Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame^] | 1 | /* |
| 2 | * This file is subject to the terms and conditions of the GNU General Public |
| 3 | * License. See the file "COPYING" in the main directory of this archive |
| 4 | * for more details. |
| 5 | * |
| 6 | * Quick'n'dirty IP checksum ... |
| 7 | * |
| 8 | * Copyright (C) 1998, 1999 Ralf Baechle |
| 9 | * Copyright (C) 1999 Silicon Graphics, Inc. |
| 10 | */ |
| 11 | #include <asm/asm.h> |
| 12 | #include <asm/regdef.h> |
| 13 | |
| 14 | #define ADDC(sum,reg) \ |
| 15 | addu sum, reg; \ |
| 16 | sltu v1, sum, reg; \ |
| 17 | addu sum, v1 |
| 18 | |
| 19 | #define CSUM_BIGCHUNK(src, offset, sum, t0, t1, t2, t3) \ |
| 20 | lw t0, (offset + 0x00)(src); \ |
| 21 | lw t1, (offset + 0x04)(src); \ |
| 22 | lw t2, (offset + 0x08)(src); \ |
| 23 | lw t3, (offset + 0x0c)(src); \ |
| 24 | ADDC(sum, t0); \ |
| 25 | ADDC(sum, t1); \ |
| 26 | ADDC(sum, t2); \ |
| 27 | ADDC(sum, t3); \ |
| 28 | lw t0, (offset + 0x10)(src); \ |
| 29 | lw t1, (offset + 0x14)(src); \ |
| 30 | lw t2, (offset + 0x18)(src); \ |
| 31 | lw t3, (offset + 0x1c)(src); \ |
| 32 | ADDC(sum, t0); \ |
| 33 | ADDC(sum, t1); \ |
| 34 | ADDC(sum, t2); \ |
| 35 | ADDC(sum, t3); \ |
| 36 | |
| 37 | /* |
| 38 | * a0: source address |
| 39 | * a1: length of the area to checksum |
| 40 | * a2: partial checksum |
| 41 | */ |
| 42 | |
| 43 | #define src a0 |
| 44 | #define sum v0 |
| 45 | |
| 46 | .text |
| 47 | .set noreorder |
| 48 | |
| 49 | /* unknown src alignment and < 8 bytes to go */ |
| 50 | small_csumcpy: |
| 51 | move a1, ta2 |
| 52 | |
| 53 | andi ta0, a1, 4 |
| 54 | beqz ta0, 1f |
| 55 | andi ta0, a1, 2 |
| 56 | |
| 57 | /* Still a full word to go */ |
| 58 | ulw ta1, (src) |
| 59 | daddiu src, 4 |
| 60 | ADDC(sum, ta1) |
| 61 | |
| 62 | 1: move ta1, zero |
| 63 | beqz ta0, 1f |
| 64 | andi ta0, a1, 1 |
| 65 | |
| 66 | /* Still a halfword to go */ |
| 67 | ulhu ta1, (src) |
| 68 | daddiu src, 2 |
| 69 | |
| 70 | 1: beqz ta0, 1f |
| 71 | sll ta1, ta1, 16 |
| 72 | |
| 73 | lbu ta2, (src) |
| 74 | nop |
| 75 | |
| 76 | #ifdef __MIPSEB__ |
| 77 | sll ta2, ta2, 8 |
| 78 | #endif |
| 79 | or ta1, ta2 |
| 80 | |
| 81 | 1: ADDC(sum, ta1) |
| 82 | |
| 83 | /* fold checksum */ |
| 84 | sll v1, sum, 16 |
| 85 | addu sum, v1 |
| 86 | sltu v1, sum, v1 |
| 87 | srl sum, sum, 16 |
| 88 | addu sum, v1 |
| 89 | |
| 90 | /* odd buffer alignment? */ |
| 91 | beqz t3, 1f |
| 92 | nop |
| 93 | sll v1, sum, 8 |
| 94 | srl sum, sum, 8 |
| 95 | or sum, v1 |
| 96 | andi sum, 0xffff |
| 97 | 1: |
| 98 | .set reorder |
| 99 | /* Add the passed partial csum. */ |
| 100 | ADDC(sum, a2) |
| 101 | jr ra |
| 102 | .set noreorder |
| 103 | |
| 104 | /* ------------------------------------------------------------------------- */ |
| 105 | |
| 106 | .align 5 |
| 107 | LEAF(csum_partial) |
| 108 | move sum, zero |
| 109 | move t3, zero |
| 110 | |
| 111 | sltiu t8, a1, 0x8 |
| 112 | bnez t8, small_csumcpy /* < 8 bytes to copy */ |
| 113 | move ta2, a1 |
| 114 | |
| 115 | beqz a1, out |
| 116 | andi t3, src, 0x1 /* odd buffer? */ |
| 117 | |
| 118 | hword_align: |
| 119 | beqz t3, word_align |
| 120 | andi t8, src, 0x2 |
| 121 | |
| 122 | lbu ta0, (src) |
| 123 | dsubu a1, a1, 0x1 |
| 124 | #ifdef __MIPSEL__ |
| 125 | sll ta0, ta0, 8 |
| 126 | #endif |
| 127 | ADDC(sum, ta0) |
| 128 | daddu src, src, 0x1 |
| 129 | andi t8, src, 0x2 |
| 130 | |
| 131 | word_align: |
| 132 | beqz t8, dword_align |
| 133 | sltiu t8, a1, 56 |
| 134 | |
| 135 | lhu ta0, (src) |
| 136 | dsubu a1, a1, 0x2 |
| 137 | ADDC(sum, ta0) |
| 138 | sltiu t8, a1, 56 |
| 139 | daddu src, src, 0x2 |
| 140 | |
| 141 | dword_align: |
| 142 | bnez t8, do_end_words |
| 143 | move t8, a1 |
| 144 | |
| 145 | andi t8, src, 0x4 |
| 146 | beqz t8, qword_align |
| 147 | andi t8, src, 0x8 |
| 148 | |
| 149 | lw ta0, 0x00(src) |
| 150 | dsubu a1, a1, 0x4 |
| 151 | ADDC(sum, ta0) |
| 152 | daddu src, src, 0x4 |
| 153 | andi t8, src, 0x8 |
| 154 | |
| 155 | qword_align: |
| 156 | beqz t8, oword_align |
| 157 | andi t8, src, 0x10 |
| 158 | |
| 159 | lw ta0, 0x00(src) |
| 160 | lw ta1, 0x04(src) |
| 161 | dsubu a1, a1, 0x8 |
| 162 | ADDC(sum, ta0) |
| 163 | ADDC(sum, ta1) |
| 164 | daddu src, src, 0x8 |
| 165 | andi t8, src, 0x10 |
| 166 | |
| 167 | oword_align: |
| 168 | beqz t8, begin_movement |
| 169 | dsrl t8, a1, 0x7 |
| 170 | |
| 171 | lw ta3, 0x08(src) |
| 172 | lw t0, 0x0c(src) |
| 173 | lw ta0, 0x00(src) |
| 174 | lw ta1, 0x04(src) |
| 175 | ADDC(sum, ta3) |
| 176 | ADDC(sum, t0) |
| 177 | ADDC(sum, ta0) |
| 178 | ADDC(sum, ta1) |
| 179 | dsubu a1, a1, 0x10 |
| 180 | daddu src, src, 0x10 |
| 181 | dsrl t8, a1, 0x7 |
| 182 | |
| 183 | begin_movement: |
| 184 | beqz t8, 1f |
| 185 | andi ta2, a1, 0x40 |
| 186 | |
| 187 | move_128bytes: |
| 188 | CSUM_BIGCHUNK(src, 0x00, sum, ta0, ta1, ta3, t0) |
| 189 | CSUM_BIGCHUNK(src, 0x20, sum, ta0, ta1, ta3, t0) |
| 190 | CSUM_BIGCHUNK(src, 0x40, sum, ta0, ta1, ta3, t0) |
| 191 | CSUM_BIGCHUNK(src, 0x60, sum, ta0, ta1, ta3, t0) |
| 192 | dsubu t8, t8, 0x01 |
| 193 | bnez t8, move_128bytes |
| 194 | daddu src, src, 0x80 |
| 195 | |
| 196 | 1: |
| 197 | beqz ta2, 1f |
| 198 | andi ta2, a1, 0x20 |
| 199 | |
| 200 | move_64bytes: |
| 201 | CSUM_BIGCHUNK(src, 0x00, sum, ta0, ta1, ta3, t0) |
| 202 | CSUM_BIGCHUNK(src, 0x20, sum, ta0, ta1, ta3, t0) |
| 203 | daddu src, src, 0x40 |
| 204 | |
| 205 | 1: |
| 206 | beqz ta2, do_end_words |
| 207 | andi t8, a1, 0x1c |
| 208 | |
| 209 | move_32bytes: |
| 210 | CSUM_BIGCHUNK(src, 0x00, sum, ta0, ta1, ta3, t0) |
| 211 | andi t8, a1, 0x1c |
| 212 | daddu src, src, 0x20 |
| 213 | |
| 214 | do_end_words: |
| 215 | beqz t8, maybe_end_cruft |
| 216 | dsrl t8, t8, 0x2 |
| 217 | |
| 218 | end_words: |
| 219 | lw ta0, (src) |
| 220 | dsubu t8, t8, 0x1 |
| 221 | ADDC(sum, ta0) |
| 222 | bnez t8, end_words |
| 223 | daddu src, src, 0x4 |
| 224 | |
| 225 | maybe_end_cruft: |
| 226 | andi ta2, a1, 0x3 |
| 227 | |
| 228 | small_memcpy: |
| 229 | j small_csumcpy; move a1, ta2 /* XXX ??? */ |
| 230 | beqz t2, out |
| 231 | move a1, ta2 |
| 232 | |
| 233 | end_bytes: |
| 234 | lb ta0, (src) |
| 235 | dsubu a1, a1, 0x1 |
| 236 | bnez a2, end_bytes |
| 237 | daddu src, src, 0x1 |
| 238 | |
| 239 | out: |
| 240 | jr ra |
| 241 | move v0, sum |
| 242 | END(csum_partial) |