nuttx/libs/libc/machine/risc-v/gnu/arch_memset.S
anjiahao 603c87977f libc/machine:Add prefixes to libc functions implemented by arch
kasan does not instrument the assembly function,
so it is checked in advance before calling.
If Kasan is not turned on, the speed and space will
be almost unchanged under compiler optimization.

Signed-off-by: anjiahao <anjiahao@xiaomi.com>
2024-10-31 18:11:16 +08:00

113 lines
2.1 KiB
ArmAsm

/****************************************************************************
* libs/libc/machine/risc-v/gnu/arch_memset.S
*
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
* SPDX-FileCopyrightText: 2017 SiFive Inc. All rights reserved.
*
* This copyrighted material is made available to anyone wishing to use,
* modify, copy, or redistribute it subject to the terms and conditions
* of the FreeBSD License. This program is distributed in the hope that
* it will be useful, but WITHOUT ANY WARRANTY expressed or implied,
* including the implied warranties of MERCHANTABILITY or FITNESS FOR
* A PARTICULAR PURPOSE. A copy of this license is available at
* http://www.opensource.org/licenses.
*
****************************************************************************/
#include "libc.h"
#ifdef LIBC_BUILD_MEMSET
.text
.global ARCH_LIBCFUN(memset)
.type ARCH_LIBCFUN(memset), @function
ARCH_LIBCFUN(memset):
li t1, 15
move a4, a0
bleu a2, t1, .Ltiny
and a5, a4, 15
bnez a5, .Lmisaligned
.Laligned:
bnez a1, .Lwordify
.Lwordified:
and a3, a2, ~15
and a2, a2, 15
add a3, a3, a4
#if __riscv_xlen == 64
1:
sd a1, 0(a4)
sd a1, 8(a4)
#else
1:
sw a1, 0(a4)
sw a1, 4(a4)
sw a1, 8(a4)
sw a1, 12(a4)
#endif
add a4, a4, 16
bltu a4, a3, 1b
bnez a2, .Ltiny
ret
.Ltiny:
sub a3, t1, a2
sll a3, a3, 2
1:
auipc t0, %pcrel_hi(.Ltable)
add a3, a3, t0
.option push
.option norvc
.Ltable_misaligned:
jr a3, %pcrel_lo(1b)
.Ltable:
sb a1,14(a4)
sb a1,13(a4)
sb a1,12(a4)
sb a1,11(a4)
sb a1,10(a4)
sb a1, 9(a4)
sb a1, 8(a4)
sb a1, 7(a4)
sb a1, 6(a4)
sb a1, 5(a4)
sb a1, 4(a4)
sb a1, 3(a4)
sb a1, 2(a4)
sb a1, 1(a4)
sb a1, 0(a4)
.option pop
ret
.Lwordify:
and a1, a1, 0xFF
sll a3, a1, 8
or a1, a1, a3
sll a3, a1, 16
or a1, a1, a3
#if __riscv_xlen == 64
sll a3, a1, 32
or a1, a1, a3
#endif
j .Lwordified
.Lmisaligned:
sll a3, a5, 2
1:
auipc t0, %pcrel_hi(.Ltable_misaligned)
add a3, a3, t0
mv t0, ra
jalr a3, %pcrel_lo(1b)
mv ra, t0
add a5, a5, -16
sub a4, a4, a5
add a2, a2, a5
bleu a2, t1, .Ltiny
j .Laligned
.size ARCH_LIBCFUN(memset), .-ARCH_LIBCFUN(memset)
#endif