From: Benjamin Marzinski <bmarzins@redhat.com>
To: cluster-devel.redhat.com
Subject: [Cluster-devel] [PATCH] GFS2: dirty inode correctly in gfs2_write_end
Date: Tue, 3 Sep 2013 16:59:42 -0500 [thread overview]
Message-ID: <20130903215942.GM2113@dhcp80-209.msp.redhat.com> (raw)
GFS2 was only setting I_DIRTY_DATASYNC on files that it wrote to, when
it actually increased the file size. If gfs2_fsync was called without
I_DIRTY_DATASYNC set, it didn't flush the incore data to the log before
returning, so any metadata or journaled data changes were not getting
fsynced. This meant that writes to the middle of files were not always
getting fsynced properly.
This patch makes gfs2 set I_DIRTY_DATASYNC whenever metadata has been
updated during a write. It also make gfs2_sync flush the incore log
if I_DIRTY_PAGES is set, and the file is using data journalling. This
will make sure that all incore logged data gets written to disk before
returning from a fsync.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
fs/gfs2/aops.c | 9 +++++++--
fs/gfs2/file.c | 4 +++-
2 files changed, 10 insertions(+), 3 deletions(-)
Index: gfs2-3.0-nmw-130830/fs/gfs2/aops.c
===================================================================
--- gfs2-3.0-nmw-130830.orig/fs/gfs2/aops.c
+++ gfs2-3.0-nmw-130830/fs/gfs2/aops.c
@@ -815,6 +815,8 @@ static int gfs2_write_end(struct file *f
unsigned int from = pos & (PAGE_CACHE_SIZE - 1);
unsigned int to = from + len;
int ret;
+ struct gfs2_trans *tr = current->journal_info;
+ BUG_ON(!tr);
BUG_ON(gfs2_glock_is_locked_by_me(ip->i_gl) == NULL);
@@ -825,8 +827,6 @@ static int gfs2_write_end(struct file *f
goto failed;
}
- gfs2_trans_add_meta(ip->i_gl, dibh);
-
if (gfs2_is_stuffed(ip))
return gfs2_stuffed_write_end(inode, dibh, pos, len, copied, page);
@@ -834,6 +834,11 @@ static int gfs2_write_end(struct file *f
gfs2_page_add_databufs(ip, page, from, to);
ret = generic_write_end(file, mapping, pos, len, copied, page, fsdata);
+ if (tr->tr_num_buf_new)
+ __mark_inode_dirty(inode, I_DIRTY_DATASYNC);
+ else
+ gfs2_trans_add_meta(ip->i_gl, dibh);
+
if (inode == sdp->sd_rindex) {
adjust_fs_space(inode);
Index: gfs2-3.0-nmw-130830/fs/gfs2/file.c
===================================================================
--- gfs2-3.0-nmw-130830.orig/fs/gfs2/file.c
+++ gfs2-3.0-nmw-130830/fs/gfs2/file.c
@@ -650,7 +650,7 @@ static int gfs2_fsync(struct file *file,
{
struct address_space *mapping = file->f_mapping;
struct inode *inode = mapping->host;
- int sync_state = inode->i_state & (I_DIRTY_SYNC|I_DIRTY_DATASYNC);
+ int sync_state = inode->i_state & I_DIRTY;
struct gfs2_inode *ip = GFS2_I(inode);
int ret = 0, ret1 = 0;
@@ -660,6 +660,8 @@ static int gfs2_fsync(struct file *file,
return ret1;
}
+ if (!gfs2_is_jdata(ip))
+ sync_state &= ~I_DIRTY_PAGES;
if (datasync)
sync_state &= ~I_DIRTY_SYNC;
next reply other threads:[~2013-09-03 21:59 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-09-03 21:59 Benjamin Marzinski [this message]
2013-09-04 13:24 ` [Cluster-devel] [PATCH] GFS2: dirty inode correctly in gfs2_write_end Bob Peterson
2013-09-04 14:55 ` Steven Whitehouse
2013-09-04 16:59 ` Benjamin Marzinski
2013-09-04 21:36 ` Steven Whitehouse
2013-09-05 3:48 ` Benjamin Marzinski
2013-09-05 8:28 ` Steven Whitehouse
2013-09-05 9:34 ` Steven Whitehouse
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20130903215942.GM2113@dhcp80-209.msp.redhat.com \
--to=bmarzins@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).