1 | /******************************************************************************* |
---|
2 | * |
---|
3 | * Copyright (c) 1993 Intel Corporation |
---|
4 | * |
---|
5 | * Intel hereby grants you permission to copy, modify, and distribute this |
---|
6 | * software and its documentation. Intel grants this permission provided |
---|
7 | * that the above copyright notice appears in all copies and that both the |
---|
8 | * copyright notice and this permission notice appear in supporting |
---|
9 | * documentation. In addition, Intel grants this permission provided that |
---|
10 | * you prominently mark as "not part of the original" any modifications |
---|
11 | * made to this software or documentation, and that the name of Intel |
---|
12 | * Corporation not be used in advertising or publicity pertaining to |
---|
13 | * distribution of the software or the documentation without specific, |
---|
14 | * written prior permission. |
---|
15 | * |
---|
16 | * Intel Corporation provides this AS IS, WITHOUT ANY WARRANTY, EXPRESS OR |
---|
17 | * IMPLIED, INCLUDING, WITHOUT LIMITATION, ANY WARRANTY OF MERCHANTABILITY |
---|
18 | * OR FITNESS FOR A PARTICULAR PURPOSE. Intel makes no guarantee or |
---|
19 | * representations regarding the use of, or the results of the use of, |
---|
20 | * the software and documentation in terms of correctness, accuracy, |
---|
21 | * reliability, currentness, or otherwise; and you rely on the software, |
---|
22 | * documentation and results solely at your own risk. |
---|
23 | * |
---|
24 | * IN NO EVENT SHALL INTEL BE LIABLE FOR ANY LOSS OF USE, LOSS OF BUSINESS, |
---|
25 | * LOSS OF PROFITS, INDIRECT, INCIDENTAL, SPECIAL OR CONSEQUENTIAL DAMAGES |
---|
26 | * OF ANY KIND. IN NO EVENT SHALL INTEL'S TOTAL LIABILITY EXCEED THE SUM |
---|
27 | * PAID TO INTEL FOR THE PRODUCT LICENSED HEREUNDER. |
---|
28 | * |
---|
29 | ******************************************************************************/ |
---|
30 | |
---|
31 | .file "sncat_ca.s" |
---|
32 | #ifdef __PIC |
---|
33 | .pic |
---|
34 | #endif |
---|
35 | #ifdef __PID |
---|
36 | .pid |
---|
37 | #endif |
---|
38 | /* |
---|
39 | * (c) copyright 1988,1993 Intel Corp., all rights reserved |
---|
40 | */ |
---|
41 | |
---|
42 | /* |
---|
43 | procedure strncat (optimized assembler version for the CA) |
---|
44 | |
---|
45 | dest_addr = strncat (dest_addr, src_addr, max_bytes) |
---|
46 | |
---|
47 | append the null terminated string pointed to by src_addr to the null |
---|
48 | terminated string pointed to by dest_addr. Return the original |
---|
49 | dest_addr. If the source string is longer than max_bytes, then |
---|
50 | append only max_bytes bytes, and tack on a null byte on the end |
---|
51 | |
---|
52 | This routine will fail if the source and destination string |
---|
53 | overlap (in particular, if the end of the source is overlapped |
---|
54 | by the beginning of the destination). The behavior is undefined. |
---|
55 | This is acceptable according to the draft C standard. |
---|
56 | |
---|
57 | Undefined behavior will also occur if the end of the source string |
---|
58 | (i.e. the terminating null byte) is in the last word of the program's |
---|
59 | allocated memory space. This is so because, in several cases, strncat |
---|
60 | will fetch ahead one word. Disallowing the fetch ahead would impose |
---|
61 | a severe performance penalty. |
---|
62 | |
---|
63 | This program handles five cases: |
---|
64 | |
---|
65 | 1) both arguments start on a word boundary |
---|
66 | 2) neither are word aligned, but they are offset by the same amount |
---|
67 | 3) source is word aligned, destination is not |
---|
68 | 4) destination is word aligned, source is not |
---|
69 | 5) neither is word aligned, and they are offset by differing amounts |
---|
70 | |
---|
71 | At the time of this writing, only g0 thru g7 and g13 are available |
---|
72 | for use in this leafproc; other registers would have to be saved and |
---|
73 | restored. These nine registers, plus tricky use of g14 are sufficient |
---|
74 | to implement the routine. The registers are used as follows: |
---|
75 | |
---|
76 | g0 original dest ptr; not modified, so that it may be returned. |
---|
77 | g1 src ptr; shift count |
---|
78 | g2 max_bytes |
---|
79 | g3 src ptr (word aligned) |
---|
80 | g4 dest ptr (word aligned) |
---|
81 | g5 0xff -- byte extraction mask |
---|
82 | Little endian: |
---|
83 | g6 lsw of double word for extraction of 4 bytes |
---|
84 | g7 msw of double word for extraction of 4 bytes |
---|
85 | Big endian: |
---|
86 | g6 msw of double word for extraction of 4 bytes |
---|
87 | g7 lsw of double word for extraction of 4 bytes |
---|
88 | g13 return address |
---|
89 | g14 byte extracted. |
---|
90 | */ |
---|
91 | |
---|
92 | #if __i960_BIG_ENDIAN__ |
---|
93 | #define MSW g6 |
---|
94 | #define LSW g7 |
---|
95 | #else |
---|
96 | #define LSW g6 |
---|
97 | #define MSW g7 |
---|
98 | #endif |
---|
99 | |
---|
100 | .globl _strncat |
---|
101 | .globl __strncat |
---|
102 | .leafproc _strncat, __strncat |
---|
103 | .align 2 |
---|
104 | _strncat: |
---|
105 | #ifndef __PIC |
---|
106 | lda Lrett,g14 |
---|
107 | #else |
---|
108 | lda Lrett-(.+8)(ip),g14 |
---|
109 | #endif |
---|
110 | __strncat: |
---|
111 | notand g0,3,g4 # extract word addr of start of dest |
---|
112 | lda (g14),g13 # preserve return address |
---|
113 | cmpibge.f 0,g2,Lexit_code # Lexit if number of bytes to move is <= zero. |
---|
114 | and g0,3,LSW # extract byte offset of dest |
---|
115 | ld (g4),MSW # fetch word containing at least first byte |
---|
116 | shlo 3,LSW,g14 # get shift count for making mask for first word |
---|
117 | subi 1,0,LSW # mask initially all ones |
---|
118 | #if __i960_BIG_ENDIAN__ |
---|
119 | shro g14,LSW,LSW # get mask for bytes needed from first word |
---|
120 | #else |
---|
121 | shlo g14,LSW,LSW # get mask for bytes needed from first word |
---|
122 | #endif |
---|
123 | notor MSW,LSW,MSW # set unneeded bytes to all ones |
---|
124 | lda 0xff,g5 # byte extraction mask |
---|
125 | Lsearch_for_word_with_null: |
---|
126 | scanbyte 0,MSW # check for null byte |
---|
127 | lda 4(g4),g4 # post-increment dest word pointer |
---|
128 | mov MSW,LSW # keep a copy of current word |
---|
129 | ld (g4),MSW # fetch next word of dest |
---|
130 | bno.t Lsearch_for_word_with_null # branch if null not found yet |
---|
131 | #if __i960_BIG_ENDIAN__ |
---|
132 | shro 24,LSW,g14 # extract byte |
---|
133 | #else |
---|
134 | and g5,LSW,g14 # extract byte |
---|
135 | #endif |
---|
136 | cmpo 0,g14 # branch if null is first byte of word |
---|
137 | subo 4,g4,g4 # move dest word ptr to word with null |
---|
138 | notand g1,3,g3 # extract word addr of start of src |
---|
139 | bne.t Lsearch_for_null |
---|
140 | |
---|
141 | Lcase_14: |
---|
142 | cmpo g1,g3 # check alignment of source |
---|
143 | ld (g3),LSW # fetch first word of source |
---|
144 | shlo 3,g1,g14 # compute shift count |
---|
145 | lda 4(g3),g3 # post-increment src addr |
---|
146 | bne.f Lcase_4 # branch if source is unaligned |
---|
147 | Lcase_1: |
---|
148 | Lcase_1_wloop: # word copying loop |
---|
149 | cmpi g2,4 # check for fewer than four bytes to move |
---|
150 | lda (LSW),g1 # keep a copy of the src word |
---|
151 | bl.f Lcase_1_cloop # branch if fewer than four bytes to copy |
---|
152 | scanbyte 0,g1 # check for null byte in src word |
---|
153 | ld (g3),LSW # pre-fetch next word of src |
---|
154 | addo 4,g3,g3 # post-increment src addr |
---|
155 | bo.f Lcase_1_cloop # branch if word contains null byte |
---|
156 | subi 4,g2,g2 # decrease max_byte count by the 4 bytes moved |
---|
157 | st g1,(g4) # store word in dest string |
---|
158 | addo 4,g4,g4 # post-increment dest addr |
---|
159 | b Lcase_1_wloop |
---|
160 | |
---|
161 | Lcase_3_cloop: |
---|
162 | Lcase_1_cloop: # character copying loop (max_bytes <= 3) |
---|
163 | cmpdeci 0,g2,g2 # is max_bytes exhausted? |
---|
164 | #if __i960_BIG_ENDIAN__ |
---|
165 | rotate 8,g1,g1 # move next byte into position for extraction |
---|
166 | #endif |
---|
167 | and g5,g1,g14 # extract next char |
---|
168 | be.f Lstore_null # if max_bytes is exhausted, store null and quit |
---|
169 | cmpo 0,g14 # check for null byte |
---|
170 | stob g14,(g4) # store the byte in dest |
---|
171 | #if ! __i960_BIG_ENDIAN__ |
---|
172 | shro 8,g1,g1 # move next byte into position for extraction |
---|
173 | #endif |
---|
174 | lda 1(g4),g4 # post-increment dest byte addr |
---|
175 | bne.t Lcase_1_cloop # branch if null not reached |
---|
176 | bx (g13) # Lexit (g14 == 0) |
---|
177 | |
---|
178 | Lstore_null: |
---|
179 | mov 0,g14 # store null, and set g14 to zero |
---|
180 | stob g14,(g4) |
---|
181 | bx (g13) |
---|
182 | |
---|
183 | |
---|
184 | Lsearch_for_null: |
---|
185 | #if __i960_BIG_ENDIAN__ |
---|
186 | shlo 8,LSW,LSW # check next byte |
---|
187 | shro 24,LSW,g14 |
---|
188 | #else |
---|
189 | shlo 8,g5,g5 # move mask up to next byte |
---|
190 | and g5,LSW,g14 # extract byte |
---|
191 | #endif |
---|
192 | lda 1(g4),g4 # move dest byte ptr to next byte |
---|
193 | cmpobne.t 0,g14,Lsearch_for_null # branch if null is not yet found |
---|
194 | |
---|
195 | Lcase_235: |
---|
196 | cmpo g1,g3 # check alignment of src |
---|
197 | ld (g3),LSW # pre-fetch word with start of src |
---|
198 | and 3,g1,g1 # compute shift count |
---|
199 | lda 0xff,g5 # load mask for byte extraction |
---|
200 | shlo 3,g1,g14 |
---|
201 | lda 4(g3),g3 # post-increment src word counter |
---|
202 | be.t Lcase_3 # branch if src is word aligned |
---|
203 | and g4,3,MSW # extract byte offset for dest string |
---|
204 | cmpo MSW,g1 # < indicates first word of dest has more bytes |
---|
205 | /* than first word of source. */ |
---|
206 | ld (g3),MSW # fetch second word of src |
---|
207 | #if __i960_BIG_ENDIAN__ |
---|
208 | subo g14,0,g14 # adjust shift count for big endian |
---|
209 | #endif |
---|
210 | eshro g14,g6,g5 # extract four bytes |
---|
211 | #if __i960_BIG_ENDIAN__ |
---|
212 | bge.f 1f |
---|
213 | #else |
---|
214 | bg.f 1f |
---|
215 | #endif |
---|
216 | mov MSW,LSW |
---|
217 | lda 4(g3),g3 # move src word addr to second word boundary |
---|
218 | 1: |
---|
219 | mov g5,MSW |
---|
220 | lda 0xff,g5 |
---|
221 | b Lcase_25 |
---|
222 | |
---|
223 | Lcase_3: # src is word aligned; dest is not |
---|
224 | mov LSW,MSW # make copy of first word of src |
---|
225 | lda 32,g14 # initialize shift count to zero (mod 32) |
---|
226 | Lcase_25: |
---|
227 | |
---|
228 | Lcase_3_cloop_at_start: # character copying loop for start of dest str |
---|
229 | cmpdeci 0,g2,g2 # is max_bytes exhausted? |
---|
230 | #if __i960_BIG_ENDIAN__ |
---|
231 | shro 24,MSW,g5 # extract next char |
---|
232 | #else |
---|
233 | and g5,MSW,g5 # extract next char |
---|
234 | #endif |
---|
235 | be.f Lstore_null # Lexit if max_bytes is exhausted |
---|
236 | cmpo 0,g5 # check for null byte |
---|
237 | stob g5,(g4) # store the byte in dest |
---|
238 | addo 1,g4,g4 # post-increment dest ptr |
---|
239 | lda 0xff,g5 # re-initialize byte extraction mask |
---|
240 | notand g4,3,g1 # extract word address |
---|
241 | be.t Lexit_code # Lexit if null byte reached |
---|
242 | cmpo g1,g4 # have we reached word boundary in dest yet? |
---|
243 | #if __i960_BIG_ENDIAN__ |
---|
244 | lda -8(g14),g14 # augment the shift counter |
---|
245 | rotate 8,MSW,MSW # move next byte into position for extraction |
---|
246 | #else |
---|
247 | lda 8(g14),g14 # augment the shift counter |
---|
248 | shro 8,MSW,MSW # move next byte into position for extraction |
---|
249 | #endif |
---|
250 | bne.t Lcase_3_cloop_at_start # branch if reached word boundary? |
---|
251 | |
---|
252 | #if __i960_BIG_ENDIAN__ |
---|
253 | cmpo 0,g14 |
---|
254 | ld (g3),MSW # fetch msw of operand for double shift |
---|
255 | bne Lcase_3_wloop # branch if src is still unaligned. |
---|
256 | |
---|
257 | Lcase_3_wloop2: |
---|
258 | cmpi g2,4 # less than four bytes to move? |
---|
259 | mov LSW,g1 # extract 4 bytes of src |
---|
260 | lda 4(g3),g3 # post-increment src word addr |
---|
261 | bl.f Lcase_3_cloop # branch if < four bytes left to move |
---|
262 | scanbyte 0,g1 # check for null byte |
---|
263 | mov MSW,LSW # move msw to lsw |
---|
264 | ld (g3),MSW # pre-fetch msw of operand for double shift |
---|
265 | bo.f Lcase_3_cloop # branch if word contains null byte |
---|
266 | subi 4,g2,g2 # decrease max_byte count by the 4 bytes moved |
---|
267 | st g1,(g4) # store 4 bytes to dest |
---|
268 | addo 4,g4,g4 # post-increment dest ptr |
---|
269 | b Lcase_3_wloop2 |
---|
270 | Lcase_4: |
---|
271 | subo g14,0,g14 # adjust shift count for big endian |
---|
272 | #else |
---|
273 | Lcase_4: |
---|
274 | #endif |
---|
275 | |
---|
276 | ld (g3),MSW # fetch msw of operand for double shift |
---|
277 | |
---|
278 | Lcase_3_wloop: |
---|
279 | cmpi g2,4 # less than four bytes to move? |
---|
280 | eshro g14,g6,g1 # extract 4 bytes of src |
---|
281 | lda 4(g3),g3 # post-increment src word addr |
---|
282 | bl.f Lcase_3_cloop # branch if < four bytes left to move |
---|
283 | scanbyte 0,g1 # check for null byte |
---|
284 | mov MSW,LSW # move msw to lsw |
---|
285 | ld (g3),MSW # pre-fetch msw of operand for double shift |
---|
286 | bo.f Lcase_3_cloop # branch if word contains null byte |
---|
287 | subi 4,g2,g2 # decrease max_byte count by the 4 bytes moved |
---|
288 | st g1,(g4) # store 4 bytes to dest |
---|
289 | addo 4,g4,g4 # post-increment dest ptr |
---|
290 | b Lcase_3_wloop |
---|
291 | |
---|
292 | |
---|
293 | Lexit_code: |
---|
294 | mov 0,g14 # conform to register conventions |
---|
295 | bx (g13) # g0 = addr of dest; g14 = 0 |
---|
296 | Lrett: |
---|
297 | ret |
---|
298 | |
---|
299 | /* end of strncat */ |
---|
300 | |
---|