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) 14 BEGIN_FTR_SECTION 15 std r3,48(r1) /* save destination pointer for return value */ 16 FTR_SECTION_ELSE 17 b memcpy_power7 18 ALT_FTR_SECTION_END_IFCLR(CPU_FTR_VMX_COPY) 19 PPC_MTOCRF(0x01,r5) 20 cmpldi cr1,r5,16 21 neg r6,r3 # LS 3 bits = # bytes to 8-byte dest bdry 22 andi. r6,r6,7 23 dcbt 0,r4 24 blt cr1,.Lshort_copy 25 /* Below we want to nop out the bne if we're on a CPU that has the 26 CPU_FTR_UNALIGNED_LD_STD bit set and the CPU_FTR_CP_USE_DCBTZ bit 27 cleared. 28 At the time of writing the only CPU that has this combination of bits 29 set is Power6. */ 30 BEGIN_FTR_SECTION 31 nop 32 FTR_SECTION_ELSE 33 bne .Ldst_unaligned 34 ALT_FTR_SECTION_END(CPU_FTR_UNALIGNED_LD_STD | CPU_FTR_CP_USE_DCBTZ, \ 35 CPU_FTR_UNALIGNED_LD_STD) 36 .Ldst_aligned: 37 addi r3,r3,-16 38 BEGIN_FTR_SECTION 39 andi. r0,r4,7 40 bne .Lsrc_unaligned 41 END_FTR_SECTION_IFCLR(CPU_FTR_UNALIGNED_LD_STD) 42 srdi r7,r5,4 43 ld r9,0(r4) 44 addi r4,r4,-8 45 mtctr r7 46 andi. r5,r5,7 47 bf cr7*4+0,2f 48 addi r3,r3,8 49 addi r4,r4,8 50 mr r8,r9 51 blt cr1,3f 52 1: ld r9,8(r4) 53 std r8,8(r3) 54 2: ldu r8,16(r4) 55 stdu r9,16(r3) 56 bdnz 1b 57 3: std r8,8(r3) 58 beq 3f 59 addi r3,r3,16 60 .Ldo_tail: 61 bf cr7*4+1,1f 62 lwz r9,8(r4) 63 addi r4,r4,4 64 stw r9,0(r3) 65 addi r3,r3,4 66 1: bf cr7*4+2,2f 67 lhz r9,8(r4) 68 addi r4,r4,2 69 sth r9,0(r3) 70 addi r3,r3,2 71 2: bf cr7*4+3,3f 72 lbz r9,8(r4) 73 stb r9,0(r3) 74 3: ld r3,48(r1) /* return dest pointer */ 75 blr 76 77 .Lsrc_unaligned: 78 srdi r6,r5,3 79 addi r5,r5,-16 80 subf r4,r0,r4 81 srdi r7,r5,4 82 sldi r10,r0,3 83 cmpdi cr6,r6,3 84 andi. r5,r5,7 85 mtctr r7 86 subfic r11,r10,64 87 add r5,r5,r0 88 89 bt cr7*4+0,0f 90 91 ld r9,0(r4) # 3+2n loads, 2+2n stores 92 ld r0,8(r4) 93 sld r6,r9,r10 94 ldu r9,16(r4) 95 srd r7,r0,r11 96 sld r8,r0,r10 97 or r7,r7,r6 98 blt cr6,4f 99 ld r0,8(r4) 100 # s1<< in r8, d0=(s0<<|s1>>) in r7, s3 in r0, s2 in r9, nix in r6 & r12 101 b 2f 102 103 0: ld r0,0(r4) # 4+2n loads, 3+2n stores 104 ldu r9,8(r4) 105 sld r8,r0,r10 106 addi r3,r3,-8 107 blt cr6,5f 108 ld r0,8(r4) 109 srd r12,r9,r11 110 sld r6,r9,r10 111 ldu r9,16(r4) 112 or r12,r8,r12 113 srd r7,r0,r11 114 sld r8,r0,r10 115 addi r3,r3,16 116 beq cr6,3f 117 118 # d0=(s0<<|s1>>) in r12, s1<< in r6, s2>> in r7, s2<< in r8, s3 in r9 119 1: or r7,r7,r6 120 ld r0,8(r4) 121 std r12,8(r3) 122 2: srd r12,r9,r11 123 sld r6,r9,r10 124 ldu r9,16(r4) 125 or r12,r8,r12 126 stdu r7,16(r3) 127 srd r7,r0,r11 128 sld r8,r0,r10 129 bdnz 1b 130 131 3: std r12,8(r3) 132 or r7,r7,r6 133 4: std r7,16(r3) 134 5: srd r12,r9,r11 135 or r12,r8,r12 136 std r12,24(r3) 137 beq 4f 138 cmpwi cr1,r5,8 139 addi r3,r3,32 140 sld r9,r9,r10 141 ble cr1,6f 142 ld r0,8(r4) 143 srd r7,r0,r11 144 or r9,r7,r9 145 6: 146 bf cr7*4+1,1f 147 rotldi r9,r9,32 148 stw r9,0(r3) 149 addi r3,r3,4 150 1: bf cr7*4+2,2f 151 rotldi r9,r9,16 152 sth r9,0(r3) 153 addi r3,r3,2 154 2: bf cr7*4+3,3f 155 rotldi r9,r9,8 156 stb r9,0(r3) 157 3: ld r3,48(r1) /* return dest pointer */ 158 blr 159 160 .Ldst_unaligned: 161 PPC_MTOCRF(0x01,r6) # put #bytes to 8B bdry into cr7 162 subf r5,r6,r5 163 li r7,0 164 cmpldi cr1,r5,16 165 bf cr7*4+3,1f 166 lbz r0,0(r4) 167 stb r0,0(r3) 168 addi r7,r7,1 169 1: bf cr7*4+2,2f 170 lhzx r0,r7,r4 171 sthx r0,r7,r3 172 addi r7,r7,2 173 2: bf cr7*4+1,3f 174 lwzx r0,r7,r4 175 stwx r0,r7,r3 176 3: PPC_MTOCRF(0x01,r5) 177 add r4,r6,r4 178 add r3,r6,r3 179 b .Ldst_aligned 180 181 .Lshort_copy: 182 bf cr7*4+0,1f 183 lwz r0,0(r4) 184 lwz r9,4(r4) 185 addi r4,r4,8 186 stw r0,0(r3) 187 stw r9,4(r3) 188 addi r3,r3,8 189 1: bf cr7*4+1,2f 190 lwz r0,0(r4) 191 addi r4,r4,4 192 stw r0,0(r3) 193 addi r3,r3,4 194 2: bf cr7*4+2,3f 195 lhz r0,0(r4) 196 addi r4,r4,2 197 sth r0,0(r3) 198 addi r3,r3,2 199 3: bf cr7*4+3,4f 200 lbz r0,0(r4) 201 stb r0,0(r3) 202 4: ld r3,48(r1) /* return dest pointer */ 203 blr 204
