1 /* SPDX-License-Identifier: GPL-2.0 */ <<
2 /* U3memcpy.S: UltraSparc-III optimized memcpy 1 /* U3memcpy.S: UltraSparc-III optimized memcpy.
3 * 2 *
4 * Copyright (C) 1999, 2000, 2004 David S. Mil 3 * Copyright (C) 1999, 2000, 2004 David S. Miller (davem@redhat.com)
5 */ 4 */
6 5
7 #ifdef __KERNEL__ 6 #ifdef __KERNEL__
8 #include <linux/linkage.h> <<
9 #include <asm/visasm.h> 7 #include <asm/visasm.h>
10 #include <asm/asi.h> 8 #include <asm/asi.h>
11 #define GLOBAL_SPARE %g7 9 #define GLOBAL_SPARE %g7
12 #else 10 #else
13 #define ASI_BLK_P 0xf0 11 #define ASI_BLK_P 0xf0
14 #define FPRS_FEF 0x04 12 #define FPRS_FEF 0x04
15 #ifdef MEMCPY_DEBUG 13 #ifdef MEMCPY_DEBUG
16 #define VISEntryHalf rd %fprs, %o5; wr %g0, FP 14 #define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs; \
17 clr %g1; clr %g2; clr %g3 15 clr %g1; clr %g2; clr %g3; subcc %g0, %g0, %g0;
18 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 16 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
19 #else 17 #else
20 #define VISEntryHalf rd %fprs, %o5; wr %g0, FP 18 #define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs
21 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 19 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
22 #endif 20 #endif
23 #define GLOBAL_SPARE %g5 21 #define GLOBAL_SPARE %g5
24 #endif 22 #endif
25 23
26 #ifndef EX_LD 24 #ifndef EX_LD
27 #define EX_LD(x,y) x !! 25 #define EX_LD(x) x
28 #endif <<
29 #ifndef EX_LD_FP <<
30 #define EX_LD_FP(x,y) x <<
31 #endif 26 #endif
32 27
33 #ifndef EX_ST 28 #ifndef EX_ST
34 #define EX_ST(x,y) x !! 29 #define EX_ST(x) x
35 #endif 30 #endif
36 #ifndef EX_ST_FP !! 31
37 #define EX_ST_FP(x,y) x !! 32 #ifndef EX_RETVAL
>> 33 #define EX_RETVAL(x) x
38 #endif 34 #endif
39 35
40 #ifndef LOAD 36 #ifndef LOAD
41 #define LOAD(type,addr,dest) type [addr], d 37 #define LOAD(type,addr,dest) type [addr], dest
42 #endif 38 #endif
43 39
44 #ifndef STORE 40 #ifndef STORE
45 #define STORE(type,src,addr) type src, [add 41 #define STORE(type,src,addr) type src, [addr]
46 #endif 42 #endif
47 43
48 #ifndef STORE_BLK 44 #ifndef STORE_BLK
49 #define STORE_BLK(src,addr) stda src, [add 45 #define STORE_BLK(src,addr) stda src, [addr] ASI_BLK_P
50 #endif 46 #endif
51 47
52 #ifndef FUNC_NAME 48 #ifndef FUNC_NAME
53 #define FUNC_NAME U3memcpy 49 #define FUNC_NAME U3memcpy
54 #endif 50 #endif
55 51
56 #ifndef PREAMBLE 52 #ifndef PREAMBLE
57 #define PREAMBLE 53 #define PREAMBLE
58 #endif 54 #endif
59 55
60 #ifndef XCC 56 #ifndef XCC
61 #define XCC xcc 57 #define XCC xcc
62 #endif 58 #endif
63 59
64 .register %g2,#scratch 60 .register %g2,#scratch
65 .register %g3,#scratch 61 .register %g3,#scratch
66 62
67 /* Special/non-trivial issues of this 63 /* Special/non-trivial issues of this code:
68 * 64 *
69 * 1) %o5 is preserved from VISEntryHa 65 * 1) %o5 is preserved from VISEntryHalf to VISExitHalf
70 * 2) Only low 32 FPU registers are us 66 * 2) Only low 32 FPU registers are used so that only the
71 * lower half of the FPU register s 67 * lower half of the FPU register set is dirtied by this
72 * code. This is especially import 68 * code. This is especially important in the kernel.
73 * 3) This code never prefetches cache 69 * 3) This code never prefetches cachelines past the end
74 * of the source buffer. 70 * of the source buffer.
75 */ 71 */
76 72
77 .text 73 .text
78 #ifndef EX_RETVAL <<
79 #define EX_RETVAL(x) x <<
80 __restore_fp: <<
81 VISExitHalf <<
82 retl <<
83 nop <<
84 ENTRY(U3_retl_o2_plus_g2_plus_g1_plus_1_fp) <<
85 add %g1, 1, %g1 <<
86 add %g2, %g1, %g2 <<
87 ba,pt %xcc, __restore_fp <<
88 add %o2, %g2, %o0 <<
89 ENDPROC(U3_retl_o2_plus_g2_plus_g1_plus_1_fp) <<
90 ENTRY(U3_retl_o2_plus_g2_fp) <<
91 ba,pt %xcc, __restore_fp <<
92 add %o2, %g2, %o0 <<
93 ENDPROC(U3_retl_o2_plus_g2_fp) <<
94 ENTRY(U3_retl_o2_plus_g2_plus_8_fp) <<
95 add %g2, 8, %g2 <<
96 ba,pt %xcc, __restore_fp <<
97 add %o2, %g2, %o0 <<
98 ENDPROC(U3_retl_o2_plus_g2_plus_8_fp) <<
99 ENTRY(U3_retl_o2) <<
100 retl <<
101 mov %o2, %o0 <<
102 ENDPROC(U3_retl_o2) <<
103 ENTRY(U3_retl_o2_plus_1) <<
104 retl <<
105 add %o2, 1, %o0 <<
106 ENDPROC(U3_retl_o2_plus_1) <<
107 ENTRY(U3_retl_o2_plus_4) <<
108 retl <<
109 add %o2, 4, %o0 <<
110 ENDPROC(U3_retl_o2_plus_4) <<
111 ENTRY(U3_retl_o2_plus_8) <<
112 retl <<
113 add %o2, 8, %o0 <<
114 ENDPROC(U3_retl_o2_plus_8) <<
115 ENTRY(U3_retl_o2_plus_g1_plus_1) <<
116 add %g1, 1, %g1 <<
117 retl <<
118 add %o2, %g1, %o0 <<
119 ENDPROC(U3_retl_o2_plus_g1_plus_1) <<
120 ENTRY(U3_retl_o2_fp) <<
121 ba,pt %xcc, __restore_fp <<
122 mov %o2, %o0 <<
123 ENDPROC(U3_retl_o2_fp) <<
124 ENTRY(U3_retl_o2_plus_o3_sll_6_plus_0x80_fp) <<
125 sll %o3, 6, %o3 <<
126 add %o3, 0x80, %o3 <<
127 ba,pt %xcc, __restore_fp <<
128 add %o2, %o3, %o0 <<
129 ENDPROC(U3_retl_o2_plus_o3_sll_6_plus_0x80_fp) <<
130 ENTRY(U3_retl_o2_plus_o3_sll_6_plus_0x40_fp) <<
131 sll %o3, 6, %o3 <<
132 add %o3, 0x40, %o3 <<
133 ba,pt %xcc, __restore_fp <<
134 add %o2, %o3, %o0 <<
135 ENDPROC(U3_retl_o2_plus_o3_sll_6_plus_0x40_fp) <<
136 ENTRY(U3_retl_o2_plus_GS_plus_0x10) <<
137 add GLOBAL_SPARE, 0x10, GLOBAL_SPA <<
138 retl <<
139 add %o2, GLOBAL_SPARE, %o0 <<
140 ENDPROC(U3_retl_o2_plus_GS_plus_0x10) <<
141 ENTRY(U3_retl_o2_plus_GS_plus_0x08) <<
142 add GLOBAL_SPARE, 0x08, GLOBAL_SPA <<
143 retl <<
144 add %o2, GLOBAL_SPARE, %o0 <<
145 ENDPROC(U3_retl_o2_plus_GS_plus_0x08) <<
146 ENTRY(U3_retl_o2_and_7_plus_GS) <<
147 and %o2, 7, %o2 <<
148 retl <<
149 add %o2, GLOBAL_SPARE, %o0 <<
150 ENDPROC(U3_retl_o2_and_7_plus_GS) <<
151 ENTRY(U3_retl_o2_and_7_plus_GS_plus_8) <<
152 add GLOBAL_SPARE, 8, GLOBAL_SPARE <<
153 and %o2, 7, %o2 <<
154 retl <<
155 add %o2, GLOBAL_SPARE, %o0 <<
156 ENDPROC(U3_retl_o2_and_7_plus_GS_plus_8) <<
157 #endif <<
158 <<
159 .align 64 74 .align 64
160 75
161 /* The cheetah's flexible spine, overs 76 /* The cheetah's flexible spine, oversized liver, enlarged heart,
162 * slender muscular body, and claws ma 77 * slender muscular body, and claws make it the swiftest hunter
163 * in Africa and the fastest animal on 78 * in Africa and the fastest animal on land. Can reach speeds
164 * of up to 2.4GB per second. 79 * of up to 2.4GB per second.
165 */ 80 */
166 81
167 .globl FUNC_NAME 82 .globl FUNC_NAME
168 .type FUNC_NAME,#function 83 .type FUNC_NAME,#function
169 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len * 84 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len */
170 srlx %o2, 31, %g2 85 srlx %o2, 31, %g2
171 cmp %g2, 0 86 cmp %g2, 0
172 <<
173 /* software trap 5 "Range Check" if ds <<
174 tne %xcc, 5 87 tne %xcc, 5
175 PREAMBLE 88 PREAMBLE
176 mov %o0, %o4 89 mov %o0, %o4
177 <<
178 /* if len == 0 */ <<
179 cmp %o2, 0 90 cmp %o2, 0
180 be,pn %XCC, end_return !! 91 be,pn %XCC, 85f
181 or %o0, %o1, %o3 92 or %o0, %o1, %o3
182 <<
183 /* if len < 16 */ <<
184 cmp %o2, 16 93 cmp %o2, 16
185 blu,a,pn %XCC, less_than_16 !! 94 blu,a,pn %XCC, 80f
186 or %o3, %o2, %o3 95 or %o3, %o2, %o3
187 96
188 /* if len < 192 */ <<
189 cmp %o2, (3 * 64) 97 cmp %o2, (3 * 64)
190 blu,pt %XCC, less_than_192 !! 98 blu,pt %XCC, 70f
191 andcc %o3, 0x7, %g0 99 andcc %o3, 0x7, %g0
192 100
193 /* Clobbers o5/g1/g2/g3/g7/icc/xcc. W 101 /* Clobbers o5/g1/g2/g3/g7/icc/xcc. We must preserve
194 * o5 from here until we hit VISExitHa 102 * o5 from here until we hit VISExitHalf.
195 */ 103 */
196 VISEntryHalf 104 VISEntryHalf
197 105
198 /* Is 'dst' already aligned on an 64-b 106 /* Is 'dst' already aligned on an 64-byte boundary? */
199 andcc %o0, 0x3f, %g2 107 andcc %o0, 0x3f, %g2
200 be,pt %XCC, 2f 108 be,pt %XCC, 2f
201 109
202 /* Compute abs((dst & 0x3f) - 0x40) in 110 /* Compute abs((dst & 0x3f) - 0x40) into %g2. This is the number
203 * of bytes to copy to make 'dst' 64-b 111 * of bytes to copy to make 'dst' 64-byte aligned. We pre-
204 * subtract this from 'len'. 112 * subtract this from 'len'.
205 */ 113 */
206 sub %o0, %o1, GLOBAL_SPARE 114 sub %o0, %o1, GLOBAL_SPARE
207 sub %g2, 0x40, %g2 115 sub %g2, 0x40, %g2
208 sub %g0, %g2, %g2 116 sub %g0, %g2, %g2
209 sub %o2, %g2, %o2 117 sub %o2, %g2, %o2
210 andcc %g2, 0x7, %g1 118 andcc %g2, 0x7, %g1
211 be,pt %icc, 2f 119 be,pt %icc, 2f
212 and %g2, 0x38, %g2 120 and %g2, 0x38, %g2
213 121
214 1: subcc %g1, 0x1, %g1 122 1: subcc %g1, 0x1, %g1
215 EX_LD_FP(LOAD(ldub, %o1 + 0x00, %o3), !! 123 EX_LD(LOAD(ldub, %o1 + 0x00, %o3))
216 EX_ST_FP(STORE(stb, %o3, %o1 + GLOBAL_ !! 124 EX_ST(STORE(stb, %o3, %o1 + GLOBAL_SPARE))
217 bgu,pt %XCC, 1b 125 bgu,pt %XCC, 1b
218 add %o1, 0x1, %o1 126 add %o1, 0x1, %o1
219 127
220 add %o1, GLOBAL_SPARE, %o0 128 add %o1, GLOBAL_SPARE, %o0
221 129
222 2: cmp %g2, 0x0 130 2: cmp %g2, 0x0
223 and %o1, 0x7, %g1 131 and %o1, 0x7, %g1
224 be,pt %icc, 3f 132 be,pt %icc, 3f
225 alignaddr %o1, %g0, %o1 133 alignaddr %o1, %g0, %o1
226 134
227 EX_LD_FP(LOAD(ldd, %o1, %f4), U3_retl_ !! 135 EX_LD(LOAD(ldd, %o1, %f4))
228 1: EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f6), U3 !! 136 1: EX_LD(LOAD(ldd, %o1 + 0x8, %f6))
229 add %o1, 0x8, %o1 137 add %o1, 0x8, %o1
230 subcc %g2, 0x8, %g2 138 subcc %g2, 0x8, %g2
231 faligndata %f4, %f6, %f0 139 faligndata %f4, %f6, %f0
232 EX_ST_FP(STORE(std, %f0, %o0), U3_retl !! 140 EX_ST(STORE(std, %f0, %o0))
233 be,pn %icc, 3f 141 be,pn %icc, 3f
234 add %o0, 0x8, %o0 142 add %o0, 0x8, %o0
235 143
236 EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f4), U3 !! 144 EX_LD(LOAD(ldd, %o1 + 0x8, %f4))
237 add %o1, 0x8, %o1 145 add %o1, 0x8, %o1
238 subcc %g2, 0x8, %g2 146 subcc %g2, 0x8, %g2
239 faligndata %f6, %f4, %f2 147 faligndata %f6, %f4, %f2
240 EX_ST_FP(STORE(std, %f2, %o0), U3_retl !! 148 EX_ST(STORE(std, %f2, %o0))
241 bne,pt %icc, 1b 149 bne,pt %icc, 1b
242 add %o0, 0x8, %o0 150 add %o0, 0x8, %o0
243 151
244 3: LOAD(prefetch, %o1 + 0x000, #one_read) 152 3: LOAD(prefetch, %o1 + 0x000, #one_read)
245 LOAD(prefetch, %o1 + 0x040, #one_read) 153 LOAD(prefetch, %o1 + 0x040, #one_read)
246 andn %o2, (0x40 - 1), GLOBA 154 andn %o2, (0x40 - 1), GLOBAL_SPARE
247 LOAD(prefetch, %o1 + 0x080, #one_read) 155 LOAD(prefetch, %o1 + 0x080, #one_read)
248 LOAD(prefetch, %o1 + 0x0c0, #one_read) 156 LOAD(prefetch, %o1 + 0x0c0, #one_read)
249 LOAD(prefetch, %o1 + 0x100, #one_read) 157 LOAD(prefetch, %o1 + 0x100, #one_read)
250 EX_LD_FP(LOAD(ldd, %o1 + 0x000, %f0), !! 158 EX_LD(LOAD(ldd, %o1 + 0x000, %f0))
251 LOAD(prefetch, %o1 + 0x140, #one_read) 159 LOAD(prefetch, %o1 + 0x140, #one_read)
252 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), !! 160 EX_LD(LOAD(ldd, %o1 + 0x008, %f2))
253 LOAD(prefetch, %o1 + 0x180, #one_read) 161 LOAD(prefetch, %o1 + 0x180, #one_read)
254 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), !! 162 EX_LD(LOAD(ldd, %o1 + 0x010, %f4))
255 LOAD(prefetch, %o1 + 0x1c0, #one_read) 163 LOAD(prefetch, %o1 + 0x1c0, #one_read)
256 faligndata %f0, %f2, %f16 164 faligndata %f0, %f2, %f16
257 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), !! 165 EX_LD(LOAD(ldd, %o1 + 0x018, %f6))
258 faligndata %f2, %f4, %f18 166 faligndata %f2, %f4, %f18
259 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), !! 167 EX_LD(LOAD(ldd, %o1 + 0x020, %f8))
260 faligndata %f4, %f6, %f20 168 faligndata %f4, %f6, %f20
261 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), !! 169 EX_LD(LOAD(ldd, %o1 + 0x028, %f10))
262 faligndata %f6, %f8, %f22 170 faligndata %f6, %f8, %f22
263 171
264 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), !! 172 EX_LD(LOAD(ldd, %o1 + 0x030, %f12))
265 faligndata %f8, %f10, %f24 173 faligndata %f8, %f10, %f24
266 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), !! 174 EX_LD(LOAD(ldd, %o1 + 0x038, %f14))
267 faligndata %f10, %f12, %f26 175 faligndata %f10, %f12, %f26
268 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), !! 176 EX_LD(LOAD(ldd, %o1 + 0x040, %f0))
269 177
270 subcc GLOBAL_SPARE, 0x80, GL 178 subcc GLOBAL_SPARE, 0x80, GLOBAL_SPARE
271 add %o1, 0x40, %o1 179 add %o1, 0x40, %o1
272 bgu,pt %XCC, 1f 180 bgu,pt %XCC, 1f
273 srl GLOBAL_SPARE, 6, %o3 181 srl GLOBAL_SPARE, 6, %o3
274 ba,pt %xcc, 2f 182 ba,pt %xcc, 2f
275 nop 183 nop
276 184
277 .align 64 185 .align 64
278 1: 186 1:
279 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), !! 187 EX_LD(LOAD(ldd, %o1 + 0x008, %f2))
280 faligndata %f12, %f14, %f28 188 faligndata %f12, %f14, %f28
281 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), !! 189 EX_LD(LOAD(ldd, %o1 + 0x010, %f4))
282 faligndata %f14, %f0, %f30 190 faligndata %f14, %f0, %f30
283 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl !! 191 EX_ST(STORE_BLK(%f16, %o0))
284 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), !! 192 EX_LD(LOAD(ldd, %o1 + 0x018, %f6))
285 faligndata %f0, %f2, %f16 193 faligndata %f0, %f2, %f16
286 add %o0, 0x40, %o0 194 add %o0, 0x40, %o0
287 195
288 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), !! 196 EX_LD(LOAD(ldd, %o1 + 0x020, %f8))
289 faligndata %f2, %f4, %f18 197 faligndata %f2, %f4, %f18
290 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), !! 198 EX_LD(LOAD(ldd, %o1 + 0x028, %f10))
291 faligndata %f4, %f6, %f20 199 faligndata %f4, %f6, %f20
292 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), !! 200 EX_LD(LOAD(ldd, %o1 + 0x030, %f12))
293 subcc %o3, 0x01, %o3 201 subcc %o3, 0x01, %o3
294 faligndata %f6, %f8, %f22 202 faligndata %f6, %f8, %f22
295 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), !! 203 EX_LD(LOAD(ldd, %o1 + 0x038, %f14))
296 204
297 faligndata %f8, %f10, %f24 205 faligndata %f8, %f10, %f24
298 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), !! 206 EX_LD(LOAD(ldd, %o1 + 0x040, %f0))
299 LOAD(prefetch, %o1 + 0x1c0, #one_read) 207 LOAD(prefetch, %o1 + 0x1c0, #one_read)
300 faligndata %f10, %f12, %f26 208 faligndata %f10, %f12, %f26
301 bg,pt %XCC, 1b 209 bg,pt %XCC, 1b
302 add %o1, 0x40, %o1 210 add %o1, 0x40, %o1
303 211
304 /* Finally we copy the last full 64-by 212 /* Finally we copy the last full 64-byte block. */
305 2: 213 2:
306 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), !! 214 EX_LD(LOAD(ldd, %o1 + 0x008, %f2))
307 faligndata %f12, %f14, %f28 215 faligndata %f12, %f14, %f28
308 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), !! 216 EX_LD(LOAD(ldd, %o1 + 0x010, %f4))
309 faligndata %f14, %f0, %f30 217 faligndata %f14, %f0, %f30
310 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl !! 218 EX_ST(STORE_BLK(%f16, %o0))
311 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), !! 219 EX_LD(LOAD(ldd, %o1 + 0x018, %f6))
312 faligndata %f0, %f2, %f16 220 faligndata %f0, %f2, %f16
313 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), !! 221 EX_LD(LOAD(ldd, %o1 + 0x020, %f8))
314 faligndata %f2, %f4, %f18 222 faligndata %f2, %f4, %f18
315 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), !! 223 EX_LD(LOAD(ldd, %o1 + 0x028, %f10))
316 faligndata %f4, %f6, %f20 224 faligndata %f4, %f6, %f20
317 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), !! 225 EX_LD(LOAD(ldd, %o1 + 0x030, %f12))
318 faligndata %f6, %f8, %f22 226 faligndata %f6, %f8, %f22
319 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), !! 227 EX_LD(LOAD(ldd, %o1 + 0x038, %f14))
320 faligndata %f8, %f10, %f24 228 faligndata %f8, %f10, %f24
321 cmp %g1, 0 229 cmp %g1, 0
322 be,pt %XCC, 1f 230 be,pt %XCC, 1f
323 add %o0, 0x40, %o0 231 add %o0, 0x40, %o0
324 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), !! 232 EX_LD(LOAD(ldd, %o1 + 0x040, %f0))
325 1: faligndata %f10, %f12, %f26 233 1: faligndata %f10, %f12, %f26
326 faligndata %f12, %f14, %f28 234 faligndata %f12, %f14, %f28
327 faligndata %f14, %f0, %f30 235 faligndata %f14, %f0, %f30
328 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl !! 236 EX_ST(STORE_BLK(%f16, %o0))
329 add %o0, 0x40, %o0 237 add %o0, 0x40, %o0
330 add %o1, 0x40, %o1 238 add %o1, 0x40, %o1
331 membar #Sync 239 membar #Sync
332 240
333 /* Now we copy the (len modulo 64) byt 241 /* Now we copy the (len modulo 64) bytes at the end.
334 * Note how we borrow the %f0 loaded a 242 * Note how we borrow the %f0 loaded above.
335 * 243 *
336 * Also notice how this code is carefu 244 * Also notice how this code is careful not to perform a
337 * load past the end of the src buffer 245 * load past the end of the src buffer.
338 */ 246 */
339 and %o2, 0x3f, %o2 247 and %o2, 0x3f, %o2
340 andcc %o2, 0x38, %g2 248 andcc %o2, 0x38, %g2
341 be,pn %XCC, 2f 249 be,pn %XCC, 2f
342 subcc %g2, 0x8, %g2 250 subcc %g2, 0x8, %g2
343 be,pn %XCC, 2f 251 be,pn %XCC, 2f
344 cmp %g1, 0 252 cmp %g1, 0
345 253
346 sub %o2, %g2, %o2 254 sub %o2, %g2, %o2
347 be,a,pt %XCC, 1f 255 be,a,pt %XCC, 1f
348 EX_LD_FP(LOAD(ldd, %o1 + 0x00, %f0), !! 256 EX_LD(LOAD(ldd, %o1 + 0x00, %f0))
349 257
350 1: EX_LD_FP(LOAD(ldd, %o1 + 0x08, %f2), U !! 258 1: EX_LD(LOAD(ldd, %o1 + 0x08, %f2))
351 add %o1, 0x8, %o1 259 add %o1, 0x8, %o1
352 subcc %g2, 0x8, %g2 260 subcc %g2, 0x8, %g2
353 faligndata %f0, %f2, %f8 261 faligndata %f0, %f2, %f8
354 EX_ST_FP(STORE(std, %f8, %o0), U3_retl !! 262 EX_ST(STORE(std, %f8, %o0))
355 be,pn %XCC, 2f 263 be,pn %XCC, 2f
356 add %o0, 0x8, %o0 264 add %o0, 0x8, %o0
357 EX_LD_FP(LOAD(ldd, %o1 + 0x08, %f0), U !! 265 EX_LD(LOAD(ldd, %o1 + 0x08, %f0))
358 add %o1, 0x8, %o1 266 add %o1, 0x8, %o1
359 subcc %g2, 0x8, %g2 267 subcc %g2, 0x8, %g2
360 faligndata %f2, %f0, %f8 268 faligndata %f2, %f0, %f8
361 EX_ST_FP(STORE(std, %f8, %o0), U3_retl !! 269 EX_ST(STORE(std, %f8, %o0))
362 bne,pn %XCC, 1b 270 bne,pn %XCC, 1b
363 add %o0, 0x8, %o0 271 add %o0, 0x8, %o0
364 272
365 /* If anything is left, we copy it one 273 /* If anything is left, we copy it one byte at a time.
366 * Note that %g1 is (src & 0x3) saved 274 * Note that %g1 is (src & 0x3) saved above before the
367 * alignaddr was performed. 275 * alignaddr was performed.
368 */ 276 */
369 2: 277 2:
370 cmp %o2, 0 278 cmp %o2, 0
371 add %o1, %g1, %o1 279 add %o1, %g1, %o1
372 VISExitHalf 280 VISExitHalf
373 be,pn %XCC, end_return !! 281 be,pn %XCC, 85f
374 sub %o0, %o1, %o3 282 sub %o0, %o1, %o3
375 283
376 andcc %g1, 0x7, %g0 284 andcc %g1, 0x7, %g0
377 bne,pn %icc, 90f 285 bne,pn %icc, 90f
378 andcc %o2, 0x8, %g0 286 andcc %o2, 0x8, %g0
379 be,pt %icc, 1f 287 be,pt %icc, 1f
380 nop 288 nop
381 EX_LD(LOAD(ldx, %o1, %o5), U3_retl_o2) !! 289 EX_LD(LOAD(ldx, %o1, %o5))
382 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_r !! 290 EX_ST(STORE(stx, %o5, %o1 + %o3))
383 add %o1, 0x8, %o1 291 add %o1, 0x8, %o1
384 sub %o2, 8, %o2 <<
385 292
386 1: andcc %o2, 0x4, %g0 293 1: andcc %o2, 0x4, %g0
387 be,pt %icc, 1f 294 be,pt %icc, 1f
388 nop 295 nop
389 EX_LD(LOAD(lduw, %o1, %o5), U3_retl_o2 !! 296 EX_LD(LOAD(lduw, %o1, %o5))
390 EX_ST(STORE(stw, %o5, %o1 + %o3), U3_r !! 297 EX_ST(STORE(stw, %o5, %o1 + %o3))
391 add %o1, 0x4, %o1 298 add %o1, 0x4, %o1
392 sub %o2, 4, %o2 <<
393 299
394 1: andcc %o2, 0x2, %g0 300 1: andcc %o2, 0x2, %g0
395 be,pt %icc, 1f 301 be,pt %icc, 1f
396 nop 302 nop
397 EX_LD(LOAD(lduh, %o1, %o5), U3_retl_o2 !! 303 EX_LD(LOAD(lduh, %o1, %o5))
398 EX_ST(STORE(sth, %o5, %o1 + %o3), U3_r !! 304 EX_ST(STORE(sth, %o5, %o1 + %o3))
399 add %o1, 0x2, %o1 305 add %o1, 0x2, %o1
400 sub %o2, 2, %o2 <<
401 306
402 1: andcc %o2, 0x1, %g0 307 1: andcc %o2, 0x1, %g0
403 be,pt %icc, end_return !! 308 be,pt %icc, 85f
404 nop 309 nop
405 EX_LD(LOAD(ldub, %o1, %o5), U3_retl_o2 !! 310 EX_LD(LOAD(ldub, %o1, %o5))
406 ba,pt %xcc, end_return !! 311 ba,pt %xcc, 85f
407 EX_ST(STORE(stb, %o5, %o1 + %o3), U3_ !! 312 EX_ST(STORE(stb, %o5, %o1 + %o3))
408 313
409 .align 64 314 .align 64
410 /* 16 <= len < 192 */ !! 315 70: /* 16 < len <= 64 */
411 less_than_192: <<
412 bne,pn %XCC, 75f 316 bne,pn %XCC, 75f
413 sub %o0, %o1, %o3 317 sub %o0, %o1, %o3
414 318
415 72: 319 72:
416 andn %o2, 0xf, GLOBAL_SPARE 320 andn %o2, 0xf, GLOBAL_SPARE
417 and %o2, 0xf, %o2 321 and %o2, 0xf, %o2
418 1: subcc GLOBAL_SPARE, 0x10, GL 322 1: subcc GLOBAL_SPARE, 0x10, GLOBAL_SPARE
419 EX_LD(LOAD(ldx, %o1 + 0x00, %o5), U3_r !! 323 EX_LD(LOAD(ldx, %o1 + 0x00, %o5))
420 EX_LD(LOAD(ldx, %o1 + 0x08, %g1), U3_r !! 324 EX_LD(LOAD(ldx, %o1 + 0x08, %g1))
421 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_r !! 325 EX_ST(STORE(stx, %o5, %o1 + %o3))
422 add %o1, 0x8, %o1 326 add %o1, 0x8, %o1
423 EX_ST(STORE(stx, %g1, %o1 + %o3), U3_r !! 327 EX_ST(STORE(stx, %g1, %o1 + %o3))
424 bgu,pt %XCC, 1b 328 bgu,pt %XCC, 1b
425 add %o1, 0x8, %o1 329 add %o1, 0x8, %o1
426 73: andcc %o2, 0x8, %g0 330 73: andcc %o2, 0x8, %g0
427 be,pt %XCC, 1f 331 be,pt %XCC, 1f
428 nop 332 nop
429 sub %o2, 0x8, %o2 333 sub %o2, 0x8, %o2
430 EX_LD(LOAD(ldx, %o1, %o5), U3_retl_o2_ !! 334 EX_LD(LOAD(ldx, %o1, %o5))
431 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_r !! 335 EX_ST(STORE(stx, %o5, %o1 + %o3))
432 add %o1, 0x8, %o1 336 add %o1, 0x8, %o1
433 1: andcc %o2, 0x4, %g0 337 1: andcc %o2, 0x4, %g0
434 be,pt %XCC, 1f 338 be,pt %XCC, 1f
435 nop 339 nop
436 sub %o2, 0x4, %o2 340 sub %o2, 0x4, %o2
437 EX_LD(LOAD(lduw, %o1, %o5), U3_retl_o2 !! 341 EX_LD(LOAD(lduw, %o1, %o5))
438 EX_ST(STORE(stw, %o5, %o1 + %o3), U3_r !! 342 EX_ST(STORE(stw, %o5, %o1 + %o3))
439 add %o1, 0x4, %o1 343 add %o1, 0x4, %o1
440 1: cmp %o2, 0 344 1: cmp %o2, 0
441 be,pt %XCC, end_return !! 345 be,pt %XCC, 85f
442 nop 346 nop
443 ba,pt %xcc, 90f 347 ba,pt %xcc, 90f
444 nop 348 nop
445 349
446 75: 350 75:
447 andcc %o0, 0x7, %g1 351 andcc %o0, 0x7, %g1
448 sub %g1, 0x8, %g1 352 sub %g1, 0x8, %g1
449 be,pn %icc, 2f 353 be,pn %icc, 2f
450 sub %g0, %g1, %g1 354 sub %g0, %g1, %g1
451 sub %o2, %g1, %o2 355 sub %o2, %g1, %o2
452 356
453 1: subcc %g1, 1, %g1 357 1: subcc %g1, 1, %g1
454 EX_LD(LOAD(ldub, %o1, %o5), U3_retl_o2 !! 358 EX_LD(LOAD(ldub, %o1, %o5))
455 EX_ST(STORE(stb, %o5, %o1 + %o3), U3_r !! 359 EX_ST(STORE(stb, %o5, %o1 + %o3))
456 bgu,pt %icc, 1b 360 bgu,pt %icc, 1b
457 add %o1, 1, %o1 361 add %o1, 1, %o1
458 362
459 2: add %o1, %o3, %o0 363 2: add %o1, %o3, %o0
460 andcc %o1, 0x7, %g1 364 andcc %o1, 0x7, %g1
461 bne,pt %icc, 8f 365 bne,pt %icc, 8f
462 sll %g1, 3, %g1 366 sll %g1, 3, %g1
463 367
464 cmp %o2, 16 368 cmp %o2, 16
465 bgeu,pt %icc, 72b 369 bgeu,pt %icc, 72b
466 nop 370 nop
467 ba,a,pt %xcc, 73b 371 ba,a,pt %xcc, 73b
468 372
469 8: mov 64, %o3 373 8: mov 64, %o3
470 andn %o1, 0x7, %o1 374 andn %o1, 0x7, %o1
471 EX_LD(LOAD(ldx, %o1, %g2), U3_retl_o2) !! 375 EX_LD(LOAD(ldx, %o1, %g2))
472 sub %o3, %g1, %o3 376 sub %o3, %g1, %o3
473 andn %o2, 0x7, GLOBAL_SPARE 377 andn %o2, 0x7, GLOBAL_SPARE
474 sllx %g2, %g1, %g2 378 sllx %g2, %g1, %g2
475 1: EX_LD(LOAD(ldx, %o1 + 0x8, %g3), U3_re !! 379 1: EX_LD(LOAD(ldx, %o1 + 0x8, %g3))
476 subcc GLOBAL_SPARE, 0x8, GLO 380 subcc GLOBAL_SPARE, 0x8, GLOBAL_SPARE
477 add %o1, 0x8, %o1 381 add %o1, 0x8, %o1
478 srlx %g3, %o3, %o5 382 srlx %g3, %o3, %o5
479 or %o5, %g2, %o5 383 or %o5, %g2, %o5
480 EX_ST(STORE(stx, %o5, %o0), U3_retl_o2 !! 384 EX_ST(STORE(stx, %o5, %o0))
481 add %o0, 0x8, %o0 385 add %o0, 0x8, %o0
482 bgu,pt %icc, 1b 386 bgu,pt %icc, 1b
483 sllx %g3, %g1, %g2 387 sllx %g3, %g1, %g2
484 388
485 srl %g1, 3, %g1 389 srl %g1, 3, %g1
486 andcc %o2, 0x7, %o2 390 andcc %o2, 0x7, %o2
487 be,pn %icc, end_return !! 391 be,pn %icc, 85f
488 add %o1, %g1, %o1 392 add %o1, %g1, %o1
489 ba,pt %xcc, 90f 393 ba,pt %xcc, 90f
490 sub %o0, %o1, %o3 394 sub %o0, %o1, %o3
491 395
492 .align 64 396 .align 64
493 /* 0 < len < 16 */ !! 397 80: /* 0 < len <= 16 */
494 less_than_16: <<
495 andcc %o3, 0x3, %g0 398 andcc %o3, 0x3, %g0
496 bne,pn %XCC, 90f 399 bne,pn %XCC, 90f
497 sub %o0, %o1, %o3 400 sub %o0, %o1, %o3
498 401
499 1: 402 1:
500 subcc %o2, 4, %o2 403 subcc %o2, 4, %o2
501 EX_LD(LOAD(lduw, %o1, %g1), U3_retl_o2 !! 404 EX_LD(LOAD(lduw, %o1, %g1))
502 EX_ST(STORE(stw, %g1, %o1 + %o3), U3_r !! 405 EX_ST(STORE(stw, %g1, %o1 + %o3))
503 bgu,pt %XCC, 1b 406 bgu,pt %XCC, 1b
504 add %o1, 4, %o1 407 add %o1, 4, %o1
505 408
506 end_return: !! 409 85: retl
507 retl <<
508 mov EX_RETVAL(%o4), %o0 410 mov EX_RETVAL(%o4), %o0
509 411
510 .align 32 412 .align 32
511 90: 413 90:
512 subcc %o2, 1, %o2 414 subcc %o2, 1, %o2
513 EX_LD(LOAD(ldub, %o1, %g1), U3_retl_o2 !! 415 EX_LD(LOAD(ldub, %o1, %g1))
514 EX_ST(STORE(stb, %g1, %o1 + %o3), U3_r !! 416 EX_ST(STORE(stb, %g1, %o1 + %o3))
515 bgu,pt %XCC, 90b 417 bgu,pt %XCC, 90b
516 add %o1, 1, %o1 418 add %o1, 1, %o1
517 retl 419 retl
518 mov EX_RETVAL(%o4), %o0 420 mov EX_RETVAL(%o4), %o0
519 421
520 .size FUNC_NAME, .-FUNC_NAME 422 .size FUNC_NAME, .-FUNC_NAME
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.