Linux/arch/hexagon/lib/memset.S

Version: ~ [ linux-6.12-rc7 ] ~ [ linux-6.11.7 ] ~ [ linux-6.10.14 ] ~ [ linux-6.9.12 ] ~ [ linux-6.8.12 ] ~ [ linux-6.7.12 ] ~ [ linux-6.6.60 ] ~ [ linux-6.5.13 ] ~ [ linux-6.4.16 ] ~ [ linux-6.3.13 ] ~ [ linux-6.2.16 ] ~ [ linux-6.1.116 ] ~ [ linux-6.0.19 ] ~ [ linux-5.19.17 ] ~ [ linux-5.18.19 ] ~ [ linux-5.17.15 ] ~ [ linux-5.16.20 ] ~ [ linux-5.15.171 ] ~ [ linux-5.14.21 ] ~ [ linux-5.13.19 ] ~ [ linux-5.12.19 ] ~ [ linux-5.11.22 ] ~ [ linux-5.10.229 ] ~ [ linux-5.9.16 ] ~ [ linux-5.8.18 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.285 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.323 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.336 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.337 ] ~ [ linux-4.4.302 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.12 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

Diff markup

1 /* SPDX-License-Identifier: GPL-2.0-only */ !! 1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 2 /* 3 * Copyright (c) 2011, The Linux Foundation. A !! 3 * linux/arch/alpha/lib/memset.S >> 4 * >> 5 * This is an efficient (and small) implementation of the C library "memset()" >> 6 * function for the alpha. >> 7 * >> 8 * (C) Copyright 1996 Linus Torvalds >> 9 * >> 10 * This routine is "moral-ware": you are free to use it any way you wish, and >> 11 * the only obligation I put on you is a moral one: if you make any improvements >> 12 * to the routine, please send me your improvements for me to use similarly. >> 13 * >> 14 * The scheduling comments are according to the EV5 documentation (and done by >> 15 * hand, so they might well be incorrect, please do tell me about it..) 4 */ 16 */ 5 !! 17 #include <linux/export.h> 6 !! 18 .set noat 7 /* HEXAGON assembly optimized memset */ !! 19 .set noreorder 8 /* Replaces the standard library function mems !! 20 .text 9 !! 21 .globl memset 10 !! 22 .globl __memset 11 .macro HEXAGON_OPT_FUNC_BEGIN name !! 23 .globl ___memset 12 .text !! 24 .globl __memset16 13 .p2align 4 !! 25 .globl __constant_c_memset 14 .globl \name !! 26 15 .type \name, @function !! 27 .ent ___memset 16 \name: !! 28 .align 5 17 .endm !! 29 ___memset: 18 !! 30 .frame $30,0,$26,0 19 .macro HEXAGON_OPT_FUNC_FINISH name !! 31 .prologue 0 20 .size \name, . - \name !! 32 21 .endm !! 33 and $17,255,$1 /* E1 */ 22 !! 34 insbl $17,1,$17 /* .. E0 */ 23 /* FUNCTION: memset (v2 version) */ !! 35 bis $17,$1,$17 /* E0 (p-c latency, next cycle) */ 24 #if __HEXAGON_ARCH__ < 3 !! 36 sll $17,16,$1 /* E1 (p-c latency, next cycle) */ 25 HEXAGON_OPT_FUNC_BEGIN memset !! 37 26 { !! 38 bis $17,$1,$17 /* E0 (p-c latency, next cycle) */ 27 r6 = #8 !! 39 sll $17,32,$1 /* E1 (p-c latency, next cycle) */ 28 r7 = extractu(r0, #3 , #0) !! 40 bis $17,$1,$17 /* E0 (p-c latency, next cycle) */ 29 p0 = cmp.eq(r2, #0) !! 41 ldq_u $31,0($30) /* .. E1 */ 30 p1 = cmp.gtu(r2, #7) !! 42 31 } !! 43 .align 5 32 { !! 44 __constant_c_memset: 33 r4 = vsplatb(r1) !! 45 addq $18,$16,$6 /* E0 */ 34 r8 = r0 /* leave r0 !! 46 bis $16,$16,$0 /* .. E1 */ 35 r9 = sub(r6, r7) /* bytes unt !! 47 xor $16,$6,$1 /* E0 */ 36 if p0 jumpr r31 /* count == !! 48 ble $18,end /* .. E1 */ 37 } !! 49 38 { !! 50 bic $1,7,$1 /* E0 */ 39 r3 = #0 !! 51 beq $1,within_one_quad /* .. E1 (note EV5 zero-latency forwarding) */ 40 r7 = #0 !! 52 and $16,7,$3 /* E0 */ 41 p0 = tstbit(r9, #0) !! 53 beq $3,aligned /* .. E1 (note EV5 zero-latency forwarding) */ 42 if p1 jump 2f /* skip byte loo !! 54 43 } !! 55 ldq_u $4,0($16) /* E0 */ 44 !! 56 bis $16,$16,$5 /* .. E1 */ 45 /* less than 8 bytes to set, so just set a byt !! 57 insql $17,$16,$2 /* E0 */ 46 !! 58 subq $3,8,$3 /* .. E1 */ 47 loop0(1f, r2) /* byte loop */ !! 59 48 .falign !! 60 addq $18,$3,$18 /* E0 $18 is new count ($3 is negative) */ 49 1: /* byte loop */ !! 61 mskql $4,$16,$4 /* .. E1 (and possible load stall) */ 50 { !! 62 subq $16,$3,$16 /* E0 $16 is new aligned destination */ 51 memb(r8++#1) = r4 !! 63 bis $2,$4,$1 /* .. E1 */ 52 }:endloop0 !! 64 53 jumpr r31 !! 65 bis $31,$31,$31 /* E0 */ 54 .falign !! 66 ldq_u $31,0($30) /* .. E1 */ 55 2: /* skip byte loop */ !! 67 stq_u $1,0($5) /* E0 */ 56 { !! 68 bis $31,$31,$31 /* .. E1 */ 57 r6 = #1 !! 69 58 p0 = tstbit(r9, #1) !! 70 .align 4 59 p1 = cmp.eq(r2, #1) !! 71 aligned: 60 if !p0 jump 3f /* skip initial !! 72 sra $18,3,$3 /* E0 */ 61 } !! 73 and $18,7,$18 /* .. E1 */ 62 { !! 74 bis $16,$16,$5 /* E0 */ 63 memb(r8++#1) = r4 !! 75 beq $3,no_quad /* .. E1 */ 64 r3:2 = sub(r3:2, r7:6) !! 76 65 if p1 jumpr r31 !! 77 .align 3 66 } !! 78 loop: 67 .falign !! 79 stq $17,0($5) /* E0 */ 68 3: /* skip initial byte store */ !! 80 subq $3,1,$3 /* .. E1 */ 69 { !! 81 addq $5,8,$5 /* E0 */ 70 r6 = #2 !! 82 bne $3,loop /* .. E1 */ 71 p0 = tstbit(r9, #2) !! 83 72 p1 = cmp.eq(r2, #2) !! 84 no_quad: 73 if !p0 jump 4f /* skip initial !! 85 bis $31,$31,$31 /* E0 */ 74 } !! 86 beq $18,end /* .. E1 */ 75 { !! 87 ldq $7,0($5) /* E0 */ 76 memh(r8++#2) = r4 !! 88 mskqh $7,$6,$2 /* .. E1 (and load stall) */ 77 r3:2 = sub(r3:2, r7:6) !! 89 78 if p1 jumpr r31 !! 90 insqh $17,$6,$4 /* E0 */ 79 } !! 91 bis $2,$4,$1 /* .. E1 */ 80 .falign !! 92 stq $1,0($5) /* E0 */ 81 4: /* skip initial half store */ !! 93 ret $31,($26),1 /* .. E1 */ 82 { !! 94 83 r6 = #4 !! 95 .align 3 84 p0 = cmp.gtu(r2, #7) !! 96 within_one_quad: 85 p1 = cmp.eq(r2, #4) !! 97 ldq_u $1,0($16) /* E0 */ 86 if !p0 jump 5f /* skip initial !! 98 insql $17,$16,$2 /* E1 */ 87 } !! 99 mskql $1,$16,$4 /* E0 (after load stall) */ 88 { !! 100 bis $2,$4,$2 /* E0 */ 89 memw(r8++#4) = r4 !! 101 90 r3:2 = sub(r3:2, r7:6) !! 102 mskql $2,$6,$4 /* E0 */ 91 p0 = cmp.gtu(r2, #11) !! 103 mskqh $1,$6,$2 /* .. E1 */ 92 if p1 jumpr r31 !! 104 bis $2,$4,$1 /* E0 */ 93 } !! 105 stq_u $1,0($16) /* E0 */ 94 .falign !! 106 95 5: /* skip initial word store */ !! 107 end: 96 { !! 108 ret $31,($26),1 /* E1 */ 97 r10 = lsr(r2, #3) !! 109 .end ___memset 98 p1 = cmp.eq(r3, #1) !! 110 EXPORT_SYMBOL(___memset) 99 if !p0 jump 7f /* skip double !! 111 EXPORT_SYMBOL(__constant_c_memset) 100 } !! 112 101 { !! 113 .align 5 102 r5 = r4 !! 114 .ent __memset16 103 r6 = #8 !! 115 __memset16: 104 loop0(6f, r10) /* double loop !! 116 .prologue 0 105 } !! 117 106 !! 118 inswl $17,0,$1 /* E0 */ 107 /* set bytes a double word at a time */ !! 119 inswl $17,2,$2 /* E0 */ 108 !! 120 inswl $17,4,$3 /* E0 */ 109 .falign !! 121 or $1,$2,$1 /* .. E1 */ 110 6: /* double loop */ !! 122 inswl $17,6,$4 /* E0 */ 111 { !! 123 or $1,$3,$1 /* .. E1 */ 112 memd(r8++#8) = r5:4 !! 124 or $1,$4,$17 /* E0 */ 113 r3:2 = sub(r3:2, r7:6) !! 125 br __constant_c_memset /* .. E1 */ 114 p1 = cmp.eq(r2, #8) !! 126 115 }:endloop0 !! 127 .end __memset16 116 .falign !! 128 EXPORT_SYMBOL(__memset16) 117 7: /* skip double loop */ !! 129 118 { !! 130 memset = ___memset 119 p0 = tstbit(r2, #2) !! 131 __memset = ___memset 120 if p1 jumpr r31 !! 132 EXPORT_SYMBOL(memset) 121 } !! 133 EXPORT_SYMBOL(__memset) 122 { << 123 r6 = #4 << 124 p0 = tstbit(r2, #1) << 125 p1 = cmp.eq(r2, #4) << 126 if !p0 jump 8f /* skip final w << 127 } << 128 { << 129 memw(r8++#4) = r4 << 130 r3:2 = sub(r3:2, r7:6) << 131 if p1 jumpr r31 << 132 } << 133 .falign << 134 8: /* skip final word store */ << 135 { << 136 p1 = cmp.eq(r2, #2) << 137 if !p0 jump 9f /* skip final h << 138 } << 139 { << 140 memh(r8++#2) = r4 << 141 if p1 jumpr r31 << 142 } << 143 .falign << 144 9: /* skip final half store */ << 145 { << 146 memb(r8++#1) = r4 << 147 jumpr r31 << 148 } << 149 HEXAGON_OPT_FUNC_FINISH memset << 150 #endif << 151 << 152 << 153 /* FUNCTION: memset (v3 and higher version) << 154 #if __HEXAGON_ARCH__ >= 3 << 155 HEXAGON_OPT_FUNC_BEGIN memset << 156 { << 157 r7=vsplatb(r1) << 158 r6 = r0 << 159 if (r2==#0) jump:nt .L1 << 160 } << 161 { << 162 r5:4=combine(r7,r7) << 163 p0 = cmp.gtu(r2,#8) << 164 if (p0.new) jump:nt .L3 << 165 } << 166 { << 167 r3 = r0 << 168 loop0(.L47,r2) << 169 } << 170 .falign << 171 .L47: << 172 { << 173 memb(r3++#1) = r1 << 174 }:endloop0 /* start=.L47 */ << 175 jumpr r31 << 176 .L3: << 177 { << 178 p0 = tstbit(r0,#0) << 179 if (!p0.new) jump:nt .L8 << 180 p1 = cmp.eq(r2, #1) << 181 } << 182 { << 183 r6 = add(r0, #1) << 184 r2 = add(r2,#-1) << 185 memb(r0) = r1 << 186 if (p1) jump .L1 << 187 } << 188 .L8: << 189 { << 190 p0 = tstbit(r6,#1) << 191 if (!p0.new) jump:nt .L10 << 192 } << 193 { << 194 r2 = add(r2,#-2) << 195 memh(r6++#2) = r7 << 196 p0 = cmp.eq(r2, #2) << 197 if (p0.new) jump:nt .L1 << 198 } << 199 .L10: << 200 { << 201 p0 = tstbit(r6,#2) << 202 if (!p0.new) jump:nt .L12 << 203 } << 204 { << 205 r2 = add(r2,#-4) << 206 memw(r6++#4) = r7 << 207 p0 = cmp.eq(r2, #4) << 208 if (p0.new) jump:nt .L1 << 209 } << 210 .L12: << 211 { << 212 p0 = cmp.gtu(r2,#127) << 213 if (!p0.new) jump:nt .L14 << 214 } << 215 r3 = and(r6,#31) << 216 if (r3==#0) jump:nt .L17 << 217 { << 218 memd(r6++#8) = r5:4 << 219 r2 = add(r2,#-8) << 220 } << 221 r3 = and(r6,#31) << 222 if (r3==#0) jump:nt .L17 << 223 { << 224 memd(r6++#8) = r5:4 << 225 r2 = add(r2,#-8) << 226 } << 227 r3 = and(r6,#31) << 228 if (r3==#0) jump:nt .L17 << 229 { << 230 memd(r6++#8) = r5:4 << 231 r2 = add(r2,#-8) << 232 } << 233 .L17: << 234 { << 235 r3 = lsr(r2,#5) << 236 if (r1!=#0) jump:nt .L18 << 237 } << 238 { << 239 r8 = r3 << 240 r3 = r6 << 241 loop0(.L46,r3) << 242 } << 243 .falign << 244 .L46: << 245 { << 246 dczeroa(r6) << 247 r6 = add(r6,#32) << 248 r2 = add(r2,#-32) << 249 }:endloop0 /* start=.L46 */ << 250 .L14: << 251 { << 252 p0 = cmp.gtu(r2,#7) << 253 if (!p0.new) jump:nt .L28 << 254 r8 = lsr(r2,#3) << 255 } << 256 loop0(.L44,r8) << 257 .falign << 258 .L44: << 259 { << 260 memd(r6++#8) = r5:4 << 261 r2 = add(r2,#-8) << 262 }:endloop0 /* start=.L44 */ << 263 .L28: << 264 { << 265 p0 = tstbit(r2,#2) << 266 if (!p0.new) jump:nt .L33 << 267 } << 268 { << 269 r2 = add(r2,#-4) << 270 memw(r6++#4) = r7 << 271 } << 272 .L33: << 273 { << 274 p0 = tstbit(r2,#1) << 275 if (!p0.new) jump:nt .L35 << 276 } << 277 { << 278 r2 = add(r2,#-2) << 279 memh(r6++#2) = r7 << 280 } << 281 .L35: << 282 p0 = cmp.eq(r2,#1) << 283 if (p0) memb(r6) = r1 << 284 .L1: << 285 jumpr r31 << 286 .L18: << 287 loop0(.L45,r3) << 288 .falign << 289 .L45: << 290 dczeroa(r6) << 291 { << 292 memd(r6++#8) = r5:4 << 293 r2 = add(r2,#-32) << 294 } << 295 memd(r6++#8) = r5:4 << 296 memd(r6++#8) = r5:4 << 297 { << 298 memd(r6++#8) = r5:4 << 299 }:endloop0 /* start=.L45 */ << 300 jump .L14 << 301 HEXAGON_OPT_FUNC_FINISH memset << 302 #endif <<

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

TOMOYO Linux Cross Reference
Linux/arch/hexagon/lib/memset.S

Diff markup

Differences between /arch/hexagon/lib/memset.S (Architecture ppc) and /arch/alpha/lib/memset.S (Architecture alpha)

TOMOYO Linux Cross Reference Linux/arch/hexagon/lib/memset.S

Diff markup

Differences between /arch/hexagon/lib/memset.S (Architecture ppc) and /arch/alpha/lib/memset.S (Architecture alpha)

TOMOYO Linux Cross Reference
Linux/arch/hexagon/lib/memset.S