1 /* SPDX-License-Identifier: GPL-2.0 */ <<
2 /* NG4memcpy.S: Niagara-4 optimized memcpy. 1 /* NG4memcpy.S: Niagara-4 optimized memcpy.
3 * 2 *
4 * Copyright (C) 2012 David S. Miller (davem@d 3 * Copyright (C) 2012 David S. Miller (davem@davemloft.net)
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_INIT_QUAD_LDD_P 0xe2 11 #define ASI_BLK_INIT_QUAD_LDD_P 0xe2
14 #define FPRS_FEF 0x04 12 #define FPRS_FEF 0x04
15 13
16 /* On T4 it is very expensive to access ASRs l 14 /* On T4 it is very expensive to access ASRs like %fprs and
17 * %asi, avoiding a read or a write can save ~ 15 * %asi, avoiding a read or a write can save ~50 cycles.
18 */ 16 */
19 #define FPU_ENTER \ 17 #define FPU_ENTER \
20 rd %fprs, %o5; \ 18 rd %fprs, %o5; \
21 andcc %o5, FPRS_FEF, %g0; \ 19 andcc %o5, FPRS_FEF, %g0; \
22 be,a,pn %icc, 999f; \ 20 be,a,pn %icc, 999f; \
23 wr %g0, FPRS_FEF, %fprs; \ 21 wr %g0, FPRS_FEF, %fprs; \
24 999: 22 999:
25 23
26 #ifdef MEMCPY_DEBUG 24 #ifdef MEMCPY_DEBUG
27 #define VISEntryHalf FPU_ENTER; \ 25 #define VISEntryHalf FPU_ENTER; \
28 clr %g1; clr %g2; clr %g3 26 clr %g1; clr %g2; clr %g3; clr %g5; subcc %g0, %g0, %g0;
29 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 27 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
30 #else 28 #else
31 #define VISEntryHalf FPU_ENTER 29 #define VISEntryHalf FPU_ENTER
32 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 30 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
33 #endif 31 #endif
34 32
35 #define GLOBAL_SPARE %g5 33 #define GLOBAL_SPARE %g5
36 #endif 34 #endif
37 35
38 #ifndef STORE_ASI 36 #ifndef STORE_ASI
39 #ifndef SIMULATE_NIAGARA_ON_NON_NIAGARA 37 #ifndef SIMULATE_NIAGARA_ON_NON_NIAGARA
40 #define STORE_ASI ASI_BLK_INIT_QUAD_LDD_ 38 #define STORE_ASI ASI_BLK_INIT_QUAD_LDD_P
41 #else 39 #else
42 #define STORE_ASI 0x80 /* ASI 40 #define STORE_ASI 0x80 /* ASI_P */
43 #endif 41 #endif
44 #endif 42 #endif
45 43
46 #if !defined(EX_LD) && !defined(EX_ST) 44 #if !defined(EX_LD) && !defined(EX_ST)
47 #define NON_USER_COPY 45 #define NON_USER_COPY
48 #endif 46 #endif
49 47
50 #ifndef EX_LD 48 #ifndef EX_LD
51 #define EX_LD(x,y) x !! 49 #define EX_LD(x) x
52 #endif <<
53 #ifndef EX_LD_FP <<
54 #define EX_LD_FP(x,y) x <<
55 #endif 50 #endif
56 51
57 #ifndef EX_ST 52 #ifndef EX_ST
58 #define EX_ST(x,y) x !! 53 #define EX_ST(x) x
59 #endif <<
60 #ifndef EX_ST_FP <<
61 #define EX_ST_FP(x,y) x <<
62 #endif 54 #endif
63 55
>> 56 #ifndef EX_RETVAL
>> 57 #define EX_RETVAL(x) x
>> 58 #endif
64 59
65 #ifndef LOAD 60 #ifndef LOAD
66 #define LOAD(type,addr,dest) type [addr], d 61 #define LOAD(type,addr,dest) type [addr], dest
67 #endif 62 #endif
68 63
69 #ifndef STORE 64 #ifndef STORE
70 #ifndef MEMCPY_DEBUG 65 #ifndef MEMCPY_DEBUG
71 #define STORE(type,src,addr) type src, [add 66 #define STORE(type,src,addr) type src, [addr]
72 #else 67 #else
73 #define STORE(type,src,addr) type##a src, [ 68 #define STORE(type,src,addr) type##a src, [addr] %asi
74 #endif 69 #endif
75 #endif 70 #endif
76 71
77 #ifndef STORE_INIT 72 #ifndef STORE_INIT
78 #define STORE_INIT(src,addr) stxa src, [add 73 #define STORE_INIT(src,addr) stxa src, [addr] STORE_ASI
79 #endif 74 #endif
80 75
81 #ifndef FUNC_NAME 76 #ifndef FUNC_NAME
82 #define FUNC_NAME NG4memcpy 77 #define FUNC_NAME NG4memcpy
83 #endif 78 #endif
84 #ifndef PREAMBLE 79 #ifndef PREAMBLE
85 #define PREAMBLE 80 #define PREAMBLE
86 #endif 81 #endif
87 82
88 #ifndef XCC 83 #ifndef XCC
89 #define XCC xcc 84 #define XCC xcc
90 #endif 85 #endif
91 86
92 .register %g2,#scratch 87 .register %g2,#scratch
93 .register %g3,#scratch 88 .register %g3,#scratch
94 89
95 .text 90 .text
96 #ifndef EX_RETVAL <<
97 #define EX_RETVAL(x) x <<
98 #endif <<
99 .align 64 91 .align 64
100 92
101 .globl FUNC_NAME 93 .globl FUNC_NAME
102 .type FUNC_NAME,#function 94 .type FUNC_NAME,#function
103 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len * 95 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len */
104 #ifdef MEMCPY_DEBUG 96 #ifdef MEMCPY_DEBUG
105 wr %g0, 0x80, %asi 97 wr %g0, 0x80, %asi
106 #endif 98 #endif
107 srlx %o2, 31, %g2 99 srlx %o2, 31, %g2
108 cmp %g2, 0 100 cmp %g2, 0
109 tne %XCC, 5 101 tne %XCC, 5
110 PREAMBLE 102 PREAMBLE
111 mov %o0, %o3 103 mov %o0, %o3
112 brz,pn %o2, .Lexit 104 brz,pn %o2, .Lexit
113 cmp %o2, 3 105 cmp %o2, 3
114 ble,pn %icc, .Ltiny 106 ble,pn %icc, .Ltiny
115 cmp %o2, 19 107 cmp %o2, 19
116 ble,pn %icc, .Lsmall 108 ble,pn %icc, .Lsmall
117 or %o0, %o1, %g2 109 or %o0, %o1, %g2
118 cmp %o2, 128 110 cmp %o2, 128
119 bl,pn %icc, .Lmedium 111 bl,pn %icc, .Lmedium
120 nop 112 nop
121 113
122 .Llarge:/* len >= 0x80 */ 114 .Llarge:/* len >= 0x80 */
123 /* First get dest 8 byte aligned. */ 115 /* First get dest 8 byte aligned. */
124 sub %g0, %o0, %g1 116 sub %g0, %o0, %g1
125 and %g1, 0x7, %g1 117 and %g1, 0x7, %g1
126 brz,pt %g1, 51f 118 brz,pt %g1, 51f
127 sub %o2, %g1, %o2 119 sub %o2, %g1, %o2
128 120
129 !! 121 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2))
130 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2), mem <<
131 add %o1, 1, %o1 122 add %o1, 1, %o1
132 subcc %g1, 1, %g1 123 subcc %g1, 1, %g1
133 add %o0, 1, %o0 124 add %o0, 1, %o0
134 bne,pt %icc, 1b 125 bne,pt %icc, 1b
135 EX_ST(STORE(stb, %g2, %o0 - 0x01), me !! 126 EX_ST(STORE(stb, %g2, %o0 - 0x01))
136 127
137 51: LOAD(prefetch, %o1 + 0x040, #n_reads_s 128 51: LOAD(prefetch, %o1 + 0x040, #n_reads_strong)
138 LOAD(prefetch, %o1 + 0x080, #n_reads_s 129 LOAD(prefetch, %o1 + 0x080, #n_reads_strong)
139 LOAD(prefetch, %o1 + 0x0c0, #n_reads_s 130 LOAD(prefetch, %o1 + 0x0c0, #n_reads_strong)
140 LOAD(prefetch, %o1 + 0x100, #n_reads_s 131 LOAD(prefetch, %o1 + 0x100, #n_reads_strong)
141 LOAD(prefetch, %o1 + 0x140, #n_reads_s 132 LOAD(prefetch, %o1 + 0x140, #n_reads_strong)
142 LOAD(prefetch, %o1 + 0x180, #n_reads_s 133 LOAD(prefetch, %o1 + 0x180, #n_reads_strong)
143 LOAD(prefetch, %o1 + 0x1c0, #n_reads_s 134 LOAD(prefetch, %o1 + 0x1c0, #n_reads_strong)
144 LOAD(prefetch, %o1 + 0x200, #n_reads_s 135 LOAD(prefetch, %o1 + 0x200, #n_reads_strong)
145 136
146 /* Check if we can use the straight fu 137 /* Check if we can use the straight fully aligned
147 * loop, or we require the alignaddr/f 138 * loop, or we require the alignaddr/faligndata variant.
148 */ 139 */
149 andcc %o1, 0x7, %o5 140 andcc %o1, 0x7, %o5
150 bne,pn %icc, .Llarge_src_unal 141 bne,pn %icc, .Llarge_src_unaligned
151 sub %g0, %o0, %g1 142 sub %g0, %o0, %g1
152 143
153 /* Legitimize the use of initializing 144 /* Legitimize the use of initializing stores by getting dest
154 * to be 64-byte aligned. 145 * to be 64-byte aligned.
155 */ 146 */
156 and %g1, 0x3f, %g1 147 and %g1, 0x3f, %g1
157 brz,pt %g1, .Llarge_aligned 148 brz,pt %g1, .Llarge_aligned
158 sub %o2, %g1, %o2 149 sub %o2, %g1, %o2
159 150
160 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g2), memc !! 151 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g2))
161 add %o1, 8, %o1 152 add %o1, 8, %o1
162 subcc %g1, 8, %g1 153 subcc %g1, 8, %g1
163 add %o0, 8, %o0 154 add %o0, 8, %o0
164 bne,pt %icc, 1b 155 bne,pt %icc, 1b
165 EX_ST(STORE(stx, %g2, %o0 - 0x08), me !! 156 EX_ST(STORE(stx, %g2, %o0 - 0x08))
166 157
167 .Llarge_aligned: 158 .Llarge_aligned:
168 /* len >= 0x80 && src 8-byte aligned & 159 /* len >= 0x80 && src 8-byte aligned && dest 8-byte aligned */
169 andn %o2, 0x3f, %o4 160 andn %o2, 0x3f, %o4
170 sub %o2, %o4, %o2 161 sub %o2, %o4, %o2
171 162
172 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memc !! 163 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1))
173 add %o1, 0x40, %o1 164 add %o1, 0x40, %o1
174 EX_LD(LOAD(ldx, %o1 - 0x38, %g2), memc !! 165 EX_LD(LOAD(ldx, %o1 - 0x38, %g2))
175 subcc %o4, 0x40, %o4 166 subcc %o4, 0x40, %o4
176 EX_LD(LOAD(ldx, %o1 - 0x30, %g3), memc !! 167 EX_LD(LOAD(ldx, %o1 - 0x30, %g3))
177 EX_LD(LOAD(ldx, %o1 - 0x28, GLOBAL_SPA !! 168 EX_LD(LOAD(ldx, %o1 - 0x28, GLOBAL_SPARE))
178 EX_LD(LOAD(ldx, %o1 - 0x20, %o5), memc !! 169 EX_LD(LOAD(ldx, %o1 - 0x20, %o5))
179 EX_ST(STORE_INIT(%g1, %o0), memcpy_ret !! 170 EX_ST(STORE_INIT(%g1, %o0))
180 add %o0, 0x08, %o0 171 add %o0, 0x08, %o0
181 EX_ST(STORE_INIT(%g2, %o0), memcpy_ret !! 172 EX_ST(STORE_INIT(%g2, %o0))
182 add %o0, 0x08, %o0 173 add %o0, 0x08, %o0
183 EX_LD(LOAD(ldx, %o1 - 0x18, %g2), memc !! 174 EX_LD(LOAD(ldx, %o1 - 0x18, %g2))
184 EX_ST(STORE_INIT(%g3, %o0), memcpy_ret !! 175 EX_ST(STORE_INIT(%g3, %o0))
185 add %o0, 0x08, %o0 176 add %o0, 0x08, %o0
186 EX_LD(LOAD(ldx, %o1 - 0x10, %g3), memc !! 177 EX_LD(LOAD(ldx, %o1 - 0x10, %g3))
187 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0), m !! 178 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0))
188 add %o0, 0x08, %o0 179 add %o0, 0x08, %o0
189 EX_LD(LOAD(ldx, %o1 - 0x08, GLOBAL_SPA !! 180 EX_LD(LOAD(ldx, %o1 - 0x08, GLOBAL_SPARE))
190 EX_ST(STORE_INIT(%o5, %o0), memcpy_ret !! 181 EX_ST(STORE_INIT(%o5, %o0))
191 add %o0, 0x08, %o0 182 add %o0, 0x08, %o0
192 EX_ST(STORE_INIT(%g2, %o0), memcpy_ret !! 183 EX_ST(STORE_INIT(%g2, %o0))
193 add %o0, 0x08, %o0 184 add %o0, 0x08, %o0
194 EX_ST(STORE_INIT(%g3, %o0), memcpy_ret !! 185 EX_ST(STORE_INIT(%g3, %o0))
195 add %o0, 0x08, %o0 186 add %o0, 0x08, %o0
196 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0), m !! 187 EX_ST(STORE_INIT(GLOBAL_SPARE, %o0))
197 add %o0, 0x08, %o0 188 add %o0, 0x08, %o0
198 bne,pt %icc, 1b 189 bne,pt %icc, 1b
199 LOAD(prefetch, %o1 + 0x200, #n_reads_ 190 LOAD(prefetch, %o1 + 0x200, #n_reads_strong)
200 191
201 membar #StoreLoad | #StoreSto 192 membar #StoreLoad | #StoreStore
202 193
203 brz,pn %o2, .Lexit 194 brz,pn %o2, .Lexit
204 cmp %o2, 19 195 cmp %o2, 19
205 ble,pn %icc, .Lsmall_unaligne 196 ble,pn %icc, .Lsmall_unaligned
206 nop 197 nop
207 ba,a,pt %icc, .Lmedium_noprefe 198 ba,a,pt %icc, .Lmedium_noprefetch
208 199
209 .Lexit: retl 200 .Lexit: retl
210 mov EX_RETVAL(%o3), %o0 201 mov EX_RETVAL(%o3), %o0
211 202
212 .Llarge_src_unaligned: 203 .Llarge_src_unaligned:
213 #ifdef NON_USER_COPY 204 #ifdef NON_USER_COPY
214 VISEntryHalfFast(.Lmedium_vis_entry_fa 205 VISEntryHalfFast(.Lmedium_vis_entry_fail)
215 #else 206 #else
216 VISEntryHalf 207 VISEntryHalf
217 #endif 208 #endif
218 andn %o2, 0x3f, %o4 209 andn %o2, 0x3f, %o4
219 sub %o2, %o4, %o2 210 sub %o2, %o4, %o2
220 alignaddr %o1, %g0, %g1 211 alignaddr %o1, %g0, %g1
221 add %o1, %o4, %o1 212 add %o1, %o4, %o1
222 EX_LD_FP(LOAD(ldd, %g1 + 0x00, %f0), m !! 213 EX_LD(LOAD(ldd, %g1 + 0x00, %f0))
223 1: EX_LD_FP(LOAD(ldd, %g1 + 0x08, %f2), m !! 214 1: EX_LD(LOAD(ldd, %g1 + 0x08, %f2))
224 subcc %o4, 0x40, %o4 215 subcc %o4, 0x40, %o4
225 EX_LD_FP(LOAD(ldd, %g1 + 0x10, %f4), m !! 216 EX_LD(LOAD(ldd, %g1 + 0x10, %f4))
226 EX_LD_FP(LOAD(ldd, %g1 + 0x18, %f6), m !! 217 EX_LD(LOAD(ldd, %g1 + 0x18, %f6))
227 EX_LD_FP(LOAD(ldd, %g1 + 0x20, %f8), m !! 218 EX_LD(LOAD(ldd, %g1 + 0x20, %f8))
228 EX_LD_FP(LOAD(ldd, %g1 + 0x28, %f10), !! 219 EX_LD(LOAD(ldd, %g1 + 0x28, %f10))
229 EX_LD_FP(LOAD(ldd, %g1 + 0x30, %f12), !! 220 EX_LD(LOAD(ldd, %g1 + 0x30, %f12))
230 EX_LD_FP(LOAD(ldd, %g1 + 0x38, %f14), !! 221 EX_LD(LOAD(ldd, %g1 + 0x38, %f14))
231 faligndata %f0, %f2, %f16 222 faligndata %f0, %f2, %f16
232 EX_LD_FP(LOAD(ldd, %g1 + 0x40, %f0), m !! 223 EX_LD(LOAD(ldd, %g1 + 0x40, %f0))
233 faligndata %f2, %f4, %f18 224 faligndata %f2, %f4, %f18
234 add %g1, 0x40, %g1 225 add %g1, 0x40, %g1
235 faligndata %f4, %f6, %f20 226 faligndata %f4, %f6, %f20
236 faligndata %f6, %f8, %f22 227 faligndata %f6, %f8, %f22
237 faligndata %f8, %f10, %f24 228 faligndata %f8, %f10, %f24
238 faligndata %f10, %f12, %f26 229 faligndata %f10, %f12, %f26
239 faligndata %f12, %f14, %f28 230 faligndata %f12, %f14, %f28
240 faligndata %f14, %f0, %f30 231 faligndata %f14, %f0, %f30
241 EX_ST_FP(STORE(std, %f16, %o0 + 0x00), !! 232 EX_ST(STORE(std, %f16, %o0 + 0x00))
242 EX_ST_FP(STORE(std, %f18, %o0 + 0x08), !! 233 EX_ST(STORE(std, %f18, %o0 + 0x08))
243 EX_ST_FP(STORE(std, %f20, %o0 + 0x10), !! 234 EX_ST(STORE(std, %f20, %o0 + 0x10))
244 EX_ST_FP(STORE(std, %f22, %o0 + 0x18), !! 235 EX_ST(STORE(std, %f22, %o0 + 0x18))
245 EX_ST_FP(STORE(std, %f24, %o0 + 0x20), !! 236 EX_ST(STORE(std, %f24, %o0 + 0x20))
246 EX_ST_FP(STORE(std, %f26, %o0 + 0x28), !! 237 EX_ST(STORE(std, %f26, %o0 + 0x28))
247 EX_ST_FP(STORE(std, %f28, %o0 + 0x30), !! 238 EX_ST(STORE(std, %f28, %o0 + 0x30))
248 EX_ST_FP(STORE(std, %f30, %o0 + 0x38), !! 239 EX_ST(STORE(std, %f30, %o0 + 0x38))
249 add %o0, 0x40, %o0 240 add %o0, 0x40, %o0
250 bne,pt %icc, 1b 241 bne,pt %icc, 1b
251 LOAD(prefetch, %g1 + 0x200, #n_reads_ 242 LOAD(prefetch, %g1 + 0x200, #n_reads_strong)
252 #ifdef NON_USER_COPY 243 #ifdef NON_USER_COPY
253 VISExitHalfFast 244 VISExitHalfFast
254 #else 245 #else
255 VISExitHalf 246 VISExitHalf
256 #endif 247 #endif
257 brz,pn %o2, .Lexit 248 brz,pn %o2, .Lexit
258 cmp %o2, 19 249 cmp %o2, 19
259 ble,pn %icc, .Lsmall_unaligne 250 ble,pn %icc, .Lsmall_unaligned
260 nop 251 nop
261 ba,a,pt %icc, .Lmedium_unalign 252 ba,a,pt %icc, .Lmedium_unaligned
262 253
263 #ifdef NON_USER_COPY 254 #ifdef NON_USER_COPY
264 .Lmedium_vis_entry_fail: 255 .Lmedium_vis_entry_fail:
265 or %o0, %o1, %g2 256 or %o0, %o1, %g2
266 #endif 257 #endif
267 .Lmedium: 258 .Lmedium:
268 LOAD(prefetch, %o1 + 0x40, #n_reads_st 259 LOAD(prefetch, %o1 + 0x40, #n_reads_strong)
269 andcc %g2, 0x7, %g0 260 andcc %g2, 0x7, %g0
270 bne,pn %icc, .Lmedium_unalign 261 bne,pn %icc, .Lmedium_unaligned
271 nop 262 nop
272 .Lmedium_noprefetch: 263 .Lmedium_noprefetch:
273 andncc %o2, 0x20 - 1, %o5 264 andncc %o2, 0x20 - 1, %o5
274 be,pn %icc, 2f 265 be,pn %icc, 2f
275 sub %o2, %o5, %o2 266 sub %o2, %o5, %o2
276 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memc !! 267 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1))
277 EX_LD(LOAD(ldx, %o1 + 0x08, %g2), memc !! 268 EX_LD(LOAD(ldx, %o1 + 0x08, %g2))
278 EX_LD(LOAD(ldx, %o1 + 0x10, GLOBAL_SPA !! 269 EX_LD(LOAD(ldx, %o1 + 0x10, GLOBAL_SPARE))
279 EX_LD(LOAD(ldx, %o1 + 0x18, %o4), memc !! 270 EX_LD(LOAD(ldx, %o1 + 0x18, %o4))
280 add %o1, 0x20, %o1 271 add %o1, 0x20, %o1
281 subcc %o5, 0x20, %o5 272 subcc %o5, 0x20, %o5
282 EX_ST(STORE(stx, %g1, %o0 + 0x00), mem !! 273 EX_ST(STORE(stx, %g1, %o0 + 0x00))
283 EX_ST(STORE(stx, %g2, %o0 + 0x08), mem !! 274 EX_ST(STORE(stx, %g2, %o0 + 0x08))
284 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0 !! 275 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0x10))
285 EX_ST(STORE(stx, %o4, %o0 + 0x18), mem !! 276 EX_ST(STORE(stx, %o4, %o0 + 0x18))
286 bne,pt %icc, 1b 277 bne,pt %icc, 1b
287 add %o0, 0x20, %o0 278 add %o0, 0x20, %o0
288 2: andcc %o2, 0x18, %o5 279 2: andcc %o2, 0x18, %o5
289 be,pt %icc, 3f 280 be,pt %icc, 3f
290 sub %o2, %o5, %o2 281 sub %o2, %o5, %o2
291 !! 282 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1))
292 1: EX_LD(LOAD(ldx, %o1 + 0x00, %g1), memc <<
293 add %o1, 0x08, %o1 283 add %o1, 0x08, %o1
294 add %o0, 0x08, %o0 284 add %o0, 0x08, %o0
295 subcc %o5, 0x08, %o5 285 subcc %o5, 0x08, %o5
296 bne,pt %icc, 1b 286 bne,pt %icc, 1b
297 EX_ST(STORE(stx, %g1, %o0 - 0x08), me !! 287 EX_ST(STORE(stx, %g1, %o0 - 0x08))
298 3: brz,pt %o2, .Lexit 288 3: brz,pt %o2, .Lexit
299 cmp %o2, 0x04 289 cmp %o2, 0x04
300 bl,pn %icc, .Ltiny 290 bl,pn %icc, .Ltiny
301 nop 291 nop
302 EX_LD(LOAD(lduw, %o1 + 0x00, %g1), mem !! 292 EX_LD(LOAD(lduw, %o1 + 0x00, %g1))
303 add %o1, 0x04, %o1 293 add %o1, 0x04, %o1
304 add %o0, 0x04, %o0 294 add %o0, 0x04, %o0
305 subcc %o2, 0x04, %o2 295 subcc %o2, 0x04, %o2
306 bne,pn %icc, .Ltiny 296 bne,pn %icc, .Ltiny
307 EX_ST(STORE(stw, %g1, %o0 - 0x04), me !! 297 EX_ST(STORE(stw, %g1, %o0 - 0x04))
308 ba,a,pt %icc, .Lexit 298 ba,a,pt %icc, .Lexit
309 .Lmedium_unaligned: 299 .Lmedium_unaligned:
310 /* First get dest 8 byte aligned. */ 300 /* First get dest 8 byte aligned. */
311 sub %g0, %o0, %g1 301 sub %g0, %o0, %g1
312 and %g1, 0x7, %g1 302 and %g1, 0x7, %g1
313 brz,pt %g1, 2f 303 brz,pt %g1, 2f
314 sub %o2, %g1, %o2 304 sub %o2, %g1, %o2
315 305
316 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2), mem !! 306 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g2))
317 add %o1, 1, %o1 307 add %o1, 1, %o1
318 subcc %g1, 1, %g1 308 subcc %g1, 1, %g1
319 add %o0, 1, %o0 309 add %o0, 1, %o0
320 bne,pt %icc, 1b 310 bne,pt %icc, 1b
321 EX_ST(STORE(stb, %g2, %o0 - 0x01), me !! 311 EX_ST(STORE(stb, %g2, %o0 - 0x01))
322 2: 312 2:
323 and %o1, 0x7, %g1 313 and %o1, 0x7, %g1
324 brz,pn %g1, .Lmedium_noprefet 314 brz,pn %g1, .Lmedium_noprefetch
325 sll %g1, 3, %g1 315 sll %g1, 3, %g1
326 mov 64, %g2 316 mov 64, %g2
327 sub %g2, %g1, %g2 317 sub %g2, %g1, %g2
328 andn %o1, 0x7, %o1 318 andn %o1, 0x7, %o1
329 EX_LD(LOAD(ldx, %o1 + 0x00, %o4), memc !! 319 EX_LD(LOAD(ldx, %o1 + 0x00, %o4))
330 sllx %o4, %g1, %o4 320 sllx %o4, %g1, %o4
331 andn %o2, 0x08 - 1, %o5 321 andn %o2, 0x08 - 1, %o5
332 sub %o2, %o5, %o2 322 sub %o2, %o5, %o2
333 1: EX_LD(LOAD(ldx, %o1 + 0x08, %g3), memc !! 323 1: EX_LD(LOAD(ldx, %o1 + 0x08, %g3))
334 add %o1, 0x08, %o1 324 add %o1, 0x08, %o1
335 subcc %o5, 0x08, %o5 325 subcc %o5, 0x08, %o5
336 srlx %g3, %g2, GLOBAL_SPARE 326 srlx %g3, %g2, GLOBAL_SPARE
337 or GLOBAL_SPARE, %o4, GLO 327 or GLOBAL_SPARE, %o4, GLOBAL_SPARE
338 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0 !! 328 EX_ST(STORE(stx, GLOBAL_SPARE, %o0 + 0x00))
339 add %o0, 0x08, %o0 329 add %o0, 0x08, %o0
340 bne,pt %icc, 1b 330 bne,pt %icc, 1b
341 sllx %g3, %g1, %o4 331 sllx %g3, %g1, %o4
342 srl %g1, 3, %g1 332 srl %g1, 3, %g1
343 add %o1, %g1, %o1 333 add %o1, %g1, %o1
344 brz,pn %o2, .Lexit 334 brz,pn %o2, .Lexit
345 nop 335 nop
346 ba,pt %icc, .Lsmall_unaligne 336 ba,pt %icc, .Lsmall_unaligned
347 337
348 .Ltiny: 338 .Ltiny:
349 EX_LD(LOAD(ldub, %o1 + 0x00, %g1), mem !! 339 EX_LD(LOAD(ldub, %o1 + 0x00, %g1))
350 subcc %o2, 1, %o2 340 subcc %o2, 1, %o2
351 be,pn %icc, .Lexit 341 be,pn %icc, .Lexit
352 EX_ST(STORE(stb, %g1, %o0 + 0x00), me !! 342 EX_ST(STORE(stb, %g1, %o0 + 0x00))
353 EX_LD(LOAD(ldub, %o1 + 0x01, %g1), mem !! 343 EX_LD(LOAD(ldub, %o1 + 0x01, %g1))
354 subcc %o2, 1, %o2 344 subcc %o2, 1, %o2
355 be,pn %icc, .Lexit 345 be,pn %icc, .Lexit
356 EX_ST(STORE(stb, %g1, %o0 + 0x01), me !! 346 EX_ST(STORE(stb, %g1, %o0 + 0x01))
357 EX_LD(LOAD(ldub, %o1 + 0x02, %g1), mem !! 347 EX_LD(LOAD(ldub, %o1 + 0x02, %g1))
358 ba,pt %icc, .Lexit 348 ba,pt %icc, .Lexit
359 EX_ST(STORE(stb, %g1, %o0 + 0x02), me !! 349 EX_ST(STORE(stb, %g1, %o0 + 0x02))
360 350
361 .Lsmall: 351 .Lsmall:
362 andcc %g2, 0x3, %g0 352 andcc %g2, 0x3, %g0
363 bne,pn %icc, .Lsmall_unaligne 353 bne,pn %icc, .Lsmall_unaligned
364 andn %o2, 0x4 - 1, %o5 354 andn %o2, 0x4 - 1, %o5
365 sub %o2, %o5, %o2 355 sub %o2, %o5, %o2
366 1: 356 1:
367 EX_LD(LOAD(lduw, %o1 + 0x00, %g1), mem !! 357 EX_LD(LOAD(lduw, %o1 + 0x00, %g1))
368 add %o1, 0x04, %o1 358 add %o1, 0x04, %o1
369 subcc %o5, 0x04, %o5 359 subcc %o5, 0x04, %o5
370 add %o0, 0x04, %o0 360 add %o0, 0x04, %o0
371 bne,pt %icc, 1b 361 bne,pt %icc, 1b
372 EX_ST(STORE(stw, %g1, %o0 - 0x04), me !! 362 EX_ST(STORE(stw, %g1, %o0 - 0x04))
373 brz,pt %o2, .Lexit 363 brz,pt %o2, .Lexit
374 nop 364 nop
375 ba,a,pt %icc, .Ltiny 365 ba,a,pt %icc, .Ltiny
376 366
377 .Lsmall_unaligned: 367 .Lsmall_unaligned:
378 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g1), mem !! 368 1: EX_LD(LOAD(ldub, %o1 + 0x00, %g1))
379 add %o1, 1, %o1 369 add %o1, 1, %o1
380 add %o0, 1, %o0 370 add %o0, 1, %o0
381 subcc %o2, 1, %o2 371 subcc %o2, 1, %o2
382 bne,pt %icc, 1b 372 bne,pt %icc, 1b
383 EX_ST(STORE(stb, %g1, %o0 - 0x01), me !! 373 EX_ST(STORE(stb, %g1, %o0 - 0x01))
384 ba,a,pt %icc, .Lexit 374 ba,a,pt %icc, .Lexit
385 nop <<
386 .size FUNC_NAME, .-FUNC_NAME 375 .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.