arch/powerpc/lib/string_64.S

   1 /*
   2  * This program is free software; you can redistribute it and/or modify
   3  * it under the terms of the GNU General Public License as published by
   4  * the Free Software Foundation; either version 2 of the License, or
   5  * (at your option) any later version.
   6  *
   7  * This program is distributed in the hope that it will be useful,
   8  * but WITHOUT ANY WARRANTY; without even the implied warranty of
   9  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  10  * GNU General Public License for more details.
  11  *
  12  * You should have received a copy of the GNU General Public License
  13  * along with this program; if not, write to the Free Software
  14  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  15  *
  16  * Copyright (C) IBM Corporation, 2012
  17  *
  18  * Author: Anton Blanchard <anton@au.ibm.com>
  19  */
  20
  21 #include <asm/ppc_asm.h>
  22 #include <asm/asm-offsets.h>
  23
  24         .section        ".toc","aw"
  25 PPC64_CACHES:
  26         .tc             ppc64_caches[TC],ppc64_caches
  27         .section        ".text"
  28
  29 /**
  30  * __clear_user: - Zero a block of memory in user space, with less checking.
  31  * @to:   Destination address, in user space.
  32  * @n:    Number of bytes to zero.
  33  *
  34  * Zero a block of memory in user space.  Caller must check
  35  * the specified block with access_ok() before calling this function.
  36  *
  37  * Returns number of bytes that could not be cleared.
  38  * On success, this will be zero.
  39  */
  40
  41         .macro err1
  42 100:
  43         .section __ex_table,"a"
  44         .align 3
  45         .llong 100b,.Ldo_err1
  46         .previous
  47         .endm
  48
  49         .macro err2
  50 200:
  51         .section __ex_table,"a"
  52         .align 3
  53         .llong 200b,.Ldo_err2
  54         .previous
  55         .endm
  56
  57         .macro err3
  58 300:
  59         .section __ex_table,"a"
  60         .align 3
  61         .llong 300b,.Ldo_err3
  62         .previous
  63         .endm
  64
  65 .Ldo_err1:
  66         mr      r3,r8
  67
  68 .Ldo_err2:
  69         mtctr   r4
  70 1:
  71 err3;   stb     r0,0(r3)
  72         addi    r3,r3,1
  73         addi    r4,r4,-1
  74         bdnz    1b
  75
  76 .Ldo_err3:
  77         mr      r3,r4
  78         blr
  79
  80 _GLOBAL(__clear_user)
  81         cmpdi   r4,32
  82         neg     r6,r3
  83         li      r0,0
  84         blt     .Lshort_clear
  85         mr      r8,r3
  86         mtocrf  0x01,r6
  87         clrldi  r6,r6,(64-3)
  88
  89         /* Get the destination 8 byte aligned */
  90         bf      cr7*4+3,1f
  91 err1;   stb     r0,0(r3)
  92         addi    r3,r3,1
  93
  94 1:      bf      cr7*4+2,2f
  95 err1;   sth     r0,0(r3)
  96         addi    r3,r3,2
  97
  98 2:      bf      cr7*4+1,3f
  99 err1;   stw     r0,0(r3)
 100         addi    r3,r3,4
 101
 102 3:      sub     r4,r4,r6
 103
 104         cmpdi   r4,32
 105         cmpdi   cr1,r4,512
 106         blt     .Lshort_clear
 107         bgt     cr1,.Llong_clear
 108
 109 .Lmedium_clear:
 110         srdi    r6,r4,5
 111         mtctr   r6
 112
 113         /* Do 32 byte chunks */
 114 4:
 115 err2;   std     r0,0(r3)
 116 err2;   std     r0,8(r3)
 117 err2;   std     r0,16(r3)
 118 err2;   std     r0,24(r3)
 119         addi    r3,r3,32
 120         addi    r4,r4,-32
 121         bdnz    4b
 122
 123 .Lshort_clear:
 124         /* up to 31 bytes to go */
 125         cmpdi   r4,16
 126         blt     6f
 127 err2;   std     r0,0(r3)
 128 err2;   std     r0,8(r3)
 129         addi    r3,r3,16
 130         addi    r4,r4,-16
 131
 132         /* Up to 15 bytes to go */
 133 6:      mr      r8,r3
 134         clrldi  r4,r4,(64-4)
 135         mtocrf  0x01,r4
 136         bf      cr7*4+0,7f
 137 err1;   std     r0,0(r3)
 138         addi    r3,r3,8
 139
 140 7:      bf      cr7*4+1,8f
 141 err1;   stw     r0,0(r3)
 142         addi    r3,r3,4
 143
 144 8:      bf      cr7*4+2,9f
 145 err1;   sth     r0,0(r3)
 146         addi    r3,r3,2
 147
 148 9:      bf      cr7*4+3,10f
 149 err1;   stb     r0,0(r3)
 150
 151 10:     li      r3,0
 152         blr
 153
 154 .Llong_clear:
 155         ld      r5,PPC64_CACHES@toc(r2)
 156
 157         bf      cr7*4+0,11f
 158 err2;   std     r0,0(r3)
 159         addi    r3,r3,8
 160         addi    r4,r4,-8
 161
 162         /* Destination is 16 byte aligned, need to get it cacheline aligned */
 163 11:     lwz     r7,DCACHEL1LOGLINESIZE(r5)
 164         lwz     r9,DCACHEL1LINESIZE(r5)
 165
 166         /*
 167          * With worst case alignment the long clear loop takes a minimum
 168          * of 1 byte less than 2 cachelines.
 169          */
 170         sldi    r10,r9,2
 171         cmpd    r4,r10
 172         blt     .Lmedium_clear
 173
 174         neg     r6,r3
 175         addi    r10,r9,-1
 176         and.    r5,r6,r10
 177         beq     13f
 178
 179         srdi    r6,r5,4
 180         mtctr   r6
 181         mr      r8,r3
 182 12:
 183 err1;   std     r0,0(r3)
 184 err1;   std     r0,8(r3)
 185         addi    r3,r3,16
 186         bdnz    12b
 187
 188         sub     r4,r4,r5
 189
 190 13:     srd     r6,r4,r7
 191         mtctr   r6
 192         mr      r8,r3
 193 14:
 194 err1;   dcbz    r0,r3
 195         add     r3,r3,r9
 196         bdnz    14b
 197
 198         and     r4,r4,r10
 199
 200         cmpdi   r4,32
 201         blt     .Lshort_clear
 202         b       .Lmedium_clear