = '*'; + if (p == 3) + c = '\n'; + putc(c, stderr); + fflush(stderr); + (void)n; + (void)arg; +} #endif diff --git a/openssl/crypto/bn/exptest.c b/openssl/crypto/bn/exptest.c index 5fa02a122..8b3a4bae4 100644 --- a/openssl/crypto/bn/exptest.c +++ b/openssl/crypto/bn/exptest.c @@ -5,21 +5,21 @@ * This package is an SSL implementation written * by Eric Young (eay@cryptsoft.com). * The implementation was written so as to conform with Netscapes SSL. - * + * * This library is free for commercial and non-commercial use as long as * the following conditions are aheared to. The following conditions * apply to all code found in this distribution, be it the RC4, RSA, * lhash, DES, etc., code; not just the SSL code. The SSL documentation * included with this distribution is covered by the same copyright terms * except that the holder is Tim Hudson (tjh@cryptsoft.com). - * + * * Copyright remains Eric Young's, and as such any Copyright notices in * the code are not to be removed. * If this package is used in a product, Eric Young should be given attribution * as the author of the parts of the library used. * This can be in the form of a textual message at program startup or * in documentation (online or textual) provided with the package. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -34,10 +34,10 @@ * Eric Young (eay@cryptsoft.com)" * The word 'cryptographic' can be left out if the rouines from the library * being used are not cryptographic related :-). - * 4. If you include any Windows specific code (or a derivative thereof) from + * 4. If you include any Windows specific code (or a derivative thereof) from * the apps directory (application code) you must include an acknowledgement: * "This product includes software written by Tim Hudson (tjh@cryptsoft.com)" - * + * * THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -49,7 +49,7 @@ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. - * + * * The licence and distribution terms for any publically available version or * derivative of this code cannot be changed. i.e. this code cannot simply be * copied and put under another distribution licence @@ -67,181 +67,183 @@ #include <openssl/rand.h> #include <openssl/err.h> -#define NUM_BITS (BN_BITS*2) - -static const char rnd_seed[] = "string to make the random number generator think it has entropy"; - -/* test_exp_mod_zero tests that x**0 mod 1 == 0. It returns zero on success. */ -static int test_exp_mod_zero() { - BIGNUM a, p, m; - BIGNUM r; - BN_CTX *ctx = BN_CTX_new(); - int ret = 1; - - BN_init(&m); - BN_one(&m); +#define NUM_BITS (BN_BITS*2) - BN_init(&a); - BN_one(&a); +static const char rnd_seed[] = + "string to make the random number generator think it has entropy"; - BN_init(&p); - BN_zero(&p); - - BN_init(&r); - BN_mod_exp(&r, &a, &p, &m, ctx); - BN_CTX_free(ctx); - - if (BN_is_zero(&r)) - ret = 0; - else - { - printf("1**0 mod 1 = "); - BN_print_fp(stdout, &r); - printf(", should be 0\n"); - } - - BN_free(&r); - BN_free(&a); - BN_free(&p); - BN_free(&m); - - return ret; +/* + * test_exp_mod_zero tests that x**0 mod 1 == 0. It returns zero on success. + */ +static int test_exp_mod_zero() +{ + BIGNUM a, p, m; + BIGNUM r; + BN_CTX *ctx = BN_CTX_new(); + int ret = 1; + + BN_init(&m); + BN_one(&m); + + BN_init(&a); + BN_one(&a); + + BN_init(&p); + BN_zero(&p); + + BN_init(&r); + BN_mod_exp(&r, &a, &p, &m, ctx); + BN_CTX_free(ctx); + + if (BN_is_zero(&r)) + ret = 0; + else { + printf("1**0 mod 1 = "); + BN_print_fp(stdout, &r); + printf(", should be 0\n"); + } + + BN_free(&r); + BN_free(&a); + BN_free(&p); + BN_free(&m); + + return ret; } int main(int argc, char *argv[]) - { - BN_CTX *ctx; - BIO *out=NULL; - int i,ret; - unsigned char c; - BIGNUM *r_mont,*r_mont_const,*r_recp,*r_simple,*a,*b,*m; - - RAND_seed(rnd_seed, sizeof rnd_seed); /* or BN_rand may fail, and we don't - * even check its return value - * (which we should) */ - - ERR_load_BN_strings(); - - ctx=BN_CTX_new(); - if (ctx == NULL) EXIT(1); - r_mont=BN_new(); - r_mont_const=BN_new(); - r_recp=BN_new(); - r_simple=BN_new(); - a=BN_new(); - b=BN_new(); - m=BN_new(); - if ( (r_mont == NULL) || (r_recp == NULL) || - (a == NULL) || (b == NULL)) - goto err; - - out=BIO_new(BIO_s_file()); - - if (out == NULL) EXIT(1); - BIO_set_fp(out,stdout,BIO_NOCLOSE); - - for (i=0; i<200; i++) - { - RAND_bytes(&c,1); - c=(c%BN_BITS)-BN_BITS2; - BN_rand(a,NUM_BITS+c,0,0); - - RAND_bytes(&c,1); - c=(c%BN_BITS)-BN_BITS2; - BN_rand(b,NUM_BITS+c,0,0); - - RAND_bytes(&c,1); - c=(c%BN_BITS)-BN_BITS2; - BN_rand(m,NUM_BITS+c,0,1); - - BN_mod(a,a,m,ctx); - BN_mod(b,b,m,ctx); - - ret=BN_mod_exp_mont(r_mont,a,b,m,ctx,NULL); - if (ret <= 0) - { - printf("BN_mod_exp_mont() problems\n"); - ERR_print_errors(out); - EXIT(1); - } - - ret=BN_mod_exp_recp(r_recp,a,b,m,ctx); - if (ret <= 0) - { - printf("BN_mod_exp_recp() problems\n"); - ERR_print_errors(out); - EXIT(1); - } - - ret=BN_mod_exp_simple(r_simple,a,b,m,ctx); - if (ret <= 0) - { - printf("BN_mod_exp_simple() problems\n"); - ERR_print_errors(out); - EXIT(1); - } - - ret=BN_mod_exp_mont_consttime(r_mont_const,a,b,m,ctx,NULL); - if (ret <= 0) - { - printf("BN_mod_exp_mont_consttime() problems\n"); - ERR_print_errors(out); - EXIT(1); - } - - if (BN_cmp(r_simple, r_mont) == 0 - && BN_cmp(r_simple,r_recp) == 0 - && BN_cmp(r_simple,r_mont_const) == 0) - { - printf("."); - fflush(stdout); - } - else - { - if (BN_cmp(r_simple,r_mont) != 0) - printf("\nsimple and mont results differ\n"); - if (BN_cmp(r_simple,r_mont_const) != 0) - printf("\nsimple and mont const time results differ\n"); - if (BN_cmp(r_simple,r_recp) != 0) - printf("\nsimple and recp results differ\n"); - - printf("a (%3d) = ",BN_num_bits(a)); BN_print(out,a); - printf("\nb (%3d) = ",BN_num_bits(b)); BN_print(out,b); - printf("\nm (%3d) = ",BN_num_bits(m)); BN_print(out,m); - printf("\nsimple ="); BN_print(out,r_simple); - printf("\nrecp ="); BN_print(out,r_recp); - printf("\nmont ="); BN_print(out,r_mont); - printf("\nmont_ct ="); BN_print(out,r_mont_const); - printf("\n"); - EXIT(1); - } - } - BN_free(r_mont); - BN_free(r_mont_const); - BN_free(r_recp); - BN_free(r_simple); - BN_free(a); - BN_free(b); - BN_free(m); - BN_CTX_free(ctx); - ERR_remove_thread_state(NULL); - CRYPTO_mem_leaks(out); - BIO_free(out); - printf("\n"); - - if (test_exp_mod_zero() != 0) - goto err; - - printf("done\n"); - - EXIT(0); -err: - ERR_load_crypto_strings(); - ERR_print_errors(out); +{ + BN_CTX *ctx; + BIO *out = NULL; + int i, ret; + unsigned char c; + BIGNUM *r_mont, *r_mont_const, *r_recp, *r_simple, *a, *b, *m; + + RAND_seed(rnd_seed, sizeof rnd_seed); /* or BN_rand may fail, and we + * don't even check its return + * value (which we should) */ + + ERR_load_BN_strings(); + + ctx = BN_CTX_new(); + if (ctx == NULL) + EXIT(1); + r_mont = BN_new(); + r_mont_const = BN_new(); + r_recp = BN_new(); + r_simple = BN_new(); + a = BN_new(); + b = BN_new(); + m = BN_new(); + if ((r_mont == NULL) || (r_recp == NULL) || (a == NULL) || (b == NULL)) + goto err; + + out = BIO_new(BIO_s_file()); + + if (out == NULL) + EXIT(1); + BIO_set_fp(out, stdout, BIO_NOCLOSE); + + for (i = 0; i < 200; i++) { + RAND_bytes(&c, 1); + c = (c % BN_BITS) - BN_BITS2; + BN_rand(a, NUM_BITS + c, 0, 0); + + RAND_bytes(&c, 1); + c = (c % BN_BITS) - BN_BITS2; + BN_rand(b, NUM_BITS + c, 0, 0); + + RAND_bytes(&c, 1); + c = (c % BN_BITS) - BN_BITS2; + BN_rand(m, NUM_BITS + c, 0, 1); + + BN_mod(a, a, m, ctx); + BN_mod(b, b, m, ctx); + + ret = BN_mod_exp_mont(r_mont, a, b, m, ctx, NULL); + if (ret <= 0) { + printf("BN_mod_exp_mont() problems\n"); + ERR_print_errors(out); + EXIT(1); + } + + ret = BN_mod_exp_recp(r_recp, a, b, m, ctx); + if (ret <= 0) { + printf("BN_mod_exp_recp() problems\n"); + ERR_print_errors(out); + EXIT(1); + } + + ret = BN_mod_exp_simple(r_simple, a, b, m, ctx); + if (ret <= 0) { + printf("BN_mod_exp_simple() problems\n"); + ERR_print_errors(out); + EXIT(1); + } + + ret = BN_mod_exp_mont_consttime(r_mont_const, a, b, m, ctx, NULL); + if (ret <= 0) { + printf("BN_mod_exp_mont_consttime() problems\n"); + ERR_print_errors(out); + EXIT(1); + } + + if (BN_cmp(r_simple, r_mont) == 0 + && BN_cmp(r_simple, r_recp) == 0 + && BN_cmp(r_simple, r_mont_const) == 0) { + printf("."); + fflush(stdout); + } else { + if (BN_cmp(r_simple, r_mont) != 0) + printf("\nsimple and mont results differ\n"); + if (BN_cmp(r_simple, r_mont_const) != 0) + printf("\nsimple and mont const time results differ\n"); + if (BN_cmp(r_simple, r_recp) != 0) + printf("\nsimple and recp results differ\n"); + + printf("a (%3d) = ", BN_num_bits(a)); + BN_print(out, a); + printf("\nb (%3d) = ", BN_num_bits(b)); + BN_print(out, b); + printf("\nm (%3d) = ", BN_num_bits(m)); + BN_print(out, m); + printf("\nsimple ="); + BN_print(out, r_simple); + printf("\nrecp ="); + BN_print(out, r_recp); + printf("\nmont ="); + BN_print(out, r_mont); + printf("\nmont_ct ="); + BN_print(out, r_mont_const); + printf("\n"); + EXIT(1); + } + } + BN_free(r_mont); + BN_free(r_mont_const); + BN_free(r_recp); + BN_free(r_simple); + BN_free(a); + BN_free(b); + BN_free(m); + BN_CTX_free(ctx); + ERR_remove_thread_state(NULL); + CRYPTO_mem_leaks(out); + BIO_free(out); + printf("\n"); + + if (test_exp_mod_zero() != 0) + goto err; + + printf("done\n"); + + EXIT(0); + err: + ERR_load_crypto_strings(); + ERR_print_errors(out); #ifdef OPENSSL_SYS_NETWARE printf("ERROR\n"); #endif - EXIT(1); - return(1); - } - + EXIT(1); + return (1); +} diff --git a/openssl/crypto/bn/rsaz_exp.c b/openssl/crypto/bn/rsaz_exp.c new file mode 100755 index 000000000..a486b154c --- /dev/null +++ b/openssl/crypto/bn/rsaz_exp.c @@ -0,0 +1,336 @@ +/***************************************************************************** +* * +* Copyright (c) 2012, Intel Corporation * +* * +* All rights reserved. * +* * +* Redistribution and use in source and binary forms, with or without * +* modification, are permitted provided that the following conditions are * +* met: * +* * +* * Redistributions of source code must retain the above copyright * +* notice, this list of conditions and the following disclaimer. * +* * +* * Redistributions in binary form must reproduce the above copyright * +* notice, this list of conditions and the following disclaimer in the * +* documentation and/or other materials provided with the * +* distribution. * +* * +* * Neither the name of the Intel Corporation nor the names of its * +* contributors may be used to endorse or promote products derived from * +* this software without specific prior written permission. * +* * +* * +* THIS SOFTWARE IS PROVIDED BY INTEL CORPORATION ""AS IS"" AND ANY * +* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * +* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR * +* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL INTEL CORPORATION OR * +* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, * +* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, * +* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR * +* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF * +* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING * +* NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS * +* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * +* * +****************************************************************************** +* Developers and authors: * +* Shay Gueron (1, 2), and Vlad Krasnov (1) * +* (1) Intel Corporation, Israel Development Center, Haifa, Israel * +* (2) University of Haifa, Israel * +*****************************************************************************/ + +#include "rsaz_exp.h" + +/* + * See crypto/bn/asm/rsaz-avx2.pl for further details. + */ +void rsaz_1024_norm2red_avx2(void *red, const void *norm); +void rsaz_1024_mul_avx2(void *ret, const void *a, const void *b, + const void *n, BN_ULONG k); +void rsaz_1024_sqr_avx2(void *ret, const void *a, const void *n, BN_ULONG k, + int cnt); +void rsaz_1024_scatter5_avx2(void *tbl, const void *val, int i); +void rsaz_1024_gather5_avx2(void *val, const void *tbl, int i); +void rsaz_1024_red2norm_avx2(void *norm, const void *red); + +#if defined(__GNUC__) +# define ALIGN64 __attribute__((aligned(64))) +#elif defined(_MSC_VER) +# define ALIGN64 __declspec(align(64)) +#elif defined(__SUNPRO_C) +# define ALIGN64 +# pragma align 64(one,two80) +#else +/* not fatal, might hurt performance a little */ +# define ALIGN64 +#endif + +ALIGN64 static const BN_ULONG one[40] = { + 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +}; + +ALIGN64 static const BN_ULONG two80[40] = { + 0, 0, 1 << 22, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +}; + +void RSAZ_1024_mod_exp_avx2(BN_ULONG result_norm[16], + const BN_ULONG base_norm[16], + const BN_ULONG exponent[16], + const BN_ULONG m_norm[16], const BN_ULONG RR[16], + BN_ULONG k0) +{ + unsigned char storage[320 * 3 + 32 * 9 * 16 + 64]; /* 5.5KB */ + unsigned char *p_str = storage + (64 - ((size_t)storage % 64)); + unsigned char *a_inv, *m, *result; + unsigned char *table_s = p_str + 320 * 3; + unsigned char *R2 = table_s; /* borrow */ + int index; + int wvalue; + + if ((((size_t)p_str & 4095) + 320) >> 12) { + result = p_str; + a_inv = p_str + 320; + m = p_str + 320 * 2; /* should not cross page */ + } else { + m = p_str; /* should not cross page */ + result = p_str + 320; + a_inv = p_str + 320 * 2; + } + + rsaz_1024_norm2red_avx2(m, m_norm); + rsaz_1024_norm2red_avx2(a_inv, base_norm); + rsaz_1024_norm2red_avx2(R2, RR); + + rsaz_1024_mul_avx2(R2, R2, R2, m, k0); + rsaz_1024_mul_avx2(R2, R2, two80, m, k0); + + /* table[0] = 1 */ + rsaz_1024_mul_avx2(result, R2, one, m, k0); + /* table[1] = a_inv^1 */ + rsaz_1024_mul_avx2(a_inv, a_inv, R2, m, k0); + + rsaz_1024_scatter5_avx2(table_s, result, 0); + rsaz_1024_scatter5_avx2(table_s, a_inv, 1); + + /* table[2] = a_inv^2 */ + rsaz_1024_sqr_avx2(result, a_inv, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 2); +#if 0 + /* this is almost 2x smaller and less than 1% slower */ + for (index = 3; index < 32; index++) { + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, index); + } +#else + /* table[4] = a_inv^4 */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 4); + /* table[8] = a_inv^8 */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 8); + /* table[16] = a_inv^16 */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 16); + /* table[17] = a_inv^17 */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 17); + + /* table[3] */ + rsaz_1024_gather5_avx2(result, table_s, 2); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 3); + /* table[6] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 6); + /* table[12] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 12); + /* table[24] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 24); + /* table[25] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 25); + + /* table[5] */ + rsaz_1024_gather5_avx2(result, table_s, 4); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 5); + /* table[10] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 10); + /* table[20] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 20); + /* table[21] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 21); + + /* table[7] */ + rsaz_1024_gather5_avx2(result, table_s, 6); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 7); + /* table[14] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 14); + /* table[28] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 28); + /* table[29] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 29); + + /* table[9] */ + rsaz_1024_gather5_avx2(result, table_s, 8); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 9); + /* table[18] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 18); + /* table[19] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 19); + + /* table[11] */ + rsaz_1024_gather5_avx2(result, table_s, 10); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 11); + /* table[22] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 22); + /* table[23] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 23); + + /* table[13] */ + rsaz_1024_gather5_avx2(result, table_s, 12); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 13); + /* table[26] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 26); + /* table[27] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 27); + + /* table[15] */ + rsaz_1024_gather5_avx2(result, table_s, 14); + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 15); + /* table[30] */ + rsaz_1024_sqr_avx2(result, result, m, k0, 1); + rsaz_1024_scatter5_avx2(table_s, result, 30); + /* table[31] */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + rsaz_1024_scatter5_avx2(table_s, result, 31); +#endif + + /* load first window */ + p_str = (unsigned char *)exponent; + wvalue = p_str[127] >> 3; + rsaz_1024_gather5_avx2(result, table_s, wvalue); + + index = 1014; + + while (index > -1) { /* loop for the remaining 127 windows */ + + rsaz_1024_sqr_avx2(result, result, m, k0, 5); + + wvalue = *((unsigned short *)&p_str[index / 8]); + wvalue = (wvalue >> (index % 8)) & 31; + index -= 5; + + rsaz_1024_gather5_avx2(a_inv, table_s, wvalue); /* borrow a_inv */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + } + + /* square four times */ + rsaz_1024_sqr_avx2(result, result, m, k0, 4); + + wvalue = p_str[0] & 15; + + rsaz_1024_gather5_avx2(a_inv, table_s, wvalue); /* borrow a_inv */ + rsaz_1024_mul_avx2(result, result, a_inv, m, k0); + + /* from Montgomery */ + rsaz_1024_mul_avx2(result, result, one, m, k0); + + rsaz_1024_red2norm_avx2(result_norm, result); + + OPENSSL_cleanse(storage, sizeof(storage)); +} + +/* + * See crypto/bn/rsaz-x86_64.pl for further details. + */ +void rsaz_512_mul(void *ret, const void *a, const void *b, const void *n, + BN_ULONG k); +void rsaz_512_mul_scatter4(void *ret, const void *a, const void *n, + BN_ULONG k, const void *tbl, unsigned int power); +void rsaz_512_mul_gather4(void *ret, const void *a, const void *tbl, + const void *n, BN_ULONG k, unsigned int power); +void rsaz_512_mul_by_one(void *ret, const void *a, const void *n, BN_ULONG k); +void rsaz_512_sqr(void *ret, const void *a, const void *n, BN_ULONG k, + int cnt); +void rsaz_512_scatter4(void *tbl, const BN_ULONG *val, int power); +void rsaz_512_gather4(BN_ULONG *val, const void *tbl, int power); + +void RSAZ_512_mod_exp(BN_ULONG result[8], + const BN_ULONG base[8], const BN_ULONG exponent[8], + const BN_ULONG m[8], BN_ULONG k0, const BN_ULONG RR[8]) +{ + unsigned char storage[16 * 8 * 8 + 64 * 2 + 64]; /* 1.2KB */ + unsigned char *table = storage + (64 - ((size_t)storage % 64)); + BN_ULONG *a_inv = (BN_ULONG *)(table + 16 * 8 * 8); + BN_ULONG *temp = (BN_ULONG *)(table + 16 * 8 * 8 + 8 * 8); + unsigned char *p_str = (unsigned char *)exponent; + int index; + unsigned int wvalue; + + /* table[0] = 1_inv */ + temp[0] = 0 - m[0]; + temp[1] = ~m[1]; + temp[2] = ~m[2]; + temp[3] = ~m[3]; + temp[4] = ~m[4]; + temp[5] = ~m[5]; + temp[6] = ~m[6]; + temp[7] = ~m[7]; + rsaz_512_scatter4(table, temp, 0); + + /* table [1] = a_inv^1 */ + rsaz_512_mul(a_inv, base, RR, m, k0); + rsaz_512_scatter4(table, a_inv, 1); + + /* table [2] = a_inv^2 */ + rsaz_512_sqr(temp, a_inv, m, k0, 1); + rsaz_512_scatter4(table, temp, 2); + + for (index = 3; index < 16; index++) + rsaz_512_mul_scatter4(temp, a_inv, m, k0, table, index); + + /* load first window */ + wvalue = p_str[63]; + + rsaz_512_gather4(temp, table, wvalue >> 4); + rsaz_512_sqr(temp, temp, m, k0, 4); + rsaz_512_mul_gather4(temp, temp, table, m, k0, wvalue & 0xf); + + for (index = 62; index >= 0; index--) { + wvalue = p_str[index]; + + rsaz_512_sqr(temp, temp, m, k0, 4); + rsaz_512_mul_gather4(temp, temp, table, m, k0, wvalue >> 4); + + rsaz_512_sqr(temp, temp, m, k0, 4); + rsaz_512_mul_gather4(temp, temp, table, m, k0, wvalue & 0x0f); + } + + /* from Montgomery */ + rsaz_512_mul_by_one(result, temp, m, k0); + + OPENSSL_cleanse(storage, sizeof(storage)); +} diff --git a/openssl/crypto/bn/rsaz_exp.h b/openssl/crypto/bn/rsaz_exp.h new file mode 100755 index 000000000..bb71fb1e1 --- /dev/null +++ b/openssl/crypto/bn/rsaz_exp.h @@ -0,0 +1,47 @@ +/****************************************************************************** +* Copyright(c) 2012, Intel Corp. +* Developers and authors: +* Shay Gueron (1, 2), and Vlad Krasnov (1) +* (1) Intel Corporation, Israel Development Center, Haifa, Israel +* (2) University of Haifa, Israel +****************************************************************************** +* LICENSE: +* This submission to OpenSSL is to be made available under the OpenSSL +* license, and only to the OpenSSL project, in order to allow integration +* into the publicly distributed code. +* The use of this code, or portions of this code, or concepts embedded in +* this code, or modification of this code and/or algorithm(s) in it, or the +* use of this code for any other purpose than stated above, requires special +* licensing. +****************************************************************************** +* DISCLAIMER: +* THIS SOFTWARE IS PROVIDED BY THE CONTRIBUTORS AND THE COPYRIGHT OWNERS +* ``AS IS''. ANY EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED +* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR +* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE CONTRIBUTORS OR THE COPYRIGHT +* OWNERS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, +* OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF +* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS +* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN +* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) +* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE +* POSSIBILITY OF SUCH DAMAGE. +******************************************************************************/ + +#ifndef RSAZ_EXP_H +# define RSAZ_EXP_H + +# include <openssl/bn.h> + +void RSAZ_1024_mod_exp_avx2(BN_ULONG result[16], + const BN_ULONG base_norm[16], + const BN_ULONG exponent[16], + const BN_ULONG m_norm[16], const BN_ULONG RR[16], + BN_ULONG k0); +int rsaz_avx2_eligible(); + +void RSAZ_512_mod_exp(BN_ULONG result[8], + const BN_ULONG base_norm[8], const BN_ULONG exponent[8], + const BN_ULONG m_norm[8], BN_ULONG k0, + const BN_ULONG RR[8]); +#endif diff --git a/openssl/crypto/bn/vms-helper.c b/openssl/crypto/bn/vms-helper.c index 4b63149bf..f342e90cb 100644 --- a/openssl/crypto/bn/vms-helper.c +++ b/openssl/crypto/bn/vms-helper.c @@ -7,7 +7,7 @@ * are met: * * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. + * notice, this list of conditions and the following disclaimer. * * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in @@ -60,9 +60,9 @@ bn_div_words_abort(int i) { #ifdef BN_DEBUG -#if !defined(OPENSSL_NO_STDIO) && !defined(OPENSSL_SYS_WIN16) - fprintf(stderr,"Division would overflow (%d)\n",i); -#endif - abort(); +# if !defined(OPENSSL_NO_STDIO) && !defined(OPENSSL_SYS_WIN16) + fprintf(stderr, "Division would overflow (%d)\n", i); +# endif + abort(); #endif } |