arch/csky/abiv2/memcmp.S

   1 /* SPDX-License-Identifier: GPL-2.0 */
   2 // Copyright (C) 2018 Hangzhou C-SKY Microsystems co.,ltd.
   3
   4 #include <linux/linkage.h>
   5 #include "sysdep.h"
   6
   7 ENTRY(memcmp)
   8         /* Test if len less than 4 bytes.  */
   9         mov     r3, r0
  10         movi    r0, 0
  11         mov     r12, r4
  12         cmplti  r2, 4
  13         bt      .L_compare_by_byte
  14
  15         andi    r13, r0, 3
  16         movi    r19, 4
  17
  18         /* Test if s1 is not 4 bytes aligned.  */
  19         bnez    r13, .L_s1_not_aligned
  20
  21         LABLE_ALIGN
  22 .L_s1_aligned:
  23         /* If dest is aligned, then copy.  */
  24         zext    r18, r2, 31, 4
  25         /* Test if len less than 16 bytes.  */
  26         bez     r18, .L_compare_by_word
  27
  28 .L_compare_by_4word:
  29         /* If aligned, load word each time.  */
  30         ldw     r20, (r3, 0)
  31         ldw     r21, (r1, 0)
  32         /* If s1[i] != s2[i], goto .L_byte_check.  */
  33         cmpne   r20, r21
  34         bt      .L_byte_check
  35
  36         ldw     r20, (r3, 4)
  37         ldw     r21, (r1, 4)
  38         cmpne   r20, r21
  39         bt      .L_byte_check
  40
  41         ldw     r20, (r3, 8)
  42         ldw     r21, (r1, 8)
  43         cmpne   r20, r21
  44         bt      .L_byte_check
  45
  46         ldw     r20, (r3, 12)
  47         ldw     r21, (r1, 12)
  48         cmpne   r20, r21
  49         bt      .L_byte_check
  50
  51         PRE_BNEZAD (r18)
  52         addi    a3, 16
  53         addi    a1, 16
  54
  55         BNEZAD (r18, .L_compare_by_4word)
  56
  57 .L_compare_by_word:
  58         zext    r18, r2, 3, 2
  59         bez     r18, .L_compare_by_byte
  60 .L_compare_by_word_loop:
  61         ldw     r20, (r3, 0)
  62         ldw     r21, (r1, 0)
  63         addi    r3, 4
  64         PRE_BNEZAD (r18)
  65         cmpne   r20, r21
  66         addi    r1, 4
  67         bt      .L_byte_check
  68         BNEZAD (r18, .L_compare_by_word_loop)
  69
  70 .L_compare_by_byte:
  71         zext    r18, r2, 1, 0
  72         bez     r18, .L_return
  73 .L_compare_by_byte_loop:
  74         ldb     r0, (r3, 0)
  75         ldb     r4, (r1, 0)
  76         addi    r3, 1
  77         subu    r0, r4
  78         PRE_BNEZAD (r18)
  79         addi    r1, 1
  80         bnez    r0, .L_return
  81         BNEZAD (r18, .L_compare_by_byte_loop)
  82
  83 .L_return:
  84         mov     r4, r12
  85         rts
  86
  87 # ifdef __CSKYBE__
  88 /* d[i] != s[i] in word, so we check byte 0.  */
  89 .L_byte_check:
  90         xtrb0   r0, r20
  91         xtrb0   r2, r21
  92         subu    r0, r2
  93         bnez    r0, .L_return
  94
  95         /* check byte 1 */
  96         xtrb1   r0, r20
  97         xtrb1   r2, r21
  98         subu    r0, r2
  99         bnez    r0, .L_return
 100
 101         /* check byte 2 */
 102         xtrb2   r0, r20
 103         xtrb2   r2, r21
 104         subu    r0, r2
 105         bnez    r0, .L_return
 106
 107         /* check byte 3 */
 108         xtrb3   r0, r20
 109         xtrb3   r2, r21
 110         subu    r0, r2
 111 # else
 112 /* s1[i] != s2[i] in word, so we check byte 3.  */
 113 .L_byte_check:
 114         xtrb3   r0, r20
 115         xtrb3   r2, r21
 116         subu    r0, r2
 117         bnez    r0, .L_return
 118
 119         /* check byte 2 */
 120         xtrb2   r0, r20
 121         xtrb2   r2, r21
 122         subu    r0, r2
 123         bnez    r0, .L_return
 124
 125         /* check byte 1 */
 126         xtrb1   r0, r20
 127         xtrb1   r2, r21
 128         subu    r0, r2
 129         bnez    r0, .L_return
 130
 131         /* check byte 0 */
 132         xtrb0   r0, r20
 133         xtrb0   r2, r21
 134         subu    r0, r2
 135         br      .L_return
 136 # endif /* !__CSKYBE__ */
 137
 138 /* Compare when s1 is not aligned.  */
 139 .L_s1_not_aligned:
 140         sub     r13, r19, r13
 141         sub     r2, r13
 142 .L_s1_not_aligned_loop:
 143         ldb     r0, (r3, 0)
 144         ldb     r4, (r1, 0)
 145         addi    r3, 1
 146         subu    r0, r4
 147         PRE_BNEZAD (r13)
 148         addi    r1, 1
 149         bnez    r0, .L_return
 150         BNEZAD (r13, .L_s1_not_aligned_loop)
 151         br      .L_s1_aligned
 152 ENDPROC(memcmp)