summaryrefslogtreecommitdiffstats
path: root/src/crypto/sha/asm
diff options
context:
space:
mode:
Diffstat (limited to 'src/crypto/sha/asm')
-rw-r--r--src/crypto/sha/asm/sha1-586.pl25
-rw-r--r--src/crypto/sha/asm/sha1-x86_64.pl26
-rw-r--r--src/crypto/sha/asm/sha256-586.pl28
-rw-r--r--src/crypto/sha/asm/sha512-x86_64.pl32
4 files changed, 69 insertions, 42 deletions
diff --git a/src/crypto/sha/asm/sha1-586.pl b/src/crypto/sha/asm/sha1-586.pl
index 09fd3fc..e0b5d83 100644
--- a/src/crypto/sha/asm/sha1-586.pl
+++ b/src/crypto/sha/asm/sha1-586.pl
@@ -118,20 +118,23 @@ require "x86asm.pl";
$xmm=$ymm=0;
for (@ARGV) { $xmm=1 if (/-DOPENSSL_IA32_SSE2/); }
-# In upstream, this is controlled by shelling out to the compiler to check
-# versions, but BoringSSL is intended to be used with pre-generated perlasm
-# output, so this isn't useful anyway.
-#
-# TODO(davidben): Enable this after testing. $ymm goes up to 1.
-$ymm = 0;
+$ymm=1 if ($xmm &&
+ `$ENV{CC} -Wa,-v -c -o /dev/null -x assembler /dev/null 2>&1`
+ =~ /GNU assembler version ([2-9]\.[0-9]+)/ &&
+ $1>=2.19); # first version supporting AVX
-$ymm = 0 unless ($xmm);
+$ymm=1 if ($xmm && !$ymm && $ARGV[0] eq "win32n" &&
+ `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/ &&
+ $1>=2.03); # first version supporting AVX
-$shaext=$xmm; ### set to zero if compiling for 1.0.1
+$ymm=1 if ($xmm && !$ymm && $ARGV[0] eq "win32" &&
+ `ml 2>&1` =~ /Version ([0-9]+)\./ &&
+ $1>=10); # first version supporting AVX
-# TODO(davidben): Consider enabling the Intel SHA Extensions code once it's
-# been tested.
-$shaext = 0;
+$ymm=1 if ($xmm && !$ymm && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9]\.[0-9]+)/ &&
+ $2>=3.0); # first version supporting AVX
+
+$shaext=$xmm; ### set to zero if compiling for 1.0.1
&external_label("OPENSSL_ia32cap_P") if ($xmm);
diff --git a/src/crypto/sha/asm/sha1-x86_64.pl b/src/crypto/sha/asm/sha1-x86_64.pl
index 59b1607..124034d 100644
--- a/src/crypto/sha/asm/sha1-x86_64.pl
+++ b/src/crypto/sha/asm/sha1-x86_64.pl
@@ -92,15 +92,25 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
die "can't locate x86_64-xlate.pl";
-# In upstream, this is controlled by shelling out to the compiler to check
-# versions, but BoringSSL is intended to be used with pre-generated perlasm
-# output, so this isn't useful anyway.
-#
-# TODO(davidben): Enable this after testing. $avx goes up to 2.
-$avx = 0;
+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) + ($1>=2.22);
+}
+
+if (!$avx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+ `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+ $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+ `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+ $avx = ($1>=10) + ($1>=11);
+}
+
+if (!$avx && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([2-9]\.[0-9]+)/) {
+ $avx = ($2>=3.0) + ($2>3.0);
+}
-# TODO(davidben): Consider enabling the Intel SHA Extensions code once it's
-# been tested.
$shaext=0; ### set to zero if compiling for 1.0.1
$avx=1 if (!$shaext && $avx);
diff --git a/src/crypto/sha/asm/sha256-586.pl b/src/crypto/sha/asm/sha256-586.pl
index 1866d5a..e907714 100644
--- a/src/crypto/sha/asm/sha256-586.pl
+++ b/src/crypto/sha/asm/sha256-586.pl
@@ -68,20 +68,26 @@ require "x86asm.pl";
$xmm=$avx=0;
for (@ARGV) { $xmm=1 if (/-DOPENSSL_IA32_SSE2/); }
-# In upstream, this is controlled by shelling out to the compiler to check
-# versions, but BoringSSL is intended to be used with pre-generated perlasm
-# output, so this isn't useful anyway.
-#
-# TODO(davidben): Enable this after testing. $avx goes up to 2.
-$avx = 0;
+if ($xmm && `$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) + ($1>=2.22);
+}
-$avx = 0 unless ($xmm);
+if ($xmm && !$avx && $ARGV[0] eq "win32n" &&
+ `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+ $avx = ($1>=2.03) + ($1>=2.10);
+}
-$shaext=$xmm; ### set to zero if compiling for 1.0.1
+if ($xmm && !$avx && $ARGV[0] eq "win32" &&
+ `ml 2>&1` =~ /Version ([0-9]+)\./) {
+ $avx = ($1>=10) + ($1>=11);
+}
+
+if ($xmm && !$avx && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9]\.[0-9]+)/) {
+ $avx = ($2>=3.0) + ($2>3.0);
+}
-# TODO(davidben): Consider enabling the Intel SHA Extensions code once it's
-# been tested.
-$shaext = 0;
+$shaext=$xmm; ### set to zero if compiling for 1.0.1
$unroll_after = 64*4; # If pre-evicted from L1P cache first spin of
# fully unrolled loop was measured to run about
diff --git a/src/crypto/sha/asm/sha512-x86_64.pl b/src/crypto/sha/asm/sha512-x86_64.pl
index 9a0d0c4..6660a88 100644
--- a/src/crypto/sha/asm/sha512-x86_64.pl
+++ b/src/crypto/sha/asm/sha512-x86_64.pl
@@ -109,15 +109,25 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or
die "can't locate x86_64-xlate.pl";
-# In upstream, this is controlled by shelling out to the compiler to check
-# versions, but BoringSSL is intended to be used with pre-generated perlasm
-# output, so this isn't useful anyway.
-#
-# TODO(davidben): Enable this after testing. $avx goes up to 2.
-$avx = 0;
+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) + ($1>=2.22);
+}
+
+if (!$avx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
+ `nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
+ $avx = ($1>=2.09) + ($1>=2.10);
+}
+
+if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
+ `ml64 2>&1` =~ /Version ([0-9]+)\./) {
+ $avx = ($1>=10) + ($1>=11);
+}
+
+if (!$avx && `$ENV{CC} -v 2>&1` =~ /(^clang version|based on LLVM) ([3-9]\.[0-9]+)/) {
+ $avx = ($2>=3.0) + ($2>3.0);
+}
-# TODO(davidben): Consider enabling the Intel SHA Extensions code once it's
-# been tested.
$shaext=0; ### set to zero if compiling for 1.0.1
$avx=1 if (!$shaext && $avx);
@@ -2254,8 +2264,7 @@ $code.=<<___;
ret
.size se_handler,.-se_handler
___
-
-$code.=<<___ if ($SZ==4 && $shaext);
+$code.=<<___ if ($SZ == 4 && $shaext);
.type shaext_handler,\@abi-omnipotent
.align 16
shaext_handler:
@@ -2289,7 +2298,6 @@ shaext_handler:
jmp .Lin_prologue
.size shaext_handler,.-shaext_handler
___
-
$code.=<<___;
.section .pdata
.align 4
@@ -2297,7 +2305,7 @@ $code.=<<___;
.rva .LSEH_end_$func
.rva .LSEH_info_$func
___
-$code.=<<___ if ($SZ==4 && $shaext);
+$code.=<<___ if ($SZ==4 && $shext);
.rva .LSEH_begin_${func}_shaext
.rva .LSEH_end_${func}_shaext
.rva .LSEH_info_${func}_shaext
generated by cgit v1.1 at 2024-06-09 08:30:14 +0000