mirror/openssl
2
0
Fork 0
mirror of https://github.com/openssl/openssl.git synced 2024-12-09 05:51:54 +08:00
Code Issues Packages Projects Releases Wiki Activity
2f1112b22a
openssl/crypto/bn/asm/sparcv9-mont.pl

623 lines
14 KiB
Perl
Raw Normal View History

Add OpenSSL copyright to .pl files Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-05-21 20:23:39 +08:00
#! /usr/bin/env perl
Update copyright year Reviewed-by: Richard Levitte <levitte@openssl.org> (Merged from https://github.com/openssl/openssl/pull/16176)
2021-07-29 22:41:35 +08:00
# Copyright 2005-2021 The OpenSSL Project Authors. All Rights Reserved.
Add OpenSSL copyright to .pl files Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-05-21 20:23:39 +08:00
#
Following the license change, modify the boilerplates in crypto/bn/ [skip ci] Reviewed-by: Matt Caswell <matt@openssl.org> (Merged from https://github.com/openssl/openssl/pull/7777)
2018-12-06 20:22:12 +08:00
# Licensed under the Apache License 2.0 (the "License"). You may not use
Add OpenSSL copyright to .pl files Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-05-21 20:23:39 +08:00
# 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
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
# ====================================================================
Remove email addresses from source code. Names were not removed. Some comments were updated. Replace Andy's address with openssl.org Reviewed-by: Andy Polyakov <appro@openssl.org> Reviewed-by: Paul Dale <paul.dale@oracle.com> (Merged from https://github.com/openssl/openssl/pull/4516)
2017-10-11 05:55:09 +08:00
# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
Eliminate conditional final subtraction in Montgomery assembler modules.
2007-06-18 01:10:03 +08:00
# project. The module is, however, dual licensed under OpenSSL and
# CRYPTOGAMS licenses depending on where you obtain it. For further
# details see http://www.openssl.org/~appro/cryptogams/.
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
# ====================================================================
# December 2005
#
# Pure SPARCv9/8+ and IALU-only bn_mul_mont implementation. The reasons
# for undertaken effort are multiple. First of all, UltraSPARC is not
# the whole SPARCv9 universe and other VIS-free implementations deserve
# optimized code as much. Secondly, newly introduced UltraSPARC T1,
spelling fixes, just comments and readme. Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/1413)
2016-08-06 01:56:58 +08:00
# a.k.a. Niagara, has shared FPU and concurrent FPU-intensive paths,
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
# such as sparcv9a-mont, will simply sink it. Yes, T1 is equipped with
# several integrated RSA/DSA accelerator circuits accessible through
# kernel driver [only(*)], but having decent user-land software
# implementation is important too. Finally, reasons like desire to
# experiment with dedicated squaring procedure. Yes, this module
# implements one, because it was easiest to draft it in SPARCv9
# instructions...
# (*) Engine accessing the driver in question is on my TODO list.
More typo fixes Fix some comments too [skip ci] Reviewed-by: Tim Hudson <tjh@openssl.org> Reviewed-by: Richard Levitte <levitte@openssl.org> (Merged from https://github.com/openssl/openssl/pull/3069)
2017-03-29 05:57:28 +08:00
# For reference, accelerator is estimated to give 6 to 10 times
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
# improvement on single-threaded RSA sign. It should be noted
# that 6-10x improvement coefficient does not actually mean
# something extraordinary in terms of absolute [single-threaded]
# performance, as SPARCv9 instruction set is by all means least
# suitable for high performance crypto among other 64 bit
# platforms. 6-10x factor simply places T1 in same performance
# domain as say AMD64 and IA-64. Improvement of RSA verify don't
# appear impressive at all, but it's the sign operation which is
# far more critical/interesting.
# You might notice that inner loops are modulo-scheduled:-) This has
# essentially negligible impact on UltraSPARC performance, it's
# Fujitsu SPARC64 V users who should notice and hopefully appreciate
# the advantage... Currently this module surpasses sparcv9a-mont.pl
# by ~20% on UltraSPARC-III and later cores, but recall that sparcv9a
# module still have hidden potential [see TODO list there], which is
# estimated to be larger than 20%...
Unify all assembler file generators They now generally conform to the following argument sequence: script.pl "$(PERLASM_SCHEME)" [ C preprocessor arguments ... ] \ $(PROCESSOR) <output file> However, in the spirit of being able to use these scripts manually, they also allow for no argument, or for only the flavour, or for only the output file. This is done by only using the last argument as output file if it's a file (it has an extension), and only using the first argument as flavour if it isn't a file (it doesn't have an extension). While we're at it, we make all $xlate calls the same, i.e. the $output argument is always quoted, and we always die on error when trying to start $xlate. There's a perl lesson in this, regarding operator priority... This will always succeed, even when it fails: open FOO, "something" || die "ERR: $!"; The reason is that '||' has higher priority than list operators (a function is essentially a list operator and gobbles up everything following it that isn't lower priority), and since a non-empty string is always true, so that ends up being exactly the same as: open FOO, "something"; This, however, will fail if "something" can't be opened: open FOO, "something" or die "ERR: $!"; The reason is that 'or' has lower priority that list operators, i.e. it's performed after the 'open' call. Reviewed-by: Matt Caswell <matt@openssl.org> (Merged from https://github.com/openssl/openssl/pull/9884)
2019-09-13 06:06:46 +08:00
$output = pop and open STDOUT,">$output";
Unified - adapt the generation of bignum assembler to use GENERATE This gets rid of the BEGINRAW..ENDRAW sections in crypto/bn/build.info. This also moves the assembler generating perl scripts to take the output file name as last command line argument, where necessary. Reviewed-by: Rich Salz <rsalz@openssl.org>
2016-03-07 22:41:33 +08:00
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
# int bn_mul_mont(
$rp="%i0"; # BN_ULONG *rp,
$ap="%i1"; # const BN_ULONG *ap,
$bp="%i2"; # const BN_ULONG *bp,
$np="%i3"; # const BN_ULONG *np,
$n0="%i4"; # const BN_ULONG *n0,
$num="%i5"; # int num);
SPARCv9 assembly pack: unify build rules and argument handling. Make all scripts produce .S, make interpretation of $(CFLAGS) pre-processor's responsibility, start accepting $(PERLASM_SCHEME). [$(PERLASM_SCHEME) is redundant in this case, because there are no deviataions between Solaris and Linux assemblers. This is purely to unify .pl->.S handling across all targets.] Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-03-08 16:46:19 +08:00
$frame="STACK_FRAME";
$bias="STACK_BIAS";
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
$car0="%o0";
$car1="%o1";
$car2="%o2"; # 1 bit
$acc0="%o3";
$acc1="%o4";
$mask="%g1"; # 32 bits, what a waste...
$tmp0="%g4";
$tmp1="%g5";
$i="%l0";
$j="%l1";
$mul0="%l2";
$mul1="%l3";
$tp="%l4";
$apj="%l5";
$npj="%l6";
$tpj="%l7";
Unify sparcv9 assembler naming and build rules among 32- and 64-bit builds. Engage run-time switch between bn_mul_mont_fpu and bn_mul_mont_int.
2005-12-17 01:39:57 +08:00
$fname="bn_mul_mont_int";
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
$code=<<___;
Split bignum code out of the sparcv9cap.c Fixes #15978 Reviewed-by: Paul Dale <pauli@openssl.org> (Merged from https://github.com/openssl/openssl/pull/16019)
2021-07-07 23:47:06 +08:00
#ifndef __ASSEMBLER__
# define __ASSEMBLER__ 1
#endif
#include "crypto/sparc_arch.h"
SPARCv9 assembly pack: unify build rules and argument handling. Make all scripts produce .S, make interpretation of $(CFLAGS) pre-processor's responsibility, start accepting $(PERLASM_SCHEME). [$(PERLASM_SCHEME) is redundant in this case, because there are no deviataions between Solaris and Linux assemblers. This is purely to unify .pl->.S handling across all targets.] Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-03-08 16:46:19 +08:00
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
.section ".text",#alloc,#execinstr
.global $fname
.align 32
$fname:
cmp %o5,4 ! 128 bits minimum
bge,pt %icc,.Lenter
sethi %hi(0xffffffff),$mask
retl
clr %o0
.align 32
.Lenter:
save %sp,-$frame,%sp
sll $num,2,$num ! num*=4
or $mask,%lo(0xffffffff),$mask
ld [$n0],$n0
cmp $ap,$bp
and $num,$mask,$num
ld [$bp],$mul0 ! bp[0]
nop
add %sp,$bias,%o7 ! real top of stack
Minor sparcv9 clean-ups.
2005-12-28 05:27:39 +08:00
ld [$ap],$car0 ! ap[0] ! redundant in squaring context
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
sub %o7,$num,%o7
ld [$ap+4],$apj ! ap[1]
and %o7,-1024,%o7
ld [$np],$car1 ! np[0]
sub %o7,$bias,%sp ! alloca
ld [$np+4],$npj ! np[1]
SPARCv9 assembly pack: unify build rules and argument handling. Make all scripts produce .S, make interpretation of $(CFLAGS) pre-processor's responsibility, start accepting $(PERLASM_SCHEME). [$(PERLASM_SCHEME) is redundant in this case, because there are no deviataions between Solaris and Linux assemblers. This is purely to unify .pl->.S handling across all targets.] Reviewed-by: Richard Levitte <levitte@openssl.org>
2016-03-08 16:46:19 +08:00
be,pt SIZE_T_CC,.Lbn_sqr_mont
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
mov 12,$j
mulx $car0,$mul0,$car0 ! ap[0]*bp[0]
mulx $apj,$mul0,$tmp0 !prologue! ap[1]*bp[0]
and $car0,$mask,$acc0
add %sp,$bias+$frame,$tp
ld [$ap+8],$apj !prologue!
mulx $n0,$acc0,$mul1 ! "t[0]"*n0
and $mul1,$mask,$mul1
mulx $car1,$mul1,$car1 ! np[0]*"t[0]"*n0
mulx $npj,$mul1,$acc1 !prologue! np[1]*"t[0]"*n0
srlx $car0,32,$car0
add $acc0,$car1,$car1
ld [$np+8],$npj !prologue!
srlx $car1,32,$car1
mov $tmp0,$acc0 !prologue!
.L1st:
mulx $apj,$mul0,$tmp0
mulx $npj,$mul1,$tmp1
add $acc0,$car0,$car0
ld [$ap+$j],$apj ! ap[j]
and $car0,$mask,$acc0
add $acc1,$car1,$car1
ld [$np+$j],$npj ! np[j]
srlx $car0,32,$car0
add $acc0,$car1,$car1
add $j,4,$j ! j++
mov $tmp0,$acc0
st $car1,[$tp]
cmp $j,$num
mov $tmp1,$acc1
srlx $car1,32,$car1
bl %icc,.L1st
add $tp,4,$tp ! tp++
!.L1st
mulx $apj,$mul0,$tmp0 !epilogue!
mulx $npj,$mul1,$tmp1
add $acc0,$car0,$car0
and $car0,$mask,$acc0
add $acc1,$car1,$car1
srlx $car0,32,$car0
add $acc0,$car1,$car1
st $car1,[$tp]
srlx $car1,32,$car1
add $tmp0,$car0,$car0
and $car0,$mask,$acc0
add $tmp1,$car1,$car1
srlx $car0,32,$car0
add $acc0,$car1,$car1
st $car1,[$tp+4]
srlx $car1,32,$car1
add $car0,$car1,$car1
st $car1,[$tp+8]
srlx $car1,32,$car2
mov 4,$i ! i++
ld [$bp+4],$mul0 ! bp[1]
.Louter:
add %sp,$bias+$frame,$tp
ld [$ap],$car0 ! ap[0]
ld [$ap+4],$apj ! ap[1]
ld [$np],$car1 ! np[0]
ld [$np+4],$npj ! np[1]
ld [$tp],$tmp1 ! tp[0]
ld [$tp+4],$tpj ! tp[1]
mov 12,$j
mulx $car0,$mul0,$car0
mulx $apj,$mul0,$tmp0 !prologue!
add $tmp1,$car0,$car0
ld [$ap+8],$apj !prologue!
and $car0,$mask,$acc0
mulx $n0,$acc0,$mul1
and $mul1,$mask,$mul1
mulx $car1,$mul1,$car1
mulx $npj,$mul1,$acc1 !prologue!
srlx $car0,32,$car0
add $acc0,$car1,$car1
ld [$np+8],$npj !prologue!
srlx $car1,32,$car1
mov $tmp0,$acc0 !prologue!
.Linner:
mulx $apj,$mul0,$tmp0
mulx $npj,$mul1,$tmp1
add $tpj,$car0,$car0
ld [$ap+$j],$apj ! ap[j]
add $acc0,$car0,$car0
add $acc1,$car1,$car1
ld [$np+$j],$npj ! np[j]
and $car0,$mask,$acc0
ld [$tp+8],$tpj ! tp[j]
srlx $car0,32,$car0
add $acc0,$car1,$car1
add $j,4,$j ! j++
mov $tmp0,$acc0
st $car1,[$tp] ! tp[j-1]
srlx $car1,32,$car1
mov $tmp1,$acc1
cmp $j,$num
bl %icc,.Linner
add $tp,4,$tp ! tp++
!.Linner
mulx $apj,$mul0,$tmp0 !epilogue!
mulx $npj,$mul1,$tmp1
add $tpj,$car0,$car0
add $acc0,$car0,$car0
ld [$tp+8],$tpj ! tp[j]
and $car0,$mask,$acc0
add $acc1,$car1,$car1
srlx $car0,32,$car0
add $acc0,$car1,$car1
st $car1,[$tp] ! tp[j-1]
srlx $car1,32,$car1
add $tpj,$car0,$car0
add $tmp0,$car0,$car0
and $car0,$mask,$acc0
add $tmp1,$car1,$car1
add $acc0,$car1,$car1
st $car1,[$tp+4] ! tp[j-1]
srlx $car0,32,$car0
add $i,4,$i ! i++
srlx $car1,32,$car1
add $car0,$car1,$car1
cmp $i,$num
add $car2,$car1,$car1
st $car1,[$tp+8]
srlx $car1,32,$car2
bl,a %icc,.Louter
ld [$bp+$i],$mul0 ! bp[i]
!.Louter
add $tp,12,$tp
.Ltail:
add $np,$num,$np
add $rp,$num,$rp
sub %g0,$num,%o7 ! k=-num
SPARC Solaris and Linux assemblers treat .align directive differently. PR: 1547
2007-06-20 20:24:22 +08:00
ba .Lsub
subcc %g0,%g0,%g0 ! clear %icc.c
.align 16
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
.Lsub:
ld [$tp+%o7],%o0
ld [$np+%o7],%o1
Eliminate conditional final subtraction in Montgomery assembler modules.
2007-06-18 01:10:03 +08:00
subccc %o0,%o1,%o1 ! tp[j]-np[j]
SPARC Solaris and Linux assemblers treat .align directive differently. PR: 1547
2007-06-20 20:24:22 +08:00
add $rp,%o7,$i
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add %o7,4,%o7
brnz %o7,.Lsub
SPARC Solaris and Linux assemblers treat .align directive differently. PR: 1547
2007-06-20 20:24:22 +08:00
st %o1,[$i]
bn/asm/*-mont.pl: harmonize with BN_from_montgomery_word. Montgomery multiplication post-conditions in some of code paths were formally non-constant time. Cache access pattern was result-neutral, but a little bit asymmetric, which might have produced a signal [if processor reordered load and stores at run-time]. Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/6141)
2018-05-01 04:59:51 +08:00
subccc $car2,0,$car2 ! handle upmost overflow bit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
sub %g0,$num,%o7
.Lcopy:
bn/asm/*-mont.pl: harmonize with BN_from_montgomery_word. Montgomery multiplication post-conditions in some of code paths were formally non-constant time. Cache access pattern was result-neutral, but a little bit asymmetric, which might have produced a signal [if processor reordered load and stores at run-time]. Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/6141)
2018-05-01 04:59:51 +08:00
ld [$tp+%o7],%o1 ! conditional copy
ld [$rp+%o7],%o0
Eliminate conditional final subtraction in Montgomery assembler modules.
2007-06-18 01:10:03 +08:00
st %g0,[$tp+%o7] ! zap tp
bn/asm/*-mont.pl: harmonize with BN_from_montgomery_word. Montgomery multiplication post-conditions in some of code paths were formally non-constant time. Cache access pattern was result-neutral, but a little bit asymmetric, which might have produced a signal [if processor reordered load and stores at run-time]. Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/6141)
2018-05-01 04:59:51 +08:00
movcs %icc,%o1,%o0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
st %o0,[$rp+%o7]
add %o7,4,%o7
brnz %o7,.Lcopy
nop
mov 1,%i0
ret
restore
___
########
Unify sparcv9 assembler naming and build rules among 32- and 64-bit builds. Engage run-time switch between bn_mul_mont_fpu and bn_mul_mont_int.
2005-12-17 01:39:57 +08:00
######## .Lbn_sqr_mont gives up to 20% *overall* improvement over
######## code without following dedicated squaring procedure.
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
########
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
$sbit="%o5";
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
$code.=<<___;
.align 32
.Lbn_sqr_mont:
mulx $mul0,$mul0,$car0 ! ap[0]*ap[0]
mulx $apj,$mul0,$tmp0 !prologue!
and $car0,$mask,$acc0
add %sp,$bias+$frame,$tp
ld [$ap+8],$apj !prologue!
mulx $n0,$acc0,$mul1 ! "t[0]"*n0
srlx $car0,32,$car0
and $mul1,$mask,$mul1
mulx $car1,$mul1,$car1 ! np[0]*"t[0]"*n0
mulx $npj,$mul1,$acc1 !prologue!
and $car0,1,$sbit
ld [$np+8],$npj !prologue!
srlx $car0,1,$car0
add $acc0,$car1,$car1
srlx $car1,32,$car1
mov $tmp0,$acc0 !prologue!
.Lsqr_1st:
mulx $apj,$mul0,$tmp0
mulx $npj,$mul1,$tmp1
add $acc0,$car0,$car0 ! ap[j]*a0+c0
add $acc1,$car1,$car1
ld [$ap+$j],$apj ! ap[j]
and $car0,$mask,$acc0
ld [$np+$j],$npj ! np[j]
srlx $car0,32,$car0
add $acc0,$acc0,$acc0
or $sbit,$acc0,$acc0
mov $tmp1,$acc1
srlx $acc0,32,$sbit
add $j,4,$j ! j++
and $acc0,$mask,$acc0
cmp $j,$num
add $acc0,$car1,$car1
st $car1,[$tp]
mov $tmp0,$acc0
srlx $car1,32,$car1
bl %icc,.Lsqr_1st
add $tp,4,$tp ! tp++
!.Lsqr_1st
mulx $apj,$mul0,$tmp0 ! epilogue
mulx $npj,$mul1,$tmp1
add $acc0,$car0,$car0 ! ap[j]*a0+c0
add $acc1,$car1,$car1
and $car0,$mask,$acc0
srlx $car0,32,$car0
add $acc0,$acc0,$acc0
or $sbit,$acc0,$acc0
srlx $acc0,32,$sbit
and $acc0,$mask,$acc0
add $acc0,$car1,$car1
st $car1,[$tp]
srlx $car1,32,$car1
add $tmp0,$car0,$car0 ! ap[j]*a0+c0
add $tmp1,$car1,$car1
and $car0,$mask,$acc0
srlx $car0,32,$car0
add $acc0,$acc0,$acc0
or $sbit,$acc0,$acc0
srlx $acc0,32,$sbit
and $acc0,$mask,$acc0
add $acc0,$car1,$car1
st $car1,[$tp+4]
srlx $car1,32,$car1
add $car0,$car0,$car0
or $sbit,$car0,$car0
add $car0,$car1,$car1
st $car1,[$tp+8]
srlx $car1,32,$car2
ld [%sp+$bias+$frame],$tmp0 ! tp[0]
ld [%sp+$bias+$frame+4],$tmp1 ! tp[1]
ld [%sp+$bias+$frame+8],$tpj ! tp[2]
ld [$ap+4],$mul0 ! ap[1]
ld [$ap+8],$apj ! ap[2]
ld [$np],$car1 ! np[0]
ld [$np+4],$npj ! np[1]
mulx $n0,$tmp0,$mul1
mulx $mul0,$mul0,$car0
and $mul1,$mask,$mul1
mulx $car1,$mul1,$car1
mulx $npj,$mul1,$acc1
add $tmp0,$car1,$car1
and $car0,$mask,$acc0
ld [$np+8],$npj ! np[2]
srlx $car1,32,$car1
add $tmp1,$car1,$car1
srlx $car0,32,$car0
add $acc0,$car1,$car1
and $car0,1,$sbit
add $acc1,$car1,$car1
srlx $car0,1,$car0
mov 12,$j
st $car1,[%sp+$bias+$frame] ! tp[0]=
srlx $car1,32,$car1
add %sp,$bias+$frame+4,$tp
.Lsqr_2nd:
mulx $apj,$mul0,$acc0
mulx $npj,$mul1,$acc1
add $acc0,$car0,$car0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $tpj,$sbit,$sbit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
ld [$ap+$j],$apj ! ap[j]
and $car0,$mask,$acc0
ld [$np+$j],$npj ! np[j]
srlx $car0,32,$car0
add $acc1,$car1,$car1
ld [$tp+8],$tpj ! tp[j]
add $acc0,$acc0,$acc0
add $j,4,$j ! j++
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$acc0,$acc0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
srlx $acc0,32,$sbit
and $acc0,$mask,$acc0
cmp $j,$num
add $acc0,$car1,$car1
st $car1,[$tp] ! tp[j-1]
srlx $car1,32,$car1
bl %icc,.Lsqr_2nd
add $tp,4,$tp ! tp++
!.Lsqr_2nd
mulx $apj,$mul0,$acc0
mulx $npj,$mul1,$acc1
add $acc0,$car0,$car0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $tpj,$sbit,$sbit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
and $car0,$mask,$acc0
srlx $car0,32,$car0
add $acc1,$car1,$car1
add $acc0,$acc0,$acc0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$acc0,$acc0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
srlx $acc0,32,$sbit
and $acc0,$mask,$acc0
add $acc0,$car1,$car1
st $car1,[$tp] ! tp[j-1]
srlx $car1,32,$car1
add $car0,$car0,$car0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$car0,$car0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $car0,$car1,$car1
add $car2,$car1,$car1
st $car1,[$tp+4]
srlx $car1,32,$car2
ld [%sp+$bias+$frame],$tmp1 ! tp[0]
ld [%sp+$bias+$frame+4],$tpj ! tp[1]
ld [$ap+8],$mul0 ! ap[2]
ld [$np],$car1 ! np[0]
ld [$np+4],$npj ! np[1]
mulx $n0,$tmp1,$mul1
and $mul1,$mask,$mul1
mov 8,$i
mulx $mul0,$mul0,$car0
mulx $car1,$mul1,$car1
and $car0,$mask,$acc0
add $tmp1,$car1,$car1
srlx $car0,32,$car0
add %sp,$bias+$frame,$tp
srlx $car1,32,$car1
and $car0,1,$sbit
srlx $car0,1,$car0
mov 4,$j
.Lsqr_outer:
.Lsqr_inner1:
mulx $npj,$mul1,$acc1
add $tpj,$car1,$car1
add $j,4,$j
ld [$tp+8],$tpj
cmp $j,$i
add $acc1,$car1,$car1
ld [$np+$j],$npj
st $car1,[$tp]
srlx $car1,32,$car1
bl %icc,.Lsqr_inner1
add $tp,4,$tp
!.Lsqr_inner1
add $j,4,$j
ld [$ap+$j],$apj ! ap[j]
mulx $npj,$mul1,$acc1
add $tpj,$car1,$car1
ld [$np+$j],$npj ! np[j]
bn/asm/sparcv9-mont.pl: iron another glitch in squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/6440)
2018-06-08 21:02:39 +08:00
srlx $car1,32,$tmp0
and $car1,$mask,$car1
add $tmp0,$sbit,$sbit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $acc0,$car1,$car1
ld [$tp+8],$tpj ! tp[j]
add $acc1,$car1,$car1
st $car1,[$tp]
srlx $car1,32,$car1
add $j,4,$j
cmp $j,$num
be,pn %icc,.Lsqr_no_inner2
add $tp,4,$tp
.Lsqr_inner2:
mulx $apj,$mul0,$acc0
mulx $npj,$mul1,$acc1
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $tpj,$sbit,$sbit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $acc0,$car0,$car0
ld [$ap+$j],$apj ! ap[j]
and $car0,$mask,$acc0
ld [$np+$j],$npj ! np[j]
srlx $car0,32,$car0
add $acc0,$acc0,$acc0
ld [$tp+8],$tpj ! tp[j]
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$acc0,$acc0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $j,4,$j ! j++
srlx $acc0,32,$sbit
and $acc0,$mask,$acc0
cmp $j,$num
add $acc0,$car1,$car1
add $acc1,$car1,$car1
st $car1,[$tp] ! tp[j-1]
srlx $car1,32,$car1
bl %icc,.Lsqr_inner2
add $tp,4,$tp ! tp++
.Lsqr_no_inner2:
mulx $apj,$mul0,$acc0
mulx $npj,$mul1,$acc1
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $tpj,$sbit,$sbit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $acc0,$car0,$car0
and $car0,$mask,$acc0
srlx $car0,32,$car0
add $acc0,$acc0,$acc0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$acc0,$acc0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
srlx $acc0,32,$sbit
and $acc0,$mask,$acc0
add $acc0,$car1,$car1
add $acc1,$car1,$car1
st $car1,[$tp] ! tp[j-1]
srlx $car1,32,$car1
add $car0,$car0,$car0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$car0,$car0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $car0,$car1,$car1
add $car2,$car1,$car1
st $car1,[$tp+4]
srlx $car1,32,$car2
add $i,4,$i ! i++
ld [%sp+$bias+$frame],$tmp1 ! tp[0]
ld [%sp+$bias+$frame+4],$tpj ! tp[1]
ld [$ap+$i],$mul0 ! ap[j]
ld [$np],$car1 ! np[0]
ld [$np+4],$npj ! np[1]
mulx $n0,$tmp1,$mul1
and $mul1,$mask,$mul1
add $i,4,$tmp0
mulx $mul0,$mul0,$car0
mulx $car1,$mul1,$car1
and $car0,$mask,$acc0
add $tmp1,$car1,$car1
srlx $car0,32,$car0
add %sp,$bias+$frame,$tp
srlx $car1,32,$car1
and $car0,1,$sbit
srlx $car0,1,$car0
cmp $tmp0,$num ! i<num-1
bl %icc,.Lsqr_outer
mov 4,$j
.Lsqr_last:
mulx $npj,$mul1,$acc1
add $tpj,$car1,$car1
add $j,4,$j
ld [$tp+8],$tpj
cmp $j,$i
add $acc1,$car1,$car1
ld [$np+$j],$npj
st $car1,[$tp]
srlx $car1,32,$car1
bl %icc,.Lsqr_last
add $tp,4,$tp
!.Lsqr_last
mulx $npj,$mul1,$acc1
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $tpj,$acc0,$acc0
srlx $acc0,32,$tmp0
and $acc0,$mask,$acc0
add $tmp0,$sbit,$sbit
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $acc0,$car1,$car1
add $acc1,$car1,$car1
st $car1,[$tp]
srlx $car1,32,$car1
add $car0,$car0,$car0 ! recover $car0
bn/asm/sparcv9-mont.pl: fix squaring code path. This module is used only with odd input lengths, i.e. not used in normal PKI cases, on contemporary processors. The problem was "illuminated" by fuzzing tests. Reviewed-by: Richard Levitte <levitte@openssl.org>
2017-03-22 17:51:25 +08:00
add $sbit,$car0,$car0
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
add $car0,$car1,$car1
add $car2,$car1,$car1
st $car1,[$tp+4]
srlx $car1,32,$car2
ba .Ltail
add $tp,8,$tp
.type $fname,#function
.size $fname,(.-$fname)
Many spelling fixes/typo's corrected. Around 138 distinct errors found and fixed; thanks! Reviewed-by: Kurt Roeckx <kurt@roeckx.be> Reviewed-by: Tim Hudson <tjh@openssl.org> Reviewed-by: Rich Salz <rsalz@openssl.org> (Merged from https://github.com/openssl/openssl/pull/3459)
2017-11-12 08:03:10 +08:00
.asciz "Montgomery Multiplication for SPARCv9, CRYPTOGAMS by <appro\@openssl.org>"
SPARC Solaris and Linux assemblers treat .align directive differently. PR: 1547
2007-06-20 20:24:22 +08:00
.align 32
Add IALU-only bn_mul_mont for SPARCv9. See commentary section for details.
2005-12-16 06:43:33 +08:00
___
$code =~ s/\`([^\`]*)\`/eval($1)/gem;
print $code;
Also check for errors in x86_64-xlate.pl. In https://github.com/openssl/openssl/pull/10883, I'd meant to exclude the perlasm drivers since they aren't opening pipes and do not particularly need it, but I only noticed x86_64-xlate.pl, so arm-xlate.pl and ppc-xlate.pl got the change. That seems to have been fine, so be consistent and also apply the change to x86_64-xlate.pl. Checking for errors is generally a good idea. Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: David Benjamin <davidben@google.com> (Merged from https://github.com/openssl/openssl/pull/10930)
2020-02-17 10:17:53 +08:00
close STDOUT or die "error closing STDOUT: $!";
Reference in New Issue Copy Permalink