aboutsummaryrefslogtreecommitdiff
path: root/crypto/sha/asm/sha1-s390x.pl
blob: 653a6cb50a8a8efbf662a0f0a291be63d53f2670 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
#! /usr/bin/env perl
# Copyright 2007-2020 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


# ====================================================================
# Written by Andy Polyakov <appro@openssl.org> for the OpenSSL
# 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/.
# ====================================================================

# SHA1 block procedure for s390x.

# April 2007.
#
# Performance is >30% better than gcc 3.3 generated code. But the real
# twist is that SHA1 hardware support is detected and utilized. In
# which case performance can reach further >4.5x for larger chunks.

# January 2009.
#
# Optimize Xupdate for amount of memory references and reschedule
# instructions to favour dual-issue z10 pipeline. On z10 hardware is
# "only" ~2.3x faster than software.

# November 2010.
#
# Adapt for -m31 build. If kernel supports what's called "highgprs"
# feature on Linux [see /proc/cpuinfo], it's possible to use 64-bit
# instructions and achieve "64-bit" performance even in 31-bit legacy
# application context. The feature is not specific to any particular
# processor, as long as it's "z-CPU". Latter implies that the code
# remains z/Architecture specific. On z990 it was measured to perform
# 23% better than code generated by gcc 4.3.

$kimdfunc=1;	# magic function code for kimd instruction

$flavour = shift;

if ($flavour =~ /3[12]/) {
	$SIZE_T=4;
	$g="";
} else {
	$SIZE_T=8;
	$g="g";
}

while (($output=shift) && ($output!~/\w[\w\-]*\.\w+$/)) {}
open STDOUT,">$output";

$K_00_39="%r0"; $K=$K_00_39;
$K_40_79="%r1";
$ctx="%r2";	$prefetch="%r2";
$inp="%r3";
$len="%r4";

$A="%r5";
$B="%r6";
$C="%r7";
$D="%r8";
$E="%r9";	@V=($A,$B,$C,$D,$E);
$t0="%r10";
$t1="%r11";
@X=("%r12","%r13","%r14");
$sp="%r15";

$stdframe=16*$SIZE_T+4*8;
$frame=$stdframe+16*4;

sub Xupdate {
my $i=shift;

$code.=<<___ if ($i==15);
	lg	$prefetch,$stdframe($sp)	### Xupdate(16) warm-up
	lr	$X[0],$X[2]
___
return if ($i&1);	# Xupdate is vectorized and executed every 2nd cycle
$code.=<<___ if ($i<16);
	lg	$X[0],`$i*4`($inp)	### Xload($i)
	rllg	$X[1],$X[0],32
___
$code.=<<___ if ($i>=16);
	xgr	$X[0],$prefetch		### Xupdate($i)
	lg	$prefetch,`$stdframe+4*(($i+2)%16)`($sp)
	xg	$X[0],`$stdframe+4*(($i+8)%16)`($sp)
	xgr	$X[0],$prefetch
	rll	$X[0],$X[0],1
	rllg	$X[1],$X[0],32
	rll	$X[1],$X[1],1
	rllg	$X[0],$X[1],32
	lr	$X[2],$X[1]		# feedback
___
$code.=<<___ if ($i<=70);
	stg	$X[0],`$stdframe+4*($i%16)`($sp)
___
unshift(@X,pop(@X));
}

sub BODY_00_19 {
my ($i,$a,$b,$c,$d,$e)=@_;
my $xi=$X[1];

	&Xupdate($i);
$code.=<<___;
	alr	$e,$K		### $i
	rll	$t1,$a,5
	lr	$t0,$d
	xr	$t0,$c
	alr	$e,$t1
	nr	$t0,$b
	alr	$e,$xi
	xr	$t0,$d
	rll	$b,$b,30
	alr	$e,$t0
___
}

sub BODY_20_39 {
my ($i,$a,$b,$c,$d,$e)=@_;
my $xi=$X[1];

	&Xupdate($i);
$code.=<<___;
	alr	$e,$K		### $i
	rll	$t1,$a,5
	lr	$t0,$b
	alr	$e,$t1
	xr	$t0,$c
	alr	$e,$xi
	xr	$t0,$d
	rll	$b,$b,30
	alr	$e,$t0
___
}

sub BODY_40_59 {
my ($i,$a,$b,$c,$d,$e)=@_;
my $xi=$X[1];

	&Xupdate($i);
$code.=<<___;
	alr	$e,$K		### $i
	rll	$t1,$a,5
	lr	$t0,$b
	alr	$e,$t1
	or	$t0,$c
	lr	$t1,$b
	nr	$t0,$d
	nr	$t1,$c
	alr	$e,$xi
	or	$t0,$t1
	rll	$b,$b,30
	alr	$e,$t0
___
}

$code.=<<___;
#include "s390x_arch.h"

.text
.align	64
.type	Ktable,\@object
Ktable: .long	0x5a827999,0x6ed9eba1,0x8f1bbcdc,0xca62c1d6
	.skip	48	#.long	0,0,0,0,0,0,0,0,0,0,0,0
.size	Ktable,.-Ktable
.globl	sha1_block_data_order
.type	sha1_block_data_order,\@function
sha1_block_data_order:
___
$code.=<<___ if ($kimdfunc);
	larl	%r1,OPENSSL_s390xcap_P
	lg	%r0,S390X_KIMD(%r1)	# check kimd capabilities
	tmhh	%r0,`0x8000>>$kimdfunc`
	jz	.Lsoftware
	lghi	%r0,$kimdfunc
	lgr	%r1,$ctx
	lgr	%r2,$inp
	sllg	%r3,$len,6
	.long	0xb93e0002	# kimd %r0,%r2
	brc	1,.-4		# pay attention to "partial completion"
	br	%r14
.align	16
.Lsoftware:
___
$code.=<<___;
	lghi	%r1,-$frame
	st${g}	$ctx,`2*$SIZE_T`($sp)
	stm${g}	%r6,%r15,`6*$SIZE_T`($sp)
	lgr	%r0,$sp
	la	$sp,0(%r1,$sp)
	st${g}	%r0,0($sp)

	larl	$t0,Ktable
	llgf	$A,0($ctx)
	llgf	$B,4($ctx)
	llgf	$C,8($ctx)
	llgf	$D,12($ctx)
	llgf	$E,16($ctx)

	lg	$K_00_39,0($t0)
	lg	$K_40_79,8($t0)

.Lloop:
	rllg	$K_00_39,$K_00_39,32
___
for ($i=0;$i<20;$i++)	{ &BODY_00_19($i,@V); unshift(@V,pop(@V)); }
$code.=<<___;
	rllg	$K_00_39,$K_00_39,32
___
for (;$i<40;$i++)	{ &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
$code.=<<___;	$K=$K_40_79;
	rllg	$K_40_79,$K_40_79,32
___
for (;$i<60;$i++)	{ &BODY_40_59($i,@V); unshift(@V,pop(@V)); }
$code.=<<___;
	rllg	$K_40_79,$K_40_79,32
___
for (;$i<80;$i++)	{ &BODY_20_39($i,@V); unshift(@V,pop(@V)); }
$code.=<<___;

	l${g}	$ctx,`$frame+2*$SIZE_T`($sp)
	la	$inp,64($inp)
	al	$A,0($ctx)
	al	$B,4($ctx)
	al	$C,8($ctx)
	al	$D,12($ctx)
	al	$E,16($ctx)
	st	$A,0($ctx)
	st	$B,4($ctx)
	st	$C,8($ctx)
	st	$D,12($ctx)
	st	$E,16($ctx)
	brct${g} $len,.Lloop

	lm${g}	%r6,%r15,`$frame+6*$SIZE_T`($sp)
	br	%r14
.size	sha1_block_data_order,.-sha1_block_data_order
.string	"SHA1 block transform for s390x, CRYPTOGAMS by <appro\@openssl.org>"
___

$code =~ s/\`([^\`]*)\`/eval $1/gem;

print $code;
close STDOUT or die "error closing STDOUT: $!";