1 /* SPDX-License-Identifier: GPL-2.0 */ 1 /* SPDX-License-Identifier: GPL-2.0 */
2 /* NG4memcpy.S: Niagara-4 optimized memcpy. 2 /* NG4memcpy.S: Niagara-4 optimized memcpy.
3 * 3 *
4 * Copyright (C) 2012 David S. Miller (davem@d 4 * Copyright (C) 2012 David S. Miller (davem@davemloft.net)
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_INIT_QUAD_LDD_P 0xe2 13 #define ASI_BLK_INIT_QUAD_LDD_P 0xe2
14 #define FPRS_FEF 0x04 14 #define FPRS_FEF 0x04
15 15
16 /* On T4 it is very expensive to access ASRs l 16 /* On T4 it is very expensive to access ASRs like %fprs and
17 * %asi, avoiding a read or a write can save ~ 17 * %asi, avoiding a read or a write can save ~50 cycles.
18 */ 18 */
19 #define FPU_ENTER \ 19 #define FPU_ENTER \
20 rd %fprs, %o5; \ 20 rd %fprs, %o5; \
21 andcc %o5, FPRS_FEF, %g0; \ 21 andcc %o5, FPRS_FEF, %g0; \
22 be,a,pn %icc, 999f; \ 22 be,a,pn %icc, 999f; \
23 wr %g0, FPRS_FEF, %fprs; \ 23 wr %g0, FPRS_FEF, %fprs; \
24 999: 24 999:
25 25
26 #ifdef MEMCPY_DEBUG 26 #ifdef MEMCPY_DEBUG
27 #define VISEntryHalf FPU_ENTER; \ 27 #define VISEntryHalf FPU_ENTER; \
28 clr %g1; clr %g2; clr %g3 28 clr %g1; clr %g2; clr %g3; clr %g5; subcc %g0, %g0, %g0;
29 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 29 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
30 #else 30 #else
31 #define VISEntryHalf FPU_ENTER 31 #define VISEntryHalf FPU_ENTER
32 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 32 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
33 #endif 33 #endif
34 34
35 #define GLOBAL_SPARE %g5 35 #define GLOBAL_SPARE %g5
36 #endif 36 #endif
37 37
38 #ifndef STORE_ASI 38 #ifndef STORE_ASI
39 #ifndef SIMULATE_NIAGARA_ON_NON_NIAGARA 39 #ifndef SIMULATE_NIAGARA_ON_NON_NIAGARA
40 #define STORE_ASI ASI_BLK_INIT_QUAD_LDD_ 40 #define STORE_ASI ASI_BLK_INIT_QUAD_LDD_P
41 #else 41 #else
42 #define STORE_ASI 0x80 /* ASI 42 #define STORE_ASI 0x80 /* ASI_P */
43 #endif 43 #endif
44 #endif 44 #endif
45 45
46 #if !defined(EX_LD) && !defined(EX_ST) 46 #if !defined(EX_LD) && !defined(EX_ST)
47 #define NON_USER_COPY 47 #define NON_USER_COPY
48 #endif 48 #endif
49 49
50 #ifndef EX_LD 50 #ifndef EX_LD
51 #define EX_LD(x,y) x 51 #define EX_LD(x,y) x
52 #endif 52 #endif
53 #ifndef EX_LD_FP 53 #ifndef EX_LD_FP
54 #define EX_LD_FP(x,y) x 54 #define EX_LD_FP(x,y) x
55 #endif 55 #endif
56 56
57 #ifndef EX_ST 57 #ifndef EX_ST
58 #define EX_ST(x,y) x 58 #define EX_ST(x,y) x
59 #endif 59 #endif
60 #ifndef EX_ST_FP 60 #ifndef EX_ST_FP
61 #define EX_ST_FP(x,y) x 61 #define EX_ST_FP(x,y) x
62 #endif 62 #endif
63 63
64 64
65 #ifndef LOAD 65 #ifndef LOAD
66 #define LOAD(type,addr,dest) type [addr], d 66 #define LOAD(type,addr,dest) type [addr], dest
67 #endif 67 #endif
68 68
69 #ifndef STORE 69 #ifndef STORE
70 #ifndef MEMCPY_DEBUG 70 #ifndef MEMCPY_DEBUG
71 #define STORE(type,src,addr) type src, [add 71 #define STORE(type,src,addr) type src, [addr]
72 #else 72 #else
73 #define STORE(type,src,addr) type##a src, [ 73 #define STORE(type,src,addr) type##a src, [addr] %asi
74 #endif 74 #endif
75 #endif 75 #endif
76 76
77 #ifndef STORE_INIT 77 #ifndef STORE_INIT
78 #define STORE_INIT(src,addr) stxa src, [add 78 #define STORE_INIT(src,addr) stxa src, [addr] STORE_ASI
79 #endif 79 #endif
80 80
81 #ifndef FUNC_NAME 81 #ifndef FUNC_NAME
82 #define FUNC_NAME NG4memcpy 82 #define FUNC_NAME NG4memcpy
83 #endif 83 #endif
84 #ifndef PREAMBLE 84 #ifndef PREAMBLE
85 #define PREAMBLE 85 #define PREAMBLE
86 #endif 86 #endif
87 87
88 #ifndef XCC 88 #ifndef XCC
89 #define XCC xcc 89 #define XCC xcc
90 #endif 90 #endif
91 91
92 .register %g2,#scratch 92 .register %g2,#scratch
93 .register %g3,#scratch 93 .register %g3,#scratch
94 94
95 .text 95 .text
96 #ifndef EX_RETVAL 96 #ifndef EX_RETVAL
97 #define EX_RETVAL(x) x 97 #define EX_RETVAL(x) x
98 #endif 98 #endif
99 .align 64 99 .align 64
100 100
101 .globl FUNC_NAME 101 .globl FUNC_NAME
102 .type FUNC_NAME,#function 102 .type FUNC_NAME,#function
103 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len * 103 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len */
104 #ifdef MEMCPY_DEBUG 104 #ifdef MEMCPY_DEBUG
105 wr %g0, 0x80, %asi 105 wr %g0, 0x80, %asi
106 #endif 106 #endif
107 srlx %o2, 31, %g2 107 srlx %o2, 31, %g2
108 cmp %g2, 0 108 cmp %g2, 0
109 tne %XCC, 5 109 tne %XCC, 5
110 PREAMBLE 110 PREAMBLE
111 mov %o0, %o3 111 mov %o0, %o3
112 brz,pn %o2, .Lexit 112 brz,pn %o2, .Lexit
113 cmp %o2, 3 113 cmp %o2, 3
114 ble,pn %icc, .Ltiny 114 ble,pn %icc, .Ltiny
115 cmp %o2, 19 115 cmp %o2, 19
116 ble,pn %icc, .Lsmall 116 ble,pn %icc, .Lsmall
117 or %o0, %o1, %g2 117 or %o0, %o1, %g2
118 cmp %o2, 128 118 cmp %o2, 128
119 bl,pn %icc, .Lmedium 119 bl,pn %icc, .Lmedium
120 nop 120 nop
121 121
122 .Llarge:/* len >= 0x80 */ 122 .Llarge:/* len >= 0x80 */
123 /* First get dest 8 byte aligned. */ 123 /* First get dest 8 byte aligned. */
124 sub %g0, %o0, %g1 124 sub %g0, %o0, %g1
125 and %g1, 0x7, %g1 125 and %g1, 0x7, %g1
126 brz,pt %g1, 51f 126 brz,pt %g1, 51f
127 sub %o2, %g1, %o2 127 sub %o2, %g1, %o2
128 128
129 129
130 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2), mem 130 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2), memcpy_retl_o2_plus_g1)
131 add %o1, 1, %o1 131 add %o1, 1, %o1
132 subcc %g1, 1, %g1 132 subcc %g1, 1, %g1
133 add %o0, 1, %o0 133 add %o0, 1, %o0
134 bne,pt %icc, 1b 134 bne,pt %icc, 1b
135 EX_ST(STORE(stb, %g2, %o0 - 0x01), me 135 EX_ST(STORE(stb, %g2, %o0 - 0x01), memcpy_retl_o2_plus_g1_plus_1)
136 136
137 51: LOAD(prefetch, %o1 + 0x040, #n_reads_s 137 51: LOAD(prefetch, %o1 + 0x040, #n_reads_strong)
138 LOAD(prefetch, %o1 + 0x080, #n_reads_s 138 LOAD(prefetch, %o1 + 0x080, #n_reads_strong)
139 LOAD(prefetch, %o1 + 0x0c0, #n_reads_s 139 LOAD(prefetch, %o1 + 0x0c0, #n_reads_strong)
140 LOAD(prefetch, %o1 + 0x100, #n_reads_s 140 LOAD(prefetch, %o1 + 0x100, #n_reads_strong)
141 LOAD(prefetch, %o1 + 0x140, #n_reads_s 141 LOAD(prefetch, %o1 + 0x140, #n_reads_strong)
142 LOAD(prefetch, %o1 + 0x180, #n_reads_s 142 LOAD(prefetch, %o1 + 0x180, #n_reads_strong)
143 LOAD(prefetch, %o1 + 0x1c0, #n_reads_s 143 LOAD(prefetch, %o1 + 0x1c0, #n_reads_strong)
144 LOAD(prefetch, %o1 + 0x200, #n_reads_s 144 LOAD(prefetch, %o1 + 0x200, #n_reads_strong)
145 145
146 /* Check if we can use the straight fu 146 /* Check if we can use the straight fully aligned
147 * loop, or we require the alignaddr/f 147 * loop, or we require the alignaddr/faligndata variant.
148 */ 148 */
149 andcc %o1, 0x7, %o5 149 andcc %o1, 0x7, %o5
150 bne,pn %icc, .Llarge_src_unal 150 bne,pn %icc, .Llarge_src_unaligned
151 sub %g0, %o0, %g1 151 sub %g0, %o0, %g1
152 152
153 /* Legitimize the use of initializing 153 /* Legitimize the use of initializing stores by getting dest
154 * to be 64-byte aligned. 154 * to be 64-byte aligned.
155 */ 155 */
156 and %g1, 0x3f, %g1 156 and %g1, 0x3f, %g1
157 brz,pt %g1, .Llarge_aligned 157 brz,pt %g1, .Llarge_aligned
158 sub %o2, %g1, %o2 158 sub %o2, %g1, %o2
159 159
160 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g2), memc 160 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g2), memcpy_retl_o2_plus_g1)
161 add %o1, 8, %o1 161 add %o1, 8, %o1
162 subcc %g1, 8, %g1 162 subcc %g1, 8, %g1
163 add %o0, 8, %o0 163 add %o0, 8, %o0
164 bne,pt %icc, 1b 164 bne,pt %icc, 1b
165 EX_ST(STORE(stx, %g2, %o0 - 0x08), me 165 EX_ST(STORE(stx, %g2, %o0 - 0x08), memcpy_retl_o2_plus_g1_plus_8)
166 166
167 .Llarge_aligned: 167 .Llarge_aligned:
168 /* len >= 0x80 && src 8-byte aligned & 168 /* len >= 0x80 && src 8-byte aligned && dest 8-byte aligned */
169 andn %o2, 0x3f, %o4 169 andn %o2, 0x3f, %o4
170 sub %o2, %o4, %o2 170 sub %o2, %o4, %o2
171 171
172 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memc 172 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memcpy_retl_o2_plus_o4)
173 add %o1, 0x40, %o1 173 add %o1, 0x40, %o1
174 EX_LD(LOAD(ldx, %o1 - 0x38, %g2), memc 174 EX_LD(LOAD(ldx, %o1 - 0x38, %g2), memcpy_retl_o2_plus_o4)
175 subcc %o4, 0x40, %o4 175 subcc %o4, 0x40, %o4
176 EX_LD(LOAD(ldx, %o1 - 0x30, %g3), memc 176 EX_LD(LOAD(ldx, %o1 - 0x30, %g3), memcpy_retl_o2_plus_o4_plus_64)
177 EX_LD(LOAD(ldx, %o1 - 0x28, GLOBAL_SPA 177 EX_LD(LOAD(ldx, %o1 - 0x28, GLOBAL_SPARE), memcpy_retl_o2_plus_o4_plus_64)
178 EX_LD(LOAD(ldx, %o1 - 0x20, %o5), memc 178 EX_LD(LOAD(ldx, %o1 - 0x20, %o5), memcpy_retl_o2_plus_o4_plus_64)
179 EX_ST(STORE_INIT(%g1, %o0), memcpy_ret 179 EX_ST(STORE_INIT(%g1, %o0), memcpy_retl_o2_plus_o4_plus_64)
180 add %o0, 0x08, %o0 180 add %o0, 0x08, %o0
181 EX_ST(STORE_INIT(%g2, %o0), memcpy_ret 181 EX_ST(STORE_INIT(%g2, %o0), memcpy_retl_o2_plus_o4_plus_56)
182 add %o0, 0x08, %o0 182 add %o0, 0x08, %o0
183 EX_LD(LOAD(ldx, %o1 - 0x18, %g2), memc 183 EX_LD(LOAD(ldx, %o1 - 0x18, %g2), memcpy_retl_o2_plus_o4_plus_48)
184 EX_ST(STORE_INIT(%g3, %o0), memcpy_ret 184 EX_ST(STORE_INIT(%g3, %o0), memcpy_retl_o2_plus_o4_plus_48)
185 add %o0, 0x08, %o0 185 add %o0, 0x08, %o0
186 EX_LD(LOAD(ldx, %o1 - 0x10, %g3), memc 186 EX_LD(LOAD(ldx, %o1 - 0x10, %g3), memcpy_retl_o2_plus_o4_plus_40)
187 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0), m 187 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0), memcpy_retl_o2_plus_o4_plus_40)
188 add %o0, 0x08, %o0 188 add %o0, 0x08, %o0
189 EX_LD(LOAD(ldx, %o1 - 0x08, GLOBAL_SPA 189 EX_LD(LOAD(ldx, %o1 - 0x08, GLOBAL_SPARE), memcpy_retl_o2_plus_o4_plus_32)
190 EX_ST(STORE_INIT(%o5, %o0), memcpy_ret 190 EX_ST(STORE_INIT(%o5, %o0), memcpy_retl_o2_plus_o4_plus_32)
191 add %o0, 0x08, %o0 191 add %o0, 0x08, %o0
192 EX_ST(STORE_INIT(%g2, %o0), memcpy_ret 192 EX_ST(STORE_INIT(%g2, %o0), memcpy_retl_o2_plus_o4_plus_24)
193 add %o0, 0x08, %o0 193 add %o0, 0x08, %o0
194 EX_ST(STORE_INIT(%g3, %o0), memcpy_ret 194 EX_ST(STORE_INIT(%g3, %o0), memcpy_retl_o2_plus_o4_plus_16)
195 add %o0, 0x08, %o0 195 add %o0, 0x08, %o0
196 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0), m 196 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0), memcpy_retl_o2_plus_o4_plus_8)
197 add %o0, 0x08, %o0 197 add %o0, 0x08, %o0
198 bne,pt %icc, 1b 198 bne,pt %icc, 1b
199 LOAD(prefetch, %o1 + 0x200, #n_reads_ 199 LOAD(prefetch, %o1 + 0x200, #n_reads_strong)
200 200
201 membar #StoreLoad | #StoreSto 201 membar #StoreLoad | #StoreStore
202 202
203 brz,pn %o2, .Lexit 203 brz,pn %o2, .Lexit
204 cmp %o2, 19 204 cmp %o2, 19
205 ble,pn %icc, .Lsmall_unaligne 205 ble,pn %icc, .Lsmall_unaligned
206 nop 206 nop
207 ba,a,pt %icc, .Lmedium_noprefe 207 ba,a,pt %icc, .Lmedium_noprefetch
208 208
209 .Lexit: retl 209 .Lexit: retl
210 mov EX_RETVAL(%o3), %o0 210 mov EX_RETVAL(%o3), %o0
211 211
212 .Llarge_src_unaligned: 212 .Llarge_src_unaligned:
213 #ifdef NON_USER_COPY 213 #ifdef NON_USER_COPY
214 VISEntryHalfFast(.Lmedium_vis_entry_fa 214 VISEntryHalfFast(.Lmedium_vis_entry_fail)
215 #else 215 #else
216 VISEntryHalf 216 VISEntryHalf
217 #endif 217 #endif
218 andn %o2, 0x3f, %o4 218 andn %o2, 0x3f, %o4
219 sub %o2, %o4, %o2 219 sub %o2, %o4, %o2
220 alignaddr %o1, %g0, %g1 220 alignaddr %o1, %g0, %g1
221 add %o1, %o4, %o1 221 add %o1, %o4, %o1
222 EX_LD_FP(LOAD(ldd, %g1 + 0x00, %f0), m 222 EX_LD_FP(LOAD(ldd, %g1 + 0x00, %f0), memcpy_retl_o2_plus_o4)
223 1: EX_LD_FP(LOAD(ldd, %g1 + 0x08, %f2), m 223 1: EX_LD_FP(LOAD(ldd, %g1 + 0x08, %f2), memcpy_retl_o2_plus_o4)
224 subcc %o4, 0x40, %o4 224 subcc %o4, 0x40, %o4
225 EX_LD_FP(LOAD(ldd, %g1 + 0x10, %f4), m 225 EX_LD_FP(LOAD(ldd, %g1 + 0x10, %f4), memcpy_retl_o2_plus_o4_plus_64)
226 EX_LD_FP(LOAD(ldd, %g1 + 0x18, %f6), m 226 EX_LD_FP(LOAD(ldd, %g1 + 0x18, %f6), memcpy_retl_o2_plus_o4_plus_64)
227 EX_LD_FP(LOAD(ldd, %g1 + 0x20, %f8), m 227 EX_LD_FP(LOAD(ldd, %g1 + 0x20, %f8), memcpy_retl_o2_plus_o4_plus_64)
228 EX_LD_FP(LOAD(ldd, %g1 + 0x28, %f10), 228 EX_LD_FP(LOAD(ldd, %g1 + 0x28, %f10), memcpy_retl_o2_plus_o4_plus_64)
229 EX_LD_FP(LOAD(ldd, %g1 + 0x30, %f12), 229 EX_LD_FP(LOAD(ldd, %g1 + 0x30, %f12), memcpy_retl_o2_plus_o4_plus_64)
230 EX_LD_FP(LOAD(ldd, %g1 + 0x38, %f14), 230 EX_LD_FP(LOAD(ldd, %g1 + 0x38, %f14), memcpy_retl_o2_plus_o4_plus_64)
231 faligndata %f0, %f2, %f16 231 faligndata %f0, %f2, %f16
232 EX_LD_FP(LOAD(ldd, %g1 + 0x40, %f0), m 232 EX_LD_FP(LOAD(ldd, %g1 + 0x40, %f0), memcpy_retl_o2_plus_o4_plus_64)
233 faligndata %f2, %f4, %f18 233 faligndata %f2, %f4, %f18
234 add %g1, 0x40, %g1 234 add %g1, 0x40, %g1
235 faligndata %f4, %f6, %f20 235 faligndata %f4, %f6, %f20
236 faligndata %f6, %f8, %f22 236 faligndata %f6, %f8, %f22
237 faligndata %f8, %f10, %f24 237 faligndata %f8, %f10, %f24
238 faligndata %f10, %f12, %f26 238 faligndata %f10, %f12, %f26
239 faligndata %f12, %f14, %f28 239 faligndata %f12, %f14, %f28
240 faligndata %f14, %f0, %f30 240 faligndata %f14, %f0, %f30
241 EX_ST_FP(STORE(std, %f16, %o0 + 0x00), 241 EX_ST_FP(STORE(std, %f16, %o0 + 0x00), memcpy_retl_o2_plus_o4_plus_64)
242 EX_ST_FP(STORE(std, %f18, %o0 + 0x08), 242 EX_ST_FP(STORE(std, %f18, %o0 + 0x08), memcpy_retl_o2_plus_o4_plus_56)
243 EX_ST_FP(STORE(std, %f20, %o0 + 0x10), 243 EX_ST_FP(STORE(std, %f20, %o0 + 0x10), memcpy_retl_o2_plus_o4_plus_48)
244 EX_ST_FP(STORE(std, %f22, %o0 + 0x18), 244 EX_ST_FP(STORE(std, %f22, %o0 + 0x18), memcpy_retl_o2_plus_o4_plus_40)
245 EX_ST_FP(STORE(std, %f24, %o0 + 0x20), 245 EX_ST_FP(STORE(std, %f24, %o0 + 0x20), memcpy_retl_o2_plus_o4_plus_32)
246 EX_ST_FP(STORE(std, %f26, %o0 + 0x28), 246 EX_ST_FP(STORE(std, %f26, %o0 + 0x28), memcpy_retl_o2_plus_o4_plus_24)
247 EX_ST_FP(STORE(std, %f28, %o0 + 0x30), 247 EX_ST_FP(STORE(std, %f28, %o0 + 0x30), memcpy_retl_o2_plus_o4_plus_16)
248 EX_ST_FP(STORE(std, %f30, %o0 + 0x38), 248 EX_ST_FP(STORE(std, %f30, %o0 + 0x38), memcpy_retl_o2_plus_o4_plus_8)
249 add %o0, 0x40, %o0 249 add %o0, 0x40, %o0
250 bne,pt %icc, 1b 250 bne,pt %icc, 1b
251 LOAD(prefetch, %g1 + 0x200, #n_reads_ 251 LOAD(prefetch, %g1 + 0x200, #n_reads_strong)
252 #ifdef NON_USER_COPY 252 #ifdef NON_USER_COPY
253 VISExitHalfFast 253 VISExitHalfFast
254 #else 254 #else
255 VISExitHalf 255 VISExitHalf
256 #endif 256 #endif
257 brz,pn %o2, .Lexit 257 brz,pn %o2, .Lexit
258 cmp %o2, 19 258 cmp %o2, 19
259 ble,pn %icc, .Lsmall_unaligne 259 ble,pn %icc, .Lsmall_unaligned
260 nop 260 nop
261 ba,a,pt %icc, .Lmedium_unalign 261 ba,a,pt %icc, .Lmedium_unaligned
262 262
263 #ifdef NON_USER_COPY 263 #ifdef NON_USER_COPY
264 .Lmedium_vis_entry_fail: 264 .Lmedium_vis_entry_fail:
265 or %o0, %o1, %g2 265 or %o0, %o1, %g2
266 #endif 266 #endif
267 .Lmedium: 267 .Lmedium:
268 LOAD(prefetch, %o1 + 0x40, #n_reads_st 268 LOAD(prefetch, %o1 + 0x40, #n_reads_strong)
269 andcc %g2, 0x7, %g0 269 andcc %g2, 0x7, %g0
270 bne,pn %icc, .Lmedium_unalign 270 bne,pn %icc, .Lmedium_unaligned
271 nop 271 nop
272 .Lmedium_noprefetch: 272 .Lmedium_noprefetch:
273 andncc %o2, 0x20 - 1, %o5 273 andncc %o2, 0x20 - 1, %o5
274 be,pn %icc, 2f 274 be,pn %icc, 2f
275 sub %o2, %o5, %o2 275 sub %o2, %o5, %o2
276 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memc 276 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memcpy_retl_o2_plus_o5)
277 EX_LD(LOAD(ldx, %o1 + 0x08, %g2), memc 277 EX_LD(LOAD(ldx, %o1 + 0x08, %g2), memcpy_retl_o2_plus_o5)
278 EX_LD(LOAD(ldx, %o1 + 0x10, GLOBAL_SPA 278 EX_LD(LOAD(ldx, %o1 + 0x10, GLOBAL_SPARE), memcpy_retl_o2_plus_o5)
279 EX_LD(LOAD(ldx, %o1 + 0x18, %o4), memc 279 EX_LD(LOAD(ldx, %o1 + 0x18, %o4), memcpy_retl_o2_plus_o5)
280 add %o1, 0x20, %o1 280 add %o1, 0x20, %o1
281 subcc %o5, 0x20, %o5 281 subcc %o5, 0x20, %o5
282 EX_ST(STORE(stx, %g1, %o0 + 0x00), mem 282 EX_ST(STORE(stx, %g1, %o0 + 0x00), memcpy_retl_o2_plus_o5_plus_32)
283 EX_ST(STORE(stx, %g2, %o0 + 0x08), mem 283 EX_ST(STORE(stx, %g2, %o0 + 0x08), memcpy_retl_o2_plus_o5_plus_24)
284 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0 284 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0x10), memcpy_retl_o2_plus_o5_plus_24)
285 EX_ST(STORE(stx, %o4, %o0 + 0x18), mem 285 EX_ST(STORE(stx, %o4, %o0 + 0x18), memcpy_retl_o2_plus_o5_plus_8)
286 bne,pt %icc, 1b 286 bne,pt %icc, 1b
287 add %o0, 0x20, %o0 287 add %o0, 0x20, %o0
288 2: andcc %o2, 0x18, %o5 288 2: andcc %o2, 0x18, %o5
289 be,pt %icc, 3f 289 be,pt %icc, 3f
290 sub %o2, %o5, %o2 290 sub %o2, %o5, %o2
291 291
292 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memc 292 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memcpy_retl_o2_plus_o5)
293 add %o1, 0x08, %o1 293 add %o1, 0x08, %o1
294 add %o0, 0x08, %o0 294 add %o0, 0x08, %o0
295 subcc %o5, 0x08, %o5 295 subcc %o5, 0x08, %o5
296 bne,pt %icc, 1b 296 bne,pt %icc, 1b
297 EX_ST(STORE(stx, %g1, %o0 - 0x08), me 297 EX_ST(STORE(stx, %g1, %o0 - 0x08), memcpy_retl_o2_plus_o5_plus_8)
298 3: brz,pt %o2, .Lexit 298 3: brz,pt %o2, .Lexit
299 cmp %o2, 0x04 299 cmp %o2, 0x04
300 bl,pn %icc, .Ltiny 300 bl,pn %icc, .Ltiny
301 nop 301 nop
302 EX_LD(LOAD(lduw, %o1 + 0x00, %g1), mem 302 EX_LD(LOAD(lduw, %o1 + 0x00, %g1), memcpy_retl_o2)
303 add %o1, 0x04, %o1 303 add %o1, 0x04, %o1
304 add %o0, 0x04, %o0 304 add %o0, 0x04, %o0
305 subcc %o2, 0x04, %o2 305 subcc %o2, 0x04, %o2
306 bne,pn %icc, .Ltiny 306 bne,pn %icc, .Ltiny
307 EX_ST(STORE(stw, %g1, %o0 - 0x04), me 307 EX_ST(STORE(stw, %g1, %o0 - 0x04), memcpy_retl_o2_plus_4)
308 ba,a,pt %icc, .Lexit 308 ba,a,pt %icc, .Lexit
309 .Lmedium_unaligned: 309 .Lmedium_unaligned:
310 /* First get dest 8 byte aligned. */ 310 /* First get dest 8 byte aligned. */
311 sub %g0, %o0, %g1 311 sub %g0, %o0, %g1
312 and %g1, 0x7, %g1 312 and %g1, 0x7, %g1
313 brz,pt %g1, 2f 313 brz,pt %g1, 2f
314 sub %o2, %g1, %o2 314 sub %o2, %g1, %o2
315 315
316 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2), mem 316 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2), memcpy_retl_o2_plus_g1)
317 add %o1, 1, %o1 317 add %o1, 1, %o1
318 subcc %g1, 1, %g1 318 subcc %g1, 1, %g1
319 add %o0, 1, %o0 319 add %o0, 1, %o0
320 bne,pt %icc, 1b 320 bne,pt %icc, 1b
321 EX_ST(STORE(stb, %g2, %o0 - 0x01), me 321 EX_ST(STORE(stb, %g2, %o0 - 0x01), memcpy_retl_o2_plus_g1_plus_1)
322 2: 322 2:
323 and %o1, 0x7, %g1 323 and %o1, 0x7, %g1
324 brz,pn %g1, .Lmedium_noprefet 324 brz,pn %g1, .Lmedium_noprefetch
325 sll %g1, 3, %g1 325 sll %g1, 3, %g1
326 mov 64, %g2 326 mov 64, %g2
327 sub %g2, %g1, %g2 327 sub %g2, %g1, %g2
328 andn %o1, 0x7, %o1 328 andn %o1, 0x7, %o1
329 EX_LD(LOAD(ldx, %o1 + 0x00, %o4), memc 329 EX_LD(LOAD(ldx, %o1 + 0x00, %o4), memcpy_retl_o2)
330 sllx %o4, %g1, %o4 330 sllx %o4, %g1, %o4
331 andn %o2, 0x08 - 1, %o5 331 andn %o2, 0x08 - 1, %o5
332 sub %o2, %o5, %o2 332 sub %o2, %o5, %o2
333 1: EX_LD(LOAD(ldx, %o1 + 0x08, %g3), memc 333 1: EX_LD(LOAD(ldx, %o1 + 0x08, %g3), memcpy_retl_o2_plus_o5)
334 add %o1, 0x08, %o1 334 add %o1, 0x08, %o1
335 subcc %o5, 0x08, %o5 335 subcc %o5, 0x08, %o5
336 srlx %g3, %g2, GLOBAL_SPARE 336 srlx %g3, %g2, GLOBAL_SPARE
337 or GLOBAL_SPARE, %o4, GLO 337 or GLOBAL_SPARE, %o4, GLOBAL_SPARE
338 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0 338 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0x00), memcpy_retl_o2_plus_o5_plus_8)
339 add %o0, 0x08, %o0 339 add %o0, 0x08, %o0
340 bne,pt %icc, 1b 340 bne,pt %icc, 1b
341 sllx %g3, %g1, %o4 341 sllx %g3, %g1, %o4
342 srl %g1, 3, %g1 342 srl %g1, 3, %g1
343 add %o1, %g1, %o1 343 add %o1, %g1, %o1
344 brz,pn %o2, .Lexit 344 brz,pn %o2, .Lexit
345 nop 345 nop
346 ba,pt %icc, .Lsmall_unaligne 346 ba,pt %icc, .Lsmall_unaligned
347 347
348 .Ltiny: 348 .Ltiny:
349 EX_LD(LOAD(ldub, %o1 + 0x00, %g1), mem 349 EX_LD(LOAD(ldub, %o1 + 0x00, %g1), memcpy_retl_o2)
350 subcc %o2, 1, %o2 350 subcc %o2, 1, %o2
351 be,pn %icc, .Lexit 351 be,pn %icc, .Lexit
352 EX_ST(STORE(stb, %g1, %o0 + 0x00), me 352 EX_ST(STORE(stb, %g1, %o0 + 0x00), memcpy_retl_o2_plus_1)
353 EX_LD(LOAD(ldub, %o1 + 0x01, %g1), mem 353 EX_LD(LOAD(ldub, %o1 + 0x01, %g1), memcpy_retl_o2)
354 subcc %o2, 1, %o2 354 subcc %o2, 1, %o2
355 be,pn %icc, .Lexit 355 be,pn %icc, .Lexit
356 EX_ST(STORE(stb, %g1, %o0 + 0x01), me 356 EX_ST(STORE(stb, %g1, %o0 + 0x01), memcpy_retl_o2_plus_1)
357 EX_LD(LOAD(ldub, %o1 + 0x02, %g1), mem 357 EX_LD(LOAD(ldub, %o1 + 0x02, %g1), memcpy_retl_o2)
358 ba,pt %icc, .Lexit 358 ba,pt %icc, .Lexit
359 EX_ST(STORE(stb, %g1, %o0 + 0x02), me 359 EX_ST(STORE(stb, %g1, %o0 + 0x02), memcpy_retl_o2)
360 360
361 .Lsmall: 361 .Lsmall:
362 andcc %g2, 0x3, %g0 362 andcc %g2, 0x3, %g0
363 bne,pn %icc, .Lsmall_unaligne 363 bne,pn %icc, .Lsmall_unaligned
364 andn %o2, 0x4 - 1, %o5 364 andn %o2, 0x4 - 1, %o5
365 sub %o2, %o5, %o2 365 sub %o2, %o5, %o2
366 1: 366 1:
367 EX_LD(LOAD(lduw, %o1 + 0x00, %g1), mem 367 EX_LD(LOAD(lduw, %o1 + 0x00, %g1), memcpy_retl_o2_plus_o5)
368 add %o1, 0x04, %o1 368 add %o1, 0x04, %o1
369 subcc %o5, 0x04, %o5 369 subcc %o5, 0x04, %o5
370 add %o0, 0x04, %o0 370 add %o0, 0x04, %o0
371 bne,pt %icc, 1b 371 bne,pt %icc, 1b
372 EX_ST(STORE(stw, %g1, %o0 - 0x04), me 372 EX_ST(STORE(stw, %g1, %o0 - 0x04), memcpy_retl_o2_plus_o5_plus_4)
373 brz,pt %o2, .Lexit 373 brz,pt %o2, .Lexit
374 nop 374 nop
375 ba,a,pt %icc, .Ltiny 375 ba,a,pt %icc, .Ltiny
376 376
377 .Lsmall_unaligned: 377 .Lsmall_unaligned:
378 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g1), mem 378 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g1), memcpy_retl_o2)
379 add %o1, 1, %o1 379 add %o1, 1, %o1
380 add %o0, 1, %o0 380 add %o0, 1, %o0
381 subcc %o2, 1, %o2 381 subcc %o2, 1, %o2
382 bne,pt %icc, 1b 382 bne,pt %icc, 1b
383 EX_ST(STORE(stb, %g1, %o0 - 0x01), me 383 EX_ST(STORE(stb, %g1, %o0 - 0x01), memcpy_retl_o2_plus_1)
384 ba,a,pt %icc, .Lexit 384 ba,a,pt %icc, .Lexit
385 nop 385 nop
386 .size FUNC_NAME, .-FUNC_NAME 386 .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.