1/* 2 * Copyright (C) 2002 Paul Mackerras, IBM Corp. 3 * 4 * This program is free software; you can redistribute it and/or 5 * modify it under the terms of the GNU General Public License 6 * as published by the Free Software Foundation; either version 7 * 2 of the License, or (at your option) any later version. 8 */ 9#include <asm/processor.h> 10#include <asm/ppc_asm.h> 11 12 .align 7 13_GLOBAL(memcpy) 14BEGIN_FTR_SECTION 15 std r3,48(r1) /* save destination pointer for return value */ 16FTR_SECTION_ELSE 17#ifndef SELFTEST 18 b memcpy_power7 19#endif 20ALT_FTR_SECTION_END_IFCLR(CPU_FTR_VMX_COPY) 21 PPC_MTOCRF(0x01,r5) 22 cmpldi cr1,r5,16 23 neg r6,r3 # LS 3 bits = # bytes to 8-byte dest bdry 24 andi. r6,r6,7 25 dcbt 0,r4 26 blt cr1,.Lshort_copy 27/* Below we want to nop out the bne if we're on a CPU that has the 28 CPU_FTR_UNALIGNED_LD_STD bit set and the CPU_FTR_CP_USE_DCBTZ bit 29 cleared. 30 At the time of writing the only CPU that has this combination of bits 31 set is Power6. */ 32BEGIN_FTR_SECTION 33 nop 34FTR_SECTION_ELSE 35 bne .Ldst_unaligned 36ALT_FTR_SECTION_END(CPU_FTR_UNALIGNED_LD_STD | CPU_FTR_CP_USE_DCBTZ, \ 37 CPU_FTR_UNALIGNED_LD_STD) 38.Ldst_aligned: 39 addi r3,r3,-16 40BEGIN_FTR_SECTION 41 andi. r0,r4,7 42 bne .Lsrc_unaligned 43END_FTR_SECTION_IFCLR(CPU_FTR_UNALIGNED_LD_STD) 44 srdi r7,r5,4 45 ld r9,0(r4) 46 addi r4,r4,-8 47 mtctr r7 48 andi. r5,r5,7 49 bf cr7*4+0,2f 50 addi r3,r3,8 51 addi r4,r4,8 52 mr r8,r9 53 blt cr1,3f 541: ld r9,8(r4) 55 std r8,8(r3) 562: ldu r8,16(r4) 57 stdu r9,16(r3) 58 bdnz 1b 593: std r8,8(r3) 60 beq 3f 61 addi r3,r3,16 62.Ldo_tail: 63 bf cr7*4+1,1f 64 lwz r9,8(r4) 65 addi r4,r4,4 66 stw r9,0(r3) 67 addi r3,r3,4 681: bf cr7*4+2,2f 69 lhz r9,8(r4) 70 addi r4,r4,2 71 sth r9,0(r3) 72 addi r3,r3,2 732: bf cr7*4+3,3f 74 lbz r9,8(r4) 75 stb r9,0(r3) 763: ld r3,48(r1) /* return dest pointer */ 77 blr 78 79.Lsrc_unaligned: 80 srdi r6,r5,3 81 addi r5,r5,-16 82 subf r4,r0,r4 83 srdi r7,r5,4 84 sldi r10,r0,3 85 cmpdi cr6,r6,3 86 andi. r5,r5,7 87 mtctr r7 88 subfic r11,r10,64 89 add r5,r5,r0 90 91 bt cr7*4+0,0f 92 93 ld r9,0(r4) # 3+2n loads, 2+2n stores 94 ld r0,8(r4) 95 sld r6,r9,r10 96 ldu r9,16(r4) 97 srd r7,r0,r11 98 sld r8,r0,r10 99 or r7,r7,r6 100 blt cr6,4f 101 ld r0,8(r4) 102 # s1<< in r8, d0=(s0<<|s1>>) in r7, s3 in r0, s2 in r9, nix in r6 & r12 103 b 2f 104 1050: ld r0,0(r4) # 4+2n loads, 3+2n stores 106 ldu r9,8(r4) 107 sld r8,r0,r10 108 addi r3,r3,-8 109 blt cr6,5f 110 ld r0,8(r4) 111 srd r12,r9,r11 112 sld r6,r9,r10 113 ldu r9,16(r4) 114 or r12,r8,r12 115 srd r7,r0,r11 116 sld r8,r0,r10 117 addi r3,r3,16 118 beq cr6,3f 119 120 # d0=(s0<<|s1>>) in r12, s1<< in r6, s2>> in r7, s2<< in r8, s3 in r9 1211: or r7,r7,r6 122 ld r0,8(r4) 123 std r12,8(r3) 1242: srd r12,r9,r11 125 sld r6,r9,r10 126 ldu r9,16(r4) 127 or r12,r8,r12 128 stdu r7,16(r3) 129 srd r7,r0,r11 130 sld r8,r0,r10 131 bdnz 1b 132 1333: std r12,8(r3) 134 or r7,r7,r6 1354: std r7,16(r3) 1365: srd r12,r9,r11 137 or r12,r8,r12 138 std r12,24(r3) 139 beq 4f 140 cmpwi cr1,r5,8 141 addi r3,r3,32 142 sld r9,r9,r10 143 ble cr1,6f 144 ld r0,8(r4) 145 srd r7,r0,r11 146 or r9,r7,r9 1476: 148 bf cr7*4+1,1f 149 rotldi r9,r9,32 150 stw r9,0(r3) 151 addi r3,r3,4 1521: bf cr7*4+2,2f 153 rotldi r9,r9,16 154 sth r9,0(r3) 155 addi r3,r3,2 1562: bf cr7*4+3,3f 157 rotldi r9,r9,8 158 stb r9,0(r3) 1593: ld r3,48(r1) /* return dest pointer */ 160 blr 161 162.Ldst_unaligned: 163 PPC_MTOCRF(0x01,r6) # put #bytes to 8B bdry into cr7 164 subf r5,r6,r5 165 li r7,0 166 cmpldi cr1,r5,16 167 bf cr7*4+3,1f 168 lbz r0,0(r4) 169 stb r0,0(r3) 170 addi r7,r7,1 1711: bf cr7*4+2,2f 172 lhzx r0,r7,r4 173 sthx r0,r7,r3 174 addi r7,r7,2 1752: bf cr7*4+1,3f 176 lwzx r0,r7,r4 177 stwx r0,r7,r3 1783: PPC_MTOCRF(0x01,r5) 179 add r4,r6,r4 180 add r3,r6,r3 181 b .Ldst_aligned 182 183.Lshort_copy: 184 bf cr7*4+0,1f 185 lwz r0,0(r4) 186 lwz r9,4(r4) 187 addi r4,r4,8 188 stw r0,0(r3) 189 stw r9,4(r3) 190 addi r3,r3,8 1911: bf cr7*4+1,2f 192 lwz r0,0(r4) 193 addi r4,r4,4 194 stw r0,0(r3) 195 addi r3,r3,4 1962: bf cr7*4+2,3f 197 lhz r0,0(r4) 198 addi r4,r4,2 199 sth r0,0(r3) 200 addi r3,r3,2 2013: bf cr7*4+3,4f 202 lbz r0,0(r4) 203 stb r0,0(r3) 2044: ld r3,48(r1) /* return dest pointer */ 205 blr 206