From: Goldwyn Rodrigues <rgoldwyn@suse.de>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: linux-fsdevel@vger.kernel.org, linux-btrfs@vger.kernel.org,
Johannes.Thumshirn@wdc.com, hch@infradead.org, dsterba@suse.cz,
fdmanana@gmail.com
Subject: Re: [PATCH] iomap: Return zero in case of unsuccessful pagecache invalidation before DIO
Date: Mon, 1 Jun 2020 10:16:14 -0500 [thread overview]
Message-ID: <20200601151614.pxy7in4jrvuuy7nx@fiona> (raw)
In-Reply-To: <20200529002319.GQ252930@magnolia>
On 17:23 28/05, Darrick J. Wong wrote:
> On Thu, May 28, 2020 at 02:21:03PM -0500, Goldwyn Rodrigues wrote:
> >
> > Filesystems such as btrfs are unable to guarantee page invalidation
> > because pages could be locked as a part of the extent. Return zero
>
> Locked for what? filemap_write_and_wait_range should have just cleaned
> them off.
>
> > in case a page cache invalidation is unsuccessful so filesystems can
> > fallback to buffered I/O. This is similar to
> > generic_file_direct_write().
> >
> > This takes care of the following invalidation warning during btrfs
> > mixed buffered and direct I/O using iomap_dio_rw():
> >
> > Page cache invalidation failure on direct I/O. Possible data
> > corruption due to collision with buffered I/O!
> >
> > Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
> >
> > diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
> > index e4addfc58107..215315be6233 100644
> > --- a/fs/iomap/direct-io.c
> > +++ b/fs/iomap/direct-io.c
> > @@ -483,9 +483,15 @@ iomap_dio_rw(struct kiocb *iocb, struct iov_iter *iter,
> > */
> > ret = invalidate_inode_pages2_range(mapping,
> > pos >> PAGE_SHIFT, end >> PAGE_SHIFT);
> > - if (ret)
> > - dio_warn_stale_pagecache(iocb->ki_filp);
> > - ret = 0;
> > + /*
> > + * If a page can not be invalidated, return 0 to fall back
> > + * to buffered write.
> > + */
> > + if (ret) {
> > + if (ret == -EBUSY)
> > + ret = 0;
> > + goto out_free_dio;
>
> XFS doesn't fall back to buffered io when directio fails, which means
> this will cause a regression there.
>
> Granted mixing write types is bogus...
>
I have not seen page invalidation failure errors on XFS, but what should
happen hypothetically if they do occur? Carry on with the direct I/O?
Would an error return like -ENOTBLK be better?
--
Goldwyn
next prev parent reply other threads:[~2020-06-01 15:16 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-05-28 19:21 [PATCH] iomap: Return zero in case of unsuccessful pagecache invalidation before DIO Goldwyn Rodrigues
2020-05-29 0:23 ` Darrick J. Wong
2020-05-29 10:55 ` Filipe Manana
2020-05-29 11:31 ` Matthew Wilcox
2020-05-29 11:50 ` Filipe Manana
2020-05-29 12:45 ` Goldwyn Rodrigues
2020-06-01 15:16 ` Goldwyn Rodrigues [this message]
2020-06-03 11:23 ` Filipe Manana
2020-06-03 11:32 ` Filipe Manana
2020-06-03 19:02 ` Darrick J. Wong
2020-06-03 19:10 ` Filipe Manana
2020-06-03 19:18 ` Matthew Wilcox
2020-06-03 21:07 ` Goldwyn Rodrigues
2020-06-04 13:55 ` David Sterba
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=20200601151614.pxy7in4jrvuuy7nx@fiona \
--to=rgoldwyn@suse.de \
--cc=Johannes.Thumshirn@wdc.com \
--cc=darrick.wong@oracle.com \
--cc=dsterba@suse.cz \
--cc=fdmanana@gmail.com \
--cc=hch@infradead.org \
--cc=linux-btrfs@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.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).