aboutsummaryrefslogtreecommitdiff
path: root/openssl/crypto/x86_64cpuid.pl
diff options
context:
space:
mode:
authormarha <marha@users.sourceforge.net>2012-04-10 11:41:26 +0200
committermarha <marha@users.sourceforge.net>2012-04-10 11:41:26 +0200
commit67326634496ef21b4acbf4cef2f05040d34aef9b (patch)
treef19fba7c7b691e44cd97482644e383e09ab98c49 /openssl/crypto/x86_64cpuid.pl
parentc6f80401dc533b04341afe8d596960d1bc25efce (diff)
downloadvcxsrv-67326634496ef21b4acbf4cef2f05040d34aef9b.tar.gz
vcxsrv-67326634496ef21b4acbf4cef2f05040d34aef9b.tar.bz2
vcxsrv-67326634496ef21b4acbf4cef2f05040d34aef9b.zip
Update to openssl-1.0.1
Diffstat (limited to 'openssl/crypto/x86_64cpuid.pl')
-rw-r--r--openssl/crypto/x86_64cpuid.pl87
1 files changed, 69 insertions, 18 deletions
diff --git a/openssl/crypto/x86_64cpuid.pl b/openssl/crypto/x86_64cpuid.pl
index c96821a3c..7b7b93b22 100644
--- a/openssl/crypto/x86_64cpuid.pl
+++ b/openssl/crypto/x86_64cpuid.pl
@@ -7,15 +7,24 @@ if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
$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";
+( $xlate="${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 STDOUT,"| $^X $xlate $flavour $output";
+
+($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") : # Win64 order
+ ("%rdi","%rsi","%rdx","%rcx"); # Unix order
-if ($win64) { $arg1="%rcx"; $arg2="%rdx"; }
-else { $arg1="%rdi"; $arg2="%rsi"; }
print<<___;
.extern OPENSSL_cpuid_setup
+.hidden OPENSSL_cpuid_setup
.section .init
call OPENSSL_cpuid_setup
+.hidden OPENSSL_ia32cap_P
+.comm OPENSSL_ia32cap_P,8,4
+
.text
.globl OPENSSL_atomic_add
@@ -46,7 +55,7 @@ OPENSSL_rdtsc:
.type OPENSSL_ia32_cpuid,\@abi-omnipotent
.align 16
OPENSSL_ia32_cpuid:
- mov %rbx,%r8
+ mov %rbx,%r8 # save %rbx
xor %eax,%eax
cpuid
@@ -78,7 +87,15 @@ OPENSSL_ia32_cpuid:
# AMD specific
mov \$0x80000000,%eax
cpuid
- cmp \$0x80000008,%eax
+ cmp \$0x80000001,%eax
+ jb .Lintel
+ mov %eax,%r10d
+ mov \$0x80000001,%eax
+ cpuid
+ or %ecx,%r9d
+ and \$0x00000801,%r9d # isolate AMD XOP bit, 1<<11
+
+ cmp \$0x80000008,%r10d
jb .Lintel
mov \$0x80000008,%eax
@@ -89,12 +106,12 @@ OPENSSL_ia32_cpuid:
mov \$1,%eax
cpuid
bt \$28,%edx # test hyper-threading bit
- jnc .Ldone
+ jnc .Lgeneric
shr \$16,%ebx # number of logical processors
cmp %r10b,%bl
- ja .Ldone
+ ja .Lgeneric
and \$0xefffffff,%edx # ~(1<<28)
- jmp .Ldone
+ jmp .Lgeneric
.Lintel:
cmp \$4,%r11d
@@ -111,30 +128,47 @@ OPENSSL_ia32_cpuid:
.Lnocacheinfo:
mov \$1,%eax
cpuid
+ and \$0xbfefffff,%edx # force reserved bits to 0
cmp \$0,%r9d
jne .Lnotintel
- or \$0x00100000,%edx # use reserved 20th bit to engage RC4_CHAR
+ or \$0x40000000,%edx # set reserved bit#30 on Intel CPUs
and \$15,%ah
cmp \$15,%ah # examine Family ID
- je .Lnotintel
- or \$0x40000000,%edx # use reserved bit to skip unrolled loop
+ jne .Lnotintel
+ or \$0x00100000,%edx # set reserved bit#20 to engage RC4_CHAR
.Lnotintel:
bt \$28,%edx # test hyper-threading bit
- jnc .Ldone
+ jnc .Lgeneric
and \$0xefffffff,%edx # ~(1<<28)
cmp \$0,%r10d
- je .Ldone
+ je .Lgeneric
or \$0x10000000,%edx # 1<<28
shr \$16,%ebx
cmp \$1,%bl # see if cache is shared
- ja .Ldone
+ ja .Lgeneric
and \$0xefffffff,%edx # ~(1<<28)
+.Lgeneric:
+ and \$0x00000800,%r9d # isolate AMD XOP flag
+ and \$0xfffff7ff,%ecx
+ or %ecx,%r9d # merge AMD XOP flag
+
+ mov %edx,%r10d # %r9d:%r10d is copy of %ecx:%edx
+ bt \$27,%r9d # check OSXSAVE bit
+ jnc .Lclear_avx
+ xor %ecx,%ecx # XCR0
+ .byte 0x0f,0x01,0xd0 # xgetbv
+ and \$6,%eax # isolate XMM and YMM state support
+ cmp \$6,%eax
+ je .Ldone
+.Lclear_avx:
+ mov \$0xefffe7ff,%eax # ~(1<<28|1<<12|1<<11)
+ and %eax,%r9d # clear AVX, FMA and AMD XOP bits
.Ldone:
- shl \$32,%rcx
- mov %edx,%eax
- mov %r8,%rbx
- or %rcx,%rax
+ shl \$32,%r9
+ mov %r10d,%eax
+ mov %r8,%rbx # restore %rbx
+ or %r9,%rax
ret
.size OPENSSL_ia32_cpuid,.-OPENSSL_ia32_cpuid
@@ -229,4 +263,21 @@ OPENSSL_wipe_cpu:
.size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
___
+print<<___;
+.globl OPENSSL_ia32_rdrand
+.type OPENSSL_ia32_rdrand,\@abi-omnipotent
+.align 16
+OPENSSL_ia32_rdrand:
+ mov \$8,%ecx
+.Loop_rdrand:
+ rdrand %rax
+ jc .Lbreak_rdrand
+ loop .Loop_rdrand
+.Lbreak_rdrand:
+ cmp \$0,%rax
+ cmove %rcx,%rax
+ ret
+.size OPENSSL_ia32_rdrand,.-OPENSSL_ia32_rdrand
+___
+
close STDOUT; # flush