1 /* SPDX-License-Identifier: GPL-2.0 */ <<
2 /* NG2memcpy.S: Niagara-2 optimized memcpy. 1 /* NG2memcpy.S: Niagara-2 optimized memcpy.
3 * 2 *
4 * Copyright (C) 2007 David S. Miller (davem@d 3 * Copyright (C) 2007 David S. Miller (davem@davemloft.net)
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_PNF 0x82 12 #define ASI_PNF 0x82
14 #define ASI_BLK_P 0xf0 13 #define ASI_BLK_P 0xf0
15 #define ASI_BLK_INIT_QUAD_LDD_P 0xe2 14 #define ASI_BLK_INIT_QUAD_LDD_P 0xe2
16 #define FPRS_FEF 0x04 15 #define FPRS_FEF 0x04
17 #ifdef MEMCPY_DEBUG 16 #ifdef MEMCPY_DEBUG
18 #define VISEntryHalf rd %fprs, %o5; wr %g0, FP 17 #define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs; \
19 clr %g1; clr %g2; clr %g3 18 clr %g1; clr %g2; clr %g3; clr %g5; subcc %g0, %g0, %g0;
20 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 19 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
21 #else 20 #else
22 #define VISEntryHalf rd %fprs, %o5; wr %g0, FP 21 #define VISEntryHalf rd %fprs, %o5; wr %g0, FPRS_FEF, %fprs
23 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr 22 #define VISExitHalf and %o5, FPRS_FEF, %o5; wr %o5, 0x0, %fprs
24 #endif 23 #endif
25 #define GLOBAL_SPARE %g5 24 #define GLOBAL_SPARE %g5
26 #endif 25 #endif
27 26
28 #ifndef STORE_ASI 27 #ifndef STORE_ASI
29 #ifndef SIMULATE_NIAGARA_ON_NON_NIAGARA 28 #ifndef SIMULATE_NIAGARA_ON_NON_NIAGARA
30 #define STORE_ASI ASI_BLK_INIT_QUAD_LDD_ 29 #define STORE_ASI ASI_BLK_INIT_QUAD_LDD_P
31 #else 30 #else
32 #define STORE_ASI 0x80 /* ASI 31 #define STORE_ASI 0x80 /* ASI_P */
33 #endif 32 #endif
34 #endif 33 #endif
35 34
36 #ifndef EX_LD 35 #ifndef EX_LD
37 #define EX_LD(x,y) x 36 #define EX_LD(x,y) x
38 #endif 37 #endif
39 #ifndef EX_LD_FP 38 #ifndef EX_LD_FP
40 #define EX_LD_FP(x,y) x 39 #define EX_LD_FP(x,y) x
41 #endif 40 #endif
42 41
43 #ifndef EX_ST 42 #ifndef EX_ST
44 #define EX_ST(x,y) x 43 #define EX_ST(x,y) x
45 #endif 44 #endif
46 #ifndef EX_ST_FP 45 #ifndef EX_ST_FP
47 #define EX_ST_FP(x,y) x 46 #define EX_ST_FP(x,y) x
48 #endif 47 #endif
49 48
50 #ifndef LOAD 49 #ifndef LOAD
51 #define LOAD(type,addr,dest) type [addr], d 50 #define LOAD(type,addr,dest) type [addr], dest
52 #endif 51 #endif
53 52
54 #ifndef LOAD_BLK 53 #ifndef LOAD_BLK
55 #define LOAD_BLK(addr,dest) ldda [addr] AS 54 #define LOAD_BLK(addr,dest) ldda [addr] ASI_BLK_P, dest
56 #endif 55 #endif
57 56
58 #ifndef STORE 57 #ifndef STORE
59 #ifndef MEMCPY_DEBUG 58 #ifndef MEMCPY_DEBUG
60 #define STORE(type,src,addr) type src, [add 59 #define STORE(type,src,addr) type src, [addr]
61 #else 60 #else
62 #define STORE(type,src,addr) type##a src, [ 61 #define STORE(type,src,addr) type##a src, [addr] 0x80
63 #endif 62 #endif
64 #endif 63 #endif
65 64
66 #ifndef STORE_BLK 65 #ifndef STORE_BLK
67 #define STORE_BLK(src,addr) stda src, [add 66 #define STORE_BLK(src,addr) stda src, [addr] ASI_BLK_P
68 #endif 67 #endif
69 68
70 #ifndef STORE_INIT 69 #ifndef STORE_INIT
71 #define STORE_INIT(src,addr) stxa src, [add 70 #define STORE_INIT(src,addr) stxa src, [addr] STORE_ASI
72 #endif 71 #endif
73 72
74 #ifndef FUNC_NAME 73 #ifndef FUNC_NAME
75 #define FUNC_NAME NG2memcpy 74 #define FUNC_NAME NG2memcpy
76 #endif 75 #endif
77 76
78 #ifndef PREAMBLE 77 #ifndef PREAMBLE
79 #define PREAMBLE 78 #define PREAMBLE
80 #endif 79 #endif
81 80
82 #ifndef XCC 81 #ifndef XCC
83 #define XCC xcc 82 #define XCC xcc
84 #endif 83 #endif
85 84
86 #define FREG_FROB(x0, x1, x2, x3, x4, x5, x6, 85 #define FREG_FROB(x0, x1, x2, x3, x4, x5, x6, x7, x8) \
87 faligndata %x0, %x1, %f0; \ 86 faligndata %x0, %x1, %f0; \
88 faligndata %x1, %x2, %f2; \ 87 faligndata %x1, %x2, %f2; \
89 faligndata %x2, %x3, %f4; \ 88 faligndata %x2, %x3, %f4; \
90 faligndata %x3, %x4, %f6; \ 89 faligndata %x3, %x4, %f6; \
91 faligndata %x4, %x5, %f8; \ 90 faligndata %x4, %x5, %f8; \
92 faligndata %x5, %x6, %f10; \ 91 faligndata %x5, %x6, %f10; \
93 faligndata %x6, %x7, %f12; \ 92 faligndata %x6, %x7, %f12; \
94 faligndata %x7, %x8, %f14; 93 faligndata %x7, %x8, %f14;
95 94
96 #define FREG_MOVE_1(x0) \ 95 #define FREG_MOVE_1(x0) \
97 fsrc2 %x0, %f0; 96 fsrc2 %x0, %f0;
98 #define FREG_MOVE_2(x0, x1) \ 97 #define FREG_MOVE_2(x0, x1) \
99 fsrc2 %x0, %f0; \ 98 fsrc2 %x0, %f0; \
100 fsrc2 %x1, %f2; 99 fsrc2 %x1, %f2;
101 #define FREG_MOVE_3(x0, x1, x2) \ 100 #define FREG_MOVE_3(x0, x1, x2) \
102 fsrc2 %x0, %f0; \ 101 fsrc2 %x0, %f0; \
103 fsrc2 %x1, %f2; \ 102 fsrc2 %x1, %f2; \
104 fsrc2 %x2, %f4; 103 fsrc2 %x2, %f4;
105 #define FREG_MOVE_4(x0, x1, x2, x3) \ 104 #define FREG_MOVE_4(x0, x1, x2, x3) \
106 fsrc2 %x0, %f0; \ 105 fsrc2 %x0, %f0; \
107 fsrc2 %x1, %f2; \ 106 fsrc2 %x1, %f2; \
108 fsrc2 %x2, %f4; \ 107 fsrc2 %x2, %f4; \
109 fsrc2 %x3, %f6; 108 fsrc2 %x3, %f6;
110 #define FREG_MOVE_5(x0, x1, x2, x3, x4) \ 109 #define FREG_MOVE_5(x0, x1, x2, x3, x4) \
111 fsrc2 %x0, %f0; \ 110 fsrc2 %x0, %f0; \
112 fsrc2 %x1, %f2; \ 111 fsrc2 %x1, %f2; \
113 fsrc2 %x2, %f4; \ 112 fsrc2 %x2, %f4; \
114 fsrc2 %x3, %f6; \ 113 fsrc2 %x3, %f6; \
115 fsrc2 %x4, %f8; 114 fsrc2 %x4, %f8;
116 #define FREG_MOVE_6(x0, x1, x2, x3, x4, x5) \ 115 #define FREG_MOVE_6(x0, x1, x2, x3, x4, x5) \
117 fsrc2 %x0, %f0; \ 116 fsrc2 %x0, %f0; \
118 fsrc2 %x1, %f2; \ 117 fsrc2 %x1, %f2; \
119 fsrc2 %x2, %f4; \ 118 fsrc2 %x2, %f4; \
120 fsrc2 %x3, %f6; \ 119 fsrc2 %x3, %f6; \
121 fsrc2 %x4, %f8; \ 120 fsrc2 %x4, %f8; \
122 fsrc2 %x5, %f10; 121 fsrc2 %x5, %f10;
123 #define FREG_MOVE_7(x0, x1, x2, x3, x4, x5, x6 122 #define FREG_MOVE_7(x0, x1, x2, x3, x4, x5, x6) \
124 fsrc2 %x0, %f0; \ 123 fsrc2 %x0, %f0; \
125 fsrc2 %x1, %f2; \ 124 fsrc2 %x1, %f2; \
126 fsrc2 %x2, %f4; \ 125 fsrc2 %x2, %f4; \
127 fsrc2 %x3, %f6; \ 126 fsrc2 %x3, %f6; \
128 fsrc2 %x4, %f8; \ 127 fsrc2 %x4, %f8; \
129 fsrc2 %x5, %f10; \ 128 fsrc2 %x5, %f10; \
130 fsrc2 %x6, %f12; 129 fsrc2 %x6, %f12;
131 #define FREG_MOVE_8(x0, x1, x2, x3, x4, x5, x6 130 #define FREG_MOVE_8(x0, x1, x2, x3, x4, x5, x6, x7) \
132 fsrc2 %x0, %f0; \ 131 fsrc2 %x0, %f0; \
133 fsrc2 %x1, %f2; \ 132 fsrc2 %x1, %f2; \
134 fsrc2 %x2, %f4; \ 133 fsrc2 %x2, %f4; \
135 fsrc2 %x3, %f6; \ 134 fsrc2 %x3, %f6; \
136 fsrc2 %x4, %f8; \ 135 fsrc2 %x4, %f8; \
137 fsrc2 %x5, %f10; \ 136 fsrc2 %x5, %f10; \
138 fsrc2 %x6, %f12; \ 137 fsrc2 %x6, %f12; \
139 fsrc2 %x7, %f14; 138 fsrc2 %x7, %f14;
140 #define FREG_LOAD_1(base, x0) \ 139 #define FREG_LOAD_1(base, x0) \
141 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 140 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1)
142 #define FREG_LOAD_2(base, x0, x1) \ 141 #define FREG_LOAD_2(base, x0, x1) \
143 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 142 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1); \
144 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), 143 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), NG2_retl_o2_plus_g1);
145 #define FREG_LOAD_3(base, x0, x1, x2) \ 144 #define FREG_LOAD_3(base, x0, x1, x2) \
146 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 145 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1); \
147 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), 146 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), NG2_retl_o2_plus_g1); \
148 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), 147 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), NG2_retl_o2_plus_g1);
149 #define FREG_LOAD_4(base, x0, x1, x2, x3) \ 148 #define FREG_LOAD_4(base, x0, x1, x2, x3) \
150 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 149 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1); \
151 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), 150 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), NG2_retl_o2_plus_g1); \
152 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), 151 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), NG2_retl_o2_plus_g1); \
153 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), 152 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), NG2_retl_o2_plus_g1);
154 #define FREG_LOAD_5(base, x0, x1, x2, x3, x4) 153 #define FREG_LOAD_5(base, x0, x1, x2, x3, x4) \
155 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 154 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1); \
156 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), 155 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), NG2_retl_o2_plus_g1); \
157 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), 156 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), NG2_retl_o2_plus_g1); \
158 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), 157 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), NG2_retl_o2_plus_g1); \
159 EX_LD_FP(LOAD(ldd, base + 0x20, %x4), 158 EX_LD_FP(LOAD(ldd, base + 0x20, %x4), NG2_retl_o2_plus_g1);
160 #define FREG_LOAD_6(base, x0, x1, x2, x3, x4, 159 #define FREG_LOAD_6(base, x0, x1, x2, x3, x4, x5) \
161 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 160 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1); \
162 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), 161 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), NG2_retl_o2_plus_g1); \
163 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), 162 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), NG2_retl_o2_plus_g1); \
164 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), 163 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), NG2_retl_o2_plus_g1); \
165 EX_LD_FP(LOAD(ldd, base + 0x20, %x4), 164 EX_LD_FP(LOAD(ldd, base + 0x20, %x4), NG2_retl_o2_plus_g1); \
166 EX_LD_FP(LOAD(ldd, base + 0x28, %x5), 165 EX_LD_FP(LOAD(ldd, base + 0x28, %x5), NG2_retl_o2_plus_g1);
167 #define FREG_LOAD_7(base, x0, x1, x2, x3, x4, 166 #define FREG_LOAD_7(base, x0, x1, x2, x3, x4, x5, x6) \
168 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), 167 EX_LD_FP(LOAD(ldd, base + 0x00, %x0), NG2_retl_o2_plus_g1); \
169 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), 168 EX_LD_FP(LOAD(ldd, base + 0x08, %x1), NG2_retl_o2_plus_g1); \
170 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), 169 EX_LD_FP(LOAD(ldd, base + 0x10, %x2), NG2_retl_o2_plus_g1); \
171 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), 170 EX_LD_FP(LOAD(ldd, base + 0x18, %x3), NG2_retl_o2_plus_g1); \
172 EX_LD_FP(LOAD(ldd, base + 0x20, %x4), 171 EX_LD_FP(LOAD(ldd, base + 0x20, %x4), NG2_retl_o2_plus_g1); \
173 EX_LD_FP(LOAD(ldd, base + 0x28, %x5), 172 EX_LD_FP(LOAD(ldd, base + 0x28, %x5), NG2_retl_o2_plus_g1); \
174 EX_LD_FP(LOAD(ldd, base + 0x30, %x6), 173 EX_LD_FP(LOAD(ldd, base + 0x30, %x6), NG2_retl_o2_plus_g1);
175 174
176 .register %g2,#scratch 175 .register %g2,#scratch
177 .register %g3,#scratch 176 .register %g3,#scratch
178 177
179 .text 178 .text
180 #ifndef EX_RETVAL 179 #ifndef EX_RETVAL
181 #define EX_RETVAL(x) x 180 #define EX_RETVAL(x) x
182 __restore_fp: 181 __restore_fp:
183 VISExitHalf 182 VISExitHalf
184 __restore_asi: 183 __restore_asi:
185 retl 184 retl
186 wr %g0, ASI_AIUS, %asi 185 wr %g0, ASI_AIUS, %asi
187 ENTRY(NG2_retl_o2) 186 ENTRY(NG2_retl_o2)
188 ba,pt %xcc, __restore_asi 187 ba,pt %xcc, __restore_asi
189 mov %o2, %o0 188 mov %o2, %o0
190 ENDPROC(NG2_retl_o2) 189 ENDPROC(NG2_retl_o2)
191 ENTRY(NG2_retl_o2_plus_1) 190 ENTRY(NG2_retl_o2_plus_1)
192 ba,pt %xcc, __restore_asi 191 ba,pt %xcc, __restore_asi
193 add %o2, 1, %o0 192 add %o2, 1, %o0
194 ENDPROC(NG2_retl_o2_plus_1) 193 ENDPROC(NG2_retl_o2_plus_1)
195 ENTRY(NG2_retl_o2_plus_4) 194 ENTRY(NG2_retl_o2_plus_4)
196 ba,pt %xcc, __restore_asi 195 ba,pt %xcc, __restore_asi
197 add %o2, 4, %o0 196 add %o2, 4, %o0
198 ENDPROC(NG2_retl_o2_plus_4) 197 ENDPROC(NG2_retl_o2_plus_4)
199 ENTRY(NG2_retl_o2_plus_8) 198 ENTRY(NG2_retl_o2_plus_8)
200 ba,pt %xcc, __restore_asi 199 ba,pt %xcc, __restore_asi
201 add %o2, 8, %o0 200 add %o2, 8, %o0
202 ENDPROC(NG2_retl_o2_plus_8) 201 ENDPROC(NG2_retl_o2_plus_8)
203 ENTRY(NG2_retl_o2_plus_o4_plus_1) 202 ENTRY(NG2_retl_o2_plus_o4_plus_1)
204 add %o4, 1, %o4 203 add %o4, 1, %o4
205 ba,pt %xcc, __restore_asi 204 ba,pt %xcc, __restore_asi
206 add %o2, %o4, %o0 205 add %o2, %o4, %o0
207 ENDPROC(NG2_retl_o2_plus_o4_plus_1) 206 ENDPROC(NG2_retl_o2_plus_o4_plus_1)
208 ENTRY(NG2_retl_o2_plus_o4_plus_8) 207 ENTRY(NG2_retl_o2_plus_o4_plus_8)
209 add %o4, 8, %o4 208 add %o4, 8, %o4
210 ba,pt %xcc, __restore_asi 209 ba,pt %xcc, __restore_asi
211 add %o2, %o4, %o0 210 add %o2, %o4, %o0
212 ENDPROC(NG2_retl_o2_plus_o4_plus_8) 211 ENDPROC(NG2_retl_o2_plus_o4_plus_8)
213 ENTRY(NG2_retl_o2_plus_o4_plus_16) 212 ENTRY(NG2_retl_o2_plus_o4_plus_16)
214 add %o4, 16, %o4 213 add %o4, 16, %o4
215 ba,pt %xcc, __restore_asi 214 ba,pt %xcc, __restore_asi
216 add %o2, %o4, %o0 215 add %o2, %o4, %o0
217 ENDPROC(NG2_retl_o2_plus_o4_plus_16) 216 ENDPROC(NG2_retl_o2_plus_o4_plus_16)
218 ENTRY(NG2_retl_o2_plus_g1_fp) 217 ENTRY(NG2_retl_o2_plus_g1_fp)
219 ba,pt %xcc, __restore_fp 218 ba,pt %xcc, __restore_fp
220 add %o2, %g1, %o0 219 add %o2, %g1, %o0
221 ENDPROC(NG2_retl_o2_plus_g1_fp) 220 ENDPROC(NG2_retl_o2_plus_g1_fp)
222 ENTRY(NG2_retl_o2_plus_g1_plus_64_fp) 221 ENTRY(NG2_retl_o2_plus_g1_plus_64_fp)
223 add %g1, 64, %g1 222 add %g1, 64, %g1
224 ba,pt %xcc, __restore_fp 223 ba,pt %xcc, __restore_fp
225 add %o2, %g1, %o0 224 add %o2, %g1, %o0
226 ENDPROC(NG2_retl_o2_plus_g1_plus_64_fp) 225 ENDPROC(NG2_retl_o2_plus_g1_plus_64_fp)
227 ENTRY(NG2_retl_o2_plus_g1_plus_1) 226 ENTRY(NG2_retl_o2_plus_g1_plus_1)
228 add %g1, 1, %g1 227 add %g1, 1, %g1
229 ba,pt %xcc, __restore_asi 228 ba,pt %xcc, __restore_asi
230 add %o2, %g1, %o0 229 add %o2, %g1, %o0
231 ENDPROC(NG2_retl_o2_plus_g1_plus_1) 230 ENDPROC(NG2_retl_o2_plus_g1_plus_1)
232 ENTRY(NG2_retl_o2_and_7_plus_o4) 231 ENTRY(NG2_retl_o2_and_7_plus_o4)
233 and %o2, 7, %o2 232 and %o2, 7, %o2
234 ba,pt %xcc, __restore_asi 233 ba,pt %xcc, __restore_asi
235 add %o2, %o4, %o0 234 add %o2, %o4, %o0
236 ENDPROC(NG2_retl_o2_and_7_plus_o4) 235 ENDPROC(NG2_retl_o2_and_7_plus_o4)
237 ENTRY(NG2_retl_o2_and_7_plus_o4_plus_8) 236 ENTRY(NG2_retl_o2_and_7_plus_o4_plus_8)
238 and %o2, 7, %o2 237 and %o2, 7, %o2
239 add %o4, 8, %o4 238 add %o4, 8, %o4
240 ba,pt %xcc, __restore_asi 239 ba,pt %xcc, __restore_asi
241 add %o2, %o4, %o0 240 add %o2, %o4, %o0
242 ENDPROC(NG2_retl_o2_and_7_plus_o4_plus_8) 241 ENDPROC(NG2_retl_o2_and_7_plus_o4_plus_8)
243 #endif 242 #endif
244 243
245 .align 64 244 .align 64
246 245
247 .globl FUNC_NAME 246 .globl FUNC_NAME
248 .type FUNC_NAME,#function 247 .type FUNC_NAME,#function
249 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len * 248 FUNC_NAME: /* %o0=dst, %o1=src, %o2=len */
250 srlx %o2, 31, %g2 249 srlx %o2, 31, %g2
251 cmp %g2, 0 250 cmp %g2, 0
252 tne %xcc, 5 251 tne %xcc, 5
253 PREAMBLE 252 PREAMBLE
254 mov %o0, %o3 253 mov %o0, %o3
255 cmp %o2, 0 254 cmp %o2, 0
256 be,pn %XCC, 85f 255 be,pn %XCC, 85f
257 or %o0, %o1, GLOBAL_SPARE 256 or %o0, %o1, GLOBAL_SPARE
258 cmp %o2, 16 257 cmp %o2, 16
259 blu,a,pn %XCC, 80f 258 blu,a,pn %XCC, 80f
260 or GLOBAL_SPARE, %o2, GLO 259 or GLOBAL_SPARE, %o2, GLOBAL_SPARE
261 260
262 /* 2 blocks (128 bytes) is the minimum 261 /* 2 blocks (128 bytes) is the minimum we can do the block
263 * copy with. We need to ensure that 262 * copy with. We need to ensure that we'll iterate at least
264 * once in the block copy loop. At wo 263 * once in the block copy loop. At worst we'll need to align
265 * the destination to a 64-byte bounda 264 * the destination to a 64-byte boundary which can chew up
266 * to (64 - 1) bytes from the length b 265 * to (64 - 1) bytes from the length before we perform the
267 * block copy loop. 266 * block copy loop.
268 * 267 *
269 * However, the cut-off point, perform 268 * However, the cut-off point, performance wise, is around
270 * 4 64-byte blocks. 269 * 4 64-byte blocks.
271 */ 270 */
272 cmp %o2, (4 * 64) 271 cmp %o2, (4 * 64)
273 blu,pt %XCC, 75f 272 blu,pt %XCC, 75f
274 andcc GLOBAL_SPARE, 0x7, %g0 273 andcc GLOBAL_SPARE, 0x7, %g0
275 274
276 /* %o0: dst 275 /* %o0: dst
277 * %o1: src 276 * %o1: src
278 * %o2: len (known to be >= 128) 277 * %o2: len (known to be >= 128)
279 * 278 *
280 * The block copy loops can use %o4, % 279 * The block copy loops can use %o4, %g2, %g3 as
281 * temporaries while copying the data. 280 * temporaries while copying the data. %o5 must
282 * be preserved between VISEntryHalf a 281 * be preserved between VISEntryHalf and VISExitHalf
283 */ 282 */
284 283
285 LOAD(prefetch, %o1 + 0x000, #one_read) 284 LOAD(prefetch, %o1 + 0x000, #one_read)
286 LOAD(prefetch, %o1 + 0x040, #one_read) 285 LOAD(prefetch, %o1 + 0x040, #one_read)
287 LOAD(prefetch, %o1 + 0x080, #one_read) 286 LOAD(prefetch, %o1 + 0x080, #one_read)
288 287
289 /* Align destination on 64-byte bounda 288 /* Align destination on 64-byte boundary. */
290 andcc %o0, (64 - 1), %o4 289 andcc %o0, (64 - 1), %o4
291 be,pt %XCC, 2f 290 be,pt %XCC, 2f
292 sub %o4, 64, %o4 291 sub %o4, 64, %o4
293 sub %g0, %o4, %o4 ! byte 292 sub %g0, %o4, %o4 ! bytes to align dst
294 sub %o2, %o4, %o2 293 sub %o2, %o4, %o2
295 1: subcc %o4, 1, %o4 294 1: subcc %o4, 1, %o4
296 EX_LD(LOAD(ldub, %o1, %g1), NG2_retl_o 295 EX_LD(LOAD(ldub, %o1, %g1), NG2_retl_o2_plus_o4_plus_1)
297 EX_ST(STORE(stb, %g1, %o0), NG2_retl_o 296 EX_ST(STORE(stb, %g1, %o0), NG2_retl_o2_plus_o4_plus_1)
298 add %o1, 1, %o1 297 add %o1, 1, %o1
299 bne,pt %XCC, 1b 298 bne,pt %XCC, 1b
300 add %o0, 1, %o0 299 add %o0, 1, %o0
301 300
302 2: 301 2:
303 /* Clobbers o5/g1/g2/g3/g7/icc/xcc. W 302 /* Clobbers o5/g1/g2/g3/g7/icc/xcc. We must preserve
304 * o5 from here until we hit VISExitHa 303 * o5 from here until we hit VISExitHalf.
305 */ 304 */
306 VISEntryHalf 305 VISEntryHalf
307 306
308 membar #Sync 307 membar #Sync
309 alignaddr %o1, %g0, %g0 308 alignaddr %o1, %g0, %g0
310 309
311 add %o1, (64 - 1), %o4 310 add %o1, (64 - 1), %o4
312 andn %o4, (64 - 1), %o4 311 andn %o4, (64 - 1), %o4
313 andn %o2, (64 - 1), %g1 312 andn %o2, (64 - 1), %g1
314 sub %o2, %g1, %o2 313 sub %o2, %g1, %o2
315 314
316 and %o1, (64 - 1), %g2 315 and %o1, (64 - 1), %g2
317 add %o1, %g1, %o1 316 add %o1, %g1, %o1
318 sub %o0, %o4, %g3 317 sub %o0, %o4, %g3
319 brz,pt %g2, 190f 318 brz,pt %g2, 190f
320 cmp %g2, 32 319 cmp %g2, 32
321 blu,a 5f 320 blu,a 5f
322 cmp %g2, 16 321 cmp %g2, 16
323 cmp %g2, 48 322 cmp %g2, 48
324 blu,a 4f 323 blu,a 4f
325 cmp %g2, 40 324 cmp %g2, 40
326 cmp %g2, 56 325 cmp %g2, 56
327 blu 170f 326 blu 170f
328 nop 327 nop
329 ba,a,pt %xcc, 180f 328 ba,a,pt %xcc, 180f
330 nop <<
331 329
332 4: /* 32 <= low bits < 48 */ 330 4: /* 32 <= low bits < 48 */
333 blu 150f 331 blu 150f
334 nop 332 nop
335 ba,a,pt %xcc, 160f 333 ba,a,pt %xcc, 160f
336 nop <<
337 5: /* 0 < low bits < 32 */ 334 5: /* 0 < low bits < 32 */
338 blu,a 6f 335 blu,a 6f
339 cmp %g2, 8 336 cmp %g2, 8
340 cmp %g2, 24 337 cmp %g2, 24
341 blu 130f 338 blu 130f
342 nop 339 nop
343 ba,a,pt %xcc, 140f 340 ba,a,pt %xcc, 140f
344 nop <<
345 6: /* 0 < low bits < 16 */ 341 6: /* 0 < low bits < 16 */
346 bgeu 120f 342 bgeu 120f
347 nop 343 nop
348 /* fall through for 0 < low bits < 8 * 344 /* fall through for 0 < low bits < 8 */
349 110: sub %o4, 64, %g2 345 110: sub %o4, 64, %g2
350 EX_LD_FP(LOAD_BLK(%g2, %f0), NG2_retl_ 346 EX_LD_FP(LOAD_BLK(%g2, %f0), NG2_retl_o2_plus_g1)
351 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 347 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
352 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 348 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
353 FREG_FROB(f0, f2, f4, f6, f8, f10, f12 349 FREG_FROB(f0, f2, f4, f6, f8, f10, f12, f14, f16)
354 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 350 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
355 FREG_MOVE_8(f16, f18, f20, f22, f24, f 351 FREG_MOVE_8(f16, f18, f20, f22, f24, f26, f28, f30)
356 subcc %g1, 64, %g1 352 subcc %g1, 64, %g1
357 add %o4, 64, %o4 353 add %o4, 64, %o4
358 bne,pt %xcc, 1b 354 bne,pt %xcc, 1b
359 LOAD(prefetch, %o4 + 64, #one_read) 355 LOAD(prefetch, %o4 + 64, #one_read)
360 ba,pt %xcc, 195f 356 ba,pt %xcc, 195f
361 nop 357 nop
362 358
363 120: sub %o4, 56, %g2 359 120: sub %o4, 56, %g2
364 FREG_LOAD_7(%g2, f0, f2, f4, f6, f8, f 360 FREG_LOAD_7(%g2, f0, f2, f4, f6, f8, f10, f12)
365 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 361 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
366 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 362 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
367 FREG_FROB(f0, f2, f4, f6, f8, f10, f12 363 FREG_FROB(f0, f2, f4, f6, f8, f10, f12, f16, f18)
368 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 364 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
369 FREG_MOVE_7(f18, f20, f22, f24, f26, f 365 FREG_MOVE_7(f18, f20, f22, f24, f26, f28, f30)
370 subcc %g1, 64, %g1 366 subcc %g1, 64, %g1
371 add %o4, 64, %o4 367 add %o4, 64, %o4
372 bne,pt %xcc, 1b 368 bne,pt %xcc, 1b
373 LOAD(prefetch, %o4 + 64, #one_read) 369 LOAD(prefetch, %o4 + 64, #one_read)
374 ba,pt %xcc, 195f 370 ba,pt %xcc, 195f
375 nop 371 nop
376 372
377 130: sub %o4, 48, %g2 373 130: sub %o4, 48, %g2
378 FREG_LOAD_6(%g2, f0, f2, f4, f6, f8, f 374 FREG_LOAD_6(%g2, f0, f2, f4, f6, f8, f10)
379 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 375 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
380 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 376 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
381 FREG_FROB(f0, f2, f4, f6, f8, f10, f16 377 FREG_FROB(f0, f2, f4, f6, f8, f10, f16, f18, f20)
382 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 378 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
383 FREG_MOVE_6(f20, f22, f24, f26, f28, f 379 FREG_MOVE_6(f20, f22, f24, f26, f28, f30)
384 subcc %g1, 64, %g1 380 subcc %g1, 64, %g1
385 add %o4, 64, %o4 381 add %o4, 64, %o4
386 bne,pt %xcc, 1b 382 bne,pt %xcc, 1b
387 LOAD(prefetch, %o4 + 64, #one_read) 383 LOAD(prefetch, %o4 + 64, #one_read)
388 ba,pt %xcc, 195f 384 ba,pt %xcc, 195f
389 nop 385 nop
390 386
391 140: sub %o4, 40, %g2 387 140: sub %o4, 40, %g2
392 FREG_LOAD_5(%g2, f0, f2, f4, f6, f8) 388 FREG_LOAD_5(%g2, f0, f2, f4, f6, f8)
393 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 389 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
394 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 390 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
395 FREG_FROB(f0, f2, f4, f6, f8, f16, f18 391 FREG_FROB(f0, f2, f4, f6, f8, f16, f18, f20, f22)
396 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 392 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
397 FREG_MOVE_5(f22, f24, f26, f28, f30) 393 FREG_MOVE_5(f22, f24, f26, f28, f30)
398 subcc %g1, 64, %g1 394 subcc %g1, 64, %g1
399 add %o4, 64, %o4 395 add %o4, 64, %o4
400 bne,pt %xcc, 1b 396 bne,pt %xcc, 1b
401 LOAD(prefetch, %o4 + 64, #one_read) 397 LOAD(prefetch, %o4 + 64, #one_read)
402 ba,pt %xcc, 195f 398 ba,pt %xcc, 195f
403 nop 399 nop
404 400
405 150: sub %o4, 32, %g2 401 150: sub %o4, 32, %g2
406 FREG_LOAD_4(%g2, f0, f2, f4, f6) 402 FREG_LOAD_4(%g2, f0, f2, f4, f6)
407 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 403 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
408 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 404 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
409 FREG_FROB(f0, f2, f4, f6, f16, f18, f2 405 FREG_FROB(f0, f2, f4, f6, f16, f18, f20, f22, f24)
410 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 406 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
411 FREG_MOVE_4(f24, f26, f28, f30) 407 FREG_MOVE_4(f24, f26, f28, f30)
412 subcc %g1, 64, %g1 408 subcc %g1, 64, %g1
413 add %o4, 64, %o4 409 add %o4, 64, %o4
414 bne,pt %xcc, 1b 410 bne,pt %xcc, 1b
415 LOAD(prefetch, %o4 + 64, #one_read) 411 LOAD(prefetch, %o4 + 64, #one_read)
416 ba,pt %xcc, 195f 412 ba,pt %xcc, 195f
417 nop 413 nop
418 414
419 160: sub %o4, 24, %g2 415 160: sub %o4, 24, %g2
420 FREG_LOAD_3(%g2, f0, f2, f4) 416 FREG_LOAD_3(%g2, f0, f2, f4)
421 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 417 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
422 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 418 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
423 FREG_FROB(f0, f2, f4, f16, f18, f20, f 419 FREG_FROB(f0, f2, f4, f16, f18, f20, f22, f24, f26)
424 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 420 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
425 FREG_MOVE_3(f26, f28, f30) 421 FREG_MOVE_3(f26, f28, f30)
426 subcc %g1, 64, %g1 422 subcc %g1, 64, %g1
427 add %o4, 64, %o4 423 add %o4, 64, %o4
428 bne,pt %xcc, 1b 424 bne,pt %xcc, 1b
429 LOAD(prefetch, %o4 + 64, #one_read) 425 LOAD(prefetch, %o4 + 64, #one_read)
430 ba,pt %xcc, 195f 426 ba,pt %xcc, 195f
431 nop 427 nop
432 428
433 170: sub %o4, 16, %g2 429 170: sub %o4, 16, %g2
434 FREG_LOAD_2(%g2, f0, f2) 430 FREG_LOAD_2(%g2, f0, f2)
435 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 431 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
436 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 432 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
437 FREG_FROB(f0, f2, f16, f18, f20, f22, 433 FREG_FROB(f0, f2, f16, f18, f20, f22, f24, f26, f28)
438 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 434 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
439 FREG_MOVE_2(f28, f30) 435 FREG_MOVE_2(f28, f30)
440 subcc %g1, 64, %g1 436 subcc %g1, 64, %g1
441 add %o4, 64, %o4 437 add %o4, 64, %o4
442 bne,pt %xcc, 1b 438 bne,pt %xcc, 1b
443 LOAD(prefetch, %o4 + 64, #one_read) 439 LOAD(prefetch, %o4 + 64, #one_read)
444 ba,pt %xcc, 195f 440 ba,pt %xcc, 195f
445 nop 441 nop
446 442
447 180: sub %o4, 8, %g2 443 180: sub %o4, 8, %g2
448 FREG_LOAD_1(%g2, f0) 444 FREG_LOAD_1(%g2, f0)
449 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 445 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
450 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl 446 EX_LD_FP(LOAD_BLK(%o4, %f16), NG2_retl_o2_plus_g1)
451 FREG_FROB(f0, f16, f18, f20, f22, f24, 447 FREG_FROB(f0, f16, f18, f20, f22, f24, f26, f28, f30)
452 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 448 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1)
453 FREG_MOVE_1(f30) 449 FREG_MOVE_1(f30)
454 subcc %g1, 64, %g1 450 subcc %g1, 64, %g1
455 add %o4, 64, %o4 451 add %o4, 64, %o4
456 bne,pt %xcc, 1b 452 bne,pt %xcc, 1b
457 LOAD(prefetch, %o4 + 64, #one_read) 453 LOAD(prefetch, %o4 + 64, #one_read)
458 ba,pt %xcc, 195f 454 ba,pt %xcc, 195f
459 nop 455 nop
460 456
461 190: 457 190:
462 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), N 458 1: EX_ST_FP(STORE_INIT(%g0, %o4 + %g3), NG2_retl_o2_plus_g1)
463 subcc %g1, 64, %g1 459 subcc %g1, 64, %g1
464 EX_LD_FP(LOAD_BLK(%o4, %f0), NG2_retl_ 460 EX_LD_FP(LOAD_BLK(%o4, %f0), NG2_retl_o2_plus_g1_plus_64)
465 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG 461 EX_ST_FP(STORE_BLK(%f0, %o4 + %g3), NG2_retl_o2_plus_g1_plus_64)
466 add %o4, 64, %o4 462 add %o4, 64, %o4
467 bne,pt %xcc, 1b 463 bne,pt %xcc, 1b
468 LOAD(prefetch, %o4 + 64, #one_read) 464 LOAD(prefetch, %o4 + 64, #one_read)
469 465
470 195: 466 195:
471 add %o4, %g3, %o0 467 add %o4, %g3, %o0
472 membar #Sync 468 membar #Sync
473 469
474 VISExitHalf 470 VISExitHalf
475 471
476 /* %o2 contains any final bytes still 472 /* %o2 contains any final bytes still needed to be copied
477 * over. If anything is left, we copy 473 * over. If anything is left, we copy it one byte at a time.
478 */ 474 */
479 brz,pt %o2, 85f 475 brz,pt %o2, 85f
480 sub %o0, %o1, GLOBAL_SPARE 476 sub %o0, %o1, GLOBAL_SPARE
481 ba,a,pt %XCC, 90f 477 ba,a,pt %XCC, 90f
482 nop <<
483 478
484 .align 64 479 .align 64
485 75: /* 16 < len <= 64 */ 480 75: /* 16 < len <= 64 */
486 bne,pn %XCC, 75f 481 bne,pn %XCC, 75f
487 sub %o0, %o1, GLOBAL_SPARE 482 sub %o0, %o1, GLOBAL_SPARE
488 483
489 72: 484 72:
490 andn %o2, 0xf, %o4 485 andn %o2, 0xf, %o4
491 and %o2, 0xf, %o2 486 and %o2, 0xf, %o2
492 1: subcc %o4, 0x10, %o4 487 1: subcc %o4, 0x10, %o4
493 EX_LD(LOAD(ldx, %o1, %o5), NG2_retl_o2 488 EX_LD(LOAD(ldx, %o1, %o5), NG2_retl_o2_plus_o4_plus_16)
494 add %o1, 0x08, %o1 489 add %o1, 0x08, %o1
495 EX_LD(LOAD(ldx, %o1, %g1), NG2_retl_o2 490 EX_LD(LOAD(ldx, %o1, %g1), NG2_retl_o2_plus_o4_plus_16)
496 sub %o1, 0x08, %o1 491 sub %o1, 0x08, %o1
497 EX_ST(STORE(stx, %o5, %o1 + GLOBAL_SPA 492 EX_ST(STORE(stx, %o5, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_o4_plus_16)
498 add %o1, 0x8, %o1 493 add %o1, 0x8, %o1
499 EX_ST(STORE(stx, %g1, %o1 + GLOBAL_SPA 494 EX_ST(STORE(stx, %g1, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_o4_plus_8)
500 bgu,pt %XCC, 1b 495 bgu,pt %XCC, 1b
501 add %o1, 0x8, %o1 496 add %o1, 0x8, %o1
502 73: andcc %o2, 0x8, %g0 497 73: andcc %o2, 0x8, %g0
503 be,pt %XCC, 1f 498 be,pt %XCC, 1f
504 nop 499 nop
505 sub %o2, 0x8, %o2 500 sub %o2, 0x8, %o2
506 EX_LD(LOAD(ldx, %o1, %o5), NG2_retl_o2 501 EX_LD(LOAD(ldx, %o1, %o5), NG2_retl_o2_plus_8)
507 EX_ST(STORE(stx, %o5, %o1 + GLOBAL_SPA 502 EX_ST(STORE(stx, %o5, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_8)
508 add %o1, 0x8, %o1 503 add %o1, 0x8, %o1
509 1: andcc %o2, 0x4, %g0 504 1: andcc %o2, 0x4, %g0
510 be,pt %XCC, 1f 505 be,pt %XCC, 1f
511 nop 506 nop
512 sub %o2, 0x4, %o2 507 sub %o2, 0x4, %o2
513 EX_LD(LOAD(lduw, %o1, %o5), NG2_retl_o 508 EX_LD(LOAD(lduw, %o1, %o5), NG2_retl_o2_plus_4)
514 EX_ST(STORE(stw, %o5, %o1 + GLOBAL_SPA 509 EX_ST(STORE(stw, %o5, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_4)
515 add %o1, 0x4, %o1 510 add %o1, 0x4, %o1
516 1: cmp %o2, 0 511 1: cmp %o2, 0
517 be,pt %XCC, 85f 512 be,pt %XCC, 85f
518 nop 513 nop
519 ba,pt %xcc, 90f 514 ba,pt %xcc, 90f
520 nop 515 nop
521 516
522 75: 517 75:
523 andcc %o0, 0x7, %g1 518 andcc %o0, 0x7, %g1
524 sub %g1, 0x8, %g1 519 sub %g1, 0x8, %g1
525 be,pn %icc, 2f 520 be,pn %icc, 2f
526 sub %g0, %g1, %g1 521 sub %g0, %g1, %g1
527 sub %o2, %g1, %o2 522 sub %o2, %g1, %o2
528 523
529 1: subcc %g1, 1, %g1 524 1: subcc %g1, 1, %g1
530 EX_LD(LOAD(ldub, %o1, %o5), NG2_retl_o 525 EX_LD(LOAD(ldub, %o1, %o5), NG2_retl_o2_plus_g1_plus_1)
531 EX_ST(STORE(stb, %o5, %o1 + GLOBAL_SPA 526 EX_ST(STORE(stb, %o5, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_g1_plus_1)
532 bgu,pt %icc, 1b 527 bgu,pt %icc, 1b
533 add %o1, 1, %o1 528 add %o1, 1, %o1
534 529
535 2: add %o1, GLOBAL_SPARE, %o0 530 2: add %o1, GLOBAL_SPARE, %o0
536 andcc %o1, 0x7, %g1 531 andcc %o1, 0x7, %g1
537 bne,pt %icc, 8f 532 bne,pt %icc, 8f
538 sll %g1, 3, %g1 533 sll %g1, 3, %g1
539 534
540 cmp %o2, 16 535 cmp %o2, 16
541 bgeu,pt %icc, 72b 536 bgeu,pt %icc, 72b
542 nop 537 nop
543 ba,a,pt %xcc, 73b 538 ba,a,pt %xcc, 73b
544 539
545 8: mov 64, GLOBAL_SPARE 540 8: mov 64, GLOBAL_SPARE
546 andn %o1, 0x7, %o1 541 andn %o1, 0x7, %o1
547 EX_LD(LOAD(ldx, %o1, %g2), NG2_retl_o2 542 EX_LD(LOAD(ldx, %o1, %g2), NG2_retl_o2)
548 sub GLOBAL_SPARE, %g1, GLO 543 sub GLOBAL_SPARE, %g1, GLOBAL_SPARE
549 andn %o2, 0x7, %o4 544 andn %o2, 0x7, %o4
550 sllx %g2, %g1, %g2 545 sllx %g2, %g1, %g2
551 1: add %o1, 0x8, %o1 546 1: add %o1, 0x8, %o1
552 EX_LD(LOAD(ldx, %o1, %g3), NG2_retl_o2 547 EX_LD(LOAD(ldx, %o1, %g3), NG2_retl_o2_and_7_plus_o4)
553 subcc %o4, 0x8, %o4 548 subcc %o4, 0x8, %o4
554 srlx %g3, GLOBAL_SPARE, %o5 549 srlx %g3, GLOBAL_SPARE, %o5
555 or %o5, %g2, %o5 550 or %o5, %g2, %o5
556 EX_ST(STORE(stx, %o5, %o0), NG2_retl_o 551 EX_ST(STORE(stx, %o5, %o0), NG2_retl_o2_and_7_plus_o4_plus_8)
557 add %o0, 0x8, %o0 552 add %o0, 0x8, %o0
558 bgu,pt %icc, 1b 553 bgu,pt %icc, 1b
559 sllx %g3, %g1, %g2 554 sllx %g3, %g1, %g2
560 555
561 srl %g1, 3, %g1 556 srl %g1, 3, %g1
562 andcc %o2, 0x7, %o2 557 andcc %o2, 0x7, %o2
563 be,pn %icc, 85f 558 be,pn %icc, 85f
564 add %o1, %g1, %o1 559 add %o1, %g1, %o1
565 ba,pt %xcc, 90f 560 ba,pt %xcc, 90f
566 sub %o0, %o1, GLOBAL_SPARE 561 sub %o0, %o1, GLOBAL_SPARE
567 562
568 .align 64 563 .align 64
569 80: /* 0 < len <= 16 */ 564 80: /* 0 < len <= 16 */
570 andcc GLOBAL_SPARE, 0x3, %g0 565 andcc GLOBAL_SPARE, 0x3, %g0
571 bne,pn %XCC, 90f 566 bne,pn %XCC, 90f
572 sub %o0, %o1, GLOBAL_SPARE 567 sub %o0, %o1, GLOBAL_SPARE
573 568
574 1: 569 1:
575 subcc %o2, 4, %o2 570 subcc %o2, 4, %o2
576 EX_LD(LOAD(lduw, %o1, %g1), NG2_retl_o 571 EX_LD(LOAD(lduw, %o1, %g1), NG2_retl_o2_plus_4)
577 EX_ST(STORE(stw, %g1, %o1 + GLOBAL_SPA 572 EX_ST(STORE(stw, %g1, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_4)
578 bgu,pt %XCC, 1b 573 bgu,pt %XCC, 1b
579 add %o1, 4, %o1 574 add %o1, 4, %o1
580 575
581 85: retl 576 85: retl
582 mov EX_RETVAL(%o3), %o0 577 mov EX_RETVAL(%o3), %o0
583 578
584 .align 32 579 .align 32
585 90: 580 90:
586 subcc %o2, 1, %o2 581 subcc %o2, 1, %o2
587 EX_LD(LOAD(ldub, %o1, %g1), NG2_retl_o 582 EX_LD(LOAD(ldub, %o1, %g1), NG2_retl_o2_plus_1)
588 EX_ST(STORE(stb, %g1, %o1 + GLOBAL_SPA 583 EX_ST(STORE(stb, %g1, %o1 + GLOBAL_SPARE), NG2_retl_o2_plus_1)
589 bgu,pt %XCC, 90b 584 bgu,pt %XCC, 90b
590 add %o1, 1, %o1 585 add %o1, 1, %o1
591 retl 586 retl
592 mov EX_RETVAL(%o3), %o0 587 mov EX_RETVAL(%o3), %o0
593 588
594 .size FUNC_NAME, .-FUNC_NAME 589 .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.