^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) /* SPDX-License-Identifier: GPL-2.0-only */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) * Copyright (c) 2011, The Linux Foundation. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) /* HEXAGON assembly optimized memset */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) /* Replaces the standard library function memset */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) .macro HEXAGON_OPT_FUNC_BEGIN name
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) .text
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) .p2align 4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) .globl \name
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) .type \name, @function
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) \name:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17) .endm
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) .macro HEXAGON_OPT_FUNC_FINISH name
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) .size \name, . - \name
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) .endm
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) /* FUNCTION: memset (v2 version) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) #if __HEXAGON_ARCH__ < 3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) HEXAGON_OPT_FUNC_BEGIN memset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) r6 = #8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) r7 = extractu(r0, #3 , #0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) p0 = cmp.eq(r2, #0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) p1 = cmp.gtu(r2, #7)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) r4 = vsplatb(r1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) r8 = r0 /* leave r0 intact for return val */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) r9 = sub(r6, r7) /* bytes until double alignment */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) if p0 jumpr r31 /* count == 0, so return */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) r3 = #0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) r7 = #0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) p0 = tstbit(r9, #0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) if p1 jump 2f /* skip byte loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45) /* less than 8 bytes to set, so just set a byte at a time and return */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) loop0(1f, r2) /* byte loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) 1: /* byte loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) memb(r8++#1) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) }:endloop0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) 2: /* skip byte loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) r6 = #1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) p0 = tstbit(r9, #1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) p1 = cmp.eq(r2, #1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) if !p0 jump 3f /* skip initial byte store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) memb(r8++#1) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) r3:2 = sub(r3:2, r7:6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) if p1 jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) 3: /* skip initial byte store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) r6 = #2
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) p0 = tstbit(r9, #2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) p1 = cmp.eq(r2, #2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) if !p0 jump 4f /* skip initial half store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) memh(r8++#2) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) r3:2 = sub(r3:2, r7:6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) if p1 jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81) 4: /* skip initial half store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) r6 = #4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84) p0 = cmp.gtu(r2, #7)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) p1 = cmp.eq(r2, #4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) if !p0 jump 5f /* skip initial word store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) memw(r8++#4) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) r3:2 = sub(r3:2, r7:6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) p0 = cmp.gtu(r2, #11)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) if p1 jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) 5: /* skip initial word store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) r10 = lsr(r2, #3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98) p1 = cmp.eq(r3, #1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) if !p0 jump 7f /* skip double loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) r5 = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) r6 = #8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) loop0(6f, r10) /* double loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) /* set bytes a double word at a time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) 6: /* double loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) memd(r8++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113) r3:2 = sub(r3:2, r7:6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) p1 = cmp.eq(r2, #8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) }:endloop0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) 7: /* skip double loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119) p0 = tstbit(r2, #2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) if p1 jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) r6 = #4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) p0 = tstbit(r2, #1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) p1 = cmp.eq(r2, #4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126) if !p0 jump 8f /* skip final word store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) memw(r8++#4) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) r3:2 = sub(r3:2, r7:6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131) if p1 jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) 8: /* skip final word store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136) p1 = cmp.eq(r2, #2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) if !p0 jump 9f /* skip final half store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) memh(r8++#2) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141) if p1 jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) 9: /* skip final half store */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) memb(r8++#1) = r4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) HEXAGON_OPT_FUNC_FINISH memset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) /* FUNCTION: memset (v3 and higher version) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) #if __HEXAGON_ARCH__ >= 3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155) HEXAGON_OPT_FUNC_BEGIN memset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) r7=vsplatb(r1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) r6 = r0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) if (r2==#0) jump:nt .L1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) r5:4=combine(r7,r7)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) p0 = cmp.gtu(r2,#8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) if (p0.new) jump:nt .L3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167) r3 = r0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) loop0(.L47,r2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) .L47:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173) memb(r3++#1) = r1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) }:endloop0 /* start=.L47 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) .L3:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) p0 = tstbit(r0,#0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) if (!p0.new) jump:nt .L8
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) p1 = cmp.eq(r2, #1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183) r6 = add(r0, #1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) r2 = add(r2,#-1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) memb(r0) = r1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) if (p1) jump .L1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) .L8:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) p0 = tstbit(r6,#1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191) if (!p0.new) jump:nt .L10
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) r2 = add(r2,#-2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195) memh(r6++#2) = r7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) p0 = cmp.eq(r2, #2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) if (p0.new) jump:nt .L1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) .L10:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) p0 = tstbit(r6,#2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) if (!p0.new) jump:nt .L12
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) r2 = add(r2,#-4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) memw(r6++#4) = r7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) p0 = cmp.eq(r2, #4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) if (p0.new) jump:nt .L1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) .L12:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) p0 = cmp.gtu(r2,#127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) if (!p0.new) jump:nt .L14
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) r3 = and(r6,#31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) if (r3==#0) jump:nt .L17
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) r2 = add(r2,#-8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) r3 = and(r6,#31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) if (r3==#0) jump:nt .L17
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) r2 = add(r2,#-8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) r3 = and(r6,#31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228) if (r3==#0) jump:nt .L17
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231) r2 = add(r2,#-8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) .L17:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) r3 = lsr(r2,#5)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236) if (r1!=#0) jump:nt .L18
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) r8 = r3
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240) r3 = r6
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) loop0(.L46,r3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244) .L46:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) dczeroa(r6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) r6 = add(r6,#32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) r2 = add(r2,#-32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) }:endloop0 /* start=.L46 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250) .L14:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) p0 = cmp.gtu(r2,#7)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) if (!p0.new) jump:nt .L28
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) r8 = lsr(r2,#3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) loop0(.L44,r8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) .L44:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) r2 = add(r2,#-8)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) }:endloop0 /* start=.L44 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) .L28:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) p0 = tstbit(r2,#2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) if (!p0.new) jump:nt .L33
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) r2 = add(r2,#-4)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) memw(r6++#4) = r7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) .L33:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) p0 = tstbit(r2,#1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) if (!p0.new) jump:nt .L35
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) r2 = add(r2,#-2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) memh(r6++#2) = r7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) .L35:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) p0 = cmp.eq(r2,#1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283) if (p0) memb(r6) = r1
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) .L1:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) jumpr r31
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) .L18:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287) loop0(.L45,r3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) .falign
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) .L45:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) dczeroa(r6)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) r2 = add(r2,#-32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) memd(r6++#8) = r5:4
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) }:endloop0 /* start=.L45 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) jump .L14
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) HEXAGON_OPT_FUNC_FINISH memset
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) #endif