arch/sparc/include/asm/xor_32.h

   1 /* SPDX-License-Identifier: GPL-2.0-or-later */
   2 /*
   3  * include/asm/xor.h
   4  *
   5  * Optimized RAID-5 checksumming functions for 32-bit Sparc.
   6  */
   7
   8 /*
   9  * High speed xor_block operation for RAID4/5 utilizing the
  10  * ldd/std SPARC instructions.
  11  *
  12  * Copyright (C) 1999 Jakub Jelinek (jj@ultra.linux.cz)
  13  */
  14
  15 static void
  16 sparc_2(unsigned long bytes, unsigned long *p1, unsigned long *p2)
  17 {
  18         int lines = bytes / (sizeof (long)) / 8;
  19
  20         do {
  21                 __asm__ __volatile__(
  22                   "ldd [%0 + 0x00], %%g2\n\t"
  23                   "ldd [%0 + 0x08], %%g4\n\t"
  24                   "ldd [%0 + 0x10], %%o0\n\t"
  25                   "ldd [%0 + 0x18], %%o2\n\t"
  26                   "ldd [%1 + 0x00], %%o4\n\t"
  27                   "ldd [%1 + 0x08], %%l0\n\t"
  28                   "ldd [%1 + 0x10], %%l2\n\t"
  29                   "ldd [%1 + 0x18], %%l4\n\t"
  30                   "xor %%g2, %%o4, %%g2\n\t"
  31                   "xor %%g3, %%o5, %%g3\n\t"
  32                   "xor %%g4, %%l0, %%g4\n\t"
  33                   "xor %%g5, %%l1, %%g5\n\t"
  34                   "xor %%o0, %%l2, %%o0\n\t"
  35                   "xor %%o1, %%l3, %%o1\n\t"
  36                   "xor %%o2, %%l4, %%o2\n\t"
  37                   "xor %%o3, %%l5, %%o3\n\t"
  38                   "std %%g2, [%0 + 0x00]\n\t"
  39                   "std %%g4, [%0 + 0x08]\n\t"
  40                   "std %%o0, [%0 + 0x10]\n\t"
  41                   "std %%o2, [%0 + 0x18]\n"
  42                 :
  43                 : "r" (p1), "r" (p2)
  44                 : "g2", "g3", "g4", "g5",
  45                   "o0", "o1", "o2", "o3", "o4", "o5",
  46                   "l0", "l1", "l2", "l3", "l4", "l5");
  47                 p1 += 8;
  48                 p2 += 8;
  49         } while (--lines > 0);
  50 }
  51
  52 static void
  53 sparc_3(unsigned long bytes, unsigned long *p1, unsigned long *p2,
  54         unsigned long *p3)
  55 {
  56         int lines = bytes / (sizeof (long)) / 8;
  57
  58         do {
  59                 __asm__ __volatile__(
  60                   "ldd [%0 + 0x00], %%g2\n\t"
  61                   "ldd [%0 + 0x08], %%g4\n\t"
  62                   "ldd [%0 + 0x10], %%o0\n\t"
  63                   "ldd [%0 + 0x18], %%o2\n\t"
  64                   "ldd [%1 + 0x00], %%o4\n\t"
  65                   "ldd [%1 + 0x08], %%l0\n\t"
  66                   "ldd [%1 + 0x10], %%l2\n\t"
  67                   "ldd [%1 + 0x18], %%l4\n\t"
  68                   "xor %%g2, %%o4, %%g2\n\t"
  69                   "xor %%g3, %%o5, %%g3\n\t"
  70                   "ldd [%2 + 0x00], %%o4\n\t"
  71                   "xor %%g4, %%l0, %%g4\n\t"
  72                   "xor %%g5, %%l1, %%g5\n\t"
  73                   "ldd [%2 + 0x08], %%l0\n\t"
  74                   "xor %%o0, %%l2, %%o0\n\t"
  75                   "xor %%o1, %%l3, %%o1\n\t"
  76                   "ldd [%2 + 0x10], %%l2\n\t"
  77                   "xor %%o2, %%l4, %%o2\n\t"
  78                   "xor %%o3, %%l5, %%o3\n\t"
  79                   "ldd [%2 + 0x18], %%l4\n\t"
  80                   "xor %%g2, %%o4, %%g2\n\t"
  81                   "xor %%g3, %%o5, %%g3\n\t"
  82                   "xor %%g4, %%l0, %%g4\n\t"
  83                   "xor %%g5, %%l1, %%g5\n\t"
  84                   "xor %%o0, %%l2, %%o0\n\t"
  85                   "xor %%o1, %%l3, %%o1\n\t"
  86                   "xor %%o2, %%l4, %%o2\n\t"
  87                   "xor %%o3, %%l5, %%o3\n\t"
  88                   "std %%g2, [%0 + 0x00]\n\t"
  89                   "std %%g4, [%0 + 0x08]\n\t"
  90                   "std %%o0, [%0 + 0x10]\n\t"
  91                   "std %%o2, [%0 + 0x18]\n"
  92                 :
  93                 : "r" (p1), "r" (p2), "r" (p3)
  94                 : "g2", "g3", "g4", "g5",
  95                   "o0", "o1", "o2", "o3", "o4", "o5",
  96                   "l0", "l1", "l2", "l3", "l4", "l5");
  97                 p1 += 8;
  98                 p2 += 8;
  99                 p3 += 8;
 100         } while (--lines > 0);
 101 }
 102
 103 static void
 104 sparc_4(unsigned long bytes, unsigned long *p1, unsigned long *p2,
 105         unsigned long *p3, unsigned long *p4)
 106 {
 107         int lines = bytes / (sizeof (long)) / 8;
 108
 109         do {
 110                 __asm__ __volatile__(
 111                   "ldd [%0 + 0x00], %%g2\n\t"
 112                   "ldd [%0 + 0x08], %%g4\n\t"
 113                   "ldd [%0 + 0x10], %%o0\n\t"
 114                   "ldd [%0 + 0x18], %%o2\n\t"
 115                   "ldd [%1 + 0x00], %%o4\n\t"
 116                   "ldd [%1 + 0x08], %%l0\n\t"
 117                   "ldd [%1 + 0x10], %%l2\n\t"
 118                   "ldd [%1 + 0x18], %%l4\n\t"
 119                   "xor %%g2, %%o4, %%g2\n\t"
 120                   "xor %%g3, %%o5, %%g3\n\t"
 121                   "ldd [%2 + 0x00], %%o4\n\t"
 122                   "xor %%g4, %%l0, %%g4\n\t"
 123                   "xor %%g5, %%l1, %%g5\n\t"
 124                   "ldd [%2 + 0x08], %%l0\n\t"
 125                   "xor %%o0, %%l2, %%o0\n\t"
 126                   "xor %%o1, %%l3, %%o1\n\t"
 127                   "ldd [%2 + 0x10], %%l2\n\t"
 128                   "xor %%o2, %%l4, %%o2\n\t"
 129                   "xor %%o3, %%l5, %%o3\n\t"
 130                   "ldd [%2 + 0x18], %%l4\n\t"
 131                   "xor %%g2, %%o4, %%g2\n\t"
 132                   "xor %%g3, %%o5, %%g3\n\t"
 133                   "ldd [%3 + 0x00], %%o4\n\t"
 134                   "xor %%g4, %%l0, %%g4\n\t"
 135                   "xor %%g5, %%l1, %%g5\n\t"
 136                   "ldd [%3 + 0x08], %%l0\n\t"
 137                   "xor %%o0, %%l2, %%o0\n\t"
 138                   "xor %%o1, %%l3, %%o1\n\t"
 139                   "ldd [%3 + 0x10], %%l2\n\t"
 140                   "xor %%o2, %%l4, %%o2\n\t"
 141                   "xor %%o3, %%l5, %%o3\n\t"
 142                   "ldd [%3 + 0x18], %%l4\n\t"
 143                   "xor %%g2, %%o4, %%g2\n\t"
 144                   "xor %%g3, %%o5, %%g3\n\t"
 145                   "xor %%g4, %%l0, %%g4\n\t"
 146                   "xor %%g5, %%l1, %%g5\n\t"
 147                   "xor %%o0, %%l2, %%o0\n\t"
 148                   "xor %%o1, %%l3, %%o1\n\t"
 149                   "xor %%o2, %%l4, %%o2\n\t"
 150                   "xor %%o3, %%l5, %%o3\n\t"
 151                   "std %%g2, [%0 + 0x00]\n\t"
 152                   "std %%g4, [%0 + 0x08]\n\t"
 153                   "std %%o0, [%0 + 0x10]\n\t"
 154                   "std %%o2, [%0 + 0x18]\n"
 155                 :
 156                 : "r" (p1), "r" (p2), "r" (p3), "r" (p4)
 157                 : "g2", "g3", "g4", "g5",
 158                   "o0", "o1", "o2", "o3", "o4", "o5",
 159                   "l0", "l1", "l2", "l3", "l4", "l5");
 160                 p1 += 8;
 161                 p2 += 8;
 162                 p3 += 8;
 163                 p4 += 8;
 164         } while (--lines > 0);
 165 }
 166
 167 static void
 168 sparc_5(unsigned long bytes, unsigned long *p1, unsigned long *p2,
 169         unsigned long *p3, unsigned long *p4, unsigned long *p5)
 170 {
 171         int lines = bytes / (sizeof (long)) / 8;
 172
 173         do {
 174                 __asm__ __volatile__(
 175                   "ldd [%0 + 0x00], %%g2\n\t"
 176                   "ldd [%0 + 0x08], %%g4\n\t"
 177                   "ldd [%0 + 0x10], %%o0\n\t"
 178                   "ldd [%0 + 0x18], %%o2\n\t"
 179                   "ldd [%1 + 0x00], %%o4\n\t"
 180                   "ldd [%1 + 0x08], %%l0\n\t"
 181                   "ldd [%1 + 0x10], %%l2\n\t"
 182                   "ldd [%1 + 0x18], %%l4\n\t"
 183                   "xor %%g2, %%o4, %%g2\n\t"
 184                   "xor %%g3, %%o5, %%g3\n\t"
 185                   "ldd [%2 + 0x00], %%o4\n\t"
 186                   "xor %%g4, %%l0, %%g4\n\t"
 187                   "xor %%g5, %%l1, %%g5\n\t"
 188                   "ldd [%2 + 0x08], %%l0\n\t"
 189                   "xor %%o0, %%l2, %%o0\n\t"
 190                   "xor %%o1, %%l3, %%o1\n\t"
 191                   "ldd [%2 + 0x10], %%l2\n\t"
 192                   "xor %%o2, %%l4, %%o2\n\t"
 193                   "xor %%o3, %%l5, %%o3\n\t"
 194                   "ldd [%2 + 0x18], %%l4\n\t"
 195                   "xor %%g2, %%o4, %%g2\n\t"
 196                   "xor %%g3, %%o5, %%g3\n\t"
 197                   "ldd [%3 + 0x00], %%o4\n\t"
 198                   "xor %%g4, %%l0, %%g4\n\t"
 199                   "xor %%g5, %%l1, %%g5\n\t"
 200                   "ldd [%3 + 0x08], %%l0\n\t"
 201                   "xor %%o0, %%l2, %%o0\n\t"
 202                   "xor %%o1, %%l3, %%o1\n\t"
 203                   "ldd [%3 + 0x10], %%l2\n\t"
 204                   "xor %%o2, %%l4, %%o2\n\t"
 205                   "xor %%o3, %%l5, %%o3\n\t"
 206                   "ldd [%3 + 0x18], %%l4\n\t"
 207                   "xor %%g2, %%o4, %%g2\n\t"
 208                   "xor %%g3, %%o5, %%g3\n\t"
 209                   "ldd [%4 + 0x00], %%o4\n\t"
 210                   "xor %%g4, %%l0, %%g4\n\t"
 211                   "xor %%g5, %%l1, %%g5\n\t"
 212                   "ldd [%4 + 0x08], %%l0\n\t"
 213                   "xor %%o0, %%l2, %%o0\n\t"
 214                   "xor %%o1, %%l3, %%o1\n\t"
 215                   "ldd [%4 + 0x10], %%l2\n\t"
 216                   "xor %%o2, %%l4, %%o2\n\t"
 217                   "xor %%o3, %%l5, %%o3\n\t"
 218                   "ldd [%4 + 0x18], %%l4\n\t"
 219                   "xor %%g2, %%o4, %%g2\n\t"
 220                   "xor %%g3, %%o5, %%g3\n\t"
 221                   "xor %%g4, %%l0, %%g4\n\t"
 222                   "xor %%g5, %%l1, %%g5\n\t"
 223                   "xor %%o0, %%l2, %%o0\n\t"
 224                   "xor %%o1, %%l3, %%o1\n\t"
 225                   "xor %%o2, %%l4, %%o2\n\t"
 226                   "xor %%o3, %%l5, %%o3\n\t"
 227                   "std %%g2, [%0 + 0x00]\n\t"
 228                   "std %%g4, [%0 + 0x08]\n\t"
 229                   "std %%o0, [%0 + 0x10]\n\t"
 230                   "std %%o2, [%0 + 0x18]\n"
 231                 :
 232                 : "r" (p1), "r" (p2), "r" (p3), "r" (p4), "r" (p5)
 233                 : "g2", "g3", "g4", "g5",
 234                   "o0", "o1", "o2", "o3", "o4", "o5",
 235                   "l0", "l1", "l2", "l3", "l4", "l5");
 236                 p1 += 8;
 237                 p2 += 8;
 238                 p3 += 8;
 239                 p4 += 8;
 240                 p5 += 8;
 241         } while (--lines > 0);
 242 }
 243
 244 static struct xor_block_template xor_block_SPARC = {
 245         .name   = "SPARC",
 246         .do_2   = sparc_2,
 247         .do_3   = sparc_3,
 248         .do_4   = sparc_4,
 249         .do_5   = sparc_5,
 250 };
 251
 252 /* For grins, also test the generic routines.  */
 253 #include <asm-generic/xor.h>
 254
 255 #undef XOR_TRY_TEMPLATES
 256 #define XOR_TRY_TEMPLATES                               \
 257         do {                                            \
 258                 xor_speed(&xor_block_8regs);            \
 259                 xor_speed(&xor_block_32regs);           \
 260                 xor_speed(&xor_block_SPARC);            \
 261         } while (0)