[thirdparty/openssl.git] / crypto / x86_64cpuid.pl

#! /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;
$output  = shift;
if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }

$win64=0; $win64=1 if ($flavour =~ /[nm]asm|mingw64/ || $output =~ /\.asm$/);

$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
( $xlate="${dir}perlasm/x86_64-xlate.pl" and -f $xlate) or
die "can't locate x86_64-xlate.pl";

open OUT,"| \"$^X\" \"$xlate\" $flavour \"$output\"";
*STDOUT=*OUT;

($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") :	# Win64 order
				 ("%rdi","%rsi","%rdx","%rcx");	# Unix order

print<<___;
.extern		OPENSSL_cpuid_setup
.hidden		OPENSSL_cpuid_setup
.section	.init
	call	OPENSSL_cpuid_setup

.hidden	OPENSSL_ia32cap_P
.comm	OPENSSL_ia32cap_P,16,4

.text

.globl	OPENSSL_atomic_add
.type	OPENSSL_atomic_add,\@abi-omnipotent
.align	16
OPENSSL_atomic_add:
	movl	($arg1),%eax
.Lspin:	leaq	($arg2,%rax),%r8
	.byte	0xf0		# lock
	cmpxchgl	%r8d,($arg1)
	jne	.Lspin
	movl	%r8d,%eax
	.byte	0x48,0x98	# cltq/cdqe
	ret
.size	OPENSSL_atomic_add,.-OPENSSL_atomic_add

.globl	OPENSSL_rdtsc
.type	OPENSSL_rdtsc,\@abi-omnipotent
.align	16
OPENSSL_rdtsc:
	rdtsc
	shl	\$32,%rdx
	or	%rdx,%rax
	ret
.size	OPENSSL_rdtsc,.-OPENSSL_rdtsc

.globl	OPENSSL_ia32_cpuid
.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

	xor	%eax,%eax
	cmp	\$0x756e6547,%ebx	# "Genu"
	setne	%al
	mov	%eax,%r9d
	cmp	\$0x49656e69,%edx	# "ineI"
	setne	%al
	or	%eax,%r9d
	cmp	\$0x6c65746e,%ecx	# "ntel"
	setne	%al
	or	%eax,%r9d		# 0 indicates Intel CPU
	jz	.Lintel

	cmp	\$0x68747541,%ebx	# "Auth"
	setne	%al
	mov	%eax,%r10d
	cmp	\$0x69746E65,%edx	# "enti"
	setne	%al
	or	%eax,%r10d
	cmp	\$0x444D4163,%ecx	# "cAMD"
	setne	%al
	or	%eax,%r10d		# 0 indicates AMD CPU
	jnz	.Lintel

	# AMD specific
	mov	\$0x80000000,%eax
	cpuid
	cmp	\$0x80000001,%eax
	jb	.Lintel
	mov	%eax,%r10d
	mov	\$0x80000001,%eax
	cpuid
	or	%ecx,%r9d
	and	\$0x00000801,%r9d	# isolate AMD XOP bit, 1<<11

	cmp	\$0x80000008,%r10d
	jb	.Lintel

	mov	\$0x80000008,%eax
	cpuid
	movzb	%cl,%r10		# number of cores - 1
	inc	%r10			# number of cores

	mov	\$1,%eax
	cpuid
	bt	\$28,%edx		# test hyper-threading bit
	jnc	.Lgeneric
	shr	\$16,%ebx		# number of logical processors
	cmp	%r10b,%bl
	ja	.Lgeneric
	and	\$0xefffffff,%edx	# ~(1<<28)
	jmp	.Lgeneric

.Lintel:
	cmp	\$4,%r11d
	mov	\$-1,%r10d
	jb	.Lnocacheinfo

	mov	\$4,%eax
	mov	\$0,%ecx		# query L1D
	cpuid
	mov	%eax,%r10d
	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
	and	\$0xbfefffff,%edx	# force reserved bits to 0
	cmp	\$0,%r9d
	jne	.Lnotintel
	or	\$0x40000000,%edx	# set reserved bit#30 on Intel CPUs
	and	\$15,%ah
	cmp	\$15,%ah		# examine Family ID
	jne	.Lnotintel
	or	\$0x00100000,%edx	# set reserved bit#20 to engage RC4_CHAR
.Lnotintel:
	bt	\$28,%edx		# test hyper-threading bit
	jnc	.Lgeneric
	and	\$0xefffffff,%edx	# ~(1<<28)
	cmp	\$0,%r10d
	je	.Lgeneric

	or	\$0x10000000,%edx	# 1<<28
	shr	\$16,%ebx
	cmp	\$1,%bl			# see if cache is shared
	ja	.Lgeneric
	and	\$0xefffffff,%edx	# ~(1<<28)
.Lgeneric:
	and	\$0x00000800,%r9d	# isolate AMD XOP flag
	and	\$0xfffff7ff,%ecx
	or	%ecx,%r9d		# merge AMD XOP flag

	mov	%edx,%r10d		# %r9d:%r10d is copy of %ecx:%edx
	bt	\$27,%r9d		# check OSXSAVE bit
	jnc	.Lclear_avx
	xor	%ecx,%ecx		# XCR0
	.byte	0x0f,0x01,0xd0		# xgetbv
	and	\$6,%eax		# isolate XMM and YMM state support
	cmp	\$6,%eax
	je	.Ldone
.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
	mov	%r8,%rbx		# restore %rbx
	or	%r9,%rax
	ret
.size	OPENSSL_ia32_cpuid,.-OPENSSL_ia32_cpuid

.globl  OPENSSL_cleanse
.type   OPENSSL_cleanse,\@abi-omnipotent
.align  16
OPENSSL_cleanse:
	xor	%rax,%rax
	cmp	\$15,$arg2
	jae	.Lot
	cmp	\$0,$arg2
	je	.Lret
.Little:
	mov	%al,($arg1)
	sub	\$1,$arg2
	lea	1($arg1),$arg1
	jnz	.Little
.Lret:
	ret
.align	16
.Lot:
	test	\$7,$arg1
	jz	.Laligned
	mov	%al,($arg1)
	lea	-1($arg2),$arg2
	lea	1($arg1),$arg1
	jmp	.Lot
.Laligned:
	mov	%rax,($arg1)
	lea	-8($arg2),$arg2
	test	\$-8,$arg2
	lea	8($arg1),$arg1
	jnz	.Laligned
	cmp	\$0,$arg2
	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 (!$win64);
.globl	OPENSSL_wipe_cpu
.type	OPENSSL_wipe_cpu,\@abi-omnipotent
.align	16
OPENSSL_wipe_cpu:
	pxor	%xmm0,%xmm0
	pxor	%xmm1,%xmm1
	pxor	%xmm2,%xmm2
	pxor	%xmm3,%xmm3
	pxor	%xmm4,%xmm4
	pxor	%xmm5,%xmm5
	pxor	%xmm6,%xmm6
	pxor	%xmm7,%xmm7
	pxor	%xmm8,%xmm8
	pxor	%xmm9,%xmm9
	pxor	%xmm10,%xmm10
	pxor	%xmm11,%xmm11
	pxor	%xmm12,%xmm12
	pxor	%xmm13,%xmm13
	pxor	%xmm14,%xmm14
	pxor	%xmm15,%xmm15
	xorq	%rcx,%rcx
	xorq	%rdx,%rdx
	xorq	%rsi,%rsi
	xorq	%rdi,%rdi
	xorq	%r8,%r8
	xorq	%r9,%r9
	xorq	%r10,%r10
	xorq	%r11,%r11
	leaq	8(%rsp),%rax
	ret
.size	OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
___
print<<___ if ($win64);
.globl	OPENSSL_wipe_cpu
.type	OPENSSL_wipe_cpu,\@abi-omnipotent
.align	16
OPENSSL_wipe_cpu:
	pxor	%xmm0,%xmm0
	pxor	%xmm1,%xmm1
	pxor	%xmm2,%xmm2
	pxor	%xmm3,%xmm3
	pxor	%xmm4,%xmm4
	pxor	%xmm5,%xmm5
	xorq	%rcx,%rcx
	xorq	%rdx,%rdx
	xorq	%r8,%r8
	xorq	%r9,%r9
	xorq	%r10,%r10
	xorq	%r11,%r11
	leaq	8(%rsp),%rax
	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
___
}

print<<___;
.globl	OPENSSL_ia32_rdrand
.type	OPENSSL_ia32_rdrand,\@abi-omnipotent
.align	16
OPENSSL_ia32_rdrand:
	mov	\$8,%ecx
.Loop_rdrand:
	rdrand	%rax
	jc	.Lbreak_rdrand
	loop	.Loop_rdrand
.Lbreak_rdrand:
	cmp	\$0,%rax
	cmove	%rcx,%rax
	ret
.size	OPENSSL_ia32_rdrand,.-OPENSSL_ia32_rdrand

.globl	OPENSSL_ia32_rdseed
.type	OPENSSL_ia32_rdseed,\@abi-omnipotent
.align	16
OPENSSL_ia32_rdseed:
	mov	\$8,%ecx
.Loop_rdseed:
	rdseed	%rax
	jc	.Lbreak_rdseed
	loop	.Loop_rdseed
.Lbreak_rdseed:
	cmp	\$0,%rax
	cmove	%rcx,%rax
	ret
.size	OPENSSL_ia32_rdseed,.-OPENSSL_ia32_rdseed
___

close STDOUT;	# flush
Commit	Line	Data
e0a65194 RS	1	#! /usr/bin/env perl
	2	# Copyright 2005-2016 The OpenSSL Project Authors. All Rights Reserved.
	3	#
	4	# Licensed under the OpenSSL license (the "License"). You may not use
	5	# this file except in compliance with the License. You can obtain a copy
	6	# in the file LICENSE in the source distribution or at
	7	# https://www.openssl.org/source/license.html
	8
14e21f86	9
aa8f38e4 AP	10	$flavour = shift;
	11	$output = shift;
	12	if ($flavour =~ /\./) { $output = $flavour; undef $flavour; }
14e21f86	13
aa8f38e4	14	$win64=0; $win64=1 if ($flavour =~ /[nm]asm\|mingw64/ \|\| $output =~ /\.asm$/);
376729e1	15
aa8f38e4	16	$0 =~ m/(.*[\/\\])[^\/\\]+$/; $dir=$1;
301799b8 AP	17	( $xlate="${dir}x86_64-xlate.pl" and -f $xlate ) or
	18	( $xlate="${dir}perlasm/x86_64-xlate.pl" and -f $xlate) or
	19	die "can't locate x86_64-xlate.pl";
	20
cfe1d992	21	open OUT,"\| \"$^X\" \"$xlate\" $flavour \"$output\"";
46bf83f0	22	STDOUT=OUT;
376729e1	23
5fabb88a AP	24	($arg1,$arg2,$arg3,$arg4)=$win64?("%rcx","%rdx","%r8", "%r9") : # Win64 order
	25	("%rdi","%rsi","%rdx","%rcx"); # Unix order
	26
aa8f38e4 AP	27	print<<___;
aa8f38e4 AP	28	.extern OPENSSL_cpuid_setup
ddc20d4d	29	.hidden OPENSSL_cpuid_setup
aa8f38e4 AP	30	.section .init
aa8f38e4 AP	31	call OPENSSL_cpuid_setup
932cc129	32
ddc20d4d	33	.hidden OPENSSL_ia32cap_P
c5cd28bd	34	.comm OPENSSL_ia32cap_P,16,4
ddc20d4d	35
14e21f86	36	.text
376729e1 AP	37
376729e1 AP	38	.globl OPENSSL_atomic_add
aa8f38e4	39	.type OPENSSL_atomic_add,\@abi-omnipotent
376729e1 AP	40	.align 16
376729e1 AP	41	OPENSSL_atomic_add:
aa8f38e4 AP	42	movl ($arg1),%eax
	43	.Lspin: leaq ($arg2,%rax),%r8
	44	.byte 0xf0 # lock
	45	cmpxchgl %r8d,($arg1)
376729e1	46	jne .Lspin
e442c362	47	movl %r8d,%eax
aa8f38e4	48	.byte 0x48,0x98 # cltq/cdqe
376729e1 AP	49	ret
	50	.size OPENSSL_atomic_add,.-OPENSSL_atomic_add
	51
932cc129 AP	52	.globl OPENSSL_rdtsc
	53	.type OPENSSL_rdtsc,\@abi-omnipotent
	54	.align 16
	55	OPENSSL_rdtsc:
	56	rdtsc
	57	shl \$32,%rdx
	58	or %rdx,%rax
	59	ret
	60	.size OPENSSL_rdtsc,.-OPENSSL_rdtsc
	61
376729e1	62	.globl OPENSSL_ia32_cpuid
c5cd28bd	63	.type OPENSSL_ia32_cpuid,\@function,1
376729e1 AP	64	.align 16
376729e1 AP	65	OPENSSL_ia32_cpuid:
b9064221	66	mov %rbx,%r8 # save %rbx
9babf392 AP	67
9babf392 AP	68	xor %eax,%eax
c5cd28bd	69	mov %eax,8(%rdi) # clear 3rd word
9babf392	70	cpuid
761393bb AP	71	mov %eax,%r11d # max value for standard query level
761393bb AP	72
9babf392 AP	73	xor %eax,%eax
	74	cmp \$0x756e6547,%ebx # "Genu"
	75	setne %al
	76	mov %eax,%r9d
	77	cmp \$0x49656e69,%edx # "ineI"
	78	setne %al
	79	or %eax,%r9d
	80	cmp \$0x6c65746e,%ecx # "ntel"
	81	setne %al
5cd91b50	82	or %eax,%r9d # 0 indicates Intel CPU
5cd91b50 AP	83	jz .Lintel
	84
	85	cmp \$0x68747541,%ebx # "Auth"
	86	setne %al
	87	mov %eax,%r10d
	88	cmp \$0x69746E65,%edx # "enti"
	89	setne %al
	90	or %eax,%r10d
	91	cmp \$0x444D4163,%ecx # "cAMD"
	92	setne %al
	93	or %eax,%r10d # 0 indicates AMD CPU
	94	jnz .Lintel
	95
761393bb	96	# AMD specific
5cd91b50 AP	97	mov \$0x80000000,%eax
5cd91b50 AP	98	cpuid
b9064221 AP	99	cmp \$0x80000001,%eax
	100	jb .Lintel
	101	mov %eax,%r10d
	102	mov \$0x80000001,%eax
	103	cpuid
	104	or %ecx,%r9d
	105	and \$0x00000801,%r9d # isolate AMD XOP bit, 1<<11
	106
	107	cmp \$0x80000008,%r10d
5cd91b50 AP	108	jb .Lintel
	109
	110	mov \$0x80000008,%eax
	111	cpuid
	112	movzb %cl,%r10 # number of cores - 1
	113	inc %r10 # number of cores
9babf392	114
761393bb AP	115	mov \$1,%eax
	116	cpuid
	117	bt \$28,%edx # test hyper-threading bit
b9064221	118	jnc .Lgeneric
761393bb AP	119	shr \$16,%ebx # number of logical processors
761393bb AP	120	cmp %r10b,%bl
b9064221	121	ja .Lgeneric
761393bb	122	and \$0xefffffff,%edx # ~(1<<28)
b9064221	123	jmp .Lgeneric
761393bb	124
5cd91b50	125	.Lintel:
761393bb AP	126	cmp \$4,%r11d
	127	mov \$-1,%r10d
	128	jb .Lnocacheinfo
	129
	130	mov \$4,%eax
	131	mov \$0,%ecx # query L1D
	132	cpuid
	133	mov %eax,%r10d
	134	shr \$14,%r10d
	135	and \$0xfff,%r10d # number of cores -1 per L1D
	136
c5cd28bd AP	137	cmp \$7,%r11d
	138	jb .Lnocacheinfo
	139
	140	mov \$7,%eax
	141	xor %ecx,%ecx
	142	cpuid
	143	mov %ebx,8(%rdi)
	144
761393bb	145	.Lnocacheinfo:
932cc129	146	mov \$1,%eax
376729e1	147	cpuid
4bb90087	148	and \$0xbfefffff,%edx # force reserved bits to 0
932cc129	149	cmp \$0,%r9d
9babf392	150	jne .Lnotintel
4bb90087	151	or \$0x40000000,%edx # set reserved bit#30 on Intel CPUs
9babf392 AP	152	and \$15,%ah
9babf392 AP	153	cmp \$15,%ah # examine Family ID
4bb90087 AP	154	jne .Lnotintel
4bb90087 AP	155	or \$0x00100000,%edx # set reserved bit#20 to engage RC4_CHAR
9babf392	156	.Lnotintel:
3df2eff4	157	bt \$28,%edx # test hyper-threading bit
b9064221	158	jnc .Lgeneric
761393bb AP	159	and \$0xefffffff,%edx # ~(1<<28)
761393bb AP	160	cmp \$0,%r10d
b9064221	161	je .Lgeneric
761393bb AP	162
761393bb AP	163	or \$0x10000000,%edx # 1<<28
9babf392	164	shr \$16,%ebx
761393bb	165	cmp \$1,%bl # see if cache is shared
b9064221	166	ja .Lgeneric
932cc129	167	and \$0xefffffff,%edx # ~(1<<28)
b9064221 AP	168	.Lgeneric:
	169	and \$0x00000800,%r9d # isolate AMD XOP flag
	170	and \$0xfffff7ff,%ecx
2bc3ad28	171	or %ecx,%r9d # merge AMD XOP flag
b9064221	172
2bc3ad28 AP	173	mov %edx,%r10d # %r9d:%r10d is copy of %ecx:%edx
2bc3ad28 AP	174	bt \$27,%r9d # check OSXSAVE bit
b9064221 AP	175	jnc .Lclear_avx
	176	xor %ecx,%ecx # XCR0
	177	.byte 0x0f,0x01,0xd0 # xgetbv
	178	and \$6,%eax # isolate XMM and YMM state support
	179	cmp \$6,%eax
	180	je .Ldone
	181	.Lclear_avx:
	182	mov \$0xefffe7ff,%eax # ~(1<<28\|1<<12\|1<<11)
2bc3ad28	183	and %eax,%r9d # clear AVX, FMA and AMD XOP bits
c5cd28bd	184	andl \$0xffffffdf,8(%rdi) # cleax AVX2, ~(1<<5)
b9064221	185	.Ldone:
2bc3ad28 AP	186	shl \$32,%r9
2bc3ad28 AP	187	mov %r10d,%eax
b9064221	188	mov %r8,%rbx # restore %rbx
2bc3ad28	189	or %r9,%rax
376729e1 AP	190	ret
376729e1 AP	191	.size OPENSSL_ia32_cpuid,.-OPENSSL_ia32_cpuid
b2dba9bf AP	192
b2dba9bf AP	193	.globl OPENSSL_cleanse
aa8f38e4	194	.type OPENSSL_cleanse,\@abi-omnipotent
b2dba9bf AP	195	.align 16
	196	OPENSSL_cleanse:
	197	xor %rax,%rax
aa8f38e4	198	cmp \$15,$arg2
b2dba9bf	199	jae .Lot
7676eebf AP	200	cmp \$0,$arg2
7676eebf AP	201	je .Lret
b2dba9bf	202	.Little:
aa8f38e4 AP	203	mov %al,($arg1)
	204	sub \$1,$arg2
	205	lea 1($arg1),$arg1
b2dba9bf	206	jnz .Little
1fd79f66 AP	207	.Lret:
1fd79f66 AP	208	ret
b2dba9bf AP	209	.align 16
b2dba9bf AP	210	.Lot:
aa8f38e4	211	test \$7,$arg1
b2dba9bf	212	jz .Laligned
aa8f38e4 AP	213	mov %al,($arg1)
	214	lea -1($arg2),$arg2
	215	lea 1($arg1),$arg1
b2dba9bf AP	216	jmp .Lot
b2dba9bf AP	217	.Laligned:
aa8f38e4 AP	218	mov %rax,($arg1)
	219	lea -8($arg2),$arg2
	220	test \$-8,$arg2
	221	lea 8($arg1),$arg1
b2dba9bf	222	jnz .Laligned
aa8f38e4	223	cmp \$0,$arg2
b2dba9bf AP	224	jne .Little
	225	ret
	226	.size OPENSSL_cleanse,.-OPENSSL_cleanse
e33826f0 AP	227
	228	.globl CRYPTO_memcmp
	229	.type CRYPTO_memcmp,\@abi-omnipotent
	230	.align 16
	231	CRYPTO_memcmp:
	232	xor %rax,%rax
	233	xor %r10,%r10
	234	cmp \$0,$arg3
	235	je .Lno_data
	236	.Loop_cmp:
	237	mov ($arg1),%r10b
	238	lea 1($arg1),$arg1
	239	xor ($arg2),%r10b
	240	lea 1($arg2),$arg2
	241	or %r10b,%al
	242	dec $arg3
	243	jnz .Loop_cmp
	244	neg %rax
	245	shr \$63,%rax
	246	.Lno_data:
	247	ret
	248	.size CRYPTO_memcmp,.-CRYPTO_memcmp
14e21f86	249	___
aa8f38e4 AP	250
	251	print<<___ if (!$win64);
	252	.globl OPENSSL_wipe_cpu
	253	.type OPENSSL_wipe_cpu,\@abi-omnipotent
	254	.align 16
	255	OPENSSL_wipe_cpu:
	256	pxor %xmm0,%xmm0
	257	pxor %xmm1,%xmm1
	258	pxor %xmm2,%xmm2
	259	pxor %xmm3,%xmm3
	260	pxor %xmm4,%xmm4
	261	pxor %xmm5,%xmm5
	262	pxor %xmm6,%xmm6
	263	pxor %xmm7,%xmm7
	264	pxor %xmm8,%xmm8
	265	pxor %xmm9,%xmm9
	266	pxor %xmm10,%xmm10
	267	pxor %xmm11,%xmm11
	268	pxor %xmm12,%xmm12
	269	pxor %xmm13,%xmm13
	270	pxor %xmm14,%xmm14
	271	pxor %xmm15,%xmm15
	272	xorq %rcx,%rcx
	273	xorq %rdx,%rdx
	274	xorq %rsi,%rsi
	275	xorq %rdi,%rdi
	276	xorq %r8,%r8
	277	xorq %r9,%r9
	278	xorq %r10,%r10
	279	xorq %r11,%r11
	280	leaq 8(%rsp),%rax
	281	ret
	282	.size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
	283	___
	284	print<<___ if ($win64);
	285	.globl OPENSSL_wipe_cpu
	286	.type OPENSSL_wipe_cpu,\@abi-omnipotent
	287	.align 16
	288	OPENSSL_wipe_cpu:
	289	pxor %xmm0,%xmm0
	290	pxor %xmm1,%xmm1
	291	pxor %xmm2,%xmm2
	292	pxor %xmm3,%xmm3
	293	pxor %xmm4,%xmm4
	294	pxor %xmm5,%xmm5
	295	xorq %rcx,%rcx
	296	xorq %rdx,%rdx
	297	xorq %r8,%r8
	298	xorq %r9,%r9
	299	xorq %r10,%r10
	300	xorq %r11,%r11
	301	leaq 8(%rsp),%rax
	302	ret
	303	.size OPENSSL_wipe_cpu,.-OPENSSL_wipe_cpu
	304	___
5fabb88a AP	305	{
	306	my $out="%r10";
	307	my $cnt="%rcx";
	308	my $max="%r11";
	309	my $lasttick="%r8d";
	310	my $lastdiff="%r9d";
	311	my $redzone=win64?8:-8;
	312
	313	print<<___;
	314	.globl OPENSSL_instrument_bus
	315	.type OPENSSL_instrument_bus,\@abi-omnipotent
	316	.align 16
	317	OPENSSL_instrument_bus:
	318	mov $arg1,$out # tribute to Win64
	319	mov $arg2,$cnt
	320	mov $arg2,$max
	321
	322	rdtsc # collect 1st tick
	323	mov %eax,$lasttick # lasttick = tick
	324	mov \$0,$lastdiff # lastdiff = 0
	325	clflush ($out)
b9064221	326	.byte 0xf0 # lock
5fabb88a AP	327	add $lastdiff,($out)
	328	jmp .Loop
	329	.align 16
	330	.Loop: rdtsc
	331	mov %eax,%edx
	332	sub $lasttick,%eax
	333	mov %edx,$lasttick
	334	mov %eax,$lastdiff
	335	clflush ($out)
b9064221	336	.byte 0xf0 # lock
5fabb88a AP	337	add %eax,($out)
	338	lea 4($out),$out
	339	sub \$1,$cnt
	340	jnz .Loop
	341
	342	mov $max,%rax
	343	ret
	344	.size OPENSSL_instrument_bus,.-OPENSSL_instrument_bus
	345
	346	.globl OPENSSL_instrument_bus2
	347	.type OPENSSL_instrument_bus2,\@abi-omnipotent
	348	.align 16
	349	OPENSSL_instrument_bus2:
	350	mov $arg1,$out # tribute to Win64
	351	mov $arg2,$cnt
	352	mov $arg3,$max
	353	mov $cnt,$redzone(%rsp)
	354
	355	rdtsc # collect 1st tick
	356	mov %eax,$lasttick # lasttick = tick
	357	mov \$0,$lastdiff # lastdiff = 0
	358
	359	clflush ($out)
b9064221	360	.byte 0xf0 # lock
5fabb88a AP	361	add $lastdiff,($out)
	362
	363	rdtsc # collect 1st diff
	364	mov %eax,%edx
	365	sub $lasttick,%eax # diff
	366	mov %edx,$lasttick # lasttick = tick
	367	mov %eax,$lastdiff # lastdiff = diff
	368	.Loop2:
	369	clflush ($out)
b9064221	370	.byte 0xf0 # lock
5fabb88a AP	371	add %eax,($out) # accumulate diff
	372
	373	sub \$1,$max
	374	jz .Ldone2
	375
	376	rdtsc
	377	mov %eax,%edx
	378	sub $lasttick,%eax # diff
	379	mov %edx,$lasttick # lasttick = tick
	380	cmp $lastdiff,%eax
	381	mov %eax,$lastdiff # lastdiff = diff
	382	mov \$0,%edx
	383	setne %dl
	384	sub %rdx,$cnt # conditional --$cnt
	385	lea ($out,%rdx,4),$out # conditional ++$out
	386	jnz .Loop2
	387
	388	.Ldone2:
	389	mov $redzone(%rsp),%rax
	390	sub $cnt,%rax
	391	ret
	392	.size OPENSSL_instrument_bus2,.-OPENSSL_instrument_bus2
	393	___
	394	}
aa8f38e4	395
301799b8 AP	396	print<<___;
	397	.globl OPENSSL_ia32_rdrand
	398	.type OPENSSL_ia32_rdrand,\@abi-omnipotent
	399	.align 16
	400	OPENSSL_ia32_rdrand:
	401	mov \$8,%ecx
	402	.Loop_rdrand:
	403	rdrand %rax
	404	jc .Lbreak_rdrand
	405	loop .Loop_rdrand
	406	.Lbreak_rdrand:
	407	cmp \$0,%rax
	408	cmove %rcx,%rax
	409	ret
4d01f276	410	.size OPENSSL_ia32_rdrand,.-OPENSSL_ia32_rdrand
f4d45640 AP	411
	412	.globl OPENSSL_ia32_rdseed
	413	.type OPENSSL_ia32_rdseed,\@abi-omnipotent
	414	.align 16
	415	OPENSSL_ia32_rdseed:
	416	mov \$8,%ecx
	417	.Loop_rdseed:
	418	rdseed %rax
	419	jc .Lbreak_rdseed
	420	loop .Loop_rdseed
	421	.Lbreak_rdseed:
	422	cmp \$0,%rax
	423	cmove %rcx,%rax
	424	ret
	425	.size OPENSSL_ia32_rdseed,.-OPENSSL_ia32_rdseed
301799b8 AP	426	___
301799b8 AP	427
5d863367	428	close STDOUT; # flush