diff options
author | Wilco Dijkstra <wilco.dijkstra@arm.com> | 2023-01-11 13:52:23 +0000 |
---|---|---|
committer | Wilco Dijkstra <wilco.dijkstra@arm.com> | 2024-04-08 16:45:31 +0100 |
commit | 8ee69bd2df74fcc18ecf7e08eb5c74809c0be0e9 (patch) | |
tree | 634b1bcfea09598033b29d8997c503a34abce8ce | |
parent | 2e4cdcbd356094d36eca234e8dcecc5149d92dba (diff) | |
download | glibc-8ee69bd2df74fcc18ecf7e08eb5c74809c0be0e9.tar.gz glibc-8ee69bd2df74fcc18ecf7e08eb5c74809c0be0e9.tar.xz glibc-8ee69bd2df74fcc18ecf7e08eb5c74809c0be0e9.zip |
AArch64: Improve strchrnul
Unroll the main loop, which improves performance slightly. Reviewed-by: Szabolcs Nagy <szabolcs.nagy@arm.com> (cherry picked from commit 09ebd8549b2ce5a3a6c0c7c5f3e62227faf50a99)
-rw-r--r-- | sysdeps/aarch64/strchrnul.S | 12 |
1 files changed, 10 insertions, 2 deletions
diff --git a/sysdeps/aarch64/strchrnul.S b/sysdeps/aarch64/strchrnul.S index ee154ab74b..94465fc088 100644 --- a/sysdeps/aarch64/strchrnul.S +++ b/sysdeps/aarch64/strchrnul.S @@ -70,14 +70,22 @@ ENTRY (__strchrnul) .p2align 4 L(loop): - ldr qdata, [src, 16]! + ldr qdata, [src, 16] + cmeq vhas_chr.16b, vdata.16b, vrepchr.16b + cmhs vhas_chr.16b, vhas_chr.16b, vdata.16b + umaxp vend.16b, vhas_chr.16b, vhas_chr.16b + fmov tmp1, dend + cbnz tmp1, L(end) + ldr qdata, [src, 32]! cmeq vhas_chr.16b, vdata.16b, vrepchr.16b cmhs vhas_chr.16b, vhas_chr.16b, vdata.16b umaxp vend.16b, vhas_chr.16b, vhas_chr.16b fmov tmp1, dend cbz tmp1, L(loop) - + sub src, src, 16 +L(end): shrn vend.8b, vhas_chr.8h, 4 /* 128->64 */ + add src, src, 16 fmov tmp1, dend #ifndef __AARCH64EB__ rbit tmp1, tmp1 |