diff options
author | Wilco Dijkstra <wilco.dijkstra@arm.com> | 2023-01-11 13:52:23 +0000 |
---|---|---|
committer | Wilco Dijkstra <wilco.dijkstra@arm.com> | 2024-04-09 17:37:15 +0100 |
commit | b2cf48dd841ede7c57443a36d42405d39b041fa1 (patch) | |
tree | 31a13469b79777ae622ed8ad03dea81e7ab4b25a | |
parent | 979cfc9dc31a6d0dd86f3133c4c4ecfea448e6b0 (diff) | |
download | glibc-b2cf48dd841ede7c57443a36d42405d39b041fa1.tar.gz glibc-b2cf48dd841ede7c57443a36d42405d39b041fa1.tar.xz glibc-b2cf48dd841ede7c57443a36d42405d39b041fa1.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 9de680b63e..9856fc5887 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 |