Blame view
lib/raid6/sse2.c
8.34 KB
1da177e4c Linux-2.6.12-rc2 |
1 2 3 4 5 6 7 |
/* -*- linux-c -*- ------------------------------------------------------- * * * Copyright 2002 H. Peter Anvin - All Rights Reserved * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation, Inc., 53 Temple Place Ste 330, |
93ed05e2a md: fix typo in F... |
8 |
* Boston MA 02111-1307, USA; either version 2 of the License, or |
1da177e4c Linux-2.6.12-rc2 |
9 10 11 12 13 |
* (at your option) any later version; incorporated herein by reference. * * ----------------------------------------------------------------------- */ /* |
a8e026c78 Further tidyup of... |
14 |
* raid6/sse2.c |
1da177e4c Linux-2.6.12-rc2 |
15 16 17 18 |
* * SSE-2 implementation of RAID-6 syndrome functions * */ |
ca5cd877a x86 merge fallout... |
19 |
#if (defined(__i386__) || defined(__x86_64__)) && !defined(__arch_um__) |
1da177e4c Linux-2.6.12-rc2 |
20 |
|
f701d589a md/raid6: move ra... |
21 |
#include <linux/raid/pq.h> |
a8e026c78 Further tidyup of... |
22 |
#include "x86.h" |
1da177e4c Linux-2.6.12-rc2 |
23 24 25 26 27 28 29 30 31 |
static const struct raid6_sse_constants { u64 x1d[2]; } raid6_sse_constants __attribute__((aligned(16))) = { { 0x1d1d1d1d1d1d1d1dULL, 0x1d1d1d1d1d1d1d1dULL }, }; static int raid6_have_sse2(void) { |
1da177e4c Linux-2.6.12-rc2 |
32 33 34 35 36 |
/* Not really boot_cpu but "all_cpus" */ return boot_cpu_has(X86_FEATURE_MMX) && boot_cpu_has(X86_FEATURE_FXSR) && boot_cpu_has(X86_FEATURE_XMM) && boot_cpu_has(X86_FEATURE_XMM2); |
1da177e4c Linux-2.6.12-rc2 |
37 38 39 40 41 42 43 44 45 46 |
} /* * Plain SSE2 implementation */ static void raid6_sse21_gen_syndrome(int disks, size_t bytes, void **ptrs) { u8 **dptr = (u8 **)ptrs; u8 *p, *q; int d, z, z0; |
1da177e4c Linux-2.6.12-rc2 |
47 48 49 50 |
z0 = disks - 3; /* Highest data disk */ p = dptr[z0+1]; /* XOR parity */ q = dptr[z0+2]; /* RS syndrome */ |
a723406c4 [PATCH] md: RAID6... |
51 |
kernel_fpu_begin(); |
1da177e4c Linux-2.6.12-rc2 |
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 |
asm volatile("movdqa %0,%%xmm0" : : "m" (raid6_sse_constants.x1d[0])); asm volatile("pxor %xmm5,%xmm5"); /* Zero temp */ for ( d = 0 ; d < bytes ; d += 16 ) { asm volatile("prefetchnta %0" : : "m" (dptr[z0][d])); asm volatile("movdqa %0,%%xmm2" : : "m" (dptr[z0][d])); /* P[0] */ asm volatile("prefetchnta %0" : : "m" (dptr[z0-1][d])); asm volatile("movdqa %xmm2,%xmm4"); /* Q[0] */ asm volatile("movdqa %0,%%xmm6" : : "m" (dptr[z0-1][d])); for ( z = z0-2 ; z >= 0 ; z-- ) { asm volatile("prefetchnta %0" : : "m" (dptr[z][d])); asm volatile("pcmpgtb %xmm4,%xmm5"); asm volatile("paddb %xmm4,%xmm4"); asm volatile("pand %xmm0,%xmm5"); asm volatile("pxor %xmm5,%xmm4"); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pxor %xmm6,%xmm2"); asm volatile("pxor %xmm6,%xmm4"); asm volatile("movdqa %0,%%xmm6" : : "m" (dptr[z][d])); } asm volatile("pcmpgtb %xmm4,%xmm5"); asm volatile("paddb %xmm4,%xmm4"); asm volatile("pand %xmm0,%xmm5"); asm volatile("pxor %xmm5,%xmm4"); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pxor %xmm6,%xmm2"); asm volatile("pxor %xmm6,%xmm4"); asm volatile("movntdq %%xmm2,%0" : "=m" (p[d])); asm volatile("pxor %xmm2,%xmm2"); asm volatile("movntdq %%xmm4,%0" : "=m" (q[d])); asm volatile("pxor %xmm4,%xmm4"); } |
1da177e4c Linux-2.6.12-rc2 |
86 |
asm volatile("sfence" : : : "memory"); |
a723406c4 [PATCH] md: RAID6... |
87 |
kernel_fpu_end(); |
1da177e4c Linux-2.6.12-rc2 |
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 |
} const struct raid6_calls raid6_sse2x1 = { raid6_sse21_gen_syndrome, raid6_have_sse2, "sse2x1", 1 /* Has cache hints */ }; /* * Unrolled-by-2 SSE2 implementation */ static void raid6_sse22_gen_syndrome(int disks, size_t bytes, void **ptrs) { u8 **dptr = (u8 **)ptrs; u8 *p, *q; int d, z, z0; |
1da177e4c Linux-2.6.12-rc2 |
105 106 107 108 |
z0 = disks - 3; /* Highest data disk */ p = dptr[z0+1]; /* XOR parity */ q = dptr[z0+2]; /* RS syndrome */ |
a723406c4 [PATCH] md: RAID6... |
109 |
kernel_fpu_begin(); |
1da177e4c Linux-2.6.12-rc2 |
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 |
asm volatile("movdqa %0,%%xmm0" : : "m" (raid6_sse_constants.x1d[0])); asm volatile("pxor %xmm5,%xmm5"); /* Zero temp */ asm volatile("pxor %xmm7,%xmm7"); /* Zero temp */ /* We uniformly assume a single prefetch covers at least 32 bytes */ for ( d = 0 ; d < bytes ; d += 32 ) { asm volatile("prefetchnta %0" : : "m" (dptr[z0][d])); asm volatile("movdqa %0,%%xmm2" : : "m" (dptr[z0][d])); /* P[0] */ asm volatile("movdqa %0,%%xmm3" : : "m" (dptr[z0][d+16])); /* P[1] */ asm volatile("movdqa %xmm2,%xmm4"); /* Q[0] */ asm volatile("movdqa %xmm3,%xmm6"); /* Q[1] */ for ( z = z0-1 ; z >= 0 ; z-- ) { asm volatile("prefetchnta %0" : : "m" (dptr[z][d])); asm volatile("pcmpgtb %xmm4,%xmm5"); asm volatile("pcmpgtb %xmm6,%xmm7"); asm volatile("paddb %xmm4,%xmm4"); asm volatile("paddb %xmm6,%xmm6"); asm volatile("pand %xmm0,%xmm5"); asm volatile("pand %xmm0,%xmm7"); asm volatile("pxor %xmm5,%xmm4"); asm volatile("pxor %xmm7,%xmm6"); asm volatile("movdqa %0,%%xmm5" : : "m" (dptr[z][d])); asm volatile("movdqa %0,%%xmm7" : : "m" (dptr[z][d+16])); asm volatile("pxor %xmm5,%xmm2"); asm volatile("pxor %xmm7,%xmm3"); asm volatile("pxor %xmm5,%xmm4"); asm volatile("pxor %xmm7,%xmm6"); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pxor %xmm7,%xmm7"); } asm volatile("movntdq %%xmm2,%0" : "=m" (p[d])); asm volatile("movntdq %%xmm3,%0" : "=m" (p[d+16])); asm volatile("movntdq %%xmm4,%0" : "=m" (q[d])); asm volatile("movntdq %%xmm6,%0" : "=m" (q[d+16])); } |
1da177e4c Linux-2.6.12-rc2 |
146 |
asm volatile("sfence" : : : "memory"); |
a723406c4 [PATCH] md: RAID6... |
147 |
kernel_fpu_end(); |
1da177e4c Linux-2.6.12-rc2 |
148 149 150 151 152 153 154 155 156 157 |
} const struct raid6_calls raid6_sse2x2 = { raid6_sse22_gen_syndrome, raid6_have_sse2, "sse2x2", 1 /* Has cache hints */ }; #endif |
ca5cd877a x86 merge fallout... |
158 |
#if defined(__x86_64__) && !defined(__arch_um__) |
1da177e4c Linux-2.6.12-rc2 |
159 160 161 162 163 164 165 166 167 |
/* * Unrolled-by-4 SSE2 implementation */ static void raid6_sse24_gen_syndrome(int disks, size_t bytes, void **ptrs) { u8 **dptr = (u8 **)ptrs; u8 *p, *q; int d, z, z0; |
1da177e4c Linux-2.6.12-rc2 |
168 169 170 171 |
z0 = disks - 3; /* Highest data disk */ p = dptr[z0+1]; /* XOR parity */ q = dptr[z0+2]; /* RS syndrome */ |
a723406c4 [PATCH] md: RAID6... |
172 |
kernel_fpu_begin(); |
1da177e4c Linux-2.6.12-rc2 |
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 |
asm volatile("movdqa %0,%%xmm0" :: "m" (raid6_sse_constants.x1d[0])); asm volatile("pxor %xmm2,%xmm2"); /* P[0] */ asm volatile("pxor %xmm3,%xmm3"); /* P[1] */ asm volatile("pxor %xmm4,%xmm4"); /* Q[0] */ asm volatile("pxor %xmm5,%xmm5"); /* Zero temp */ asm volatile("pxor %xmm6,%xmm6"); /* Q[1] */ asm volatile("pxor %xmm7,%xmm7"); /* Zero temp */ asm volatile("pxor %xmm10,%xmm10"); /* P[2] */ asm volatile("pxor %xmm11,%xmm11"); /* P[3] */ asm volatile("pxor %xmm12,%xmm12"); /* Q[2] */ asm volatile("pxor %xmm13,%xmm13"); /* Zero temp */ asm volatile("pxor %xmm14,%xmm14"); /* Q[3] */ asm volatile("pxor %xmm15,%xmm15"); /* Zero temp */ for ( d = 0 ; d < bytes ; d += 64 ) { for ( z = z0 ; z >= 0 ; z-- ) { /* The second prefetch seems to improve performance... */ asm volatile("prefetchnta %0" :: "m" (dptr[z][d])); asm volatile("prefetchnta %0" :: "m" (dptr[z][d+32])); asm volatile("pcmpgtb %xmm4,%xmm5"); asm volatile("pcmpgtb %xmm6,%xmm7"); asm volatile("pcmpgtb %xmm12,%xmm13"); asm volatile("pcmpgtb %xmm14,%xmm15"); asm volatile("paddb %xmm4,%xmm4"); asm volatile("paddb %xmm6,%xmm6"); asm volatile("paddb %xmm12,%xmm12"); asm volatile("paddb %xmm14,%xmm14"); asm volatile("pand %xmm0,%xmm5"); asm volatile("pand %xmm0,%xmm7"); asm volatile("pand %xmm0,%xmm13"); asm volatile("pand %xmm0,%xmm15"); asm volatile("pxor %xmm5,%xmm4"); asm volatile("pxor %xmm7,%xmm6"); asm volatile("pxor %xmm13,%xmm12"); asm volatile("pxor %xmm15,%xmm14"); asm volatile("movdqa %0,%%xmm5" :: "m" (dptr[z][d])); asm volatile("movdqa %0,%%xmm7" :: "m" (dptr[z][d+16])); asm volatile("movdqa %0,%%xmm13" :: "m" (dptr[z][d+32])); asm volatile("movdqa %0,%%xmm15" :: "m" (dptr[z][d+48])); asm volatile("pxor %xmm5,%xmm2"); asm volatile("pxor %xmm7,%xmm3"); asm volatile("pxor %xmm13,%xmm10"); asm volatile("pxor %xmm15,%xmm11"); asm volatile("pxor %xmm5,%xmm4"); asm volatile("pxor %xmm7,%xmm6"); asm volatile("pxor %xmm13,%xmm12"); asm volatile("pxor %xmm15,%xmm14"); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pxor %xmm7,%xmm7"); asm volatile("pxor %xmm13,%xmm13"); asm volatile("pxor %xmm15,%xmm15"); } asm volatile("movntdq %%xmm2,%0" : "=m" (p[d])); asm volatile("pxor %xmm2,%xmm2"); asm volatile("movntdq %%xmm3,%0" : "=m" (p[d+16])); asm volatile("pxor %xmm3,%xmm3"); asm volatile("movntdq %%xmm10,%0" : "=m" (p[d+32])); asm volatile("pxor %xmm10,%xmm10"); asm volatile("movntdq %%xmm11,%0" : "=m" (p[d+48])); asm volatile("pxor %xmm11,%xmm11"); asm volatile("movntdq %%xmm4,%0" : "=m" (q[d])); asm volatile("pxor %xmm4,%xmm4"); asm volatile("movntdq %%xmm6,%0" : "=m" (q[d+16])); asm volatile("pxor %xmm6,%xmm6"); asm volatile("movntdq %%xmm12,%0" : "=m" (q[d+32])); asm volatile("pxor %xmm12,%xmm12"); asm volatile("movntdq %%xmm14,%0" : "=m" (q[d+48])); asm volatile("pxor %xmm14,%xmm14"); } |
a723406c4 [PATCH] md: RAID6... |
243 |
|
1da177e4c Linux-2.6.12-rc2 |
244 |
asm volatile("sfence" : : : "memory"); |
a723406c4 [PATCH] md: RAID6... |
245 |
kernel_fpu_end(); |
1da177e4c Linux-2.6.12-rc2 |
246 247 248 249 250 251 252 253 254 255 |
} const struct raid6_calls raid6_sse2x4 = { raid6_sse24_gen_syndrome, raid6_have_sse2, "sse2x4", 1 /* Has cache hints */ }; #endif |