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