X-Git-Url: https://code.wpia.club/?a=blobdiff_plain;f=lib%2Fopenssl%2Fcrypto%2Fx86_64cpuid.pl;h=6cb152148b5b6eae2d61da23e18dd442e3edfff5;hb=02ed66432c92de70694700164f986190aad3cbc5;hp=6ebfd017ea563e04642d693a50806d3912d74ad3;hpb=89016837dcbf2775cd15dc8cbaba00dc6379f86e;p=cassiopeia.git diff --git a/lib/openssl/crypto/x86_64cpuid.pl b/lib/openssl/crypto/x86_64cpuid.pl index 6ebfd01..6cb1521 100644 --- a/lib/openssl/crypto/x86_64cpuid.pl +++ b/lib/openssl/crypto/x86_64cpuid.pl @@ -1,4 +1,11 @@ -#!/usr/bin/env perl +#! /usr/bin/env perl +# Copyright 2005-2016 The OpenSSL Project Authors. All Rights Reserved. +# +# Licensed under the OpenSSL license (the "License"). You may not use +# this file except in compliance with the License. You can obtain a copy +# in the file LICENSE in the source distribution or at +# https://www.openssl.org/source/license.html + $flavour = shift; $output = shift; @@ -11,7 +18,7 @@ $0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1; ( $xlate="${dir}perlasm/x86_64-xlate.pl" and -f $xlate) or die "can't locate x86_64-xlate.pl"; -open OUT,"| \"$^X\" $xlate $flavour $output"; +open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\""; *STDOUT=*OUT; ($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") : # Win64 order @@ -24,7 +31,7 @@ print<<___; call OPENSSL_cpuid_setup .hidden OPENSSL_ia32cap_P -.comm OPENSSL_ia32cap_P,8,4 +.comm OPENSSL_ia32cap_P,16,4 .text @@ -53,12 +60,13 @@ OPENSSL_rdtsc: .size OPENSSL_rdtsc,.-OPENSSL_rdtsc .globl OPENSSL_ia32_cpuid -.type OPENSSL_ia32_cpuid,\@abi-omnipotent +.type OPENSSL_ia32_cpuid,\@function,1 .align 16 OPENSSL_ia32_cpuid: mov %rbx,%r8 # save %rbx xor %eax,%eax + mov %eax,8(%rdi) # clear 3rd word cpuid mov %eax,%r11d # max value for standard query level @@ -126,6 +134,14 @@ OPENSSL_ia32_cpuid: shr \$14,%r10d and \$0xfff,%r10d # number of cores -1 per L1D + cmp \$7,%r11d + jb .Lnocacheinfo + + mov \$7,%eax + xor %ecx,%ecx + cpuid + mov %ebx,8(%rdi) + .Lnocacheinfo: mov \$1,%eax cpuid @@ -165,6 +181,7 @@ OPENSSL_ia32_cpuid: .Lclear_avx: mov \$0xefffe7ff,%eax # ~(1<<28|1<<12|1<<11) and %eax,%r9d # clear AVX, FMA and AMD XOP bits + andl \$0xffffffdf,8(%rdi) # cleax AVX2, ~(1<<5) .Ldone: shl \$32,%r9 mov %r10d,%eax @@ -207,6 +224,28 @@ OPENSSL_cleanse: jne .Little ret .size OPENSSL_cleanse,.-OPENSSL_cleanse + +.globl CRYPTO_memcmp +.type CRYPTO_memcmp,\@abi-omnipotent +.align 16 +CRYPTO_memcmp: + xor %rax,%rax + xor %r10,%r10 + cmp \$0,$arg3 + je .Lno_data +.Loop_cmp: + mov ($arg1),%r10b + lea 1($arg1),$arg1 + xor ($arg2),%r10b + lea 1($arg2),$arg2 + or %r10b,%al + dec $arg3 + jnz .Loop_cmp + neg %rax + shr \$63,%rax +.Lno_data: + ret +.size CRYPTO_memcmp,.-CRYPTO_memcmp ___ print<<___ if (!$win64); @@ -263,22 +302,158 @@ OPENSSL_wipe_cpu: ret .size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu ___ +{ +my $out="%r10"; +my $cnt="%rcx"; +my $max="%r11"; +my $lasttick="%r8d"; +my $lastdiff="%r9d"; +my $redzone=win64?8:-8; + +print<<___; +.globl OPENSSL_instrument_bus +.type OPENSSL_instrument_bus,\@abi-omnipotent +.align 16 +OPENSSL_instrument_bus: + mov $arg1,$out # tribute to Win64 + mov $arg2,$cnt + mov $arg2,$max + + rdtsc # collect 1st tick + mov %eax,$lasttick # lasttick = tick + mov \$0,$lastdiff # lastdiff = 0 + clflush ($out) + .byte 0xf0 # lock + add $lastdiff,($out) + jmp .Loop +.align 16 +.Loop: rdtsc + mov %eax,%edx + sub $lasttick,%eax + mov %edx,$lasttick + mov %eax,$lastdiff + clflush ($out) + .byte 0xf0 # lock + add %eax,($out) + lea 4($out),$out + sub \$1,$cnt + jnz .Loop + + mov $max,%rax + ret +.size OPENSSL_instrument_bus,.-OPENSSL_instrument_bus + +.globl OPENSSL_instrument_bus2 +.type OPENSSL_instrument_bus2,\@abi-omnipotent +.align 16 +OPENSSL_instrument_bus2: + mov $arg1,$out # tribute to Win64 + mov $arg2,$cnt + mov $arg3,$max + mov $cnt,$redzone(%rsp) + + rdtsc # collect 1st tick + mov %eax,$lasttick # lasttick = tick + mov \$0,$lastdiff # lastdiff = 0 + + clflush ($out) + .byte 0xf0 # lock + add $lastdiff,($out) + + rdtsc # collect 1st diff + mov %eax,%edx + sub $lasttick,%eax # diff + mov %edx,$lasttick # lasttick = tick + mov %eax,$lastdiff # lastdiff = diff +.Loop2: + clflush ($out) + .byte 0xf0 # lock + add %eax,($out) # accumulate diff + + sub \$1,$max + jz .Ldone2 + + rdtsc + mov %eax,%edx + sub $lasttick,%eax # diff + mov %edx,$lasttick # lasttick = tick + cmp $lastdiff,%eax + mov %eax,$lastdiff # lastdiff = diff + mov \$0,%edx + setne %dl + sub %rdx,$cnt # conditional --$cnt + lea ($out,%rdx,4),$out # conditional ++$out + jnz .Loop2 + +.Ldone2: + mov $redzone(%rsp),%rax + sub $cnt,%rax + ret +.size OPENSSL_instrument_bus2,.-OPENSSL_instrument_bus2 +___ +} +sub gen_random { +my $rdop = shift; print<<___; -.globl OPENSSL_ia32_rdrand -.type OPENSSL_ia32_rdrand,\@abi-omnipotent +.globl OPENSSL_ia32_${rdop} +.type OPENSSL_ia32_${rdop},\@abi-omnipotent .align 16 -OPENSSL_ia32_rdrand: +OPENSSL_ia32_${rdop}: mov \$8,%ecx -.Loop_rdrand: - rdrand %rax - jc .Lbreak_rdrand - loop .Loop_rdrand -.Lbreak_rdrand: +.Loop_${rdop}: + ${rdop} %rax + jc .Lbreak_${rdop} + loop .Loop_${rdop} +.Lbreak_${rdop}: cmp \$0,%rax cmove %rcx,%rax ret -.size OPENSSL_ia32_rdrand,.-OPENSSL_ia32_rdrand +.size OPENSSL_ia32_${rdop},.-OPENSSL_ia32_${rdop} + +.globl OPENSSL_ia32_${rdop}_bytes +.type OPENSSL_ia32_${rdop}_bytes,\@abi-omnipotent +.align 16 +OPENSSL_ia32_${rdop}_bytes: + xor %rax, %rax # return value + cmp \$0,$arg2 + je .Ldone_${rdop}_bytes + + mov \$8,%r11 +.Loop_${rdop}_bytes: + ${rdop} %r10 + jc .Lbreak_${rdop}_bytes + dec %r11 + jnz .Loop_${rdop}_bytes + jmp .Ldone_${rdop}_bytes + +.align 16 +.Lbreak_${rdop}_bytes: + cmp \$8,$arg2 + jb .Ltail_${rdop}_bytes + mov %r10,($arg1) + lea 8($arg1),$arg1 + add \$8,%rax + sub \$8,$arg2 + jz .Ldone_${rdop}_bytes + mov \$8,%r11 + jmp .Loop_${rdop}_bytes + +.align 16 +.Ltail_${rdop}_bytes: + mov %r10b,($arg1) + lea 1($arg1),$arg1 + inc %rax + shr \$8,%r8 + dec $arg2 + jnz .Ltail_${rdop}_bytes + +.Ldone_${rdop}_bytes: + ret +.size OPENSSL_ia32_${rdop}_bytes,.-OPENSSL_ia32_${rdop}_bytes ___ +} +gen_random("rdrand"); +gen_random("rdseed"); close STDOUT; # flush