From 2f3f7b9da238269397d8dd8868ed494eff2c9064 Mon Sep 17 00:00:00 2001 From: Ulrich Drepper Date: Thu, 4 Jun 2009 16:45:35 -0700 Subject: [PATCH] More small optimizations for x86-64 strlen. --- ChangeLog | 2 +- sysdeps/x86_64/strlen.S | 16 +++++++--------- 2 files changed, 8 insertions(+), 10 deletions(-) diff --git a/ChangeLog b/ChangeLog index cb094d1863..7d99a207fe 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,6 +1,6 @@ 2009-06-04 Ulrich Drepper - * sysdeps/x86_64/strlen.S: Remove useless instruction. + * sysdeps/x86_64/strlen.S: Minor optimizations. 2009-06-02 H.J. Lu diff --git a/sysdeps/x86_64/strlen.S b/sysdeps/x86_64/strlen.S index 572f3c2e41..93aee6bef1 100644 --- a/sysdeps/x86_64/strlen.S +++ b/sysdeps/x86_64/strlen.S @@ -23,29 +23,27 @@ .text ENTRY(strlen) + pxor %xmm2, %xmm2 movq %rdi, %rcx movq %rdi, %r8 andq $~15, %rdi - pxor %xmm1, %xmm1 + movdqa %xmm2, %xmm1 + pcmpeqb (%rdi), %xmm2 orl $0xffffffff, %esi - movdqa (%rdi), %xmm0 subq %rdi, %rcx - leaq 16(%rdi), %rdi - pcmpeqb %xmm1, %xmm0 - shl %cl, %esi - pmovmskb %xmm0, %edx - negq %r8 + shll %cl, %esi + pmovmskb %xmm2, %edx andl %esi, %edx jnz 1f -2: movdqa (%rdi), %xmm0 +2: movdqa 16(%rdi), %xmm0 leaq 16(%rdi), %rdi pcmpeqb %xmm1, %xmm0 pmovmskb %xmm0, %edx testl %edx, %edx jz 2b -1: leaq -16(%rdi,%r8), %rdi +1: subq %r8, %rdi bsfl %edx, %eax addq %rdi, %rax ret