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
|
/* Copyright (C) 1996-2016 Free Software Foundation, Inc.
Contributed by Richard Henderson (rth@tamu.edu)
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library. If not, see
<http://www.gnu.org/licenses/>. */
/* Bytewise compare two null-terminated strings of length no longer than N. */
#include <sysdep.h>
.set noat
.set noreorder
/* EV6 only predicts one branch per octaword. We'll use these to push
subsequent branches back to the next bundle. This will generally add
a fetch+decode cycle to older machines, so skip in that case. */
#ifdef __alpha_fix__
# define ev6_unop unop
#else
# define ev6_unop
#endif
.text
ENTRY(strncmp)
#ifdef PROF
ldgp gp, 0(pv)
lda AT, _mcount
jsr AT, (AT), _mcount
.prologue 1
#else
.prologue 0
#endif
xor a0, a1, t2 # are s1 and s2 co-aligned?
beq a2, $zerolength
ldq_u t0, 0(a0) # load asap to give cache time to catch up
ldq_u t1, 0(a1)
lda t3, -1
and t2, 7, t2
srl t3, 1, t6
and a0, 7, t4 # find s1 misalignment
and a1, 7, t5 # find s2 misalignment
cmovlt a2, t6, a2 # bound neg count to LONG_MAX
addq a1, a2, a3 # s2+count
addq a2, t4, a2 # bias count by s1 misalignment
and a2, 7, t10 # ofs of last byte in s1 last word
srl a2, 3, a2 # remaining full words in s1 count
bne t2, $unaligned
/* On entry to this basic block:
t0 == the first word of s1.
t1 == the first word of s2.
t3 == -1. */
$aligned:
mskqh t3, a1, t8 # mask off leading garbage
ornot t1, t8, t1
ornot t0, t8, t0
cmpbge zero, t1, t7 # bits set iff null found
beq a2, $eoc # check end of count
bne t7, $eos
beq t10, $ant_loop
/* Aligned compare main loop.
On entry to this basic block:
t0 == an s1 word.
t1 == an s2 word not containing a null. */
.align 4
$a_loop:
xor t0, t1, t2 # e0 :
bne t2, $wordcmp # .. e1 (zdb)
ldq_u t1, 8(a1) # e0 :
ldq_u t0, 8(a0) # .. e1 :
subq a2, 1, a2 # e0 :
addq a1, 8, a1 # .. e1 :
addq a0, 8, a0 # e0 :
beq a2, $eoc # .. e1 :
cmpbge zero, t1, t7 # e0 :
beq t7, $a_loop # .. e1 :
br $eos
/* Alternate aligned compare loop, for when there's no trailing
bytes on the count. We have to avoid reading too much data. */
.align 4
$ant_loop:
xor t0, t1, t2 # e0 :
ev6_unop
ev6_unop
bne t2, $wordcmp # .. e1 (zdb)
subq a2, 1, a2 # e0 :
beq a2, $zerolength # .. e1 :
ldq_u t1, 8(a1) # e0 :
ldq_u t0, 8(a0) # .. e1 :
addq a1, 8, a1 # e0 :
addq a0, 8, a0 # .. e1 :
cmpbge zero, t1, t7 # e0 :
beq t7, $ant_loop # .. e1 :
br $eos
/* The two strings are not co-aligned. Align s1 and cope. */
/* On entry to this basic block:
t0 == the first word of s1.
t1 == the first word of s2.
t3 == -1.
t4 == misalignment of s1.
t5 == misalignment of s2.
t10 == misalignment of s1 end. */
.align 4
$unaligned:
/* If s1 misalignment is larger than s2 misalignment, we need
extra startup checks to avoid SEGV. */
subq a1, t4, a1 # adjust s2 for s1 misalignment
cmpult t4, t5, t9
subq a3, 1, a3 # last byte of s2
bic a1, 7, t8
mskqh t3, t5, t7 # mask garbage in s2
subq a3, t8, a3
ornot t1, t7, t7
srl a3, 3, a3 # remaining full words in s2 count
beq t9, $u_head
/* Failing that, we need to look for both eos and eoc within the
first word of s2. If we find either, we can continue by
pretending that the next word of s2 is all zeros. */
lda t2, 0 # next = zero
cmpeq a3, 0, t8 # eoc in the first word of s2?
cmpbge zero, t7, t7 # eos in the first word of s2?
or t7, t8, t8
bne t8, $u_head_nl
/* We know just enough now to be able to assemble the first
full word of s2. We can still find a zero at the end of it.
On entry to this basic block:
t0 == first word of s1
t1 == first partial word of s2.</
|
