packages/glibc
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

strcmp: delete align for loop_aligned

Browse Source 
In Kunpeng-920, the performance of strcmp deteriorates only
when the 16 to 23 characters are different.Or the string is
only 16-23 characters.That shows 2 misses per iteration which
means this is a branch predictor issue indeed.
In the preceding scenario, strcmp performance is 300% worse than expected.

Fortunately, this problem can be solved by modifying the alignment of the functions.

Signed-off-by: Yang Yanchao <yangyanchao6@huawei.com>
This commit is contained in:
Yang Yanchao 2022-02-23 15:34:35 +08:00
parent 14ed19693b
commit 03b9b27526
2 changed files with 37 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
glibc.spec
View File

@ -66,7 +66,7 @@
##############################################################################
Name: glibc
Version: 2.34
Release: 58
Release: 59
Summary: The GNU libc libraries
License: %{all_license}
URL: http://www.gnu.org/software/glibc/
@ -205,6 +205,7 @@ Patch9011: delete-check-installed-headers-c-and-check-installed.patch
Patch9012: fix-CVE-2019-1010023.patch
Patch9013: fix-tst-glibcsyscalls-due-to-kernel-reserved-some-sy.patch
Patch9014: use-region-to-instead-of-country-for-extract-timezon.patch
Patch9015: strcmp-delete-align-for-loop_aligned.patch
Obsoletes: nscd < 2.35
Provides: ldconfig rtld(GNU_HASH) bundled(gnulib)
@ -1219,6 +1220,9 @@ fi
%endif
%changelog
* Wed Feb 23 2022 Yang Yanchao<yangyanchao6@huawei.com> - 2.34-59
- strcmp: delete align for loop_aligned
* Wed Feb 23 2022 Yang Yanchao<yangyanchao6@huawei.com> - 2.34-58
- The release of glibc.src.rpm in OpenEuler is not based on the architecture.
Developers only have glibc.src.rpm in the ARM, so add all testsuite_whitelist in glibc.src.rpm.

32
strcmp-delete-align-for-loop_aligned.patch Normal file
View File

@ -0,0 +1,32 @@
From 9bbffed83b93f633b272368fc536a4f24e9942e6 Mon Sep 17 00:00:00 2001
From: Yang Yanchao <yangyanchao6@huawei.com>
Date: Mon, 21 Feb 2022 14:25:25 +0800
Subject: [PATCH] strcmp: delete align for loop_aligned
In Kunpeng-920, the performance of strcmp deteriorates only
when the 16 to 23 characters are different.Or the string is
only 16-23 characters.That shows 2 misses per iteration which
means this is a branch predictor issue indeed.
In the preceding scenario, strcmp performance is 300% worse than expected.
Fortunately, this problem can be solved by modifying the alignment of the functions.
---
sysdeps/aarch64/strcmp.S | 2 --
1 file changed, 2 deletions(-)
diff --git a/sysdeps/aarch64/strcmp.S b/sysdeps/aarch64/strcmp.S
index f225d718..7a048b66 100644
--- a/sysdeps/aarch64/strcmp.S
+++ b/sysdeps/aarch64/strcmp.S
@@ -71,8 +71,6 @@ ENTRY(strcmp)
b.ne L(misaligned8)
cbnz tmp, L(mutual_align)
- .p2align 4
-
L(loop_aligned):
ldr data2, [src1, off2]
ldr data1, [src1], 8
--
2.33.0