linux-fsdevel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Theodore Ts'o <tytso@mit.edu>
To: Christoph Hellwig <hch@infradead.org>
Cc: Linux Filesystem Development List <linux-fsdevel@vger.kernel.org>,
	Ext4 Developers List <linux-ext4@vger.kernel.org>,
	Jan Kara <jack@suse.cz>,
	Linux btrfs Developers List <linux-btrfs@vger.kernel.org>,
	XFS Developers <xfs@oss.sgi.com>
Subject: Re: [PATCH-v4 1/7] vfs: split update_time() into update_time() and write_time()
Date: Tue, 2 Dec 2014 10:09:12 -0500	[thread overview]
Message-ID: <20141202150912.GA3496@thunk.org> (raw)
In-Reply-To: <20141202092033.GA29712@infradead.org>

On Tue, Dec 02, 2014 at 01:20:33AM -0800, Christoph Hellwig wrote:
> Why do you need the additional I_DIRTY flag?  A "lesser"
> __mark_inode_dirty should never override a stronger one.

Agreed, will fix.

> Otherwise this looks fine to me, except that I would split the default
> implementation into a new generic_update_time helper.

Sure, I can do that.

> > XFS doesn't have a ->dirty_time yet, but that way XFS would be able to
> > use the I_DIRTY_TIME flag to log the journal timestamps if it so
> > desires, and perhaps drop the need for it to use update_time().
> 
> We will probably always need a ->update_time to proide proper locking
> around the timestamp updates.

Couldn't you let the VFS set the inode timesstamps and then have xfs's
->dirty_time(inode, I_DIRTY_TIME) copy the timestamps to the on-disk
inode structure under the appropriate lock, or am I missing something?

> In the current from the generic lazytime might even be a loss for XFS as
> we're already really good at batching updates from multiple inodes in
> the same cluster for the in-place writeback, so I really don't want
> to just enable it without those optimizations without a lot of testing.

Fair enough; it's not surprising that this might be much more
effective as an optimization for ext4, for no other reason that
timestamp updates are so much heavyweight for us.  I suspect that it
should be a win for btrfs, though, and it should definitely be a win
for those file systems that don't use journalling at all.

    	       	       	    	      		     - Ted

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

  reply	other threads:[~2014-12-02 15:09 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-11-26 10:23 [PATCH-v4 0/7] add support for a lazytime mount option Theodore Ts'o
2014-11-26 10:23 ` [PATCH-v4 1/7] vfs: split update_time() into update_time() and write_time() Theodore Ts'o
2014-11-26 19:23   ` Christoph Hellwig
2014-11-27 12:34     ` Jan Kara
2014-11-27 15:25       ` Christoph Hellwig
2014-11-27 14:41     ` Theodore Ts'o
2014-11-27 15:28       ` Christoph Hellwig
2014-11-27 15:33       ` Theodore Ts'o
2014-11-27 16:49         ` Christoph Hellwig
2014-11-27 20:27           ` Theodore Ts'o
2014-12-01  9:28             ` Christoph Hellwig
2014-12-01 15:04               ` Theodore Ts'o
2014-12-01 17:18                 ` David Sterba
2014-12-02  9:20                 ` Christoph Hellwig
2014-12-02 15:09                   ` Theodore Ts'o [this message]
2014-11-26 10:23 ` [PATCH-v4 2/7] vfs: add support for a lazytime mount option Theodore Ts'o
2014-11-27 13:14   ` Jan Kara
2014-11-27 20:19     ` Theodore Ts'o
2014-11-28 12:41       ` Jan Kara
2014-11-27 23:00     ` Theodore Ts'o
2014-11-28  5:36       ` Theodore Ts'o
2014-11-28 16:24       ` Jan Kara
2014-11-26 10:23 ` [PATCH-v4 3/7] vfs: don't let the dirty time inodes get more than a day stale Theodore Ts'o
2014-11-26 10:23 ` [PATCH-v4 4/7] vfs: add lazytime tracepoints for better debugging Theodore Ts'o
2014-11-26 10:23 ` [PATCH-v4 5/7] vfs: add find_active_inode_nowait() function Theodore Ts'o
2014-11-26 10:23 ` [PATCH-v4 6/7] ext4: add support for a lazytime mount option Theodore Ts'o
2014-11-26 19:24   ` Christoph Hellwig
2014-11-26 22:48   ` Dave Chinner
2014-11-26 23:10     ` Andreas Dilger
2014-11-26 23:35       ` Dave Chinner
2014-11-27 13:27         ` Jan Kara
2014-11-27 13:32           ` Jan Kara
2014-11-27 15:25             ` Theodore Ts'o
2014-11-27 15:41               ` Jan Kara
2014-11-27 20:13                 ` Theodore Ts'o
2014-11-26 10:23 ` [PATCH-v4 7/7] btrfs: add an is_readonly() so btrfs can use common code for update_time() Theodore Ts'o

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=20141202150912.GA3496@thunk.org \
    --to=tytso@mit.edu \
    --cc=hch@infradead.org \
    --cc=jack@suse.cz \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=linux-ext4@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --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).