| blob | d6e742d240075a05c35902d21f86979b054fb928 |
1 /*
2 * linux/arch/arm/lib/csumpartialcopygeneric.S
3 *
4 * Copyright (C) 1995-2001 Russell King
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 as
8 * published by the Free Software Foundation.
9 */
11 /*
12 * unsigned int
13 * csum_partial_copy_xxx(const char *src, char *dst, int len, int sum, )
14 * r0 = src, r1 = dst, r2 = len, r3 = sum
15 * Returns : r0 = checksum
16 *
17 * Note that 'tst' and 'teq' preserve the carry flag.
18 */
20 src .req r0
21 dst .req r1
22 len .req r2
23 sum .req r3
25 .Lzero: mov r0, sum
26 load_regs
28 /*
29 * Align an unaligned destination pointer. We know that
30 * we have >= 8 bytes here, so we don't need to check
31 * the length. Note that the source pointer hasn't been
32 * aligned yet.
33 */
34 .Ldst_unaligned:
35 tst dst, #1
36 beq .Ldst_16bit
38 load1b ip
39 sub len, len, #1
40 adcs sum, sum, ip, put_byte_1 @ update checksum
41 strb ip, [dst], #1
42 tst dst, #2
43 moveq pc, lr @ dst is now 32bit aligned
45 .Ldst_16bit: load2b r8, ip
46 sub len, len, #2
47 adcs sum, sum, r8, put_byte_0
48 strb r8, [dst], #1
49 adcs sum, sum, ip, put_byte_1
50 strb ip, [dst], #1
51 mov pc, lr @ dst is now 32bit aligned
53 /*
54 * Handle 0 to 7 bytes, with any alignment of source and
55 * destination pointers. Note that when we get here, C = 0
56 */
57 .Lless8: teq len, #0 @ check for zero count
58 beq .Lzero
60 /* we must have at least one byte. */
61 tst dst, #1 @ dst 16-bit aligned
62 beq .Lless8_aligned
64 /* Align dst */
65 load1b ip
66 sub len, len, #1
67 adcs sum, sum, ip, put_byte_1 @ update checksum
68 strb ip, [dst], #1
69 tst len, #6
70 beq .Lless8_byteonly
72 1: load2b r8, ip
73 sub len, len, #2
74 adcs sum, sum, r8, put_byte_0
75 strb r8, [dst], #1
76 adcs sum, sum, ip, put_byte_1
77 strb ip, [dst], #1
78 .Lless8_aligned:
79 tst len, #6
80 bne 1b
81 .Lless8_byteonly:
82 tst len, #1
83 beq .Ldone
84 load1b r8
85 adcs sum, sum, r8, put_byte_0 @ update checksum
86 strb r8, [dst], #1
87 b .Ldone
89 FN_ENTRY
90 save_regs
92 cmp len, #8 @ Ensure that we have at least
93 blo .Lless8 @ 8 bytes to copy.
95 adds sum, sum, #0 @ C = 0
96 tst dst, #3 @ Test destination alignment
97 blne .Ldst_unaligned @ align destination, return here
99 /*
100 * Ok, the dst pointer is now 32bit aligned, and we know
101 * that we must have more than 4 bytes to copy. Note
102 * that C contains the carry from the dst alignment above.
103 */
105 tst src, #3 @ Test source alignment
106 bne .Lsrc_not_aligned
108 /* Routine for src & dst aligned */
110 bics ip, len, #15
111 beq 2f
113 1: load4l r4, r5, r6, r7
114 stmia dst!, {r4, r5, r6, r7}
115 adcs sum, sum, r4
116 adcs sum, sum, r5
117 adcs sum, sum, r6
118 adcs sum, sum, r7
119 sub ip, ip, #16
120 teq ip, #0
121 bne 1b
123 2: ands ip, len, #12
124 beq 4f
125 tst ip, #8
126 beq 3f
127 load2l r4, r5
128 stmia dst!, {r4, r5}
129 adcs sum, sum, r4
130 adcs sum, sum, r5
131 tst ip, #4
132 beq 4f
134 3: load1l r4
135 str r4, [dst], #4
136 adcs sum, sum, r4
138 4: ands len, len, #3
139 beq .Ldone
140 load1l r4
141 tst len, #2
142 mov r5, r4, get_byte_0
143 beq .Lexit
144 adcs sum, sum, r4, lspush #16
145 strb r5, [dst], #1
146 mov r5, r4, get_byte_1
147 strb r5, [dst], #1
148 mov r5, r4, get_byte_2
149 .Lexit: tst len, #1
150 strneb r5, [dst], #1
151 andne r5, r5, #255
152 adcnes sum, sum, r5, put_byte_0
154 /*
155 * If the dst pointer was not 16-bit aligned, we
156 * need to rotate the checksum here to get around
157 * the inefficient byte manipulations in the
158 * architecture independent code.
159 */
160 .Ldone: adc r0, sum, #0
161 ldr sum, [sp, #0] @ dst
162 tst sum, #1
163 movne r0, r0, ror #8
164 load_regs
166 .Lsrc_not_aligned:
167 adc sum, sum, #0 @ include C from dst alignment
168 and ip, src, #3
169 bic src, src, #3
170 load1l r5
171 cmp ip, #2
172 beq .Lsrc2_aligned
173 bhi .Lsrc3_aligned
174 mov r4, r5, lspull #8 @ C = 0
175 bics ip, len, #15
176 beq 2f
177 1: load4l r5, r6, r7, r8
178 orr r4, r4, r5, lspush #24
179 mov r5, r5, lspull #8
180 orr r5, r5, r6, lspush #24
181 mov r6, r6, lspull #8
182 orr r6, r6, r7, lspush #24
183 mov r7, r7, lspull #8
184 orr r7, r7, r8, lspush #24
185 stmia dst!, {r4, r5, r6, r7}
186 adcs sum, sum, r4
187 adcs sum, sum, r5
188 adcs sum, sum, r6
189 adcs sum, sum, r7
190 mov r4, r8, lspull #8
191 sub ip, ip, #16
192 teq ip, #0
193 bne 1b
194 2: ands ip, len, #12
195 beq 4f
196 tst ip, #8
197 beq 3f
198 load2l r5, r6
199 orr r4, r4, r5, lspush #24
200 mov r5, r5, lspull #8
201 orr r5, r5, r6, lspush #24
202 stmia dst!, {r4, r5}
203 adcs sum, sum, r4
204 adcs sum, sum, r5
205 mov r4, r6, lspull #8
206 tst ip, #4
207 beq 4f
208 3: load1l r5
209 orr r4, r4, r5, lspush #24
210 str r4, [dst], #4
211 adcs sum, sum, r4
212 mov r4, r5, lspull #8
213 4: ands len, len, #3
214 beq .Ldone
215 mov r5, r4, get_byte_0
216 tst len, #2
217 beq .Lexit
218 adcs sum, sum, r4, lspush #16
219 strb r5, [dst], #1
220 mov r5, r4, get_byte_1
221 strb r5, [dst], #1
222 mov r5, r4, get_byte_2
223 b .Lexit
225 .Lsrc2_aligned: mov r4, r5, lspull #16
226 adds sum, sum, #0
227 bics ip, len, #15
228 beq 2f
229 1: load4l r5, r6, r7, r8
230 orr r4, r4, r5, lspush #16
231 mov r5, r5, lspull #16
232 orr r5, r5, r6, lspush #16
233 mov r6, r6, lspull #16
234 orr r6, r6, r7, lspush #16
235 mov r7, r7, lspull #16
236 orr r7, r7, r8, lspush #16
237 stmia dst!, {r4, r5, r6, r7}
238 adcs sum, sum, r4
239 adcs sum, sum, r5
240 adcs sum, sum, r6
241 adcs sum, sum, r7
242 mov r4, r8, lspull #16
243 sub ip, ip, #16
244 teq ip, #0
245 bne 1b
246 2: ands ip, len, #12
247 beq 4f
248 tst ip, #8
249 beq 3f
250 load2l r5, r6
251 orr r4, r4, r5, lspush #16
252 mov r5, r5, lspull #16
253 orr r5, r5, r6, lspush #16
254 stmia dst!, {r4, r5}
255 adcs sum, sum, r4
256 adcs sum, sum, r5
257 mov r4, r6, lspull #16
258 tst ip, #4
259 beq 4f
260 3: load1l r5
261 orr r4, r4, r5, lspush #16
262 str r4, [dst], #4
263 adcs sum, sum, r4
264 mov r4, r5, lspull #16
265 4: ands len, len, #3
266 beq .Ldone
267 mov r5, r4, get_byte_0
268 tst len, #2
269 beq .Lexit
270 adcs sum, sum, r4
271 strb r5, [dst], #1
272 mov r5, r4, get_byte_1
273 strb r5, [dst], #1
274 tst len, #1
275 beq .Ldone
276 load1b r5
277 b .Lexit
279 .Lsrc3_aligned: mov r4, r5, lspull #24
280 adds sum, sum, #0
281 bics ip, len, #15
282 beq 2f
283 1: load4l r5, r6, r7, r8
284 orr r4, r4, r5, lspush #8
285 mov r5, r5, lspull #24
286 orr r5, r5, r6, lspush #8
287 mov r6, r6, lspull #24
288 orr r6, r6, r7, lspush #8
289 mov r7, r7, lspull #24
290 orr r7, r7, r8, lspush #8
291 stmia dst!, {r4, r5, r6, r7}
292 adcs sum, sum, r4
293 adcs sum, sum, r5
294 adcs sum, sum, r6
295 adcs sum, sum, r7
296 mov r4, r8, lspull #24
297 sub ip, ip, #16
298 teq ip, #0
299 bne 1b
300 2: ands ip, len, #12
301 beq 4f
302 tst ip, #8
303 beq 3f
304 load2l r5, r6
305 orr r4, r4, r5, lspush #8
306 mov r5, r5, lspull #24
307 orr r5, r5, r6, lspush #8
308 stmia dst!, {r4, r5}
309 adcs sum, sum, r4
310 adcs sum, sum, r5
311 mov r4, r6, lspull #24
312 tst ip, #4
313 beq 4f
314 3: load1l r5
315 orr r4, r4, r5, lspush #8
316 str r4, [dst], #4
317 adcs sum, sum, r4
318 mov r4, r5, lspull #24
319 4: ands len, len, #3
320 beq .Ldone
321 mov r5, r4, get_byte_0
322 tst len, #2
323 beq .Lexit
324 strb r5, [dst], #1
325 adcs sum, sum, r4
326 load1l r4
327 mov r5, r4, get_byte_0
328 strb r5, [dst], #1
329 adcs sum, sum, r4, lspush #24
330 mov r5, r4, get_byte_1
331 b .Lexit
332 FN_EXIT