arch/arm/lib/sha1.S

   1 /*
   2  *  linux/arch/arm/lib/sha1.S
   3  *
   4  *  SHA transform optimized for ARM
   5  *
   6  *  Copyright:  (C) 2005 by Nicolas Pitre <nico@fluxnic.net>
   7  *  Created:    September 17, 2005
   8  *
   9  *  This program is free software; you can redistribute it and/or modify
  10  *  it under the terms of the GNU General Public License version 2 as
  11  *  published by the Free Software Foundation.
  12  *
  13  *  The reference implementation for this code is linux/lib/sha1.c
  14  */
  15
  16 #include <linux/linkage.h>
  17
  18         .text
  19
  20
  21 /*
  22  * void sha_transform(__u32 *digest, const char *in, __u32 *W)
  23  *
  24  * Note: the "in" ptr may be unaligned.
  25  */
  26
  27 ENTRY(sha_transform)
  28
  29         stmfd   sp!, {r4 - r8, lr}
  30
  31         @ for (i = 0; i < 16; i++)
  32         @         W[i] = be32_to_cpu(in[i]);
  33
  34 #ifdef __ARMEB__
  35         mov     r4, r0
  36         mov     r0, r2
  37         mov     r2, #64
  38         bl      memcpy
  39         mov     r2, r0
  40         mov     r0, r4
  41 #else
  42         mov     r3, r2
  43         mov     lr, #16
  44 1:      ldrb    r4, [r1], #1
  45         ldrb    r5, [r1], #1
  46         ldrb    r6, [r1], #1
  47         ldrb    r7, [r1], #1
  48         subs    lr, lr, #1
  49         orr     r5, r5, r4, lsl #8
  50         orr     r6, r6, r5, lsl #8
  51         orr     r7, r7, r6, lsl #8
  52         str     r7, [r3], #4
  53         bne     1b
  54 #endif
  55
  56         @ for (i = 0; i < 64; i++)
  57         @         W[i+16] = ror(W[i+13] ^ W[i+8] ^ W[i+2] ^ W[i], 31);
  58
  59         sub     r3, r2, #4
  60         mov     lr, #64
  61 2:      ldr     r4, [r3, #4]!
  62         subs    lr, lr, #1
  63         ldr     r5, [r3, #8]
  64         ldr     r6, [r3, #32]
  65         ldr     r7, [r3, #52]
  66         eor     r4, r4, r5
  67         eor     r4, r4, r6
  68         eor     r4, r4, r7
  69         mov     r4, r4, ror #31
  70         str     r4, [r3, #64]
  71         bne     2b
  72
  73         /*
  74          * The SHA functions are:
  75          *
  76          * f1(B,C,D) = (D ^ (B & (C ^ D)))
  77          * f2(B,C,D) = (B ^ C ^ D)
  78          * f3(B,C,D) = ((B & C) | (D & (B | C)))
  79          *
  80          * Then the sub-blocks are processed as follows:
  81          *
  82          * A' = ror(A, 27) + f(B,C,D) + E + K + *W++
  83          * B' = A
  84          * C' = ror(B, 2)
  85          * D' = C
  86          * E' = D
  87          *
  88          * We therefore unroll each loop 5 times to avoid register shuffling.
  89          * Also the ror for C (and also D and E which are successivelyderived
  90          * from it) is applied in place to cut on an additional mov insn for
  91          * each round.
  92          */
  93
  94         .macro  sha_f1, A, B, C, D, E
  95         ldr     r3, [r2], #4
  96         eor     ip, \C, \D
  97         add     \E, r1, \E, ror #2
  98         and     ip, \B, ip, ror #2
  99         add     \E, \E, \A, ror #27
 100         eor     ip, ip, \D, ror #2
 101         add     \E, \E, r3
 102         add     \E, \E, ip
 103         .endm
 104
 105         .macro  sha_f2, A, B, C, D, E
 106         ldr     r3, [r2], #4
 107         add     \E, r1, \E, ror #2
 108         eor     ip, \B, \C, ror #2
 109         add     \E, \E, \A, ror #27
 110         eor     ip, ip, \D, ror #2
 111         add     \E, \E, r3
 112         add     \E, \E, ip
 113         .endm
 114
 115         .macro  sha_f3, A, B, C, D, E
 116         ldr     r3, [r2], #4
 117         add     \E, r1, \E, ror #2
 118         orr     ip, \B, \C, ror #2
 119         add     \E, \E, \A, ror #27
 120         and     ip, ip, \D, ror #2
 121         add     \E, \E, r3
 122         and     r3, \B, \C, ror #2
 123         orr     ip, ip, r3
 124         add     \E, \E, ip
 125         .endm
 126
 127         ldmia   r0, {r4 - r8}
 128
 129         mov     lr, #4
 130         ldr     r1, .L_sha_K + 0
 131
 132         /* adjust initial values */
 133         mov     r6, r6, ror #30
 134         mov     r7, r7, ror #30
 135         mov     r8, r8, ror #30
 136
 137 3:      subs    lr, lr, #1
 138         sha_f1  r4, r5, r6, r7, r8
 139         sha_f1  r8, r4, r5, r6, r7
 140         sha_f1  r7, r8, r4, r5, r6
 141         sha_f1  r6, r7, r8, r4, r5
 142         sha_f1  r5, r6, r7, r8, r4
 143         bne     3b
 144
 145         ldr     r1, .L_sha_K + 4
 146         mov     lr, #4
 147
 148 4:      subs    lr, lr, #1
 149         sha_f2  r4, r5, r6, r7, r8
 150         sha_f2  r8, r4, r5, r6, r7
 151         sha_f2  r7, r8, r4, r5, r6
 152         sha_f2  r6, r7, r8, r4, r5
 153         sha_f2  r5, r6, r7, r8, r4
 154         bne     4b
 155
 156         ldr     r1, .L_sha_K + 8
 157         mov     lr, #4
 158
 159 5:      subs    lr, lr, #1
 160         sha_f3  r4, r5, r6, r7, r8
 161         sha_f3  r8, r4, r5, r6, r7
 162         sha_f3  r7, r8, r4, r5, r6
 163         sha_f3  r6, r7, r8, r4, r5
 164         sha_f3  r5, r6, r7, r8, r4
 165         bne     5b
 166
 167         ldr     r1, .L_sha_K + 12
 168         mov     lr, #4
 169
 170 6:      subs    lr, lr, #1
 171         sha_f2  r4, r5, r6, r7, r8
 172         sha_f2  r8, r4, r5, r6, r7
 173         sha_f2  r7, r8, r4, r5, r6
 174         sha_f2  r6, r7, r8, r4, r5
 175         sha_f2  r5, r6, r7, r8, r4
 176         bne     6b
 177
 178         ldmia   r0, {r1, r2, r3, ip, lr}
 179         add     r4, r1, r4
 180         add     r5, r2, r5
 181         add     r6, r3, r6, ror #2
 182         add     r7, ip, r7, ror #2
 183         add     r8, lr, r8, ror #2
 184         stmia   r0, {r4 - r8}
 185
 186         ldmfd   sp!, {r4 - r8, pc}
 187
 188 ENDPROC(sha_transform)
 189
 190         .align  2
 191 .L_sha_K:
 192         .word   0x5a827999, 0x6ed9eba1, 0x8f1bbcdc, 0xca62c1d6
 193
 194
 195 /*
 196  * void sha_init(__u32 *buf)
 197  */
 198
 199         .align  2
 200 .L_sha_initial_digest:
 201         .word   0x67452301, 0xefcdab89, 0x98badcfe, 0x10325476, 0xc3d2e1f0
 202
 203 ENTRY(sha_init)
 204
 205         str     lr, [sp, #-4]!
 206         adr     r1, .L_sha_initial_digest
 207         ldmia   r1, {r1, r2, r3, ip, lr}
 208         stmia   r0, {r1, r2, r3, ip, lr}
 209         ldr     pc, [sp], #4
 210
 211 ENDPROC(sha_init)