From 46bf83f07ae1ba7fda435c90af93960e77159f4b Mon Sep 17 00:00:00 2001 From: Andy Polyakov Date: Tue, 22 Jan 2013 22:11:31 +0100 Subject: [PATCH] x86_64 assembly pack: make Windows build more robust. PR: 2963 and a number of others --- crypto/aes/asm/aes-x86_64.pl | 3 ++- crypto/aes/asm/aesni-sha1-x86_64.pl | 3 ++- crypto/aes/asm/aesni-x86_64.pl | 3 ++- crypto/aes/asm/bsaes-x86_64.pl | 3 ++- crypto/aes/asm/vpaes-x86_64.pl | 3 ++- crypto/bn/asm/modexp512-x86_64.pl | 3 ++- crypto/camellia/asm/cmll-x86_64.pl | 3 ++- crypto/md5/asm/md5-x86_64.pl | 3 ++- crypto/modes/asm/ghash-x86_64.pl | 3 ++- crypto/rc4/asm/rc4-md5-x86_64.pl | 3 ++- crypto/rc4/asm/rc4-x86_64.pl | 3 ++- crypto/sha/asm/sha1-x86_64.pl | 3 ++- crypto/sha/asm/sha512-x86_64.pl | 3 ++- crypto/whrlpool/asm/wp-x86_64.pl | 3 ++- crypto/x86_64cpuid.pl | 3 ++- engines/asm/e_padlock-x86_64.pl | 3 ++- ms/uplink-x86_64.pl | 3 ++- 17 files changed, 34 insertions(+), 17 deletions(-) diff --git a/crypto/aes/asm/aes-x86_64.pl b/crypto/aes/asm/aes-x86_64.pl index 01f9cb7f8e..e95593e35e 100755 --- a/crypto/aes/asm/aes-x86_64.pl +++ b/crypto/aes/asm/aes-x86_64.pl @@ -37,7 +37,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; $verticalspin=1; # unlike 32-bit version $verticalspin performs # ~15% better on both AMD and Intel cores diff --git a/crypto/aes/asm/aesni-sha1-x86_64.pl b/crypto/aes/asm/aesni-sha1-x86_64.pl index f8f133b231..2a5071d880 100644 --- a/crypto/aes/asm/aesni-sha1-x86_64.pl +++ b/crypto/aes/asm/aesni-sha1-x86_64.pl @@ -77,7 +77,8 @@ $avx=1 if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) && `ml64 2>&1` =~ /Version ([0-9]+)\./ && $1>=10); -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; # void aesni_cbc_sha1_enc(const void *inp, # void *out, diff --git a/crypto/aes/asm/aesni-x86_64.pl b/crypto/aes/asm/aesni-x86_64.pl index 318e52fe9d..8a30904916 100644 --- a/crypto/aes/asm/aesni-x86_64.pl +++ b/crypto/aes/asm/aesni-x86_64.pl @@ -179,7 +179,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; $movkey = $PREFIX eq "aesni" ? "movups" : "movups"; @_4args=$win64? ("%rcx","%rdx","%r8", "%r9") : # Win64 order diff --git a/crypto/aes/asm/bsaes-x86_64.pl b/crypto/aes/asm/bsaes-x86_64.pl index f5f05c0e73..ceb02b50dd 100644 --- a/crypto/aes/asm/bsaes-x86_64.pl +++ b/crypto/aes/asm/bsaes-x86_64.pl @@ -105,7 +105,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; my ($inp,$out,$len,$key,$ivp)=("%rdi","%rsi","%rdx","%rcx"); my @XMM=map("%xmm$_",(15,0..14)); # best on Atom, +10% over (0..15) diff --git a/crypto/aes/asm/vpaes-x86_64.pl b/crypto/aes/asm/vpaes-x86_64.pl index bca97fdcdf..6e41066344 100644 --- a/crypto/aes/asm/vpaes-x86_64.pl +++ b/crypto/aes/asm/vpaes-x86_64.pl @@ -56,7 +56,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; $PREFIX="vpaes"; diff --git a/crypto/bn/asm/modexp512-x86_64.pl b/crypto/bn/asm/modexp512-x86_64.pl index 28b6dab2f0..bfd6e97541 100644 --- a/crypto/bn/asm/modexp512-x86_64.pl +++ b/crypto/bn/asm/modexp512-x86_64.pl @@ -68,7 +68,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; use strict; my $code=".text\n\n"; diff --git a/crypto/camellia/asm/cmll-x86_64.pl b/crypto/camellia/asm/cmll-x86_64.pl index 53e41f2e5e..d94f46b887 100644 --- a/crypto/camellia/asm/cmll-x86_64.pl +++ b/crypto/camellia/asm/cmll-x86_64.pl @@ -40,7 +40,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; sub hi() { my $r=shift; $r =~ s/%[er]([a-d])x/%\1h/; $r; } sub lo() { my $r=shift; $r =~ s/%[er]([a-d])x/%\1l/; diff --git a/crypto/md5/asm/md5-x86_64.pl b/crypto/md5/asm/md5-x86_64.pl index eea6802a8a..a79d0e46ea 100755 --- a/crypto/md5/asm/md5-x86_64.pl +++ b/crypto/md5/asm/md5-x86_64.pl @@ -117,7 +117,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; my $dir=$1; my $xlate; die "can't locate x86_64-xlate.pl"; no warnings qw(uninitialized); -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; $code .= <&1` =~ /Version ([0-9]+)\./ && $1>=10); -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; $ctx="%rdi"; # 1st arg $inp="%rsi"; # 2nd arg diff --git a/crypto/sha/asm/sha512-x86_64.pl b/crypto/sha/asm/sha512-x86_64.pl index ab9a0b472d..8f10f973e9 100755 --- a/crypto/sha/asm/sha512-x86_64.pl +++ b/crypto/sha/asm/sha512-x86_64.pl @@ -103,7 +103,8 @@ $avx=1 if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) && `ml64 2>&1` =~ /Version ([0-9]+)\./ && $1>=10); -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; if ($output =~ /512/) { $func="sha512_block_data_order"; diff --git a/crypto/whrlpool/asm/wp-x86_64.pl b/crypto/whrlpool/asm/wp-x86_64.pl index 855aabc892..5a3bdbcf20 100644 --- a/crypto/whrlpool/asm/wp-x86_64.pl +++ b/crypto/whrlpool/asm/wp-x86_64.pl @@ -41,7 +41,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; my $dir=$1; my $xlate; ( $xlate="${dir}../../perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; sub L() { $code.=".byte ".join(',',@_)."\n"; } sub LL(){ $code.=".byte ".join(',',@_).",".join(',',@_)."\n"; } diff --git a/crypto/x86_64cpuid.pl b/crypto/x86_64cpuid.pl index 3a1adeeccc..0cf68f1360 100644 --- a/crypto/x86_64cpuid.pl +++ b/crypto/x86_64cpuid.pl @@ -11,7 +11,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; ($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") : # Win64 order ("%rdi","%rsi","%rdx","%rcx"); # Unix order diff --git a/engines/asm/e_padlock-x86_64.pl b/engines/asm/e_padlock-x86_64.pl index 5d19757961..f8ba1e909f 100644 --- a/engines/asm/e_padlock-x86_64.pl +++ b/engines/asm/e_padlock-x86_64.pl @@ -23,7 +23,8 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}../../crypto/perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open STDOUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" $xlate $flavour $output"; +*STDOUT=*OUT; $code=".text\n"; diff --git a/ms/uplink-x86_64.pl b/ms/uplink-x86_64.pl index 2758c86d1e..48bf559ee1 100755 --- a/ms/uplink-x86_64.pl +++ b/ms/uplink-x86_64.pl @@ -2,7 +2,8 @@ $output=shift; $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; -open STDOUT,"| \"$^X\" ${dir}../crypto/perlasm/x86_64-xlate.pl $output"; +open OUT,"| \"$^X\" ${dir}../crypto/perlasm/x86_64-xlate.pl $output"; +*STDOUT=*OUT; push(@INC,"${dir}."); require "uplink-common.pl";