From: Jan Kara <jack@suse.cz>
To: Jeff Moyer <jmoyer@redhat.com>
Cc: Jan Kara <jack@suse.cz>,
"Darrick J. Wong" <darrick.wong@oracle.com>,
axboe@kernel.dk, tytso@mit.edu, david@fromorbit.com, bpm@sgi.com,
viro@zeniv.linux.org.uk, linux-fsdevel@vger.kernel.org,
hch@infradead.org, linux-ext4@vger.kernel.org,
linux-kernel@vger.kernel.org, xfs@oss.sgi.com
Subject: Re: [PATCH 2/9] ext4: honor the O_SYNC flag for aysnchronous direct I/O requests
Date: Wed, 21 Nov 2012 01:56:26 +0100 [thread overview]
Message-ID: <20121121005626.GC10507@quack.suse.cz> (raw)
In-Reply-To: <x49sj84hwl4.fsf@segfault.boston.devel.redhat.com>
[-- Attachment #1: Type: text/plain, Size: 2231 bytes --]
On Tue 20-11-12 15:02:15, Jeff Moyer wrote:
> Jan Kara <jack@suse.cz> writes:
>
> >> @@ -1279,6 +1280,9 @@ struct ext4_sb_info {
> >> /* workqueue for dio unwritten */
> >> struct workqueue_struct *dio_unwritten_wq;
> >>
> >> + /* workqueue for aio+dio+o_sync disk cache flushing */
> >> + struct workqueue_struct *aio_dio_flush_wq;
> >> +
> > Umm, I'm not completely decided whether we really need a separate
> > workqueue. But it doesn't cost too much so I guess it makes some sense -
> > fsync() is rather heavy so syncing won't starve extent conversion...
>
> I'm assuming you'd like me to convert the names from flush to fsync,
> yes?
Would be nicer, yes.
> >> +
> >> + /*
> >> + * If we are running in nojournal mode, just flush the disk
> >> + * cache and return.
> >> + */
> >> + if (!journal)
> >> + return blkdev_issue_flush(inode->i_sb->s_bdev, GFP_NOIO, NULL);
> > And this is wrong as well - you need to do work similar to what
> > ext4_sync_file() does. Actually it would be *much* better if these two
> > sites used the same helper function. Which also poses an interesting
> > question about locking - do we need i_mutex or not? Forcing a transaction
> > commit is definitely OK without it, similarly as grabbing transaction ids
> > from inode or ext4_should_journal_data() test. __sync_inode() call seems
> > to be OK without i_mutex as well so I believe we can just get rid of it
> > (getting i_mutex from the workqueue is a locking nightmare we don't want to
> > return to).
>
> Just to be clear, are you saying you would like me to remove the
> mutex_lock/unlock pair from ext4_sync_file? (I had already factored out
> the common code between this new code path and the fsync path in my tree.)
Yes, after some thinking I came to that conclusion. We actually need to
keep i_mutex around ext4_flush_unwritten_io() to avoid livelocks but the
rest doesn't need it. The change should be definitely a separate patch just
in case there's something subtle I missed and we need to bisect in
future... I've attached a patch for that so that blame for bugs goes my way
;) Compile tested only so far. I'll give it some more testing overnight.
Honza
--
Jan Kara <jack@suse.cz>
SUSE Labs, CR
[-- Attachment #2: 0001-ext4-Reduce-i_mutex-usage-in-ext4_file_sync.patch --]
[-- Type: text/x-patch, Size: 1862 bytes --]
>From 98f02e76b90e278e9688b4311a8889cec7095601 Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Wed, 21 Nov 2012 01:46:51 +0100
Subject: [PATCH] ext4: Reduce i_mutex usage in ext4_file_sync()
ext4_file_sync() needs i_mutex only to avoid livelocks of
ext4_flush_unwritten_io() all other code doesn't need it. In particular
syncing of inode & metadata in non-journal case is safe (writeback doesn't
hold i_mutex either) and forcing of transaction commits doesn't need i_mutex
either (there's nothing inode specific in that code apart from grabbing
transaction ids from the inode). So shorten the span where i_mutex is held.
Signed-off-by: Jan Kara <jack@suse.cz>
---
fs/ext4/fsync.c | 6 ++----
1 files changed, 2 insertions(+), 4 deletions(-)
diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c
index be1d89f..2268114 100644
--- a/fs/ext4/fsync.c
+++ b/fs/ext4/fsync.c
@@ -113,8 +113,6 @@ static int __sync_inode(struct inode *inode, int datasync)
*
* What we do is just kick off a commit and wait on it. This will snapshot the
* inode to disk.
- *
- * i_mutex lock is held when entering and exiting this function
*/
int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
@@ -133,12 +131,13 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
if (ret)
return ret;
- mutex_lock(&inode->i_mutex);
if (inode->i_sb->s_flags & MS_RDONLY)
goto out;
+ mutex_lock(&inode->i_mutex);
ret = ext4_flush_unwritten_io(inode);
+ mutex_unlock(&inode->i_mutex);
if (ret < 0)
goto out;
@@ -180,7 +179,6 @@ int ext4_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
ret = err;
}
out:
- mutex_unlock(&inode->i_mutex);
trace_ext4_sync_file_exit(inode, ret);
return ret;
}
--
1.7.1
next prev parent reply other threads:[~2012-11-21 0:56 UTC|newest]
Thread overview: 36+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-11-20 7:41 [PATCH v1 0/9] fs: fix up AIO+DIO+O_SYNC to actually do the sync part Darrick J. Wong
2012-11-20 7:41 ` [PATCH 1/9] vfs: Handle O_SYNC AIO DIO in generic code properly Darrick J. Wong
2012-11-21 10:08 ` Christoph Hellwig
2012-11-21 16:58 ` Jeff Moyer
2012-11-21 18:29 ` Christoph Hellwig
2012-11-21 18:38 ` Jeff Moyer
2012-11-21 21:37 ` Jan Kara
2012-11-21 23:09 ` Jeffrey Ellis
2012-11-20 7:41 ` [PATCH 2/9] ext4: honor the O_SYNC flag for aysnchronous direct I/O requests Darrick J. Wong
2012-11-20 10:07 ` Jan Kara
2012-11-20 20:02 ` Jeff Moyer
2012-11-21 0:56 ` Jan Kara [this message]
2012-11-21 14:09 ` Jeff Moyer
2012-11-21 16:54 ` Jan Kara
2012-11-20 7:41 ` [PATCH 3/9] xfs: factor out everything but the filemap_write_and_wait from xfs_file_fsync Darrick J. Wong
2012-11-20 10:47 ` Dave Chinner
2012-11-21 10:09 ` Christoph Hellwig
2012-11-21 14:10 ` Jeff Moyer
2012-11-20 7:51 ` [PATCH 7/9] ocfs2: Use generic handlers of O_SYNC AIO DIO Darrick J. Wong
2012-11-21 19:32 ` Joel Becker
2012-11-20 7:51 ` [PATCH 5/9] btrfs: " Darrick J. Wong
2012-11-20 7:51 ` [PATCH 6/9] gfs2: " Darrick J. Wong
2012-11-20 7:51 ` [PATCH 4/9] xfs: honor the O_SYNC flag for aysnchronous direct I/O requests Darrick J. Wong
2012-11-20 10:24 ` Jan Kara
2012-11-20 11:20 ` Dave Chinner
2012-11-20 19:42 ` Jeff Moyer
2012-11-20 20:08 ` Dave Chinner
2012-11-20 7:51 ` [PATCH 8/9] filemap: don't call generic_write_sync for -EIOCBQUEUED Darrick J. Wong
2012-11-20 7:51 ` [PATCH 9/9] blkdev: Fix up AIO+DIO+O_SYNC to do the sync part correctly Darrick J. Wong
2012-11-20 10:15 ` Jan Kara
2012-11-20 20:47 ` Jeff Moyer
2012-11-21 0:57 ` Jan Kara
2012-11-20 8:38 ` [PATCH v1 0/9] fs: fix up AIO+DIO+O_SYNC to actually do the sync part Darrick J. Wong
2012-11-20 14:23 ` Jeff Moyer
2012-11-20 18:57 ` Darrick J. Wong
2012-11-20 19:05 ` Jeff Moyer
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=20121121005626.GC10507@quack.suse.cz \
--to=jack@suse.cz \
--cc=axboe@kernel.dk \
--cc=bpm@sgi.com \
--cc=darrick.wong@oracle.com \
--cc=david@fromorbit.com \
--cc=hch@infradead.org \
--cc=jmoyer@redhat.com \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=tytso@mit.edu \
--cc=viro@zeniv.linux.org.uk \
--cc=xfs@oss.sgi.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).