diff options
| author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-16 19:19:13 +0000 |
|---|---|---|
| committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-16 19:19:13 +0000 |
| commit | ccd992355df7192993c666236047820244914598 (patch) | |
| tree | f00fea65147227b7743083c6148396f74cd66935 /src/runtime/memclr_mips64x.s | |
| parent | Initial commit. (diff) | |
| download | golang-1.21-ccd992355df7192993c666236047820244914598.tar.xz golang-1.21-ccd992355df7192993c666236047820244914598.zip | |
Adding upstream version 1.21.8.upstream/1.21.8
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'src/runtime/memclr_mips64x.s')
| -rw-r--r-- | src/runtime/memclr_mips64x.s | 99 |
1 files changed, 99 insertions, 0 deletions
diff --git a/src/runtime/memclr_mips64x.s b/src/runtime/memclr_mips64x.s new file mode 100644 index 0000000..cf3a9c4 --- /dev/null +++ b/src/runtime/memclr_mips64x.s @@ -0,0 +1,99 @@ +// Copyright 2015 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +//go:build mips64 || mips64le + +#include "go_asm.h" +#include "textflag.h" + +// See memclrNoHeapPointers Go doc for important implementation constraints. + +// func memclrNoHeapPointers(ptr unsafe.Pointer, n uintptr) +TEXT runtime·memclrNoHeapPointers(SB),NOSPLIT,$0-16 + MOVV ptr+0(FP), R1 + MOVV n+8(FP), R2 + ADDV R1, R2, R4 + + // if less than 16 bytes or no MSA, do words check + SGTU $16, R2, R3 + BNE R3, no_msa + MOVBU internal∕cpu·MIPS64X+const_offsetMIPS64XHasMSA(SB), R3 + BEQ R3, R0, no_msa + + VMOVB $0, W0 + + SGTU $128, R2, R3 + BEQ R3, msa_large + + AND $15, R2, R5 + XOR R2, R5, R6 + ADDVU R1, R6 + +msa_small: + VMOVB W0, (R1) + ADDVU $16, R1 + SGTU R6, R1, R3 + BNE R3, R0, msa_small + BEQ R5, R0, done + VMOVB W0, -16(R4) + JMP done + +msa_large: + AND $127, R2, R5 + XOR R2, R5, R6 + ADDVU R1, R6 + +msa_large_loop: + VMOVB W0, (R1) + VMOVB W0, 16(R1) + VMOVB W0, 32(R1) + VMOVB W0, 48(R1) + VMOVB W0, 64(R1) + VMOVB W0, 80(R1) + VMOVB W0, 96(R1) + VMOVB W0, 112(R1) + + ADDVU $128, R1 + SGTU R6, R1, R3 + BNE R3, R0, msa_large_loop + BEQ R5, R0, done + VMOVB W0, -128(R4) + VMOVB W0, -112(R4) + VMOVB W0, -96(R4) + VMOVB W0, -80(R4) + VMOVB W0, -64(R4) + VMOVB W0, -48(R4) + VMOVB W0, -32(R4) + VMOVB W0, -16(R4) + JMP done + +no_msa: + // if less than 8 bytes, do one byte at a time + SGTU $8, R2, R3 + BNE R3, out + + // do one byte at a time until 8-aligned + AND $7, R1, R3 + BEQ R3, words + MOVB R0, (R1) + ADDV $1, R1 + JMP -4(PC) + +words: + // do 8 bytes at a time if there is room + ADDV $-7, R4, R2 + + SGTU R2, R1, R3 + BEQ R3, out + MOVV R0, (R1) + ADDV $8, R1 + JMP -4(PC) + +out: + BEQ R1, R4, done + MOVB R0, (R1) + ADDV $1, R1 + JMP -3(PC) +done: + RET |