aboutsummaryrefslogtreecommitdiff
path: root/openssl/crypto/sha/asm/sha256-armv4.pl
diff options
context:
space:
mode:
authormarha <marha@users.sourceforge.net>2012-04-10 11:54:31 +0200
committermarha <marha@users.sourceforge.net>2012-04-10 11:54:31 +0200
commit5564e91e3cf4ba5cb2fbebbc2d63d18f588016b8 (patch)
treec800a66664ea3af61eb13928db45a26275930b0b /openssl/crypto/sha/asm/sha256-armv4.pl
parentd79e641dea89c0d5d651b11971c4c9e14df34629 (diff)
parent67326634496ef21b4acbf4cef2f05040d34aef9b (diff)
downloadvcxsrv-5564e91e3cf4ba5cb2fbebbc2d63d18f588016b8.tar.gz
vcxsrv-5564e91e3cf4ba5cb2fbebbc2d63d18f588016b8.tar.bz2
vcxsrv-5564e91e3cf4ba5cb2fbebbc2d63d18f588016b8.zip
Merge remote-tracking branch 'origin/released'
Conflicts: openssl/Configure openssl/Makefile openssl/crypto/opensslconf.h openssl/util/mk1mf.pl openssl/util/pl/VC-32.pl
Diffstat (limited to 'openssl/crypto/sha/asm/sha256-armv4.pl')
-rw-r--r--openssl/crypto/sha/asm/sha256-armv4.pl55
1 files changed, 40 insertions, 15 deletions
diff --git a/openssl/crypto/sha/asm/sha256-armv4.pl b/openssl/crypto/sha/asm/sha256-armv4.pl
index 492cb62bc..9c84e8d93 100644
--- a/openssl/crypto/sha/asm/sha256-armv4.pl
+++ b/openssl/crypto/sha/asm/sha256-armv4.pl
@@ -18,11 +18,16 @@
# Rescheduling for dual-issue pipeline resulted in 22% improvement on
# Cortex A8 core and ~20 cycles per processed byte.
+# February 2011.
+#
+# Profiler-assisted and platform-specific optimization resulted in 16%
+# improvement on Cortex A8 core and ~17 cycles per processed byte.
+
while (($output=shift) && ($output!~/^\w[\w\-]*\.\w+$/)) {}
open STDOUT,">$output";
$ctx="r0"; $t0="r0";
-$inp="r1";
+$inp="r1"; $t3="r1";
$len="r2"; $t1="r2";
$T1="r3";
$A="r4";
@@ -46,6 +51,9 @@ sub BODY_00_15 {
my ($i,$a,$b,$c,$d,$e,$f,$g,$h) = @_;
$code.=<<___ if ($i<16);
+#if __ARM_ARCH__>=7
+ ldr $T1,[$inp],#4
+#else
ldrb $T1,[$inp,#3] @ $i
ldrb $t2,[$inp,#2]
ldrb $t1,[$inp,#1]
@@ -53,16 +61,24 @@ $code.=<<___ if ($i<16);
orr $T1,$T1,$t2,lsl#8
orr $T1,$T1,$t1,lsl#16
orr $T1,$T1,$t0,lsl#24
- `"str $inp,[sp,#17*4]" if ($i==15)`
+#endif
___
$code.=<<___;
- ldr $t2,[$Ktbl],#4 @ *K256++
mov $t0,$e,ror#$Sigma1[0]
- str $T1,[sp,#`$i%16`*4]
+ ldr $t2,[$Ktbl],#4 @ *K256++
eor $t0,$t0,$e,ror#$Sigma1[1]
eor $t1,$f,$g
+#if $i>=16
+ add $T1,$T1,$t3 @ from BODY_16_xx
+#elif __ARM_ARCH__>=7 && defined(__ARMEL__)
+ rev $T1,$T1
+#endif
+#if $i==15
+ str $inp,[sp,#17*4] @ leave room for $t3
+#endif
eor $t0,$t0,$e,ror#$Sigma1[2] @ Sigma1(e)
and $t1,$t1,$e
+ str $T1,[sp,#`$i%16`*4]
add $T1,$T1,$t0
eor $t1,$t1,$g @ Ch(e,f,g)
add $T1,$T1,$h
@@ -71,6 +87,9 @@ $code.=<<___;
eor $h,$h,$a,ror#$Sigma0[1]
add $T1,$T1,$t2
eor $h,$h,$a,ror#$Sigma0[2] @ Sigma0(a)
+#if $i>=15
+ ldr $t3,[sp,#`($i+2)%16`*4] @ from BODY_16_xx
+#endif
orr $t0,$a,$b
and $t1,$a,$b
and $t0,$t0,$c
@@ -85,24 +104,26 @@ sub BODY_16_XX {
my ($i,$a,$b,$c,$d,$e,$f,$g,$h) = @_;
$code.=<<___;
- ldr $t1,[sp,#`($i+1)%16`*4] @ $i
+ @ ldr $t3,[sp,#`($i+1)%16`*4] @ $i
ldr $t2,[sp,#`($i+14)%16`*4]
+ mov $t0,$t3,ror#$sigma0[0]
ldr $T1,[sp,#`($i+0)%16`*4]
- mov $t0,$t1,ror#$sigma0[0]
- ldr $inp,[sp,#`($i+9)%16`*4]
- eor $t0,$t0,$t1,ror#$sigma0[1]
- eor $t0,$t0,$t1,lsr#$sigma0[2] @ sigma0(X[i+1])
- mov $t1,$t2,ror#$sigma1[0]
+ eor $t0,$t0,$t3,ror#$sigma0[1]
+ ldr $t1,[sp,#`($i+9)%16`*4]
+ eor $t0,$t0,$t3,lsr#$sigma0[2] @ sigma0(X[i+1])
+ mov $t3,$t2,ror#$sigma1[0]
add $T1,$T1,$t0
- eor $t1,$t1,$t2,ror#$sigma1[1]
- add $T1,$T1,$inp
- eor $t1,$t1,$t2,lsr#$sigma1[2] @ sigma1(X[i+14])
+ eor $t3,$t3,$t2,ror#$sigma1[1]
add $T1,$T1,$t1
+ eor $t3,$t3,$t2,lsr#$sigma1[2] @ sigma1(X[i+14])
+ @ add $T1,$T1,$t3
___
&BODY_00_15(@_);
}
$code=<<___;
+#include "arm_arch.h"
+
.text
.code 32
@@ -132,7 +153,7 @@ K256:
sha256_block_data_order:
sub r3,pc,#8 @ sha256_block_data_order
add $len,$inp,$len,lsl#6 @ len to point at the end of inp
- stmdb sp!,{$ctx,$inp,$len,r4-r12,lr}
+ stmdb sp!,{$ctx,$inp,$len,r4-r11,lr}
ldmia $ctx,{$A,$B,$C,$D,$E,$F,$G,$H}
sub $Ktbl,r3,#256 @ K256
sub sp,sp,#16*4 @ alloca(X[16])
@@ -171,10 +192,14 @@ $code.=<<___;
bne .Loop
add sp,sp,#`16+3`*4 @ destroy frame
- ldmia sp!,{r4-r12,lr}
+#if __ARM_ARCH__>=5
+ ldmia sp!,{r4-r11,pc}
+#else
+ ldmia sp!,{r4-r11,lr}
tst lr,#1
moveq pc,lr @ be binary compatible with V4, yet
bx lr @ interoperable with Thumb ISA:-)
+#endif
.size sha256_block_data_order,.-sha256_block_data_order
.asciz "SHA256 block transform for ARMv4, CRYPTOGAMS by <appro\@openssl.org>"
.align 2