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
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
|
#! /usr/bin/env perl
# Copyright 2012-2016 The OpenSSL Project Authors. All Rights Reserved.
#
# Licensed under the Apache License 2.0 (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 for C64x+.
#
# November 2011
#
# If compared to compiler-generated code with similar characteristics,
# i.e. compiled with OPENSSL_SMALL_FOOTPRINT and utilizing SPLOOPs,
# this implementation is 25% smaller and >2x faster. In absolute terms
# performance is (quite impressive) ~6.5 cycles per processed byte.
# Fully unrolled assembler would be ~5x larger and is likely to be
# ~15% faster. It would be free from references to intermediate ring
# buffer, but put more pressure on L1P [both because the code would be
# larger and won't be using SPLOOP buffer]. There are no plans to
# realize fully unrolled variant though...
#
# !!! Note that this module uses AMR, which means that all interrupt
# service routines are expected to preserve it and for own well-being
# zero it upon entry.
$output = pop and open STDOUT,">$output";
($CTX,$INP,$NUM) = ("A4","B4","A6"); # arguments
($A,$B,$C,$D,$E, $Arot,$F,$F0,$T,$K) = map("A$_",(16..20, 21..25));
($X0,$X2,$X8,$X13) = ("A26","B26","A27","B27");
($TX0,$TX1,$TX2,$TX3) = map("B$_",(28..31));
($XPA,$XPB) = ("A5","B5"); # X circular buffer
($Actx,$Bctx,$Cctx,$Dctx,$Ectx) = map("A$_",(3,6..9)); # zaps $NUM
$code=<<___;
.text
.if .ASSEMBLER_VERSION<7000000
.asg 0,__TI_EABI__
.endif
.if __TI_EABI__
.asg sha1_block_data_order,_sha1_block_data_order
.endif
.asg B3,RA
.asg A15,FP
.asg B15,SP
.if .BIG_ENDIAN
.asg MV,SWAP2
.asg MV,SWAP4
.endif
.global _sha1_block_data_order
_sha1_block_data_order:
.asmfunc stack_usage(64)
MV $NUM,A0 ; reassign $NUM
|| MVK -64,B0
[!A0] BNOP RA ; if ($NUM==0) return;
|| [A0] STW FP,*SP--[16] ; save frame pointer and alloca(64)
|| [A0] MV SP,FP
[A0] LDW *${CTX}[0],$A ; load A-E...
|| [A0] AND B0,SP,SP ; align stack at 64 bytes
[A0] LDW *${CTX}[1],$B
|| [A0] SUBAW SP,2,SP ; reserve two words above buffer
[A0] LDW *${CTX}[2],$C
|| [A0] MVK 0x00404,B0
[A0] LDW *${CTX}[3],$D
|| [A0] MVKH 0x50000,B0 ; 0x050404, 64 bytes for $XP[AB]
[A0] LDW *${CTX}[4],$E
|| [A0] MVC B0,AMR ; setup circular addressing
LDNW *${INP}++,$TX1 ; pre-fetch input
NOP 1
loop?:
MVK 0x00007999,$K
|| ADDAW SP,2,$XPA
|| SUB A0,1,A0
|| MVK 13,B0
MVKH 0x5a820000,$K ; K_00_19
|| ADDAW SP,2,$XPB
|| MV $A,$Actx
|| MV $B,$Bctx
;;==================================================
SPLOOPD 5 ; BODY_00_13
|| MV $C,$Cctx
|| MV $D,$Dctx
|| MV $E,$Ectx
|| MVC B0,ILC
ROTL $A,5,$Arot
|| AND $C,$B,$F
|| ANDN $D,$B,$F0
|| ADD $K,$E,$T ; T=E+K
XOR $F0,$F,$F ; F_00_19(B,C,D)
|| MV $D,$E ; E=D
|| MV $C,$D ; D=C
|| SWAP2 $TX1,$TX2
|| LDNW *${INP}++,$TX1
ADD $F,$T,$T ; T+=F_00_19(B,C,D)
|| ROTL $B,30,$C ; C=ROL(B,30)
|| SWAP4 $TX2,$TX3 ; byte swap
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| MV $A,$B ; B=A
ADD $TX3,$T,$A ; A=T+Xi
|| STW $TX3,*${XPB}++
SPKERNEL
;;==================================================
ROTL $A,5,$Arot ; BODY_14
|| AND $C,$B,$F
|| ANDN $D,$B,$F0
|| ADD $K,$E,$T ; T=E+K
XOR $F0,$F,$F ; F_00_19(B,C,D)
|| MV $D,$E ; E=D
|| MV $C,$D ; D=C
|| SWAP2 $TX1,$TX2
|| LDNW *${INP}++,$TX1
ADD $F,$T,$T ; T+=F_00_19(B,C,D)
|| ROTL $B,30,$C ; C=ROL(B,30)
|| SWAP4 $TX2,$TX2 ; byte swap
|| LDW *${XPA}++,$X0 ; fetches from X ring buffer are
|| LDW *${XPB}[4],$X2 ; 2 iterations ahead
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| MV $A,$B ; B=A
|| LDW *${XPA}[7],$X8
|| MV $TX3,$X13 ; || LDW *${XPB}[15],$X13
|| MV $TX2,$TX3
ADD $TX2,$T,$A ; A=T+Xi
|| STW $TX2,*${XPB}++
;;==================================================
ROTL $A,5,$Arot ; BODY_15
|| AND $C,$B,$F
|| ANDN $D,$B,$F0
|| ADD $K,$E,$T ; T=E+K
XOR $F0,$F,$F ; F_00_19(B,C,D)
|| MV $D,$E ; E=D
|| MV $C,$D ; D=C
|| SWAP2 $TX1,$TX2
ADD $F,$T,$T ; T+=F_00_19(B,C,D)
|| ROTL $B,30,$C ; C=ROL(B,30)
|| SWAP4 $TX2,$TX2 ; byte swap
|| XOR $X0,$X2,$TX0 ; Xupdate XORs are 1 iteration ahead
|| LDW *${XPA}++,$X0
|| LDW *${XPB}[4],$X2
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| MV $A,$B ; B=A
|| XOR $X8,$X13,$TX1
|| LDW *${XPA}[7],$X8
|| MV $TX3,$X13 ; || LDW *${XPB}[15],$X13
|| MV $TX2,$TX3
ADD $TX2,$T,$A ; A=T+Xi
|| STW $TX2,*${XPB}++
|| XOR $TX0,$TX1,$TX1
|| MVK 3,B0
;;==================================================
SPLOOPD 5 ; BODY_16_19
|| MVC B0,ILC
ROTL $A,5,$Arot
|| AND $C,$B,$F
|| ANDN $D,$B,$F0
|| ADD $K,$E,$T ; T=E+K
|| ROTL $TX1,1,$TX2 ; Xupdate output
XOR $F0,$F,$F ; F_00_19(B,C,D)
|| MV $D,$E ; E=D
|| MV $C,$D ; D=C
ADD $F,$T,$T ; T+=F_00_19(B,C,D)
|| ROTL $B,30,$C ; C=ROL(B,30)
|| XOR $X0,$X2,$TX0
|| LDW *${XPA}++,$X0
|| LDW *${XPB}[4],$X2
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| MV $A,$B ; B=A
|| XOR $X8,$X13,$TX1
|| LDW *${XPA}[7],$X8
|| MV $TX3,$X13 ; || LDW *${XPB}[15],$X13
|| MV $TX2,$TX3
ADD $TX2,$T,$A ; A=T+Xi
|| STW $TX2,*${XPB}++
|| XOR $TX0,$TX1,$TX1
SPKERNEL
MVK 0xffffeba1,$K
|| MVK 19,B0
MVKH 0x6ed90000,$K ; K_20_39
___
sub BODY_20_39 {
$code.=<<___;
;;==================================================
SPLOOPD 5 ; BODY_20_39
|| MVC B0,ILC
ROTL $A,5,$Arot
|| XOR $B,$C,$F
|| ADD $K,$E,$T ; T=E+K
|| ROTL $TX1,1,$TX2 ; Xupdate output
XOR $D,$F,$F ; F_20_39(B,C,D)
|| MV $D,$E ; E=D
|| MV $C,$D ; D=C
ADD $F,$T,$T ; T+=F_20_39(B,C,D)
|| ROTL $B,30,$C ; C=ROL(B,30)
|| XOR $X0,$X2,$TX0
|| LDW *${XPA}++,$X0
|| LDW *${XPB}[4],$X2
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| MV $A,$B ; B=A
|| XOR $X8,$X13,$TX1
|| LDW *${XPA}[7],$X8
|| MV $TX3,$X13 ; || LDW *${XPB}[15],$X13
|| MV $TX2,$TX3
ADD $TX2,$T,$A ; A=T+Xi
|| STW $TX2,*${XPB}++ ; last one is redundant
|| XOR $TX0,$TX1,$TX1
SPKERNEL
___
$code.=<<___ if (!shift);
MVK 0xffffbcdc,$K
MVKH 0x8f1b0000,$K ; K_40_59
___
} &BODY_20_39();
$code.=<<___;
;;==================================================
SPLOOPD 5 ; BODY_40_59
|| MVC B0,ILC
|| AND $B,$C,$F
|| AND $B,$D,$F0
ROTL $A,5,$Arot
|| XOR $F0,$F,$F
|| AND $C,$D,$F0
|| ADD $K,$E,$T ; T=E+K
|| ROTL $TX1,1,$TX2 ; Xupdate output
XOR $F0,$F,$F ; F_40_59(B,C,D)
|| MV $D,$E ; E=D
|| MV $C,$D ; D=C
ADD $F,$T,$T ; T+=F_40_59(B,C,D)
|| ROTL $B,30,$C ; C=ROL(B,30)
|| XOR $X0,$X2,$TX0
|| LDW *${XPA}++,$X0
|| LDW *${XPB}[4],$X2
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| MV $A,$B ; B=A
|| XOR $X8,$X13,$TX1
|| LDW *${XPA}[7],$X8
|| MV $TX3,$X13 ; || LDW *${XPB}[15],$X13
|| MV $TX2,$TX3
ADD $TX2,$T,$A ; A=T+Xi
|| STW $TX2,*${XPB}++
|| XOR $TX0,$TX1,$TX1
|| AND $B,$C,$F
|| AND $B,$D,$F0
SPKERNEL
MVK 0xffffc1d6,$K
|| MVK 18,B0
MVKH 0xca620000,$K ; K_60_79
___
&BODY_20_39(-1); # BODY_60_78
$code.=<<___;
;;==================================================
[A0] B loop?
|| ROTL $A,5,$Arot ; BODY_79
|| XOR $B,$C,$F
|| ROTL $TX1,1,$TX2 ; Xupdate output
[A0] LDNW *${INP}++,$TX1 ; pre-fetch input
|| ADD $K,$E,$T ; T=E+K
|| XOR $D,$F,$F ; F_20_39(B,C,D)
ADD $F,$T,$T ; T+=F_20_39(B,C,D)
|| ADD $Ectx,$D,$E ; E=D,E+=Ectx
|| ADD $Dctx,$C,$D ; D=C,D+=Dctx
|| ROTL $B,30,$C ; C=ROL(B,30)
ADD $Arot,$T,$T ; T+=ROL(A,5)
|| ADD $Bctx,$A,$B ; B=A,B+=Bctx
ADD $TX2,$T,$A ; A=T+Xi
ADD $Actx,$A,$A ; A+=Actx
|| ADD $Cctx,$C,$C ; C+=Cctx
;; end of loop?
BNOP RA ; return
|| MV FP,SP ; restore stack pointer
|| LDW *FP[0],FP ; restore frame pointer
STW $A,*${CTX}[0] ; emit A-E...
|| MVK 0,B0
STW $B,*${CTX}[1]
|| MVC B0,AMR ; clear AMR
STW $C,*${CTX}[2]
STW $D,*${CTX}[3]
STW $E,*${CTX}[4]
.endasmfunc
.sect .const
.cstring "SHA1 block transform for C64x+, CRYPTOGAMS by <appro\@openssl.org>"
.align 4
___
print $code;
close STDOUT or die "error closing STDOUT: $!";
|
