AuroraMiddleware/glibc
1
0
Fork 0
mirror of https://sourceware.org/git/glibc.git synced 2024-11-29 16:21:07 +00:00
Code Issues Packages Projects Releases Wiki Activity
f8e8effa26
glibc/sysdeps/aarch64/strnlen.S

118 lines
2.7 KiB
ArmAsm
Raw Normal View History

AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
/* strnlen - calculate the length of a string with limit.
Update copyright dates with scripts/update-copyrights
2023-01-06 21:08:04 +00:00
Copyright (C) 2013-2023 Free Software Foundation, Inc.
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library. If not, see
Prefer https to http for gnu.org and fsf.org URLs Also, change sources.redhat.com to sourceware.org. This patch was automatically generated by running the following shell script, which uses GNU sed, and which avoids modifying files imported from upstream: sed -ri ' s,(http|ftp)(://(.*\.)?(gnu|fsf|sourceware)\.org($|[^.]|\.[^a-z])),https\2,g s,(http|ftp)(://(.*\.)?)sources\.redhat\.com($|[^.]|\.[^a-z]),https\2sourceware.org\4,g ' \ $(find $(git ls-files) -prune -type f \ ! -name '*.po' \ ! -name 'ChangeLog*' \ ! -path COPYING ! -path COPYING.LIB \ ! -path manual/fdl-1.3.texi ! -path manual/lgpl-2.1.texi \ ! -path manual/texinfo.tex ! -path scripts/config.guess \ ! -path scripts/config.sub ! -path scripts/install-sh \ ! -path scripts/mkinstalldirs ! -path scripts/move-if-change \ ! -path INSTALL ! -path locale/programs/charmap-kw.h \ ! -path po/libc.pot ! -path sysdeps/gnu/errlist.c \ ! '(' -name configure \ -execdir test -f configure.ac -o -f configure.in ';' ')' \ ! '(' -name preconfigure \ -execdir test -f preconfigure.ac ';' ')' \ -print) and then by running 'make dist-prepare' to regenerate files built from the altered files, and then executing the following to cleanup: chmod a+x sysdeps/unix/sysv/linux/riscv/configure # Omit irrelevant whitespace and comment-only changes, # perhaps from a slightly-different Autoconf version. git checkout -f \ sysdeps/csky/configure \ sysdeps/hppa/configure \ sysdeps/riscv/configure \ sysdeps/unix/sysv/linux/csky/configure # Omit changes that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/powerpc/powerpc64/ppc-mcount.S: trailing lines git checkout -f \ sysdeps/powerpc/powerpc64/ppc-mcount.S \ sysdeps/unix/sysv/linux/s390/s390-64/syscall.S # Omit change that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S: last line does not end in newline git checkout -f sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
2019-09-07 05:40:42 +00:00
<https://www.gnu.org/licenses/>. */
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
#include <sysdep.h>
/* Assumptions:
*
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
* ARMv8-a, AArch64, Advanced SIMD.
* MTE compatible.
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
*/
#define srcin x0
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
#define cntin x1
#define result x0
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
#define src x2
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
#define synd x3
#define shift x4
#define tmp x4
#define cntrem x5
#define qdata q0
#define vdata v0
#define vhas_chr v1
aarch64: Optimize string functions with shrn instruction We found that string functions were using AND+ADDP to find the nibble/syndrome mask but there is an easier opportunity through `SHRN dst.8b, src.8h, 4` (shift right every 2 bytes by 4 and narrow to 1 byte) and has same latency on all SIMD ARMv8 targets as ADDP. There are also possible gaps for memcmp but that's for another patch. We see 10-20% savings for small-mid size cases (<=128) which are primary cases for general workloads.
2022-06-27 16:12:13 +00:00
#define vend v2
#define dend d2
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
/*
Core algorithm:
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
Process the string in 16-byte aligned chunks. Compute a 64-bit mask with
four bits per byte using the shrn instruction. A count trailing zeros then
identifies the first zero byte. */
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
ENTRY (__strnlen)
aarch64: use PTR_ARG and SIZE_ARG instead of DELOUSE DELOUSE was added to asm code to make them compatible with non-LP64 ABIs, but it is an unfortunate name and the code was not compatible with ABIs where pointer and size_t are different. Glibc currently only supports the LP64 ABI so these macros are not really needed or tested, but for now the name is changed to be more meaningful instead of removing them completely. Some DELOUSE macros were dropped: clone, strlen and strnlen used it unnecessarily. The out of tree ILP32 patches are currently not maintained and will likely need a rework to rebase them on top of the time64 changes.
2020-12-17 10:03:05 +00:00
PTR_ARG (0)
SIZE_ARG (1)
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
bic src, srcin, 15
cbz cntin, L(nomatch)
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
ld1 {vdata.16b}, [src]
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
cmeq vhas_chr.16b, vdata.16b, 0
lsl shift, srcin, 2
aarch64: Optimize string functions with shrn instruction We found that string functions were using AND+ADDP to find the nibble/syndrome mask but there is an easier opportunity through `SHRN dst.8b, src.8h, 4` (shift right every 2 bytes by 4 and narrow to 1 byte) and has same latency on all SIMD ARMv8 targets as ADDP. There are also possible gaps for memcmp but that's for another patch. We see 10-20% savings for small-mid size cases (<=128) which are primary cases for general workloads.
2022-06-27 16:12:13 +00:00
shrn vend.8b, vhas_chr.8h, 4 /* 128->64 */
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
fmov synd, dend
lsr synd, synd, shift
cbz synd, L(start_loop)
L(finish):
rbit synd, synd
clz synd, synd
lsr result, synd, 2
cmp cntin, result
csel result, cntin, result, ls
ret
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
L(nomatch):
mov result, cntin
ret
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
L(start_loop):
sub tmp, src, srcin
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
add tmp, tmp, 17
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
subs cntrem, cntin, tmp
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
b.lo L(nomatch)
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
/* Make sure that it won't overread by a 16-byte chunk */
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
tbz cntrem, 4, L(loop32_2)
sub src, src, 16
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
.p2align 5
L(loop32):
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
ldr qdata, [src, 32]!
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
cmeq vhas_chr.16b, vdata.16b, 0
umaxp vend.16b, vhas_chr.16b, vhas_chr.16b /* 128->64 */
fmov synd, dend
cbnz synd, L(end)
L(loop32_2):
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
ldr qdata, [src, 16]
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
subs cntrem, cntrem, 32
cmeq vhas_chr.16b, vdata.16b, 0
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
b.lo L(end_2)
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
umaxp vend.16b, vhas_chr.16b, vhas_chr.16b /* 128->64 */
fmov synd, dend
cbz synd, L(loop32)
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
L(end_2):
add src, src, 16
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
L(end):
aarch64: Optimize string functions with shrn instruction We found that string functions were using AND+ADDP to find the nibble/syndrome mask but there is an easier opportunity through `SHRN dst.8b, src.8h, 4` (shift right every 2 bytes by 4 and narrow to 1 byte) and has same latency on all SIMD ARMv8 targets as ADDP. There are also possible gaps for memcmp but that's for another patch. We see 10-20% savings for small-mid size cases (<=128) which are primary cases for general workloads.
2022-06-27 16:12:13 +00:00
shrn vend.8b, vhas_chr.8h, 4 /* 128->64 */
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
sub result, src, srcin
AArch64: Optimize strnlen Optimize strnlen using the shrn instruction and improve the main loop. Small strings are around 10% faster, large strings are 40% faster on modern CPUs. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2023-01-11 13:53:05 +00:00
fmov synd, dend
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
#ifndef __AARCH64EB__
rbit synd, synd
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
#endif
AArch64: Improve strnlen performance Optimize strnlen by avoiding UMINV which is slow on most cores. On Neoverse N1 large strings are 1.8x faster than the current version, and bench-strnlen is 50% faster overall. This version is MTE compatible. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com>
2021-07-01 14:30:42 +00:00
clz synd, synd
add result, result, synd, lsr 2
cmp cntin, result
csel result, cntin, result, ls
ret
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
END (__strnlen)
Use libc_hidden_proto / libc_hidden_def with __strnlen. Various code in glibc uses __strnlen instead of strnlen for namespace reasons. However, __strnlen does not use libc_hidden_proto / libc_hidden_def (as is normally done for any function defined and called within the same library, whether or not exported from the library and whatever namespace it is in), so the compiler does not know that those calls are to a function within libc. This patch uses libc_hidden_proto / libc_hidden_def with __strnlen. On x86_64, it makes no difference to the installed stripped shared libraries. On 32-bit x86, it causes __strnlen calls to go to the same place as strnlen calls (the fallback strnlen implementation), rather than through a PLT entry for the strnlen IFUNC; I'm not sure of the logic behind when calls from within libc should use IFUNCs versus when they should go direct to a particular function implementation, but clearly it doesn't make sense for strnlen and __strnlen to be handled differently in this regard. Tested for x86_64 and x86 (testsuite, and comparison of installed shared libraries as described above). * string/strnlen.c [!STRNLEN] (__strnlen): Use libc_hidden_def. * include/string.h (__strnlen): Use libc_hidden_proto. * sysdeps/aarch64/strnlen.S (__strnlen): Use libc_hidden_def. * sysdeps/i386/i686/multiarch/strnlen-c.c [SHARED] (libc_hidden_def): Define __GI___strnlen as well as __GI_strnlen. * sysdeps/powerpc/powerpc32/power4/multiarch/strnlen-power7.S (libc_hidden_def): Undefine and redefine. * sysdeps/powerpc/powerpc32/power4/multiarch/strnlen-ppc32.c [SHARED] (libc_hidden_def): Define __GI___strnlen as well as __GI_strnlen. * sysdeps/powerpc/powerpc32/power7/strnlen.S (__strnlen): Use libc_hidden_def. * sysdeps/tile/tilegx/strnlen.c (__strnlen): Likewise.
2015-06-02 20:24:25 +00:00
libc_hidden_def (__strnlen)
AArch64: Adding optimized strnlen implementation.
2013-01-23 16:32:33 +00:00
weak_alias (__strnlen, strnlen)
libc_hidden_def (strnlen)
Reference in New Issue Copy Permalink