aboutsummaryrefslogtreecommitdiff
path: root/openssl/crypto/x86_64cpuid.pl
diff options
context:
space:
mode:
authormarha <marha@users.sourceforge.net>2010-03-29 17:08:02 +0000
committermarha <marha@users.sourceforge.net>2010-03-29 17:08:02 +0000
commit15272ab4ed1e6250412fccd48200ed9eae59608f (patch)
treea5996ea67966a778a16565f19dfc2e7c7f49b376 /openssl/crypto/x86_64cpuid.pl
parent3827301b2ea5a45ac009c3bf9f08586ff40b8506 (diff)
downloadvcxsrv-15272ab4ed1e6250412fccd48200ed9eae59608f.tar.gz
vcxsrv-15272ab4ed1e6250412fccd48200ed9eae59608f.tar.bz2
vcxsrv-15272ab4ed1e6250412fccd48200ed9eae59608f.zip
Updated to openssl 1.0.0
Diffstat (limited to 'openssl/crypto/x86_64cpuid.pl')
-rw-r--r--openssl/crypto/x86_64cpuid.pl256
1 files changed, 164 insertions, 92 deletions
diff --git a/openssl/crypto/x86_64cpuid.pl b/openssl/crypto/x86_64cpuid.pl
index 2616a03da..a7f98b3fd 100644
--- a/openssl/crypto/x86_64cpuid.pl
+++ b/openssl/crypto/x86_64cpuid.pl
@@ -1,108 +1,37 @@
#!/usr/bin/env perl
-$output=shift;
-$masm=1 if ($output =~ /\.asm/);
-open STDOUT,">$output" || die "can't open $output: $!";
-
-print<<___ if(defined($masm));
-_TEXT SEGMENT
-PUBLIC OPENSSL_rdtsc
-
-PUBLIC OPENSSL_atomic_add
-ALIGN 16
-OPENSSL_atomic_add PROC
- mov eax,DWORD PTR[rcx]
-\$Lspin: lea r8,DWORD PTR[rdx+rax]
-lock cmpxchg DWORD PTR[rcx],r8d
- jne \$Lspin
- mov eax,r8d
- cdqe
- ret
-OPENSSL_atomic_add ENDP
-
-PUBLIC OPENSSL_wipe_cpu
-ALIGN 16
-OPENSSL_wipe_cpu PROC
- pxor xmm0,xmm0
- pxor xmm1,xmm1
- pxor xmm2,xmm2
- pxor xmm3,xmm3
- pxor xmm4,xmm4
- pxor xmm5,xmm5
- xor rcx,rcx
- xor rdx,rdx
- xor r8,r8
- xor r9,r9
- xor r10,r10
- xor r11,r11
- lea rax,QWORD PTR[rsp+8]
- ret
-OPENSSL_wipe_cpu ENDP
-_TEXT ENDS
+$flavour = shift;
+$output = shift;
+if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
-CRT\$XIU SEGMENT
-EXTRN OPENSSL_cpuid_setup:PROC
-DQ OPENSSL_cpuid_setup
-CRT\$XIU ENDS
+$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);
+
+$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
+open STDOUT,"| $^X ${dir}perlasm/x86_64-xlate.pl $flavour $output";
+
+if ($win64) { $arg1="%rcx"; $arg2="%rdx"; }
+else { $arg1="%rdi"; $arg2="%rsi"; }
+print<<___;
+.extern OPENSSL_cpuid_setup
+.section .init
+ call OPENSSL_cpuid_setup
-___
-print<<___ if(!defined($masm));
.text
.globl OPENSSL_atomic_add
-.type OPENSSL_atomic_add,\@function
+.type OPENSSL_atomic_add,\@abi-omnipotent
.align 16
OPENSSL_atomic_add:
- movl (%rdi),%eax
-.Lspin: leaq (%rsi,%rax),%r8
-lock; cmpxchgl %r8d,(%rdi)
+ movl ($arg1),%eax
+.Lspin: leaq ($arg2,%rax),%r8
+ .byte 0xf0 # lock
+ cmpxchgl %r8d,($arg1)
jne .Lspin
movl %r8d,%eax
- .byte 0x48,0x98
+ .byte 0x48,0x98 # cltq/cdqe
ret
.size OPENSSL_atomic_add,.-OPENSSL_atomic_add
-.globl OPENSSL_wipe_cpu
-.type OPENSSL_wipe_cpu,\@function
-.align 16
-OPENSSL_wipe_cpu:
- pxor %xmm0,%xmm0
- pxor %xmm1,%xmm1
- pxor %xmm2,%xmm2
- pxor %xmm3,%xmm3
- pxor %xmm4,%xmm4
- pxor %xmm5,%xmm5
- pxor %xmm6,%xmm6
- pxor %xmm7,%xmm7
- pxor %xmm8,%xmm8
- pxor %xmm9,%xmm9
- pxor %xmm10,%xmm10
- pxor %xmm11,%xmm11
- pxor %xmm12,%xmm12
- pxor %xmm13,%xmm13
- pxor %xmm14,%xmm14
- pxor %xmm15,%xmm15
- xorq %rcx,%rcx
- xorq %rdx,%rdx
- xorq %rsi,%rsi
- xorq %rdi,%rdi
- xorq %r8,%r8
- xorq %r9,%r9
- xorq %r10,%r10
- xorq %r11,%r11
- leaq 8(%rsp),%rax
- ret
-.size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
-
-.section .init
- call OPENSSL_cpuid_setup
-
-___
-
-open STDOUT,"| $^X perlasm/x86_64-xlate.pl $output";
-print<<___;
-.text
-
.globl OPENSSL_rdtsc
.type OPENSSL_rdtsc,\@abi-omnipotent
.align 16
@@ -121,6 +50,8 @@ OPENSSL_ia32_cpuid:
xor %eax,%eax
cpuid
+ mov %eax,%r11d # max value for standard query level
+
xor %eax,%eax
cmp \$0x756e6547,%ebx # "Genu"
setne %al
@@ -130,8 +61,54 @@ OPENSSL_ia32_cpuid:
or %eax,%r9d
cmp \$0x6c65746e,%ecx # "ntel"
setne %al
- or %eax,%r9d
+ or %eax,%r9d # 0 indicates Intel CPU
+ jz .Lintel
+
+ cmp \$0x68747541,%ebx # "Auth"
+ setne %al
+ mov %eax,%r10d
+ cmp \$0x69746E65,%edx # "enti"
+ setne %al
+ or %eax,%r10d
+ cmp \$0x444D4163,%ecx # "cAMD"
+ setne %al
+ or %eax,%r10d # 0 indicates AMD CPU
+ jnz .Lintel
+
+ # AMD specific
+ mov \$0x80000000,%eax
+ cpuid
+ cmp \$0x80000008,%eax
+ jb .Lintel
+
+ mov \$0x80000008,%eax
+ cpuid
+ movzb %cl,%r10 # number of cores - 1
+ inc %r10 # number of cores
+
+ mov \$1,%eax
+ cpuid
+ bt \$28,%edx # test hyper-threading bit
+ jnc .Ldone
+ shr \$16,%ebx # number of logical processors
+ cmp %r10b,%bl
+ ja .Ldone
+ and \$0xefffffff,%edx # ~(1<<28)
+ jmp .Ldone
+
+.Lintel:
+ cmp \$4,%r11d
+ mov \$-1,%r10d
+ jb .Lnocacheinfo
+
+ mov \$4,%eax
+ mov \$0,%ecx # query L1D
+ cpuid
+ mov %eax,%r10d
+ shr \$14,%r10d
+ and \$0xfff,%r10d # number of cores -1 per L1D
+.Lnocacheinfo:
mov \$1,%eax
cpuid
cmp \$0,%r9d
@@ -144,6 +121,11 @@ OPENSSL_ia32_cpuid:
.Lnotintel:
bt \$28,%edx # test hyper-threading bit
jnc .Ldone
+ and \$0xefffffff,%edx # ~(1<<28)
+ cmp \$0,%r10d
+ je .Ldone
+
+ or \$0x10000000,%edx # 1<<28
shr \$16,%ebx
cmp \$1,%bl # see if cache is shared
ja .Ldone
@@ -155,5 +137,95 @@ OPENSSL_ia32_cpuid:
or %rcx,%rax
ret
.size OPENSSL_ia32_cpuid,.-OPENSSL_ia32_cpuid
+
+.globl OPENSSL_cleanse
+.type OPENSSL_cleanse,\@abi-omnipotent
+.align 16
+OPENSSL_cleanse:
+ xor %rax,%rax
+ cmp \$15,$arg2
+ jae .Lot
+ cmp \$0,$arg2
+ je .Lret
+.Little:
+ mov %al,($arg1)
+ sub \$1,$arg2
+ lea 1($arg1),$arg1
+ jnz .Little
+.Lret: ret
+.align 16
+.Lot:
+ test \$7,$arg1
+ jz .Laligned
+ mov %al,($arg1)
+ lea -1($arg2),$arg2
+ lea 1($arg1),$arg1
+ jmp .Lot
+.Laligned:
+ mov %rax,($arg1)
+ lea -8($arg2),$arg2
+ test \$-8,$arg2
+ lea 8($arg1),$arg1
+ jnz .Laligned
+ cmp \$0,$arg2
+ jne .Little
+ ret
+.size OPENSSL_cleanse,.-OPENSSL_cleanse
+___
+
+print<<___ if (!$win64);
+.globl OPENSSL_wipe_cpu
+.type OPENSSL_wipe_cpu,\@abi-omnipotent
+.align 16
+OPENSSL_wipe_cpu:
+ pxor %xmm0,%xmm0
+ pxor %xmm1,%xmm1
+ pxor %xmm2,%xmm2
+ pxor %xmm3,%xmm3
+ pxor %xmm4,%xmm4
+ pxor %xmm5,%xmm5
+ pxor %xmm6,%xmm6
+ pxor %xmm7,%xmm7
+ pxor %xmm8,%xmm8
+ pxor %xmm9,%xmm9
+ pxor %xmm10,%xmm10
+ pxor %xmm11,%xmm11
+ pxor %xmm12,%xmm12
+ pxor %xmm13,%xmm13
+ pxor %xmm14,%xmm14
+ pxor %xmm15,%xmm15
+ xorq %rcx,%rcx
+ xorq %rdx,%rdx
+ xorq %rsi,%rsi
+ xorq %rdi,%rdi
+ xorq %r8,%r8
+ xorq %r9,%r9
+ xorq %r10,%r10
+ xorq %r11,%r11
+ leaq 8(%rsp),%rax
+ ret
+.size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
___
+print<<___ if ($win64);
+.globl OPENSSL_wipe_cpu
+.type OPENSSL_wipe_cpu,\@abi-omnipotent
+.align 16
+OPENSSL_wipe_cpu:
+ pxor %xmm0,%xmm0
+ pxor %xmm1,%xmm1
+ pxor %xmm2,%xmm2
+ pxor %xmm3,%xmm3
+ pxor %xmm4,%xmm4
+ pxor %xmm5,%xmm5
+ xorq %rcx,%rcx
+ xorq %rdx,%rdx
+ xorq %r8,%r8
+ xorq %r9,%r9
+ xorq %r10,%r10
+ xorq %r11,%r11
+ leaq 8(%rsp),%rax
+ ret
+.size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
+___
+
close STDOUT; # flush