From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: Matthew Bobrowski <mbobrowski@mbobrowski.org>
Cc: Dave Chinner <david@fromorbit.com>,
Christoph Hellwig <hch@lst.de>,
Damien Le Moal <Damien.LeMoal@wdc.com>,
Andreas Gruenbacher <agruenba@redhat.com>,
linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
linux-kernel@vger.kernel.org, Dave Chinner <dchinner@redhat.com>
Subject: Re: [PATCH 01/14] iomap: iomap that extends beyond EOF should be marked dirty
Date: Thu, 17 Oct 2019 18:02:20 -0700 [thread overview]
Message-ID: <20191018010220.GR13108@magnolia> (raw)
In-Reply-To: <20191017230814.GB31874@bobrowski>
On Fri, Oct 18, 2019 at 10:08:14AM +1100, Matthew Bobrowski wrote:
> On Fri, Oct 18, 2019 at 08:56:13AM +1100, Dave Chinner wrote:
> > On Thu, Oct 17, 2019 at 11:39:17AM -0700, Darrick J. Wong wrote:
> > > On Thu, Oct 17, 2019 at 07:56:11PM +0200, Christoph Hellwig wrote:
> > > > From: Dave Chinner <dchinner@redhat.com>
> > > >
> > > > When doing a direct IO that spans the current EOF, and there are
> > > > written blocks beyond EOF that extend beyond the current write, the
> > > > only metadata update that needs to be done is a file size extension.
> > > >
> > > > However, we don't mark such iomaps as IOMAP_F_DIRTY to indicate that
> > > > there is IO completion metadata updates required, and hence we may
> > > > fail to correctly sync file size extensions made in IO completion
> > > > when O_DSYNC writes are being used and the hardware supports FUA.
> > > >
> > > > Hence when setting IOMAP_F_DIRTY, we need to also take into account
> > > > whether the iomap spans the current EOF. If it does, then we need to
> > > > mark it dirty so that IO completion will call generic_write_sync()
> > > > to flush the inode size update to stable storage correctly.
> > > >
> > > > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> > > > Signed-off-by: Christoph Hellwig <hch@lst.de>
> > >
> > > Looks ok, but need fixes tag. Also, might it be wise to split off the
> > > ext4 section into a separate patch so that it can be backported
> > > separately?
> >
> > I 've done a bit more digging on this, and the ext4 part is not
> > needed for DAX as IOMAP_F_DIRTY is only used in the page fault path
> > and hence can't change the file size. As such, this only affects
> > direct IO. Hence the ext4 hunk can be added to the ext4 iomap-dio
> > patchset that is being developed rather than being in this patch.
>
> Noted, thanks Dave. I've incorporated the ext4 related change into my patch
> series.
Ok, I've dropped the ext4 hunk from my branch.
--D
> --<M>--
next prev parent reply other threads:[~2019-10-18 1:02 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-10-17 17:56 lift the xfs writepage code into iomap v8 Christoph Hellwig
2019-10-17 17:56 ` [PATCH 01/14] iomap: iomap that extends beyond EOF should be marked dirty Christoph Hellwig
2019-10-17 18:39 ` Darrick J. Wong
2019-10-17 21:56 ` Dave Chinner
2019-10-17 23:08 ` Matthew Bobrowski
2019-10-18 1:02 ` Darrick J. Wong [this message]
2019-10-18 7:20 ` Christoph Hellwig
2019-10-17 17:56 ` [PATCH 02/14] xfs: initialize iomap->flags in xfs_bmbt_to_iomap Christoph Hellwig
2019-10-17 17:56 ` [PATCH 03/14] xfs: set IOMAP_F_NEW more carefully Christoph Hellwig
2019-10-17 17:56 ` [PATCH 04/14] xfs: use a struct iomap in xfs_writepage_ctx Christoph Hellwig
2019-10-17 17:56 ` [PATCH 05/14] xfs: refactor the ioend merging code Christoph Hellwig
2019-10-17 18:40 ` Darrick J. Wong
2019-10-17 17:56 ` [PATCH 06/14] xfs: turn io_append_trans into an io_private void pointer Christoph Hellwig
2019-10-17 17:56 ` [PATCH 07/14] xfs: remove the fork fields in the writepage_ctx and ioend Christoph Hellwig
2019-10-17 17:56 ` [PATCH 08/14] iomap: zero newly allocated mapped blocks Christoph Hellwig
2019-10-17 17:56 ` [PATCH 09/14] iomap: lift common tracing code from xfs to iomap Christoph Hellwig
2019-10-17 17:56 ` [PATCH 10/14] iomap: lift the xfs writeback code " Christoph Hellwig
2019-10-17 18:48 ` Darrick J. Wong
2019-10-17 22:01 ` Dave Chinner
2019-10-17 17:56 ` [PATCH 11/14] iomap: warn on inline maps in iomap_writepage_map Christoph Hellwig
2019-10-17 17:56 ` [PATCH 12/14] iomap: move struct iomap_page out of iomap.h Christoph Hellwig
2019-10-17 17:56 ` [PATCH 13/14] iomap: cleanup iomap_ioend_compare Christoph Hellwig
2019-10-17 17:56 ` [PATCH 14/14] iomap: pass a struct page to iomap_finish_page_writeback Christoph Hellwig
2019-10-17 18:41 ` Darrick J. Wong
2019-10-17 20:49 ` lift the xfs writepage code into iomap v8 Darrick J. Wong
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=20191018010220.GR13108@magnolia \
--to=darrick.wong@oracle.com \
--cc=Damien.LeMoal@wdc.com \
--cc=agruenba@redhat.com \
--cc=david@fromorbit.com \
--cc=dchinner@redhat.com \
--cc=hch@lst.de \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=mbobrowski@mbobrowski.org \
/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).