From: Jeff Layton <jlayton@redhat.com>
To: Alan Cox <gnomes@lxorguk.ukuu.org.uk>
Cc: 焦晓冬 <milestonejxd@gmail.com>,
linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
"Rogier Wolff" <R.E.Wolff@BitWizard.nl>
Subject: Re: POSIX violation by writeback error
Date: Tue, 25 Sep 2018 07:15:34 -0400 [thread overview]
Message-ID: <0662a4c5d2e164d651a6a116d06da380f317100f.camel@redhat.com> (raw)
In-Reply-To: <20180925003044.239531c7@alans-desktop>
On Tue, 2018-09-25 at 00:30 +0100, Alan Cox wrote:
> > write()
> > kernel attempts to write back page and fails
> > page is marked clean and evicted from the cache
> > read()
> >
> > Now your write is gone and there were no calls between the write and
> > read.
> >
> > The question we still need to answer is this:
> >
> > When we attempt to write back some data from the cache and that fails,
> > what should happen to the dirty pages?
>
> Why do you care about the content of the pages at that point. The only
> options are to use the data (todays model), or to report that you are on
> fire.
>
The data itself doesn't matter much. What does matter is consistent
behavior in the face of such an error. The issue (IMO) is that
currently, the result of a read that takes place after a write but
before an fsync is indeterminate.
If writeback succeeded (or hasn't been done yet) you'll get back the
data you wrote, but if there was a writeback error you may or may not.
The behavior in that case mostly depends on the whim of the filesystem
developer, and they all behave somewhat differently.
> If you are going to error you don't need to use the data so you could in
> fact compress dramatically the amount of stuff you need to save
> somewhere. You need the page information so you can realize what page
> this is, but you can point the data into oblivion somewhere because you
> are no longer going to give it to anyone (assuming you can successfully
> force unmap it from everyone once it's not locked by a DMA or similar).
>
> In the real world though it's fairly unusual to just lose a bit of I/O.
> Flash devices in particular have a nasty tendancy to simply go *poof* and
> the first you know about an I/O error is the last data the drive ever
> gives you short of jtag. NFS is an exception and NFS soft timeouts are
> nasty.
>
Linux has dozens of filesystems and they all behave differently in this
regard. A catastrophic failure (paradoxically) makes things simpler for
the fs developer, but even on local filesystems isolated errors can
occur. It's also not just NFS -- what mostly started me down this road
was working on ENOSPC handling for CephFS.
I think it'd be good to at least establish a "gold standard" for what
filesystems ought to do in this situation. We might not be able to
achieve that in all cases, but we could then document the exceptions.
--
Jeff Layton <jlayton@redhat.com>
next prev parent reply other threads:[~2018-09-25 17:22 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <CAJDTihx2yaR-_-9Ks1PoFcrKNZgUOoLdN-wRTTMV76Jg_dCLrw@mail.gmail.com>
2018-09-04 10:56 ` POSIX violation by writeback error Jeff Layton
2018-09-24 23:30 ` Alan Cox
2018-09-25 11:15 ` Jeff Layton [this message]
2018-09-25 15:46 ` Theodore Y. Ts'o
2018-09-25 16:17 ` Rogier Wolff
2018-09-25 16:39 ` Alan Cox
2018-09-25 16:41 ` Jeff Layton
2018-09-25 22:30 ` Theodore Y. Ts'o
2018-09-26 18:10 ` Alan Cox
2018-09-26 21:49 ` Theodore Y. Ts'o
2018-09-27 22:48 ` Alan Cox
2018-09-27 7:18 ` Rogier Wolff
2018-09-27 12:43 ` Jeff Layton
2018-09-27 14:27 ` Theodore Y. Ts'o
2018-09-25 17:35 ` Adam Borowski
2018-09-25 22:46 ` Theodore Y. Ts'o
2018-09-04 6:32 焦晓冬
2018-09-04 7:53 ` Rogier Wolff
2018-09-04 8:58 ` 焦晓冬
2018-09-04 9:29 ` Rogier Wolff
2018-09-04 10:45 ` 焦晓冬
2018-09-04 11:09 ` Jeff Layton
2018-09-04 14:56 ` 焦晓冬
2018-09-04 15:44 ` Jeff Layton
2018-09-04 16:12 ` J. Bruce Fields
2018-09-04 16:23 ` Rogier Wolff
2018-09-04 18:54 ` J. Bruce Fields
2018-09-04 20:18 ` Jeff Layton
2018-09-04 20:35 ` Vito Caputo
2018-09-04 21:02 ` Matthew Wilcox
2018-09-05 0:51 ` Dave Chinner
2018-09-05 8:24 ` 焦晓冬
2018-09-05 10:55 ` Jeff Layton
2018-09-05 12:07 ` Rogier Wolff
2018-09-06 2:57 ` Dave Chinner
2018-09-06 9:17 ` Rogier Wolff
2018-09-24 23:09 ` Alan Cox
2018-09-05 13:53 ` J. Bruce Fields
2018-09-05 7:08 ` Rogier Wolff
2018-09-05 7:39 ` Martin Steigerwald
2018-09-05 8:04 ` Rogier Wolff
2018-09-05 8:37 ` 焦晓冬
2018-09-05 12:07 ` Austin S. Hemmelgarn
2018-09-05 12:46 ` Rogier Wolff
2018-09-05 9:32 ` Martin Steigerwald
2018-09-05 7:37 ` Martin Steigerwald
2018-09-05 11:42 ` Jeff Layton
2018-09-05 8:09 ` 焦晓冬
2018-09-05 13:08 ` Theodore Y. Ts'o
2018-09-24 23:21 ` Alan Cox
2018-09-06 7:28 ` 焦晓冬
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=0662a4c5d2e164d651a6a116d06da380f317100f.camel@redhat.com \
--to=jlayton@redhat.com \
--cc=R.E.Wolff@BitWizard.nl \
--cc=gnomes@lxorguk.ukuu.org.uk \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=milestonejxd@gmail.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).