1 /* SPDX-License-Identifier: GPL-2.0 */ 1 /* SPDX-License-Identifier: GPL-2.0 */
2 /* U3memcpy.S: UltraSparc-III optimized memcpy 2 /* U3memcpy.S: UltraSparc-III optimized memcpy.
3 * 3 *
4 * Copyright (C) 1999, 2000, 2004 David S. Mil 4 * Copyright (C) 1999, 2000, 2004 David S. Miller (davem@redhat.com)
5 */ 5 */
6 6
7 #ifdef __KERNEL__ 7 #ifdef __KERNEL__
8 #include <linux/linkage.h> 8 #include <linux/linkage.h>
9 #include <asm/visasm.h> 9 #include <asm/visasm.h>
10 #include <asm/asi.h> 10 #include <asm/asi.h>
11 #define GLOBAL_SPARE %g7 11 #define GLOBAL_SPARE %g7
12 #else 12 #else
13 #define ASI_BLK_P 0xf0 13 #define ASI_BLK_P 0xf0
14 #define FPRS_FEF 0x04 14 #define FPRS_FEF 0x04
15 #ifdef MEMCPY_DEBUG 15 #ifdef MEMCPY_DEBUG
16 #define VISEntryHalf rd %fprs, %o5; wr %g0, FP 16 #define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs; \
17 clr %g1; clr %g2; clr %g3 17 clr %g1; clr %g2; clr %g3; subcc %g0, %g0, %g0;
18 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 18 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
19 #else 19 #else
20 #define VISEntryHalf rd %fprs, %o5; wr %g0, FP 20 #define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs
21 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 21 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
22 #endif 22 #endif
23 #define GLOBAL_SPARE %g5 23 #define GLOBAL_SPARE %g5
24 #endif 24 #endif
25 25
26 #ifndef EX_LD 26 #ifndef EX_LD
27 #define EX_LD(x,y) x 27 #define EX_LD(x,y) x
28 #endif 28 #endif
29 #ifndef EX_LD_FP 29 #ifndef EX_LD_FP
30 #define EX_LD_FP(x,y) x 30 #define EX_LD_FP(x,y) x
31 #endif 31 #endif
32 32
33 #ifndef EX_ST 33 #ifndef EX_ST
34 #define EX_ST(x,y) x 34 #define EX_ST(x,y) x
35 #endif 35 #endif
36 #ifndef EX_ST_FP 36 #ifndef EX_ST_FP
37 #define EX_ST_FP(x,y) x 37 #define EX_ST_FP(x,y) x
38 #endif 38 #endif
39 39
40 #ifndef LOAD 40 #ifndef LOAD
41 #define LOAD(type,addr,dest) type [addr], d 41 #define LOAD(type,addr,dest) type [addr], dest
42 #endif 42 #endif
43 43
44 #ifndef STORE 44 #ifndef STORE
45 #define STORE(type,src,addr) type src, [add 45 #define STORE(type,src,addr) type src, [addr]
46 #endif 46 #endif
47 47
48 #ifndef STORE_BLK 48 #ifndef STORE_BLK
49 #define STORE_BLK(src,addr) stda src, [add 49 #define STORE_BLK(src,addr) stda src, [addr] ASI_BLK_P
50 #endif 50 #endif
51 51
52 #ifndef FUNC_NAME 52 #ifndef FUNC_NAME
53 #define FUNC_NAME U3memcpy 53 #define FUNC_NAME U3memcpy
54 #endif 54 #endif
55 55
56 #ifndef PREAMBLE 56 #ifndef PREAMBLE
57 #define PREAMBLE 57 #define PREAMBLE
58 #endif 58 #endif
59 59
60 #ifndef XCC 60 #ifndef XCC
61 #define XCC xcc 61 #define XCC xcc
62 #endif 62 #endif
63 63
64 .register %g2,#scratch 64 .register %g2,#scratch
65 .register %g3,#scratch 65 .register %g3,#scratch
66 66
67 /* Special/non-trivial issues of this 67 /* Special/non-trivial issues of this code:
68 * 68 *
69 * 1) %o5 is preserved from VISEntryHa 69 * 1) %o5 is preserved from VISEntryHalf to VISExitHalf
70 * 2) Only low 32 FPU registers are us 70 * 2) Only low 32 FPU registers are used so that only the
71 * lower half of the FPU register s 71 * lower half of the FPU register set is dirtied by this
72 * code. This is especially import 72 * code. This is especially important in the kernel.
73 * 3) This code never prefetches cache 73 * 3) This code never prefetches cachelines past the end
74 * of the source buffer. 74 * of the source buffer.
75 */ 75 */
76 76
77 .text 77 .text
78 #ifndef EX_RETVAL 78 #ifndef EX_RETVAL
79 #define EX_RETVAL(x) x 79 #define EX_RETVAL(x) x
80 __restore_fp: 80 __restore_fp:
81 VISExitHalf 81 VISExitHalf
82 retl 82 retl
83 nop 83 nop
84 ENTRY(U3_retl_o2_plus_g2_plus_g1_plus_1_fp) 84 ENTRY(U3_retl_o2_plus_g2_plus_g1_plus_1_fp)
85 add %g1, 1, %g1 85 add %g1, 1, %g1
86 add %g2, %g1, %g2 86 add %g2, %g1, %g2
87 ba,pt %xcc, __restore_fp 87 ba,pt %xcc, __restore_fp
88 add %o2, %g2, %o0 88 add %o2, %g2, %o0
89 ENDPROC(U3_retl_o2_plus_g2_plus_g1_plus_1_fp) 89 ENDPROC(U3_retl_o2_plus_g2_plus_g1_plus_1_fp)
90 ENTRY(U3_retl_o2_plus_g2_fp) 90 ENTRY(U3_retl_o2_plus_g2_fp)
91 ba,pt %xcc, __restore_fp 91 ba,pt %xcc, __restore_fp
92 add %o2, %g2, %o0 92 add %o2, %g2, %o0
93 ENDPROC(U3_retl_o2_plus_g2_fp) 93 ENDPROC(U3_retl_o2_plus_g2_fp)
94 ENTRY(U3_retl_o2_plus_g2_plus_8_fp) 94 ENTRY(U3_retl_o2_plus_g2_plus_8_fp)
95 add %g2, 8, %g2 95 add %g2, 8, %g2
96 ba,pt %xcc, __restore_fp 96 ba,pt %xcc, __restore_fp
97 add %o2, %g2, %o0 97 add %o2, %g2, %o0
98 ENDPROC(U3_retl_o2_plus_g2_plus_8_fp) 98 ENDPROC(U3_retl_o2_plus_g2_plus_8_fp)
99 ENTRY(U3_retl_o2) 99 ENTRY(U3_retl_o2)
100 retl 100 retl
101 mov %o2, %o0 101 mov %o2, %o0
102 ENDPROC(U3_retl_o2) 102 ENDPROC(U3_retl_o2)
103 ENTRY(U3_retl_o2_plus_1) 103 ENTRY(U3_retl_o2_plus_1)
104 retl 104 retl
105 add %o2, 1, %o0 105 add %o2, 1, %o0
106 ENDPROC(U3_retl_o2_plus_1) 106 ENDPROC(U3_retl_o2_plus_1)
107 ENTRY(U3_retl_o2_plus_4) 107 ENTRY(U3_retl_o2_plus_4)
108 retl 108 retl
109 add %o2, 4, %o0 109 add %o2, 4, %o0
110 ENDPROC(U3_retl_o2_plus_4) 110 ENDPROC(U3_retl_o2_plus_4)
111 ENTRY(U3_retl_o2_plus_8) 111 ENTRY(U3_retl_o2_plus_8)
112 retl 112 retl
113 add %o2, 8, %o0 113 add %o2, 8, %o0
114 ENDPROC(U3_retl_o2_plus_8) 114 ENDPROC(U3_retl_o2_plus_8)
115 ENTRY(U3_retl_o2_plus_g1_plus_1) 115 ENTRY(U3_retl_o2_plus_g1_plus_1)
116 add %g1, 1, %g1 116 add %g1, 1, %g1
117 retl 117 retl
118 add %o2, %g1, %o0 118 add %o2, %g1, %o0
119 ENDPROC(U3_retl_o2_plus_g1_plus_1) 119 ENDPROC(U3_retl_o2_plus_g1_plus_1)
120 ENTRY(U3_retl_o2_fp) 120 ENTRY(U3_retl_o2_fp)
121 ba,pt %xcc, __restore_fp 121 ba,pt %xcc, __restore_fp
122 mov %o2, %o0 122 mov %o2, %o0
123 ENDPROC(U3_retl_o2_fp) 123 ENDPROC(U3_retl_o2_fp)
124 ENTRY(U3_retl_o2_plus_o3_sll_6_plus_0x80_fp) 124 ENTRY(U3_retl_o2_plus_o3_sll_6_plus_0x80_fp)
125 sll %o3, 6, %o3 125 sll %o3, 6, %o3
126 add %o3, 0x80, %o3 126 add %o3, 0x80, %o3
127 ba,pt %xcc, __restore_fp 127 ba,pt %xcc, __restore_fp
128 add %o2, %o3, %o0 128 add %o2, %o3, %o0
129 ENDPROC(U3_retl_o2_plus_o3_sll_6_plus_0x80_fp) 129 ENDPROC(U3_retl_o2_plus_o3_sll_6_plus_0x80_fp)
130 ENTRY(U3_retl_o2_plus_o3_sll_6_plus_0x40_fp) 130 ENTRY(U3_retl_o2_plus_o3_sll_6_plus_0x40_fp)
131 sll %o3, 6, %o3 131 sll %o3, 6, %o3
132 add %o3, 0x40, %o3 132 add %o3, 0x40, %o3
133 ba,pt %xcc, __restore_fp 133 ba,pt %xcc, __restore_fp
134 add %o2, %o3, %o0 134 add %o2, %o3, %o0
135 ENDPROC(U3_retl_o2_plus_o3_sll_6_plus_0x40_fp) 135 ENDPROC(U3_retl_o2_plus_o3_sll_6_plus_0x40_fp)
136 ENTRY(U3_retl_o2_plus_GS_plus_0x10) 136 ENTRY(U3_retl_o2_plus_GS_plus_0x10)
137 add GLOBAL_SPARE, 0x10, GLOBAL_SPA 137 add GLOBAL_SPARE, 0x10, GLOBAL_SPARE
138 retl 138 retl
139 add %o2, GLOBAL_SPARE, %o0 139 add %o2, GLOBAL_SPARE, %o0
140 ENDPROC(U3_retl_o2_plus_GS_plus_0x10) 140 ENDPROC(U3_retl_o2_plus_GS_plus_0x10)
141 ENTRY(U3_retl_o2_plus_GS_plus_0x08) 141 ENTRY(U3_retl_o2_plus_GS_plus_0x08)
142 add GLOBAL_SPARE, 0x08, GLOBAL_SPA 142 add GLOBAL_SPARE, 0x08, GLOBAL_SPARE
143 retl 143 retl
144 add %o2, GLOBAL_SPARE, %o0 144 add %o2, GLOBAL_SPARE, %o0
145 ENDPROC(U3_retl_o2_plus_GS_plus_0x08) 145 ENDPROC(U3_retl_o2_plus_GS_plus_0x08)
146 ENTRY(U3_retl_o2_and_7_plus_GS) 146 ENTRY(U3_retl_o2_and_7_plus_GS)
147 and %o2, 7, %o2 147 and %o2, 7, %o2
148 retl 148 retl
149 add %o2, GLOBAL_SPARE, %o0 149 add %o2, GLOBAL_SPARE, %o0
150 ENDPROC(U3_retl_o2_and_7_plus_GS) 150 ENDPROC(U3_retl_o2_and_7_plus_GS)
151 ENTRY(U3_retl_o2_and_7_plus_GS_plus_8) 151 ENTRY(U3_retl_o2_and_7_plus_GS_plus_8)
152 add GLOBAL_SPARE, 8, GLOBAL_SPARE 152 add GLOBAL_SPARE, 8, GLOBAL_SPARE
153 and %o2, 7, %o2 153 and %o2, 7, %o2
154 retl 154 retl
155 add %o2, GLOBAL_SPARE, %o0 155 add %o2, GLOBAL_SPARE, %o0
156 ENDPROC(U3_retl_o2_and_7_plus_GS_plus_8) 156 ENDPROC(U3_retl_o2_and_7_plus_GS_plus_8)
157 #endif 157 #endif
158 158
159 .align 64 159 .align 64
160 160
161 /* The cheetah's flexible spine, overs 161 /* The cheetah's flexible spine, oversized liver, enlarged heart,
162 * slender muscular body, and claws ma 162 * slender muscular body, and claws make it the swiftest hunter
163 * in Africa and the fastest animal on 163 * in Africa and the fastest animal on land. Can reach speeds
164 * of up to 2.4GB per second. 164 * of up to 2.4GB per second.
165 */ 165 */
166 166
167 .globl FUNC_NAME 167 .globl FUNC_NAME
168 .type FUNC_NAME,#function 168 .type FUNC_NAME,#function
169 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len * 169 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len */
170 srlx %o2, 31, %g2 170 srlx %o2, 31, %g2
171 cmp %g2, 0 171 cmp %g2, 0
172 172
173 /* software trap 5 "Range Check" if ds 173 /* software trap 5 "Range Check" if dst >= 0x80000000 */
174 tne %xcc, 5 174 tne %xcc, 5
175 PREAMBLE 175 PREAMBLE
176 mov %o0, %o4 176 mov %o0, %o4
177 177
178 /* if len == 0 */ 178 /* if len == 0 */
179 cmp %o2, 0 179 cmp %o2, 0
180 be,pn %XCC, end_return 180 be,pn %XCC, end_return
181 or %o0, %o1, %o3 181 or %o0, %o1, %o3
182 182
183 /* if len < 16 */ 183 /* if len < 16 */
184 cmp %o2, 16 184 cmp %o2, 16
185 blu,a,pn %XCC, less_than_16 185 blu,a,pn %XCC, less_than_16
186 or %o3, %o2, %o3 186 or %o3, %o2, %o3
187 187
188 /* if len < 192 */ 188 /* if len < 192 */
189 cmp %o2, (3 * 64) 189 cmp %o2, (3 * 64)
190 blu,pt %XCC, less_than_192 190 blu,pt %XCC, less_than_192
191 andcc %o3, 0x7, %g0 191 andcc %o3, 0x7, %g0
192 192
193 /* Clobbers o5/g1/g2/g3/g7/icc/xcc. W 193 /* Clobbers o5/g1/g2/g3/g7/icc/xcc. We must preserve
194 * o5 from here until we hit VISExitHa 194 * o5 from here until we hit VISExitHalf.
195 */ 195 */
196 VISEntryHalf 196 VISEntryHalf
197 197
198 /* Is 'dst' already aligned on an 64-b 198 /* Is 'dst' already aligned on an 64-byte boundary? */
199 andcc %o0, 0x3f, %g2 199 andcc %o0, 0x3f, %g2
200 be,pt %XCC, 2f 200 be,pt %XCC, 2f
201 201
202 /* Compute abs((dst & 0x3f) - 0x40) in 202 /* Compute abs((dst & 0x3f) - 0x40) into %g2. This is the number
203 * of bytes to copy to make 'dst' 64-b 203 * of bytes to copy to make 'dst' 64-byte aligned. We pre-
204 * subtract this from 'len'. 204 * subtract this from 'len'.
205 */ 205 */
206 sub %o0, %o1, GLOBAL_SPARE 206 sub %o0, %o1, GLOBAL_SPARE
207 sub %g2, 0x40, %g2 207 sub %g2, 0x40, %g2
208 sub %g0, %g2, %g2 208 sub %g0, %g2, %g2
209 sub %o2, %g2, %o2 209 sub %o2, %g2, %o2
210 andcc %g2, 0x7, %g1 210 andcc %g2, 0x7, %g1
211 be,pt %icc, 2f 211 be,pt %icc, 2f
212 and %g2, 0x38, %g2 212 and %g2, 0x38, %g2
213 213
214 1: subcc %g1, 0x1, %g1 214 1: subcc %g1, 0x1, %g1
215 EX_LD_FP(LOAD(ldub, %o1 + 0x00, %o3), 215 EX_LD_FP(LOAD(ldub, %o1 + 0x00, %o3), U3_retl_o2_plus_g2_plus_g1_plus_1)
216 EX_ST_FP(STORE(stb, %o3, %o1 + GLOBAL_ 216 EX_ST_FP(STORE(stb, %o3, %o1 + GLOBAL_SPARE), U3_retl_o2_plus_g2_plus_g1_plus_1)
217 bgu,pt %XCC, 1b 217 bgu,pt %XCC, 1b
218 add %o1, 0x1, %o1 218 add %o1, 0x1, %o1
219 219
220 add %o1, GLOBAL_SPARE, %o0 220 add %o1, GLOBAL_SPARE, %o0
221 221
222 2: cmp %g2, 0x0 222 2: cmp %g2, 0x0
223 and %o1, 0x7, %g1 223 and %o1, 0x7, %g1
224 be,pt %icc, 3f 224 be,pt %icc, 3f
225 alignaddr %o1, %g0, %o1 225 alignaddr %o1, %g0, %o1
226 226
227 EX_LD_FP(LOAD(ldd, %o1, %f4), U3_retl_ 227 EX_LD_FP(LOAD(ldd, %o1, %f4), U3_retl_o2_plus_g2)
228 1: EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f6), U3 228 1: EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f6), U3_retl_o2_plus_g2)
229 add %o1, 0x8, %o1 229 add %o1, 0x8, %o1
230 subcc %g2, 0x8, %g2 230 subcc %g2, 0x8, %g2
231 faligndata %f4, %f6, %f0 231 faligndata %f4, %f6, %f0
232 EX_ST_FP(STORE(std, %f0, %o0), U3_retl 232 EX_ST_FP(STORE(std, %f0, %o0), U3_retl_o2_plus_g2_plus_8)
233 be,pn %icc, 3f 233 be,pn %icc, 3f
234 add %o0, 0x8, %o0 234 add %o0, 0x8, %o0
235 235
236 EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f4), U3 236 EX_LD_FP(LOAD(ldd, %o1 + 0x8, %f4), U3_retl_o2_plus_g2)
237 add %o1, 0x8, %o1 237 add %o1, 0x8, %o1
238 subcc %g2, 0x8, %g2 238 subcc %g2, 0x8, %g2
239 faligndata %f6, %f4, %f2 239 faligndata %f6, %f4, %f2
240 EX_ST_FP(STORE(std, %f2, %o0), U3_retl 240 EX_ST_FP(STORE(std, %f2, %o0), U3_retl_o2_plus_g2_plus_8)
241 bne,pt %icc, 1b 241 bne,pt %icc, 1b
242 add %o0, 0x8, %o0 242 add %o0, 0x8, %o0
243 243
244 3: LOAD(prefetch, %o1 + 0x000, #one_read) 244 3: LOAD(prefetch, %o1 + 0x000, #one_read)
245 LOAD(prefetch, %o1 + 0x040, #one_read) 245 LOAD(prefetch, %o1 + 0x040, #one_read)
246 andn %o2, (0x40 - 1), GLOBA 246 andn %o2, (0x40 - 1), GLOBAL_SPARE
247 LOAD(prefetch, %o1 + 0x080, #one_read) 247 LOAD(prefetch, %o1 + 0x080, #one_read)
248 LOAD(prefetch, %o1 + 0x0c0, #one_read) 248 LOAD(prefetch, %o1 + 0x0c0, #one_read)
249 LOAD(prefetch, %o1 + 0x100, #one_read) 249 LOAD(prefetch, %o1 + 0x100, #one_read)
250 EX_LD_FP(LOAD(ldd, %o1 + 0x000, %f0), 250 EX_LD_FP(LOAD(ldd, %o1 + 0x000, %f0), U3_retl_o2)
251 LOAD(prefetch, %o1 + 0x140, #one_read) 251 LOAD(prefetch, %o1 + 0x140, #one_read)
252 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), 252 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), U3_retl_o2)
253 LOAD(prefetch, %o1 + 0x180, #one_read) 253 LOAD(prefetch, %o1 + 0x180, #one_read)
254 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), 254 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), U3_retl_o2)
255 LOAD(prefetch, %o1 + 0x1c0, #one_read) 255 LOAD(prefetch, %o1 + 0x1c0, #one_read)
256 faligndata %f0, %f2, %f16 256 faligndata %f0, %f2, %f16
257 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), 257 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), U3_retl_o2)
258 faligndata %f2, %f4, %f18 258 faligndata %f2, %f4, %f18
259 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), 259 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), U3_retl_o2)
260 faligndata %f4, %f6, %f20 260 faligndata %f4, %f6, %f20
261 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), 261 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), U3_retl_o2)
262 faligndata %f6, %f8, %f22 262 faligndata %f6, %f8, %f22
263 263
264 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), 264 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), U3_retl_o2)
265 faligndata %f8, %f10, %f24 265 faligndata %f8, %f10, %f24
266 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), 266 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), U3_retl_o2)
267 faligndata %f10, %f12, %f26 267 faligndata %f10, %f12, %f26
268 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), 268 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), U3_retl_o2)
269 269
270 subcc GLOBAL_SPARE, 0x80, GL 270 subcc GLOBAL_SPARE, 0x80, GLOBAL_SPARE
271 add %o1, 0x40, %o1 271 add %o1, 0x40, %o1
272 bgu,pt %XCC, 1f 272 bgu,pt %XCC, 1f
273 srl GLOBAL_SPARE, 6, %o3 273 srl GLOBAL_SPARE, 6, %o3
274 ba,pt %xcc, 2f 274 ba,pt %xcc, 2f
275 nop 275 nop
276 276
277 .align 64 277 .align 64
278 1: 278 1:
279 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), 279 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), U3_retl_o2_plus_o3_sll_6_plus_0x80)
280 faligndata %f12, %f14, %f28 280 faligndata %f12, %f14, %f28
281 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), 281 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), U3_retl_o2_plus_o3_sll_6_plus_0x80)
282 faligndata %f14, %f0, %f30 282 faligndata %f14, %f0, %f30
283 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl 283 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl_o2_plus_o3_sll_6_plus_0x80)
284 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), 284 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), U3_retl_o2_plus_o3_sll_6_plus_0x40)
285 faligndata %f0, %f2, %f16 285 faligndata %f0, %f2, %f16
286 add %o0, 0x40, %o0 286 add %o0, 0x40, %o0
287 287
288 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), 288 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), U3_retl_o2_plus_o3_sll_6_plus_0x40)
289 faligndata %f2, %f4, %f18 289 faligndata %f2, %f4, %f18
290 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), 290 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), U3_retl_o2_plus_o3_sll_6_plus_0x40)
291 faligndata %f4, %f6, %f20 291 faligndata %f4, %f6, %f20
292 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), 292 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), U3_retl_o2_plus_o3_sll_6_plus_0x40)
293 subcc %o3, 0x01, %o3 293 subcc %o3, 0x01, %o3
294 faligndata %f6, %f8, %f22 294 faligndata %f6, %f8, %f22
295 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), 295 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), U3_retl_o2_plus_o3_sll_6_plus_0x80)
296 296
297 faligndata %f8, %f10, %f24 297 faligndata %f8, %f10, %f24
298 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), 298 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), U3_retl_o2_plus_o3_sll_6_plus_0x80)
299 LOAD(prefetch, %o1 + 0x1c0, #one_read) 299 LOAD(prefetch, %o1 + 0x1c0, #one_read)
300 faligndata %f10, %f12, %f26 300 faligndata %f10, %f12, %f26
301 bg,pt %XCC, 1b 301 bg,pt %XCC, 1b
302 add %o1, 0x40, %o1 302 add %o1, 0x40, %o1
303 303
304 /* Finally we copy the last full 64-by 304 /* Finally we copy the last full 64-byte block. */
305 2: 305 2:
306 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), 306 EX_LD_FP(LOAD(ldd, %o1 + 0x008, %f2), U3_retl_o2_plus_o3_sll_6_plus_0x80)
307 faligndata %f12, %f14, %f28 307 faligndata %f12, %f14, %f28
308 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), 308 EX_LD_FP(LOAD(ldd, %o1 + 0x010, %f4), U3_retl_o2_plus_o3_sll_6_plus_0x80)
309 faligndata %f14, %f0, %f30 309 faligndata %f14, %f0, %f30
310 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl 310 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl_o2_plus_o3_sll_6_plus_0x80)
311 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), 311 EX_LD_FP(LOAD(ldd, %o1 + 0x018, %f6), U3_retl_o2_plus_o3_sll_6_plus_0x40)
312 faligndata %f0, %f2, %f16 312 faligndata %f0, %f2, %f16
313 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), 313 EX_LD_FP(LOAD(ldd, %o1 + 0x020, %f8), U3_retl_o2_plus_o3_sll_6_plus_0x40)
314 faligndata %f2, %f4, %f18 314 faligndata %f2, %f4, %f18
315 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), 315 EX_LD_FP(LOAD(ldd, %o1 + 0x028, %f10), U3_retl_o2_plus_o3_sll_6_plus_0x40)
316 faligndata %f4, %f6, %f20 316 faligndata %f4, %f6, %f20
317 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), 317 EX_LD_FP(LOAD(ldd, %o1 + 0x030, %f12), U3_retl_o2_plus_o3_sll_6_plus_0x40)
318 faligndata %f6, %f8, %f22 318 faligndata %f6, %f8, %f22
319 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), 319 EX_LD_FP(LOAD(ldd, %o1 + 0x038, %f14), U3_retl_o2_plus_o3_sll_6_plus_0x40)
320 faligndata %f8, %f10, %f24 320 faligndata %f8, %f10, %f24
321 cmp %g1, 0 321 cmp %g1, 0
322 be,pt %XCC, 1f 322 be,pt %XCC, 1f
323 add %o0, 0x40, %o0 323 add %o0, 0x40, %o0
324 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), 324 EX_LD_FP(LOAD(ldd, %o1 + 0x040, %f0), U3_retl_o2_plus_o3_sll_6_plus_0x40)
325 1: faligndata %f10, %f12, %f26 325 1: faligndata %f10, %f12, %f26
326 faligndata %f12, %f14, %f28 326 faligndata %f12, %f14, %f28
327 faligndata %f14, %f0, %f30 327 faligndata %f14, %f0, %f30
328 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl 328 EX_ST_FP(STORE_BLK(%f16, %o0), U3_retl_o2_plus_o3_sll_6_plus_0x40)
329 add %o0, 0x40, %o0 329 add %o0, 0x40, %o0
330 add %o1, 0x40, %o1 330 add %o1, 0x40, %o1
331 membar #Sync 331 membar #Sync
332 332
333 /* Now we copy the (len modulo 64) byt 333 /* Now we copy the (len modulo 64) bytes at the end.
334 * Note how we borrow the %f0 loaded a 334 * Note how we borrow the %f0 loaded above.
335 * 335 *
336 * Also notice how this code is carefu 336 * Also notice how this code is careful not to perform a
337 * load past the end of the src buffer 337 * load past the end of the src buffer.
338 */ 338 */
339 and %o2, 0x3f, %o2 339 and %o2, 0x3f, %o2
340 andcc %o2, 0x38, %g2 340 andcc %o2, 0x38, %g2
341 be,pn %XCC, 2f 341 be,pn %XCC, 2f
342 subcc %g2, 0x8, %g2 342 subcc %g2, 0x8, %g2
343 be,pn %XCC, 2f 343 be,pn %XCC, 2f
344 cmp %g1, 0 344 cmp %g1, 0
345 345
346 sub %o2, %g2, %o2 346 sub %o2, %g2, %o2
347 be,a,pt %XCC, 1f 347 be,a,pt %XCC, 1f
348 EX_LD_FP(LOAD(ldd, %o1 + 0x00, %f0), 348 EX_LD_FP(LOAD(ldd, %o1 + 0x00, %f0), U3_retl_o2_plus_g2)
349 349
350 1: EX_LD_FP(LOAD(ldd, %o1 + 0x08, %f2), U 350 1: EX_LD_FP(LOAD(ldd, %o1 + 0x08, %f2), U3_retl_o2_plus_g2)
351 add %o1, 0x8, %o1 351 add %o1, 0x8, %o1
352 subcc %g2, 0x8, %g2 352 subcc %g2, 0x8, %g2
353 faligndata %f0, %f2, %f8 353 faligndata %f0, %f2, %f8
354 EX_ST_FP(STORE(std, %f8, %o0), U3_retl 354 EX_ST_FP(STORE(std, %f8, %o0), U3_retl_o2_plus_g2_plus_8)
355 be,pn %XCC, 2f 355 be,pn %XCC, 2f
356 add %o0, 0x8, %o0 356 add %o0, 0x8, %o0
357 EX_LD_FP(LOAD(ldd, %o1 + 0x08, %f0), U 357 EX_LD_FP(LOAD(ldd, %o1 + 0x08, %f0), U3_retl_o2_plus_g2)
358 add %o1, 0x8, %o1 358 add %o1, 0x8, %o1
359 subcc %g2, 0x8, %g2 359 subcc %g2, 0x8, %g2
360 faligndata %f2, %f0, %f8 360 faligndata %f2, %f0, %f8
361 EX_ST_FP(STORE(std, %f8, %o0), U3_retl 361 EX_ST_FP(STORE(std, %f8, %o0), U3_retl_o2_plus_g2_plus_8)
362 bne,pn %XCC, 1b 362 bne,pn %XCC, 1b
363 add %o0, 0x8, %o0 363 add %o0, 0x8, %o0
364 364
365 /* If anything is left, we copy it one 365 /* If anything is left, we copy it one byte at a time.
366 * Note that %g1 is (src & 0x3) saved 366 * Note that %g1 is (src & 0x3) saved above before the
367 * alignaddr was performed. 367 * alignaddr was performed.
368 */ 368 */
369 2: 369 2:
370 cmp %o2, 0 370 cmp %o2, 0
371 add %o1, %g1, %o1 371 add %o1, %g1, %o1
372 VISExitHalf 372 VISExitHalf
373 be,pn %XCC, end_return 373 be,pn %XCC, end_return
374 sub %o0, %o1, %o3 374 sub %o0, %o1, %o3
375 375
376 andcc %g1, 0x7, %g0 376 andcc %g1, 0x7, %g0
377 bne,pn %icc, 90f 377 bne,pn %icc, 90f
378 andcc %o2, 0x8, %g0 378 andcc %o2, 0x8, %g0
379 be,pt %icc, 1f 379 be,pt %icc, 1f
380 nop 380 nop
381 EX_LD(LOAD(ldx, %o1, %o5), U3_retl_o2) 381 EX_LD(LOAD(ldx, %o1, %o5), U3_retl_o2)
382 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_r 382 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_retl_o2)
383 add %o1, 0x8, %o1 383 add %o1, 0x8, %o1
384 sub %o2, 8, %o2 384 sub %o2, 8, %o2
385 385
386 1: andcc %o2, 0x4, %g0 386 1: andcc %o2, 0x4, %g0
387 be,pt %icc, 1f 387 be,pt %icc, 1f
388 nop 388 nop
389 EX_LD(LOAD(lduw, %o1, %o5), U3_retl_o2 389 EX_LD(LOAD(lduw, %o1, %o5), U3_retl_o2)
390 EX_ST(STORE(stw, %o5, %o1 + %o3), U3_r 390 EX_ST(STORE(stw, %o5, %o1 + %o3), U3_retl_o2)
391 add %o1, 0x4, %o1 391 add %o1, 0x4, %o1
392 sub %o2, 4, %o2 392 sub %o2, 4, %o2
393 393
394 1: andcc %o2, 0x2, %g0 394 1: andcc %o2, 0x2, %g0
395 be,pt %icc, 1f 395 be,pt %icc, 1f
396 nop 396 nop
397 EX_LD(LOAD(lduh, %o1, %o5), U3_retl_o2 397 EX_LD(LOAD(lduh, %o1, %o5), U3_retl_o2)
398 EX_ST(STORE(sth, %o5, %o1 + %o3), U3_r 398 EX_ST(STORE(sth, %o5, %o1 + %o3), U3_retl_o2)
399 add %o1, 0x2, %o1 399 add %o1, 0x2, %o1
400 sub %o2, 2, %o2 400 sub %o2, 2, %o2
401 401
402 1: andcc %o2, 0x1, %g0 402 1: andcc %o2, 0x1, %g0
403 be,pt %icc, end_return 403 be,pt %icc, end_return
404 nop 404 nop
405 EX_LD(LOAD(ldub, %o1, %o5), U3_retl_o2 405 EX_LD(LOAD(ldub, %o1, %o5), U3_retl_o2)
406 ba,pt %xcc, end_return 406 ba,pt %xcc, end_return
407 EX_ST(STORE(stb, %o5, %o1 + %o3), U3_ 407 EX_ST(STORE(stb, %o5, %o1 + %o3), U3_retl_o2)
408 408
409 .align 64 409 .align 64
410 /* 16 <= len < 192 */ 410 /* 16 <= len < 192 */
411 less_than_192: 411 less_than_192:
412 bne,pn %XCC, 75f 412 bne,pn %XCC, 75f
413 sub %o0, %o1, %o3 413 sub %o0, %o1, %o3
414 414
415 72: 415 72:
416 andn %o2, 0xf, GLOBAL_SPARE 416 andn %o2, 0xf, GLOBAL_SPARE
417 and %o2, 0xf, %o2 417 and %o2, 0xf, %o2
418 1: subcc GLOBAL_SPARE, 0x10, GL 418 1: subcc GLOBAL_SPARE, 0x10, GLOBAL_SPARE
419 EX_LD(LOAD(ldx, %o1 + 0x00, %o5), U3_r 419 EX_LD(LOAD(ldx, %o1 + 0x00, %o5), U3_retl_o2_plus_GS_plus_0x10)
420 EX_LD(LOAD(ldx, %o1 + 0x08, %g1), U3_r 420 EX_LD(LOAD(ldx, %o1 + 0x08, %g1), U3_retl_o2_plus_GS_plus_0x10)
421 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_r 421 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_retl_o2_plus_GS_plus_0x10)
422 add %o1, 0x8, %o1 422 add %o1, 0x8, %o1
423 EX_ST(STORE(stx, %g1, %o1 + %o3), U3_r 423 EX_ST(STORE(stx, %g1, %o1 + %o3), U3_retl_o2_plus_GS_plus_0x08)
424 bgu,pt %XCC, 1b 424 bgu,pt %XCC, 1b
425 add %o1, 0x8, %o1 425 add %o1, 0x8, %o1
426 73: andcc %o2, 0x8, %g0 426 73: andcc %o2, 0x8, %g0
427 be,pt %XCC, 1f 427 be,pt %XCC, 1f
428 nop 428 nop
429 sub %o2, 0x8, %o2 429 sub %o2, 0x8, %o2
430 EX_LD(LOAD(ldx, %o1, %o5), U3_retl_o2_ 430 EX_LD(LOAD(ldx, %o1, %o5), U3_retl_o2_plus_8)
431 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_r 431 EX_ST(STORE(stx, %o5, %o1 + %o3), U3_retl_o2_plus_8)
432 add %o1, 0x8, %o1 432 add %o1, 0x8, %o1
433 1: andcc %o2, 0x4, %g0 433 1: andcc %o2, 0x4, %g0
434 be,pt %XCC, 1f 434 be,pt %XCC, 1f
435 nop 435 nop
436 sub %o2, 0x4, %o2 436 sub %o2, 0x4, %o2
437 EX_LD(LOAD(lduw, %o1, %o5), U3_retl_o2 437 EX_LD(LOAD(lduw, %o1, %o5), U3_retl_o2_plus_4)
438 EX_ST(STORE(stw, %o5, %o1 + %o3), U3_r 438 EX_ST(STORE(stw, %o5, %o1 + %o3), U3_retl_o2_plus_4)
439 add %o1, 0x4, %o1 439 add %o1, 0x4, %o1
440 1: cmp %o2, 0 440 1: cmp %o2, 0
441 be,pt %XCC, end_return 441 be,pt %XCC, end_return
442 nop 442 nop
443 ba,pt %xcc, 90f 443 ba,pt %xcc, 90f
444 nop 444 nop
445 445
446 75: 446 75:
447 andcc %o0, 0x7, %g1 447 andcc %o0, 0x7, %g1
448 sub %g1, 0x8, %g1 448 sub %g1, 0x8, %g1
449 be,pn %icc, 2f 449 be,pn %icc, 2f
450 sub %g0, %g1, %g1 450 sub %g0, %g1, %g1
451 sub %o2, %g1, %o2 451 sub %o2, %g1, %o2
452 452
453 1: subcc %g1, 1, %g1 453 1: subcc %g1, 1, %g1
454 EX_LD(LOAD(ldub, %o1, %o5), U3_retl_o2 454 EX_LD(LOAD(ldub, %o1, %o5), U3_retl_o2_plus_g1_plus_1)
455 EX_ST(STORE(stb, %o5, %o1 + %o3), U3_r 455 EX_ST(STORE(stb, %o5, %o1 + %o3), U3_retl_o2_plus_g1_plus_1)
456 bgu,pt %icc, 1b 456 bgu,pt %icc, 1b
457 add %o1, 1, %o1 457 add %o1, 1, %o1
458 458
459 2: add %o1, %o3, %o0 459 2: add %o1, %o3, %o0
460 andcc %o1, 0x7, %g1 460 andcc %o1, 0x7, %g1
461 bne,pt %icc, 8f 461 bne,pt %icc, 8f
462 sll %g1, 3, %g1 462 sll %g1, 3, %g1
463 463
464 cmp %o2, 16 464 cmp %o2, 16
465 bgeu,pt %icc, 72b 465 bgeu,pt %icc, 72b
466 nop 466 nop
467 ba,a,pt %xcc, 73b 467 ba,a,pt %xcc, 73b
468 468
469 8: mov 64, %o3 469 8: mov 64, %o3
470 andn %o1, 0x7, %o1 470 andn %o1, 0x7, %o1
471 EX_LD(LOAD(ldx, %o1, %g2), U3_retl_o2) 471 EX_LD(LOAD(ldx, %o1, %g2), U3_retl_o2)
472 sub %o3, %g1, %o3 472 sub %o3, %g1, %o3
473 andn %o2, 0x7, GLOBAL_SPARE 473 andn %o2, 0x7, GLOBAL_SPARE
474 sllx %g2, %g1, %g2 474 sllx %g2, %g1, %g2
475 1: EX_LD(LOAD(ldx, %o1 + 0x8, %g3), U3_re 475 1: EX_LD(LOAD(ldx, %o1 + 0x8, %g3), U3_retl_o2_and_7_plus_GS)
476 subcc GLOBAL_SPARE, 0x8, GLO 476 subcc GLOBAL_SPARE, 0x8, GLOBAL_SPARE
477 add %o1, 0x8, %o1 477 add %o1, 0x8, %o1
478 srlx %g3, %o3, %o5 478 srlx %g3, %o3, %o5
479 or %o5, %g2, %o5 479 or %o5, %g2, %o5
480 EX_ST(STORE(stx, %o5, %o0), U3_retl_o2 480 EX_ST(STORE(stx, %o5, %o0), U3_retl_o2_and_7_plus_GS_plus_8)
481 add %o0, 0x8, %o0 481 add %o0, 0x8, %o0
482 bgu,pt %icc, 1b 482 bgu,pt %icc, 1b
483 sllx %g3, %g1, %g2 483 sllx %g3, %g1, %g2
484 484
485 srl %g1, 3, %g1 485 srl %g1, 3, %g1
486 andcc %o2, 0x7, %o2 486 andcc %o2, 0x7, %o2
487 be,pn %icc, end_return 487 be,pn %icc, end_return
488 add %o1, %g1, %o1 488 add %o1, %g1, %o1
489 ba,pt %xcc, 90f 489 ba,pt %xcc, 90f
490 sub %o0, %o1, %o3 490 sub %o0, %o1, %o3
491 491
492 .align 64 492 .align 64
493 /* 0 < len < 16 */ 493 /* 0 < len < 16 */
494 less_than_16: 494 less_than_16:
495 andcc %o3, 0x3, %g0 495 andcc %o3, 0x3, %g0
496 bne,pn %XCC, 90f 496 bne,pn %XCC, 90f
497 sub %o0, %o1, %o3 497 sub %o0, %o1, %o3
498 498
499 1: 499 1:
500 subcc %o2, 4, %o2 500 subcc %o2, 4, %o2
501 EX_LD(LOAD(lduw, %o1, %g1), U3_retl_o2 501 EX_LD(LOAD(lduw, %o1, %g1), U3_retl_o2_plus_4)
502 EX_ST(STORE(stw, %g1, %o1 + %o3), U3_r 502 EX_ST(STORE(stw, %g1, %o1 + %o3), U3_retl_o2_plus_4)
503 bgu,pt %XCC, 1b 503 bgu,pt %XCC, 1b
504 add %o1, 4, %o1 504 add %o1, 4, %o1
505 505
506 end_return: 506 end_return:
507 retl 507 retl
508 mov EX_RETVAL(%o4), %o0 508 mov EX_RETVAL(%o4), %o0
509 509
510 .align 32 510 .align 32
511 90: 511 90:
512 subcc %o2, 1, %o2 512 subcc %o2, 1, %o2
513 EX_LD(LOAD(ldub, %o1, %g1), U3_retl_o2 513 EX_LD(LOAD(ldub, %o1, %g1), U3_retl_o2_plus_1)
514 EX_ST(STORE(stb, %g1, %o1 + %o3), U3_r 514 EX_ST(STORE(stb, %g1, %o1 + %o3), U3_retl_o2_plus_1)
515 bgu,pt %XCC, 90b 515 bgu,pt %XCC, 90b
516 add %o1, 1, %o1 516 add %o1, 1, %o1
517 retl 517 retl
518 mov EX_RETVAL(%o4), %o0 518 mov EX_RETVAL(%o4), %o0
519 519
520 .size FUNC_NAME, .-FUNC_NAME 520 .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.