All of lore.kernel.org
 help / color / mirror / Atom feed
From: Marc Lehmann <schmorp@schmorp.de>
To: Chao Yu <chao2.yu@samsung.com>
Cc: linux-f2fs-devel@lists.sourceforge.net
Subject: Re: f2fs stability problems keep me from testing
Date: Thu, 19 Nov 2015 22:01:30 +0100	[thread overview]
Message-ID: <20151119210130.GC2499@schmorp.de> (raw)
In-Reply-To: <20151119205614.GA10406@jaegeuk.local> <00a101d122a0$fb9e6ab0$f2db4010$@samsung.com>

On Thu, Nov 19, 2015 at 04:04:21PM +0800, Chao Yu <chao2.yu@samsung.com> wrote:
> >    openat(AT_FDCWD, "/cold1", O_RDONLY|O_NONBLOCK|O_DIRECTORY|O_CLOEXEC) = 5
> >    fchdir(5)                               = 0
> >    getdents(5, /* 0 entries */, 32768)     = 0
> 
> I found an issue here.
> 
> If the dir entry of '/cold1/var' is removed from dentry page in '/cold1', at
> least we should return entries of '.' and '..' for the getdents invoking, but
> we didn't. So here it looks like we keep silent about some kind of error (i.e.
> ENOMEM/EIO...) when grabing&updating the dentry page of '/cold1', obviously
> it's better to report such error to user rather than ignoring it.

As for EIO, I would expect the kernel to log something. As for ENOMEM,
unless f2fs tries higher order or zone-specific allocations, that's
unlikely because the box had over 20GB memory free (according to top).

As for other errors, sure, could be.

Would that explain the sync lockup though?

> I'd like to send a patch to fix this issue, could you include the following
> patch in your f2fs module, so when it reproduces we can catch more details
> about this problem.

I can, but I guess you forgot to attach the patch.

On Fri, Nov 20, 2015 at 04:56:14AM +0800, Jaegeuk Kim <jaegeuk@kernel.org> wrote:
> > Ah, sorry if that might not have been clear - after a reboot, fsck showed
> > no errors and all files and directories (apparently) were there, i.e. it
> > looks like a transient problem, not a problem of the on-disk structure.
> 
> So, it seems that the system suffers from memory pressure significantly, which
> can incur ENOMEM, as I guess.
> Can you catch /sys/kernel/debug/f2fs/status at that moment?

If I can, I will - I didn't attempt to stress the filesystem since then, but
will presdumably start doing tests next week again.

-- 
                The choice of a       Deliantra, the free code+content MORPG
      -----==-     _GNU_              http://www.deliantra.net
      ----==-- _       generation
      ---==---(_)__  __ ____  __      Marc Lehmann
      --==---/ / _ \/ // /\ \/ /      schmorp@schmorp.de
      -=====/_/_//_/\_,_/ /_/\_\

------------------------------------------------------------------------------

  reply	other threads:[~2015-11-19 21:01 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-11-17 17:24 f2fs stability problems keep me from testing Marc Lehmann
2015-11-18 10:00 ` Chao Yu
2015-11-19  0:38   ` Marc Lehmann
2015-11-19  1:29     ` Chao Yu
2015-11-19  2:23       ` Marc Lehmann
2015-11-19 20:56         ` Jaegeuk Kim
2015-11-19  1:42   ` Marc Lehmann
2015-11-19  8:04 ` Chao Yu
2015-11-19 21:01   ` Marc Lehmann [this message]
2015-11-20  1:48     ` Chao Yu

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=20151119210130.GC2499@schmorp.de \
    --to=schmorp@schmorp.de \
    --cc=chao2.yu@samsung.com \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.