-#!/usr/bin/env perl
-
-#******************************************************************************
-#* Copyright(c) 2012, Intel Corp.
-#* Developers and authors:
-#* Shay Gueron (1, 2), and Vlad Krasnov (1)
-#* (1) Intel Corporation, Israel Development Center, Haifa, Israel
-#* (2) University of Haifa, Israel
-#******************************************************************************
-#* LICENSE:
-#* This submission to OpenSSL is to be made available under the OpenSSL
-#* license, and only to the OpenSSL project, in order to allow integration
-#* into the publicly distributed code.
-#* The use of this code, or portions of this code, or concepts embedded in
-#* this code, or modification of this code and/or algorithm(s) in it, or the
-#* use of this code for any other purpose than stated above, requires special
-#* licensing.
-#******************************************************************************
-#* DISCLAIMER:
-#* THIS SOFTWARE IS PROVIDED BY THE CONTRIBUTORS AND THE COPYRIGHT OWNERS
-#* ``AS IS''. ANY EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
-#* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
-#* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE CONTRIBUTORS OR THE COPYRIGHT
-#* OWNERS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
-#* OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
-#* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
-#* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
-#* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
-#* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
-#* POSSIBILITY OF SUCH DAMAGE.
-#******************************************************************************
-#* Reference:
-#* [1] S. Gueron, V. Krasnov: "Software Implementation of Modular
-#* Exponentiation, Using Advanced Vector Instructions Architectures",
-#* F. Ozbudak and F. Rodriguez-Henriquez (Eds.): WAIFI 2012, LNCS 7369,
-#* pp. 119?135, 2012. Springer-Verlag Berlin Heidelberg 2012
-#* [2] S. Gueron: "Efficient Software Implementations of Modular
-#* Exponentiation", Journal of Cryptographic Engineering 2:31-43 (2012).
-#* [3] S. Gueron, V. Krasnov: "Speeding up Big-numbers Squaring",IEEE
-#* Proceedings of 9th International Conference on Information Technology:
-#* New Generations (ITNG 2012), pp.821-823 (2012)
-#* [4] S. Gueron, V. Krasnov: "[PATCH] Efficient and side channel analysis
-#* resistant 1024-bit modular exponentiation, for optimizing RSA2048
-#* on AVX2 capable x86_64 platforms",
-#* http://rt.openssl.org/Ticket/Display.html?id=2850&user=guest&pass=guest
-#******************************************************************************
-
-# +10% improvement by <appro@openssl.org>
+#! /usr/bin/env perl
+# Copyright 2013-2019 The OpenSSL Project Authors. All Rights Reserved.
+# Copyright (c) 2012, Intel Corporation. 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
+#
+# Originally written by Shay Gueron (1, 2), and Vlad Krasnov (1)
+# (1) Intel Corporation, Israel Development Center, Haifa, Israel
+# (2) University of Haifa, Israel
+#
+# References:
+# [1] S. Gueron, V. Krasnov: "Software Implementation of Modular
+# Exponentiation, Using Advanced Vector Instructions Architectures",
+# F. Ozbudak and F. Rodriguez-Henriquez (Eds.): WAIFI 2012, LNCS 7369,
+# pp. 119?135, 2012. Springer-Verlag Berlin Heidelberg 2012
+# [2] S. Gueron: "Efficient Software Implementations of Modular
+# Exponentiation", Journal of Cryptographic Engineering 2:31-43 (2012).
+# [3] S. Gueron, V. Krasnov: "Speeding up Big-numbers Squaring",IEEE
+# Proceedings of 9th International Conference on Information Technology:
+# New Generations (ITNG 2012), pp.821-823 (2012)
+# [4] S. Gueron, V. Krasnov: "[PATCH] Efficient and side channel analysis
+# resistant 1024-bit modular exponentiation, for optimizing RSA2048
+# on AVX2 capable x86_64 platforms",
+# http://rt.openssl.org/Ticket/Display.html?id=2850&user=guest&pass=guest
+#
+# +13% improvement over original submission by <appro@openssl.org>
#
# rsa2048 sign/sec OpenSSL 1.0.1 scalar(*) this
-# 2GHz Haswell 544 632/+16% 947/+74%
+# 2.3GHz Haswell 621 765/+23% 1113/+79%
+# 2.3GHz Broadwell(**) 688 1200(***)/+74% 1120/+63%
#
# (*) if system doesn't support AVX2, for reference purposes;
+# (**) scaled to 2.3GHz to simplify comparison;
+# (***) scalar AD*X code is faster than AVX2 and is preferred code
+# path for Broadwell;
$flavour = shift;
$output = shift;
if (`$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);
+ $addx = ($1>=2.23);
}
if (!$avx && $win64 && ($flavour =~ /nasm/ || $ENV{ASM} =~ /nasm/) &&
`nasm -v 2>&1` =~ /NASM version ([2-9]\.[0-9]+)/) {
- $avx = ($1>=2.09) + ($1>=2.11);
+ $avx = ($1>=2.09) + ($1>=2.10);
+ $addx = ($1>=2.10);
}
if (!$avx && $win64 && ($flavour =~ /masm/ || $ENV{ASM} =~ /ml64/) &&
`ml64 2>&1` =~ /Version ([0-9]+)\./) {
$avx = ($1>=10) + ($1>=11);
+ $addx = ($1>=11);
}
-open OUT,"| $^X $xlate $flavour $output";
+if (!$avx && `$ENV{CC} -v 2>&1` =~ /((?:^clang|LLVM) version|based on LLVM) ([3-9])\.([0-9]+)/) {
+ my $ver = $2 + $3/100.0; # 3.1->3.01, 3.10->3.10
+ $avx = ($ver>=3.0) + ($ver>=3.01);
+ $addx = ($ver>=3.03);
+}
+
+open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\"";
*STDOUT = *OUT;
if ($avx>1) {{{
my $aap=$r0;
my $tp0="%rbx";
my $tp1=$r3;
+my $tpa=$tmp;
$np="%r13"; # reassigned argument
$code.=<<___;
+.text
+
.globl rsaz_1024_sqr_avx2
.type rsaz_1024_sqr_avx2,\@function,5
.align 64
rsaz_1024_sqr_avx2: # 702 cycles, 14% faster than rsaz_1024_mul_avx2
+.cfi_startproc
lea (%rsp), %rax
+.cfi_def_cfa_register %rax
push %rbx
+.cfi_push %rbx
push %rbp
+.cfi_push %rbp
push %r12
+.cfi_push %r12
push %r13
+.cfi_push %r13
push %r14
+.cfi_push %r14
push %r15
+.cfi_push %r15
+ vzeroupper
___
$code.=<<___ if ($win64);
lea -0xa8(%rsp),%rsp
- movaps %xmm6,-0xd8(%rax)
- movaps %xmm7,-0xc8(%rax)
- movaps %xmm8,-0xb8(%rax)
- movaps %xmm9,-0xa8(%rax)
- movaps %xmm10,-0x98(%rax)
- movaps %xmm11,-0x88(%rax)
- movaps %xmm12,-0x78(%rax)
- movaps %xmm13,-0x68(%rax)
- movaps %xmm14,-0x58(%rax)
- movaps %xmm15,-0x48(%rax)
+ vmovaps %xmm6,-0xd8(%rax)
+ vmovaps %xmm7,-0xc8(%rax)
+ vmovaps %xmm8,-0xb8(%rax)
+ vmovaps %xmm9,-0xa8(%rax)
+ vmovaps %xmm10,-0x98(%rax)
+ vmovaps %xmm11,-0x88(%rax)
+ vmovaps %xmm12,-0x78(%rax)
+ vmovaps %xmm13,-0x68(%rax)
+ vmovaps %xmm14,-0x58(%rax)
+ vmovaps %xmm15,-0x48(%rax)
.Lsqr_1024_body:
___
$code.=<<___;
mov %rax,%rbp
- vzeroall
+.cfi_def_cfa_register %rbp
mov %rdx, $np # reassigned argument
sub \$$FrameSize, %rsp
mov $np, $tmp
and \$4095, $tmp # see if $np crosses page
add \$32*10, $tmp
shr \$12, $tmp
+ vpxor $ACC9,$ACC9,$ACC9
jz .Lsqr_1024_no_n_copy
# unaligned 256-bit load that crosses page boundary can
vmovdqu $ACC6, 32*6-128($np)
vmovdqu $ACC7, 32*7-128($np)
vmovdqu $ACC8, 32*8-128($np)
- vmovdqu $ACC9, 32*9-128($np) # $ACC9 is zero after vzeroall
+ vmovdqu $ACC9, 32*9-128($np) # $ACC9 is zero
.Lsqr_1024_no_n_copy:
and \$-1024, %rsp
vmovdqu 32*8-128($ap), $ACC8
lea 192(%rsp), $tp0 # 64+128=192
- vpbroadcastq .Land_mask(%rip), $AND_MASK
+ vmovdqu .Land_mask(%rip), $AND_MASK
jmp .LOOP_GRANDE_SQR_1024
.align 32
vpbroadcastq 32*2-128($ap), $B1
vmovdqu $ACC9, 32*17-448($tp1)
- xor $tmp, $tmp
+ mov $ap, $tpa
mov \$4, $i
- jmp .Lentry_1024
+ jmp .Lsqr_entry_1024
___
$TEMP0=$Y1;
$TEMP2=$Y2;
$code.=<<___;
.align 32
.LOOP_SQR_1024:
- vmovdqu 32*0(%rsp,$tmp), $TEMP0 # 32*0-192($tp0,$tmp)
- vmovdqu 32*1(%rsp,$tmp), $TEMP1 # 32*1-192($tp0,$tmp)
- vpbroadcastq 32*1-128($ap,$tmp), $B2
+ vpbroadcastq 32*1-128($tpa), $B2
vpmuludq 32*0-128($ap), $B1, $ACC0
- vmovdqu 32*2-192($tp0,$tmp), $TEMP2
- vpaddq $TEMP0, $ACC0, $ACC0
+ vpaddq 32*0-192($tp0), $ACC0, $ACC0
vpmuludq 32*0-128($aap), $B1, $ACC1
- vmovdqu 32*3-192($tp0,$tmp), $TEMP0
- vpaddq $TEMP1, $ACC1, $ACC1
+ vpaddq 32*1-192($tp0), $ACC1, $ACC1
vpmuludq 32*1-128($aap), $B1, $ACC2
- vmovdqu 32*4-192($tp0,$tmp), $TEMP1
- vpaddq $TEMP2, $ACC2, $ACC2
+ vpaddq 32*2-192($tp0), $ACC2, $ACC2
vpmuludq 32*2-128($aap), $B1, $ACC3
- vmovdqu 32*5-192($tp0,$tmp), $TEMP2
- vpaddq $TEMP0, $ACC3, $ACC3
+ vpaddq 32*3-192($tp0), $ACC3, $ACC3
vpmuludq 32*3-128($aap), $B1, $ACC4
- vmovdqu 32*6-192($tp0,$tmp), $TEMP0
- vpaddq $TEMP1, $ACC4, $ACC4
+ vpaddq 32*4-192($tp0), $ACC4, $ACC4
vpmuludq 32*4-128($aap), $B1, $ACC5
- vmovdqu 32*7-192($tp0,$tmp), $TEMP1
- vpaddq $TEMP2, $ACC5, $ACC5
+ vpaddq 32*5-192($tp0), $ACC5, $ACC5
vpmuludq 32*5-128($aap), $B1, $ACC6
- vmovdqu 32*8-192($tp0,$tmp), $TEMP2
- vpaddq $TEMP0, $ACC6, $ACC6
+ vpaddq 32*6-192($tp0), $ACC6, $ACC6
vpmuludq 32*6-128($aap), $B1, $ACC7
- vpaddq $TEMP1, $ACC7, $ACC7
+ vpaddq 32*7-192($tp0), $ACC7, $ACC7
vpmuludq 32*7-128($aap), $B1, $ACC8
- vpbroadcastq 32*2-128($ap,$tmp), $B1
- vpaddq $TEMP2, $ACC8, $ACC8
-.Lentry_1024:
- vmovdqu $ACC0, 32*0(%rsp,$tmp) # 32*0-192($tp0,$tmp)
- vmovdqu $ACC1, 32*1(%rsp,$tmp) # 32*1-192($tp0,$tmp)
+ vpbroadcastq 32*2-128($tpa), $B1
+ vpaddq 32*8-192($tp0), $ACC8, $ACC8
+.Lsqr_entry_1024:
+ vmovdqu $ACC0, 32*0-192($tp0)
+ vmovdqu $ACC1, 32*1-192($tp0)
vpmuludq 32*1-128($ap), $B2, $TEMP0
vpaddq $TEMP0, $ACC2, $ACC2
vpmuludq 32*4-128($aap), $B2, $TEMP1
vpaddq $TEMP1, $ACC6, $ACC6
vpmuludq 32*5-128($aap), $B2, $TEMP2
- vmovdqu 32*9-192($tp0,$tmp), $TEMP1
vpaddq $TEMP2, $ACC7, $ACC7
vpmuludq 32*6-128($aap), $B2, $TEMP0
vpaddq $TEMP0, $ACC8, $ACC8
vpmuludq 32*7-128($aap), $B2, $ACC0
- vpbroadcastq 32*3-128($ap,$tmp), $B2
- vpaddq $TEMP1, $ACC0, $ACC0
+ vpbroadcastq 32*3-128($tpa), $B2
+ vpaddq 32*9-192($tp0), $ACC0, $ACC0
- vmovdqu $ACC2, 32*2-192($tp0,$tmp)
- vmovdqu $ACC3, 32*3-192($tp0,$tmp)
+ vmovdqu $ACC2, 32*2-192($tp0)
+ vmovdqu $ACC3, 32*3-192($tp0)
vpmuludq 32*2-128($ap), $B1, $TEMP2
vpaddq $TEMP2, $ACC4, $ACC4
vpmuludq 32*4-128($aap), $B1, $TEMP2
vpaddq $TEMP2, $ACC7, $ACC7
vpmuludq 32*5-128($aap), $B1, $TEMP0
- vmovdqu 32*10-448($tp1,$tmp), $TEMP2
vpaddq $TEMP0, $ACC8, $ACC8
vpmuludq 32*6-128($aap), $B1, $TEMP1
vpaddq $TEMP1, $ACC0, $ACC0
vpmuludq 32*7-128($aap), $B1, $ACC1
- vpbroadcastq 32*4-128($ap,$tmp), $B1
- vpaddq $TEMP2, $ACC1, $ACC1
+ vpbroadcastq 32*4-128($tpa), $B1
+ vpaddq 32*10-448($tp1), $ACC1, $ACC1
- vmovdqu $ACC4, 32*4-192($tp0,$tmp)
- vmovdqu $ACC5, 32*5-192($tp0,$tmp)
+ vmovdqu $ACC4, 32*4-192($tp0)
+ vmovdqu $ACC5, 32*5-192($tp0)
vpmuludq 32*3-128($ap), $B2, $TEMP0
vpaddq $TEMP0, $ACC6, $ACC6
vpmuludq 32*4-128($aap), $B2, $TEMP2
vpaddq $TEMP2, $ACC8, $ACC8
vpmuludq 32*5-128($aap), $B2, $TEMP0
- vmovdqu 32*11-448($tp1,$tmp), $TEMP2
vpaddq $TEMP0, $ACC0, $ACC0
vpmuludq 32*6-128($aap), $B2, $TEMP1
vpaddq $TEMP1, $ACC1, $ACC1
vpmuludq 32*7-128($aap), $B2, $ACC2
- vpbroadcastq 32*5-128($ap,$tmp), $B2
- vpaddq $TEMP2, $ACC2, $ACC2
+ vpbroadcastq 32*5-128($tpa), $B2
+ vpaddq 32*11-448($tp1), $ACC2, $ACC2
- vmovdqu $ACC6, 32*6-192($tp0,$tmp)
- vmovdqu $ACC7, 32*7-192($tp0,$tmp)
+ vmovdqu $ACC6, 32*6-192($tp0)
+ vmovdqu $ACC7, 32*7-192($tp0)
vpmuludq 32*4-128($ap), $B1, $TEMP0
vpaddq $TEMP0, $ACC8, $ACC8
vpmuludq 32*4-128($aap), $B1, $TEMP1
vpaddq $TEMP1, $ACC0, $ACC0
vpmuludq 32*5-128($aap), $B1, $TEMP2
- vmovdqu 32*12-448($tp1,$tmp), $TEMP1
vpaddq $TEMP2, $ACC1, $ACC1
vpmuludq 32*6-128($aap), $B1, $TEMP0
vpaddq $TEMP0, $ACC2, $ACC2
vpmuludq 32*7-128($aap), $B1, $ACC3
- vpbroadcastq 32*6-128($ap,$tmp), $B1
- vpaddq $TEMP1, $ACC3, $ACC3
+ vpbroadcastq 32*6-128($tpa), $B1
+ vpaddq 32*12-448($tp1), $ACC3, $ACC3
- vmovdqu $ACC8, 32*8-192($tp0,$tmp)
- vmovdqu $ACC0, 32*9-192($tp0,$tmp)
+ vmovdqu $ACC8, 32*8-192($tp0)
+ vmovdqu $ACC0, 32*9-192($tp0)
+ lea 8($tp0), $tp0
vpmuludq 32*5-128($ap), $B2, $TEMP2
vpaddq $TEMP2, $ACC1, $ACC1
vpmuludq 32*5-128($aap), $B2, $TEMP0
- vmovdqu 32*13-448($tp1,$tmp), $TEMP2
vpaddq $TEMP0, $ACC2, $ACC2
vpmuludq 32*6-128($aap), $B2, $TEMP1
vpaddq $TEMP1, $ACC3, $ACC3
vpmuludq 32*7-128($aap), $B2, $ACC4
- vpbroadcastq 32*7-128($ap,$tmp), $B2
- vpaddq $TEMP2, $ACC4, $ACC4
+ vpbroadcastq 32*7-128($tpa), $B2
+ vpaddq 32*13-448($tp1), $ACC4, $ACC4
- vmovdqu $ACC1, 32*10-448($tp1,$tmp)
- vmovdqu $ACC2, 32*11-448($tp1,$tmp)
+ vmovdqu $ACC1, 32*10-448($tp1)
+ vmovdqu $ACC2, 32*11-448($tp1)
vpmuludq 32*6-128($ap), $B1, $TEMP0
- vmovdqu 32*14-448($tp1,$tmp), $TEMP2
vpaddq $TEMP0, $ACC3, $ACC3
vpmuludq 32*6-128($aap), $B1, $TEMP1
- vpbroadcastq 32*8-128($ap,$tmp), $ACC0 # borrow $ACC0 for $B1
+ vpbroadcastq 32*8-128($tpa), $ACC0 # borrow $ACC0 for $B1
vpaddq $TEMP1, $ACC4, $ACC4
vpmuludq 32*7-128($aap), $B1, $ACC5
- vpbroadcastq 32*0+8-128($ap,$tmp), $B1 # for next iteration
- vpaddq $TEMP2, $ACC5, $ACC5
- vmovdqu 32*15-448($tp1,$tmp), $TEMP1
+ vpbroadcastq 32*0+8-128($tpa), $B1 # for next iteration
+ vpaddq 32*14-448($tp1), $ACC5, $ACC5
- vmovdqu $ACC3, 32*12-448($tp1,$tmp)
- vmovdqu $ACC4, 32*13-448($tp1,$tmp)
+ vmovdqu $ACC3, 32*12-448($tp1)
+ vmovdqu $ACC4, 32*13-448($tp1)
+ lea 8($tpa), $tpa
vpmuludq 32*7-128($ap), $B2, $TEMP0
- vmovdqu 32*16-448($tp1,$tmp), $TEMP2
vpaddq $TEMP0, $ACC5, $ACC5
vpmuludq 32*7-128($aap), $B2, $ACC6
- vpaddq $TEMP1, $ACC6, $ACC6
+ vpaddq 32*15-448($tp1), $ACC6, $ACC6
vpmuludq 32*8-128($ap), $ACC0, $ACC7
- vmovdqu $ACC5, 32*14-448($tp1,$tmp)
- vpaddq $TEMP2, $ACC7, $ACC7
- vmovdqu $ACC6, 32*15-448($tp1,$tmp)
- vmovdqu $ACC7, 32*16-448($tp1,$tmp)
+ vmovdqu $ACC5, 32*14-448($tp1)
+ vpaddq 32*16-448($tp1), $ACC7, $ACC7
+ vmovdqu $ACC6, 32*15-448($tp1)
+ vmovdqu $ACC7, 32*16-448($tp1)
+ lea 8($tp1), $tp1
- lea 8($tmp), $tmp
- dec $i
+ dec $i
jnz .LOOP_SQR_1024
___
$ZERO = $ACC9;
$TEMP3 = $Y1;
$TEMP4 = $Y2;
$code.=<<___;
- #we need to fix indexes 32-39 to avoid overflow
- vmovdqu 32*8-192($tp0), $ACC8
- vmovdqu 32*9-192($tp0), $ACC1
- vmovdqu 32*10-448($tp1), $ACC2
+ # we need to fix indices 32-39 to avoid overflow
+ vmovdqu 32*8(%rsp), $ACC8 # 32*8-192($tp0),
+ vmovdqu 32*9(%rsp), $ACC1 # 32*9-192($tp0)
+ vmovdqu 32*10(%rsp), $ACC2 # 32*10-192($tp0)
+ lea 192(%rsp), $tp0 # 64+128=192
vpsrlq \$29, $ACC8, $TEMP1
vpand $AND_MASK, $ACC8, $ACC8
vpaddq $TEMP1, $ACC1, $ACC1
vpaddq $TEMP2, $ACC2, $ACC2
vmovdqu $ACC1, 32*9-192($tp0)
- vmovdqu $ACC2, 32*10-448($tp1)
+ vmovdqu $ACC2, 32*10-192($tp0)
mov (%rsp), %rax
mov 8(%rsp), $r1
mov %rax, %rdx
imulq -128($np), %rax
vpaddq $TEMP0, $ACC1, $ACC1
- vpmuludq 32*2-128($np), $Y1, $TEMP1
add %rax, $r1
+ vpmuludq 32*2-128($np), $Y1, $TEMP1
mov %rdx, %rax
imulq 8-128($np), %rax
vpaddq $TEMP1, $ACC2, $ACC2
vpmuludq 32*3-128($np), $Y1, $TEMP2
+ .byte 0x67
add %rax, $r2
+ .byte 0x67
mov %rdx, %rax
imulq 16-128($np), %rax
shr \$29, $r1
vpaddq $TEMP0, $ACC7, $ACC7
vpmuludq 32*8-128($np), $Y1, $TEMP1
vmovd %eax, $Y1
- vmovdqu 32*1-8-128($np), $TEMP2
+ #vmovdqu 32*1-8-128($np), $TEMP2 # moved below
vpaddq $TEMP1, $ACC8, $ACC8
- vmovdqu 32*2-8-128($np), $TEMP0
+ #vmovdqu 32*2-8-128($np), $TEMP0 # moved below
vpbroadcastq $Y1, $Y1
- vpmuludq $Y2, $TEMP2, $TEMP2
+ vpmuludq 32*1-8-128($np), $Y2, $TEMP2 # see above
vmovdqu 32*3-8-128($np), $TEMP1
mov %rax, %rdx
imulq -128($np), %rax
vpaddq $TEMP2, $ACC1, $ACC1
- vpmuludq $Y2, $TEMP0, $TEMP0
+ vpmuludq 32*2-8-128($np), $Y2, $TEMP0 # see above
vmovdqu 32*4-8-128($np), $TEMP2
add %rax, $r2
mov %rdx, %rax
vpaddq $TEMP1, $ACC3, $ACC3
vpmuludq $Y2, $TEMP2, $TEMP2
vmovdqu 32*6-8-128($np), $TEMP1
+ .byte 0x67
mov %rax, $r3
imull $n0, %eax
vpaddq $TEMP2, $ACC4, $ACC4
vpmuludq $Y2, $TEMP0, $TEMP0
- vmovdqu 32*7-8-128($np), $TEMP2
+ .byte 0xc4,0x41,0x7e,0x6f,0x9d,0x58,0x00,0x00,0x00 # vmovdqu 32*7-8-128($np), $TEMP2
and \$0x1fffffff, %eax
vpaddq $TEMP0, $ACC5, $ACC5
vpmuludq $Y2, $TEMP1, $TEMP1
vpaddq $TEMP1, $ACC1, $ACC1
vpmuludq $Y2, $ACC0, $ACC0
vpmuludq $Y1, $TEMP2, $TEMP2
- vmovdqu 32*4-16-128($np), $TEMP1
+ .byte 0xc4,0x41,0x7e,0x6f,0xb5,0xf0,0xff,0xff,0xff # vmovdqu 32*4-16-128($np), $TEMP1
vpaddq $ACC1, $ACC0, $ACC0
vpaddq $TEMP2, $ACC2, $ACC2
vpmuludq $Y1, $TEMP0, $TEMP0
vmovdqu 32*5-16-128($np), $TEMP2
+ .byte 0x67
vmovq $ACC0, %rax
vmovdqu $ACC0, (%rsp) # transfer $r0-$r3
vpaddq $TEMP0, $ACC3, $ACC3
vmovdqu 32*8-16-128($np), $TEMP2
vpaddq $TEMP0, $ACC6, $ACC6
vpmuludq $Y1, $TEMP1, $TEMP1
- vmovdqu 32*9-16-128($np), $TEMP0
shr \$29, $r3
+ vmovdqu 32*9-16-128($np), $TEMP0
+ add $r3, %rax
vpaddq $TEMP1, $ACC7, $ACC7
vpmuludq $Y1, $TEMP2, $TEMP2
- vmovdqu 32*2-24-128($np), $TEMP1
- add $r3, %rax
+ #vmovdqu 32*2-24-128($np), $TEMP1 # moved below
mov %rax, $r0
imull $n0, %eax
vpaddq $TEMP2, $ACC8, $ACC8
and \$0x1fffffff, %eax
vmovd %eax, $Y1
vmovdqu 32*3-24-128($np), $TEMP2
+ .byte 0x67
vpaddq $TEMP0, $ACC9, $ACC9
vpbroadcastq $Y1, $Y1
- vpmuludq $Y2, $TEMP1, $TEMP1
+ vpmuludq 32*2-24-128($np), $Y2, $TEMP1 # see above
vmovdqu 32*4-24-128($np), $TEMP0
mov %rax, %rdx
imulq -128($np), %rax
add %rax, $r0
mov %rdx, %rax
imulq 8-128($np), %rax
+ .byte 0x67
shr \$29, $r0
mov 16(%rsp), $r2
vpaddq $TEMP2, $ACC3, $ACC2
vmovdqu 32*8-24-128($np), $TEMP1
mov %rax, $r1
imull $n0, %eax
- vpaddq $TEMP2, $ACC6, $ACC5
vpmuludq $Y2, $TEMP0, $TEMP0
+ vpaddq $TEMP2, $ACC6, $ACC5
vmovdqu 32*9-24-128($np), $TEMP2
and \$0x1fffffff, %eax
vpaddq $TEMP0, $ACC7, $ACC6
vpblendd \$3, $TEMP4, $TEMP5, $TEMP4
vpaddq $TEMP3, $ACC7, $ACC7
vpaddq $TEMP4, $ACC8, $ACC8
-
+
vpsrlq \$29, $ACC4, $TEMP1
vpand $AND_MASK, $ACC4, $ACC4
vpsrlq \$29, $ACC5, $TEMP2
vzeroall
mov %rbp, %rax
+.cfi_def_cfa_register %rax
___
$code.=<<___ if ($win64);
+.Lsqr_1024_in_tail:
movaps -0xd8(%rax),%xmm6
movaps -0xc8(%rax),%xmm7
movaps -0xb8(%rax),%xmm8
___
$code.=<<___;
mov -48(%rax),%r15
+.cfi_restore %r15
mov -40(%rax),%r14
+.cfi_restore %r14
mov -32(%rax),%r13
+.cfi_restore %r13
mov -24(%rax),%r12
+.cfi_restore %r12
mov -16(%rax),%rbp
+.cfi_restore %rbp
mov -8(%rax),%rbx
+.cfi_restore %rbx
lea (%rax),%rsp # restore %rsp
+.cfi_def_cfa_register %rsp
.Lsqr_1024_epilogue:
ret
+.cfi_endproc
.size rsaz_1024_sqr_avx2,.-rsaz_1024_sqr_avx2
___
}
.type rsaz_1024_mul_avx2,\@function,5
.align 64
rsaz_1024_mul_avx2:
+.cfi_startproc
lea (%rsp), %rax
+.cfi_def_cfa_register %rax
push %rbx
+.cfi_push %rbx
push %rbp
+.cfi_push %rbp
push %r12
+.cfi_push %r12
push %r13
+.cfi_push %r13
push %r14
+.cfi_push %r14
push %r15
+.cfi_push %r15
___
$code.=<<___ if ($win64);
+ vzeroupper
lea -0xa8(%rsp),%rsp
- movaps %xmm6,-0xd8(%rax)
- movaps %xmm7,-0xc8(%rax)
- movaps %xmm8,-0xb8(%rax)
- movaps %xmm9,-0xa8(%rax)
- movaps %xmm10,-0x98(%rax)
- movaps %xmm11,-0x88(%rax)
- movaps %xmm12,-0x78(%rax)
- movaps %xmm13,-0x68(%rax)
- movaps %xmm14,-0x58(%rax)
- movaps %xmm15,-0x48(%rax)
+ vmovaps %xmm6,-0xd8(%rax)
+ vmovaps %xmm7,-0xc8(%rax)
+ vmovaps %xmm8,-0xb8(%rax)
+ vmovaps %xmm9,-0xa8(%rax)
+ vmovaps %xmm10,-0x98(%rax)
+ vmovaps %xmm11,-0x88(%rax)
+ vmovaps %xmm12,-0x78(%rax)
+ vmovaps %xmm13,-0x68(%rax)
+ vmovaps %xmm14,-0x58(%rax)
+ vmovaps %xmm15,-0x48(%rax)
.Lmul_1024_body:
___
$code.=<<___;
mov %rax,%rbp
+.cfi_def_cfa_register %rbp
vzeroall
mov %rdx, $bp # reassigned argument
sub \$64,%rsp
# cross page boundary, swap it with $bp [meaning that caller
# is advised to lay down $ap and $bp next to each other, so
# that only one can cross page boundary].
+ .byte 0x67,0x67
mov $ap, $tmp
and \$4095, $tmp
add \$32*10, $tmp
and \$4095, $tmp # see if $np crosses page
add \$32*10, $tmp
+ .byte 0x67,0x67
shr \$12, $tmp
jz .Lmul_1024_no_n_copy
vpbroadcastq ($bp), $Bi
vmovdqu $ACC0, (%rsp) # clear top of stack
xor $r0, $r0
+ .byte 0x67
xor $r1, $r1
xor $r2, $r2
xor $r3, $r3
vmovdqu .Land_mask(%rip), $AND_MASK
mov \$9, $i
+ vmovdqu $ACC9, 32*9-128($rp) # $ACC9 is zero after vzeroall
jmp .Loop_mul_1024
.align 32
vpmuludq 32*6-128($np),$Yi,$TEMP1
vpaddq $TEMP1,$ACC6,$ACC6
vpmuludq 32*7-128($np),$Yi,$TEMP2
- vpblendd \$3, $ZERO, $ACC9, $ACC9 # correct $ACC3
+ vpblendd \$3, $ZERO, $ACC9, $TEMP1 # correct $ACC3
vpaddq $TEMP2,$ACC7,$ACC7
vpmuludq 32*8-128($np),$Yi,$TEMP0
- vpaddq $ACC9, $ACC3, $ACC3 # correct $ACC3
+ vpaddq $TEMP1, $ACC3, $ACC3 # correct $ACC3
vpaddq $TEMP0,$ACC8,$ACC8
mov %rbx, %rax
vmovdqu -8+32*2-128($ap),$TEMP2
mov $r1, %rax
+ vpblendd \$0xfc, $ZERO, $ACC9, $ACC9 # correct $ACC3
imull $n0, %eax
+ vpaddq $ACC9,$ACC4,$ACC4 # correct $ACC3
and \$0x1fffffff, %eax
imulq 16-128($ap),%rbx
# But as we underutilize resources, it's possible to correct in
# each iteration with marginal performance loss. But then, as
# we do it in each iteration, we can correct less digits, and
-# avoid performance penalties completely. Also note that we
-# correct only three digits out of four. This works because
-# most significant digit is subjected to less additions.
+# avoid performance penalties completely.
$TEMP0 = $ACC9;
$TEMP3 = $Bi;
$TEMP4 = $Yi;
$code.=<<___;
- vpermq \$0, $AND_MASK, $AND_MASK
vpaddq (%rsp), $TEMP1, $ACC0
vpsrlq \$29, $ACC0, $TEMP1
vpaddq $TEMP4, $ACC8, $ACC8
vmovdqu $ACC4, 128-128($rp)
- vmovdqu $ACC5, 160-128($rp)
+ vmovdqu $ACC5, 160-128($rp)
vmovdqu $ACC6, 192-128($rp)
vmovdqu $ACC7, 224-128($rp)
vmovdqu $ACC8, 256-128($rp)
vzeroupper
mov %rbp, %rax
+.cfi_def_cfa_register %rax
___
$code.=<<___ if ($win64);
+.Lmul_1024_in_tail:
movaps -0xd8(%rax),%xmm6
movaps -0xc8(%rax),%xmm7
movaps -0xb8(%rax),%xmm8
___
$code.=<<___;
mov -48(%rax),%r15
+.cfi_restore %r15
mov -40(%rax),%r14
+.cfi_restore %r14
mov -32(%rax),%r13
+.cfi_restore %r13
mov -24(%rax),%r12
+.cfi_restore %r12
mov -16(%rax),%rbp
+.cfi_restore %rbp
mov -8(%rax),%rbx
+.cfi_restore %rbx
lea (%rax),%rsp # restore %rsp
+.cfi_def_cfa_register %rsp
.Lmul_1024_epilogue:
ret
+.cfi_endproc
.size rsaz_1024_mul_avx2,.-rsaz_1024_mul_avx2
___
}
.type rsaz_1024_red2norm_avx2,\@abi-omnipotent
.align 32
rsaz_1024_red2norm_avx2:
+.cfi_startproc
sub \$-128,$inp # size optimization
xor %rax,%rax
___
}
$code.=<<___;
ret
+.cfi_endproc
.size rsaz_1024_red2norm_avx2,.-rsaz_1024_red2norm_avx2
.globl rsaz_1024_norm2red_avx2
.type rsaz_1024_norm2red_avx2,\@abi-omnipotent
.align 32
rsaz_1024_norm2red_avx2:
+.cfi_startproc
sub \$-128,$out # size optimization
mov ($inp),@T[0]
mov \$0x1fffffff,%eax
mov @T[0],`8*($j+2)-128`($out)
mov @T[0],`8*($j+3)-128`($out)
ret
+.cfi_endproc
.size rsaz_1024_norm2red_avx2,.-rsaz_1024_norm2red_avx2
___
}
.type rsaz_1024_scatter5_avx2,\@abi-omnipotent
.align 32
rsaz_1024_scatter5_avx2:
+.cfi_startproc
vzeroupper
vmovdqu .Lscatter_permd(%rip),%ymm5
shl \$4,$power
vzeroupper
ret
+.cfi_endproc
.size rsaz_1024_scatter5_avx2,.-rsaz_1024_scatter5_avx2
.globl rsaz_1024_gather5_avx2
.type rsaz_1024_gather5_avx2,\@abi-omnipotent
.align 32
rsaz_1024_gather5_avx2:
+.cfi_startproc
+ vzeroupper
+ mov %rsp,%r11
+.cfi_def_cfa_register %r11
___
$code.=<<___ if ($win64);
lea -0x88(%rsp),%rax
.LSEH_begin_rsaz_1024_gather5:
# I can't trust assembler to use specific encoding:-(
- .byte 0x48,0x8d,0x60,0xe0 #lea -0x20(%rax),%rsp
- .byte 0x0f,0x29,0x70,0xe0 #movaps %xmm6,-0x20(%rax)
- .byte 0x0f,0x29,0x78,0xf0 #movaps %xmm7,-0x10(%rax)
- .byte 0x44,0x0f,0x29,0x00 #movaps %xmm8,0(%rax)
- .byte 0x44,0x0f,0x29,0x48,0x10 #movaps %xmm9,0x10(%rax)
- .byte 0x44,0x0f,0x29,0x50,0x20 #movaps %xmm10,0x20(%rax)
- .byte 0x44,0x0f,0x29,0x58,0x30 #movaps %xmm11,0x30(%rax)
- .byte 0x44,0x0f,0x29,0x60,0x40 #movaps %xmm12,0x40(%rax)
- .byte 0x44,0x0f,0x29,0x68,0x50 #movaps %xmm13,0x50(%rax)
- .byte 0x44,0x0f,0x29,0x70,0x60 #movaps %xmm14,0x60(%rax)
- .byte 0x44,0x0f,0x29,0x78,0x70 #movaps %xmm15,0x70(%rax)
+ .byte 0x48,0x8d,0x60,0xe0 # lea -0x20(%rax),%rsp
+ .byte 0xc5,0xf8,0x29,0x70,0xe0 # vmovaps %xmm6,-0x20(%rax)
+ .byte 0xc5,0xf8,0x29,0x78,0xf0 # vmovaps %xmm7,-0x10(%rax)
+ .byte 0xc5,0x78,0x29,0x40,0x00 # vmovaps %xmm8,0(%rax)
+ .byte 0xc5,0x78,0x29,0x48,0x10 # vmovaps %xmm9,0x10(%rax)
+ .byte 0xc5,0x78,0x29,0x50,0x20 # vmovaps %xmm10,0x20(%rax)
+ .byte 0xc5,0x78,0x29,0x58,0x30 # vmovaps %xmm11,0x30(%rax)
+ .byte 0xc5,0x78,0x29,0x60,0x40 # vmovaps %xmm12,0x40(%rax)
+ .byte 0xc5,0x78,0x29,0x68,0x50 # vmovaps %xmm13,0x50(%rax)
+ .byte 0xc5,0x78,0x29,0x70,0x60 # vmovaps %xmm14,0x60(%rax)
+ .byte 0xc5,0x78,0x29,0x78,0x70 # vmovaps %xmm15,0x70(%rax)
___
$code.=<<___;
- vzeroupper
- lea .Lgather_table(%rip),%r11
- mov $power,%eax
- and \$3,$power
- shr \$2,%eax # cache line number
- shl \$4,$power # offset within cache line
-
- vmovdqu -32(%r11),%ymm7 # .Lgather_permd
- vpbroadcastb 8(%r11,%rax), %xmm8
- vpbroadcastb 7(%r11,%rax), %xmm9
- vpbroadcastb 6(%r11,%rax), %xmm10
- vpbroadcastb 5(%r11,%rax), %xmm11
- vpbroadcastb 4(%r11,%rax), %xmm12
- vpbroadcastb 3(%r11,%rax), %xmm13
- vpbroadcastb 2(%r11,%rax), %xmm14
- vpbroadcastb 1(%r11,%rax), %xmm15
-
- lea ($inp,$power),$inp
- mov \$64,%r11 # size optimization
- mov \$9,%eax
- jmp .Loop_gather_1024
+ lea -0x100(%rsp),%rsp
+ and \$-32, %rsp
+ lea .Linc(%rip), %r10
+ lea -128(%rsp),%rax # control u-op density
+
+ vmovd $power, %xmm4
+ vmovdqa (%r10),%ymm0
+ vmovdqa 32(%r10),%ymm1
+ vmovdqa 64(%r10),%ymm5
+ vpbroadcastd %xmm4,%ymm4
+
+ vpaddd %ymm5, %ymm0, %ymm2
+ vpcmpeqd %ymm4, %ymm0, %ymm0
+ vpaddd %ymm5, %ymm1, %ymm3
+ vpcmpeqd %ymm4, %ymm1, %ymm1
+ vmovdqa %ymm0, 32*0+128(%rax)
+ vpaddd %ymm5, %ymm2, %ymm0
+ vpcmpeqd %ymm4, %ymm2, %ymm2
+ vmovdqa %ymm1, 32*1+128(%rax)
+ vpaddd %ymm5, %ymm3, %ymm1
+ vpcmpeqd %ymm4, %ymm3, %ymm3
+ vmovdqa %ymm2, 32*2+128(%rax)
+ vpaddd %ymm5, %ymm0, %ymm2
+ vpcmpeqd %ymm4, %ymm0, %ymm0
+ vmovdqa %ymm3, 32*3+128(%rax)
+ vpaddd %ymm5, %ymm1, %ymm3
+ vpcmpeqd %ymm4, %ymm1, %ymm1
+ vmovdqa %ymm0, 32*4+128(%rax)
+ vpaddd %ymm5, %ymm2, %ymm8
+ vpcmpeqd %ymm4, %ymm2, %ymm2
+ vmovdqa %ymm1, 32*5+128(%rax)
+ vpaddd %ymm5, %ymm3, %ymm9
+ vpcmpeqd %ymm4, %ymm3, %ymm3
+ vmovdqa %ymm2, 32*6+128(%rax)
+ vpaddd %ymm5, %ymm8, %ymm10
+ vpcmpeqd %ymm4, %ymm8, %ymm8
+ vmovdqa %ymm3, 32*7+128(%rax)
+ vpaddd %ymm5, %ymm9, %ymm11
+ vpcmpeqd %ymm4, %ymm9, %ymm9
+ vpaddd %ymm5, %ymm10, %ymm12
+ vpcmpeqd %ymm4, %ymm10, %ymm10
+ vpaddd %ymm5, %ymm11, %ymm13
+ vpcmpeqd %ymm4, %ymm11, %ymm11
+ vpaddd %ymm5, %ymm12, %ymm14
+ vpcmpeqd %ymm4, %ymm12, %ymm12
+ vpaddd %ymm5, %ymm13, %ymm15
+ vpcmpeqd %ymm4, %ymm13, %ymm13
+ vpcmpeqd %ymm4, %ymm14, %ymm14
+ vpcmpeqd %ymm4, %ymm15, %ymm15
+
+ vmovdqa -32(%r10),%ymm7 # .Lgather_permd
+ lea 128($inp), $inp
+ mov \$9,$power
-.align 32
.Loop_gather_1024:
- vpand ($inp), %xmm8,%xmm0
- vpand ($inp,%r11), %xmm9,%xmm1
- vpand ($inp,%r11,2), %xmm10,%xmm2
- vpand 64($inp,%r11,2), %xmm11,%xmm3
- vpor %xmm0,%xmm1,%xmm1
- vpand ($inp,%r11,4), %xmm12,%xmm4
- vpor %xmm2,%xmm3,%xmm3
- vpand 64($inp,%r11,4), %xmm13,%xmm5
- vpor %xmm1,%xmm3,%xmm3
- vpand -128($inp,%r11,8), %xmm14,%xmm6
- vpor %xmm4,%xmm5,%xmm5
- vpand -64($inp,%r11,8), %xmm15,%xmm2
- lea ($inp,%r11,8),$inp
- vpor %xmm3,%xmm5,%xmm5
- vpor %xmm2,%xmm6,%xmm6
- vpor %xmm5,%xmm6,%xmm6
- vpermd %ymm6,%ymm7,%ymm6
- vmovdqu %ymm6,($out)
+ vmovdqa 32*0-128($inp), %ymm0
+ vmovdqa 32*1-128($inp), %ymm1
+ vmovdqa 32*2-128($inp), %ymm2
+ vmovdqa 32*3-128($inp), %ymm3
+ vpand 32*0+128(%rax), %ymm0, %ymm0
+ vpand 32*1+128(%rax), %ymm1, %ymm1
+ vpand 32*2+128(%rax), %ymm2, %ymm2
+ vpor %ymm0, %ymm1, %ymm4
+ vpand 32*3+128(%rax), %ymm3, %ymm3
+ vmovdqa 32*4-128($inp), %ymm0
+ vmovdqa 32*5-128($inp), %ymm1
+ vpor %ymm2, %ymm3, %ymm5
+ vmovdqa 32*6-128($inp), %ymm2
+ vmovdqa 32*7-128($inp), %ymm3
+ vpand 32*4+128(%rax), %ymm0, %ymm0
+ vpand 32*5+128(%rax), %ymm1, %ymm1
+ vpand 32*6+128(%rax), %ymm2, %ymm2
+ vpor %ymm0, %ymm4, %ymm4
+ vpand 32*7+128(%rax), %ymm3, %ymm3
+ vpand 32*8-128($inp), %ymm8, %ymm0
+ vpor %ymm1, %ymm5, %ymm5
+ vpand 32*9-128($inp), %ymm9, %ymm1
+ vpor %ymm2, %ymm4, %ymm4
+ vpand 32*10-128($inp),%ymm10, %ymm2
+ vpor %ymm3, %ymm5, %ymm5
+ vpand 32*11-128($inp),%ymm11, %ymm3
+ vpor %ymm0, %ymm4, %ymm4
+ vpand 32*12-128($inp),%ymm12, %ymm0
+ vpor %ymm1, %ymm5, %ymm5
+ vpand 32*13-128($inp),%ymm13, %ymm1
+ vpor %ymm2, %ymm4, %ymm4
+ vpand 32*14-128($inp),%ymm14, %ymm2
+ vpor %ymm3, %ymm5, %ymm5
+ vpand 32*15-128($inp),%ymm15, %ymm3
+ lea 32*16($inp), $inp
+ vpor %ymm0, %ymm4, %ymm4
+ vpor %ymm1, %ymm5, %ymm5
+ vpor %ymm2, %ymm4, %ymm4
+ vpor %ymm3, %ymm5, %ymm5
+
+ vpor %ymm5, %ymm4, %ymm4
+ vextracti128 \$1, %ymm4, %xmm5 # upper half is cleared
+ vpor %xmm4, %xmm5, %xmm5
+ vpermd %ymm5,%ymm7,%ymm5
+ vmovdqu %ymm5,($out)
lea 32($out),$out
- dec %eax
+ dec $power
jnz .Loop_gather_1024
vpxor %ymm0,%ymm0,%ymm0
vzeroupper
___
$code.=<<___ if ($win64);
- movaps (%rsp),%xmm6
- movaps 0x10(%rsp),%xmm7
- movaps 0x20(%rsp),%xmm8
- movaps 0x30(%rsp),%xmm9
- movaps 0x40(%rsp),%xmm10
- movaps 0x50(%rsp),%xmm11
- movaps 0x60(%rsp),%xmm12
- movaps 0x70(%rsp),%xmm13
- movaps 0x80(%rsp),%xmm14
- movaps 0x90(%rsp),%xmm15
- lea 0xa8(%rsp),%rsp
-.LSEH_end_rsaz_1024_gather5:
+ movaps -0xa8(%r11),%xmm6
+ movaps -0x98(%r11),%xmm7
+ movaps -0x88(%r11),%xmm8
+ movaps -0x78(%r11),%xmm9
+ movaps -0x68(%r11),%xmm10
+ movaps -0x58(%r11),%xmm11
+ movaps -0x48(%r11),%xmm12
+ movaps -0x38(%r11),%xmm13
+ movaps -0x28(%r11),%xmm14
+ movaps -0x18(%r11),%xmm15
___
$code.=<<___;
+ lea (%r11),%rsp
+.cfi_def_cfa_register %rsp
ret
+.cfi_endproc
+.LSEH_end_rsaz_1024_gather5:
.size rsaz_1024_gather5_avx2,.-rsaz_1024_gather5_avx2
___
}
.align 32
rsaz_avx2_eligible:
mov OPENSSL_ia32cap_P+8(%rip),%eax
+___
+$code.=<<___ if ($addx);
+ mov \$`1<<8|1<<19`,%ecx
+ mov \$0,%edx
+ and %eax,%ecx
+ cmp \$`1<<8|1<<19`,%ecx # check for BMI2+AD*X
+ cmove %edx,%eax
+___
+$code.=<<___;
and \$`1<<5`,%eax
shr \$5,%eax
ret
.align 64
.Land_mask:
- .quad 0x1fffffff,0x1fffffff,0x1fffffff,-1
+ .quad 0x1fffffff,0x1fffffff,0x1fffffff,0x1fffffff
.Lscatter_permd:
.long 0,2,4,6,7,7,7,7
.Lgather_permd:
.long 0,7,1,7,2,7,3,7
-.Lgather_table:
- .byte 0,0,0,0,0,0,0,0, 0xff,0,0,0,0,0,0,0
+.Linc:
+ .long 0,0,0,0, 1,1,1,1
+ .long 2,2,2,2, 3,3,3,3
+ .long 4,4,4,4, 4,4,4,4
.align 64
___
cmp %r10,%rbx # context->Rip<prologue label
jb .Lcommon_seh_tail
- mov 152($context),%rax # pull context->Rsp
-
mov 4(%r11),%r10d # HandlerData[1]
lea (%rsi,%r10),%r10 # epilogue label
cmp %r10,%rbx # context->Rip>=epilogue label
jae .Lcommon_seh_tail
- mov 160($context),%rax # pull context->Rbp
+ mov 160($context),%rbp # pull context->Rbp
+
+ mov 8(%r11),%r10d # HandlerData[2]
+ lea (%rsi,%r10),%r10 # "in tail" label
+ cmp %r10,%rbx # context->Rip>="in tail" label
+ cmovc %rbp,%rax
mov -48(%rax),%r15
mov -40(%rax),%r14
.LSEH_info_rsaz_1024_sqr_avx2:
.byte 9,0,0,0
.rva rsaz_se_handler
- .rva .Lsqr_1024_body,.Lsqr_1024_epilogue
+ .rva .Lsqr_1024_body,.Lsqr_1024_epilogue,.Lsqr_1024_in_tail
+ .long 0
.LSEH_info_rsaz_1024_mul_avx2:
.byte 9,0,0,0
.rva rsaz_se_handler
- .rva .Lmul_1024_body,.Lmul_1024_epilogue
+ .rva .Lmul_1024_body,.Lmul_1024_epilogue,.Lmul_1024_in_tail
+ .long 0
.LSEH_info_rsaz_1024_gather5:
- .byte 0x01,0x33,0x16,0x00
- .byte 0x33,0xf8,0x09,0x00 #movaps 0x90(rsp),xmm15
- .byte 0x2e,0xe8,0x08,0x00 #movaps 0x80(rsp),xmm14
- .byte 0x29,0xd8,0x07,0x00 #movaps 0x70(rsp),xmm13
- .byte 0x24,0xc8,0x06,0x00 #movaps 0x60(rsp),xmm12
- .byte 0x1f,0xb8,0x05,0x00 #movaps 0x50(rsp),xmm11
- .byte 0x1a,0xa8,0x04,0x00 #movaps 0x40(rsp),xmm10
- .byte 0x15,0x98,0x03,0x00 #movaps 0x30(rsp),xmm9
- .byte 0x10,0x88,0x02,0x00 #movaps 0x20(rsp),xmm8
- .byte 0x0c,0x78,0x01,0x00 #movaps 0x10(rsp),xmm7
- .byte 0x08,0x68,0x00,0x00 #movaps 0x00(rsp),xmm6
- .byte 0x04,0x01,0x15,0x00 #sub rsp,0xa8
+ .byte 0x01,0x36,0x17,0x0b
+ .byte 0x36,0xf8,0x09,0x00 # vmovaps 0x90(rsp),xmm15
+ .byte 0x31,0xe8,0x08,0x00 # vmovaps 0x80(rsp),xmm14
+ .byte 0x2c,0xd8,0x07,0x00 # vmovaps 0x70(rsp),xmm13
+ .byte 0x27,0xc8,0x06,0x00 # vmovaps 0x60(rsp),xmm12
+ .byte 0x22,0xb8,0x05,0x00 # vmovaps 0x50(rsp),xmm11
+ .byte 0x1d,0xa8,0x04,0x00 # vmovaps 0x40(rsp),xmm10
+ .byte 0x18,0x98,0x03,0x00 # vmovaps 0x30(rsp),xmm9
+ .byte 0x13,0x88,0x02,0x00 # vmovaps 0x20(rsp),xmm8
+ .byte 0x0e,0x78,0x01,0x00 # vmovaps 0x10(rsp),xmm7
+ .byte 0x09,0x68,0x00,0x00 # vmovaps 0x00(rsp),xmm6
+ .byte 0x04,0x01,0x15,0x00 # sub rsp,0xa8
+ .byte 0x00,0xb3,0x00,0x00 # set_frame r11
___
}
.globl rsaz_avx2_eligible
.type rsaz_avx2_eligible,\@abi-omnipotent
-rsaz_eligible:
+rsaz_avx2_eligible:
xor %eax,%eax
ret
.size rsaz_avx2_eligible,.-rsaz_avx2_eligible
projects / openssl.git / blobdiff