arch/ia64/lib/xor.S

   1 /*
   2  * arch/ia64/lib/xor.S
   3  *
   4  * Optimized RAID-5 checksumming functions for IA-64.
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2, or (at your option)
   9  * any later version.
  10  *
  11  * You should have received a copy of the GNU General Public License
  12  * (for example /usr/src/linux/COPYING); if not, write to the Free
  13  * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  14  */
  15
  16 #include <asm/asmmacro.h>
  17
  18 GLOBAL_ENTRY(xor_ia64_2)
  19         .prologue
  20         .fframe 0
  21         .save ar.pfs, r31
  22         alloc r31 = ar.pfs, 3, 0, 13, 16
  23         .save ar.lc, r30
  24         mov r30 = ar.lc
  25         .save pr, r29
  26         mov r29 = pr
  27         ;;
  28         .body
  29         mov r8 = in1
  30         mov ar.ec = 6 + 2
  31         shr in0 = in0, 3
  32         ;;
  33         adds in0 = -1, in0
  34         mov r16 = in1
  35         mov r17 = in2
  36         ;;
  37         mov ar.lc = in0
  38         mov pr.rot = 1 << 16
  39         ;;
  40         .rotr s1[6+1], s2[6+1], d[2]
  41         .rotp p[6+2]
  42 0:
  43 (p[0])  ld8.nta s1[0] = [r16], 8
  44 (p[0])  ld8.nta s2[0] = [r17], 8
  45 (p[6])  xor d[0] = s1[6], s2[6]
  46 (p[6+1])st8.nta [r8] = d[1], 8
  47         nop.f 0
  48         br.ctop.dptk.few 0b
  49         ;;
  50         mov ar.lc = r30
  51         mov pr = r29, -1
  52         br.ret.sptk.few rp
  53 END(xor_ia64_2)
  54
  55 GLOBAL_ENTRY(xor_ia64_3)
  56         .prologue
  57         .fframe 0
  58         .save ar.pfs, r31
  59         alloc r31 = ar.pfs, 4, 0, 20, 24
  60         .save ar.lc, r30
  61         mov r30 = ar.lc
  62         .save pr, r29
  63         mov r29 = pr
  64         ;;
  65         .body
  66         mov r8 = in1
  67         mov ar.ec = 6 + 2
  68         shr in0 = in0, 3
  69         ;;
  70         adds in0 = -1, in0
  71         mov r16 = in1
  72         mov r17 = in2
  73         ;;
  74         mov r18 = in3
  75         mov ar.lc = in0
  76         mov pr.rot = 1 << 16
  77         ;;
  78         .rotr s1[6+1], s2[6+1], s3[6+1], d[2]
  79         .rotp p[6+2]
  80 0:
  81 (p[0])  ld8.nta s1[0] = [r16], 8
  82 (p[0])  ld8.nta s2[0] = [r17], 8
  83 (p[6])  xor d[0] = s1[6], s2[6]
  84         ;;
  85 (p[0])  ld8.nta s3[0] = [r18], 8
  86 (p[6+1])st8.nta [r8] = d[1], 8
  87 (p[6])  xor d[0] = d[0], s3[6]
  88         br.ctop.dptk.few 0b
  89         ;;
  90         mov ar.lc = r30
  91         mov pr = r29, -1
  92         br.ret.sptk.few rp
  93 END(xor_ia64_3)
  94
  95 GLOBAL_ENTRY(xor_ia64_4)
  96         .prologue
  97         .fframe 0
  98         .save ar.pfs, r31
  99         alloc r31 = ar.pfs, 5, 0, 27, 32
 100         .save ar.lc, r30
 101         mov r30 = ar.lc
 102         .save pr, r29
 103         mov r29 = pr
 104         ;;
 105         .body
 106         mov r8 = in1
 107         mov ar.ec = 6 + 2
 108         shr in0 = in0, 3
 109         ;;
 110         adds in0 = -1, in0
 111         mov r16 = in1
 112         mov r17 = in2
 113         ;;
 114         mov r18 = in3
 115         mov ar.lc = in0
 116         mov pr.rot = 1 << 16
 117         mov r19 = in4
 118         ;;
 119         .rotr s1[6+1], s2[6+1], s3[6+1], s4[6+1], d[2]
 120         .rotp p[6+2]
 121 0:
 122 (p[0])  ld8.nta s1[0] = [r16], 8
 123 (p[0])  ld8.nta s2[0] = [r17], 8
 124 (p[6])  xor d[0] = s1[6], s2[6]
 125 (p[0])  ld8.nta s3[0] = [r18], 8
 126 (p[0])  ld8.nta s4[0] = [r19], 8
 127 (p[6])  xor r20 = s3[6], s4[6]
 128         ;;
 129 (p[6+1])st8.nta [r8] = d[1], 8
 130 (p[6])  xor d[0] = d[0], r20
 131         br.ctop.dptk.few 0b
 132         ;;
 133         mov ar.lc = r30
 134         mov pr = r29, -1
 135         br.ret.sptk.few rp
 136 END(xor_ia64_4)
 137
 138 GLOBAL_ENTRY(xor_ia64_5)
 139         .prologue
 140         .fframe 0
 141         .save ar.pfs, r31
 142         alloc r31 = ar.pfs, 6, 0, 34, 40
 143         .save ar.lc, r30
 144         mov r30 = ar.lc
 145         .save pr, r29
 146         mov r29 = pr
 147         ;;
 148         .body
 149         mov r8 = in1
 150         mov ar.ec = 6 + 2
 151         shr in0 = in0, 3
 152         ;;
 153         adds in0 = -1, in0
 154         mov r16 = in1
 155         mov r17 = in2
 156         ;;
 157         mov r18 = in3
 158         mov ar.lc = in0
 159         mov pr.rot = 1 << 16
 160         mov r19 = in4
 161         mov r20 = in5
 162         ;;
 163         .rotr s1[6+1], s2[6+1], s3[6+1], s4[6+1], s5[6+1], d[2]
 164         .rotp p[6+2]
 165 0:
 166 (p[0])  ld8.nta s1[0] = [r16], 8
 167 (p[0])  ld8.nta s2[0] = [r17], 8
 168 (p[6])  xor d[0] = s1[6], s2[6]
 169 (p[0])  ld8.nta s3[0] = [r18], 8
 170 (p[0])  ld8.nta s4[0] = [r19], 8
 171 (p[6])  xor r21 = s3[6], s4[6]
 172         ;;
 173 (p[0])  ld8.nta s5[0] = [r20], 8
 174 (p[6+1])st8.nta [r8] = d[1], 8
 175 (p[6])  xor d[0] = d[0], r21
 176         ;;
 177 (p[6])    xor d[0] = d[0], s5[6]
 178         nop.f 0
 179         br.ctop.dptk.few 0b
 180         ;;
 181         mov ar.lc = r30
 182         mov pr = r29, -1
 183         br.ret.sptk.few rp
 184 END(xor_ia64_5)