From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 21020 invoked by alias); 23 May 2006 22:28:58 -0000 Received: (qmail 21007 invoked by uid 9572); 23 May 2006 22:28:57 -0000 Date: Tue, 23 May 2006 22:28:00 -0000 Message-ID: <20060523222857.21005.qmail@sourceware.org> From: wcheng@sourceware.org To: cluster-cvs@sources.redhat.com Subject: cluster/gfs-kernel/src/gfs ops_file.c Mailing-List: contact cluster-cvs-help@sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Post: List-Help: , Sender: cluster-cvs-owner@sourceware.org X-SW-Source: 2006-q2/txt/msg00321.txt.bz2 List-Id: CVSROOT: /cvs/cluster Module name: cluster Branch: RHEL4 Changes by: wcheng@sourceware.org 2006-05-23 22:28:57 Modified files: gfs-kernel/src/gfs: ops_file.c Log message: We miss c/mtime update in fsync() via previous check-in. This patch fixes that. We also forks gfs's fsync() and fdatasync() implementations in this patch. The fdatasync, where it flushes data but not meta data like the mtime or atime, will have the same performance as previous check-in. For fsync, we will flush everything as the old code did but it is still able to gain about 3x performance when compared with the original code due to the removal of repeated writepage(). Data obtained from the test machine via fstest.c: Original code: both fsync and fdatasync: 240.94 KB/s This chech in: fdatasync: 2.67 MB/s fsync: 799.21 KB/s Patches: http://sourceware.org/cgi-bin/cvsweb.cgi/cluster/gfs-kernel/src/gfs/ops_file.c.diff?cvsroot=cluster&only_with_tag=RHEL4&r1=1.16.2.10&r2=1.16.2.11