From: "Darrick J. Wong" <djwong@kernel.org>
To: Miklos Szeredi <miklos@szeredi.hu>
Cc: bernd@bsbernd.com, linux-xfs@vger.kernel.org, John@groves.net,
linux-fsdevel@vger.kernel.org, neal@gompa.dev,
joannelkoong@gmail.com
Subject: Re: [PATCH 2/8] fuse: flush pending fuse events before aborting the connection
Date: Tue, 23 Sep 2025 15:34:47 -0700 [thread overview]
Message-ID: <20250923223447.GJ1587915@frogsfrogsfrogs> (raw)
In-Reply-To: <20250923205936.GI1587915@frogsfrogsfrogs>
On Tue, Sep 23, 2025 at 01:59:36PM -0700, Darrick J. Wong wrote:
> On Tue, Sep 23, 2025 at 08:56:47PM +0200, Miklos Szeredi wrote:
> > On Tue, 23 Sept 2025 at 16:54, Darrick J. Wong <djwong@kernel.org> wrote:
> >
> > > I'm not sure what you're referring to by "special" -- are you asking
> > > about why I added the touch_softlockup_watchdog() call here but not in
> > > fuse_wait_aborted()? I think it could use that treatment too, but once
> > > you abort all the pending requests they tend to go away very quickly.
> > > It might be the case that nobody's gotten a warning simply because the
> > > aborted requests all go away in under 30 seconds.
> >
> > Maybe I'm not understanding how the softlockup detector works. I
> > thought that it triggers if task is spinning in a tight loop. That
> > precludes any timeouts, since that means that the task went to sleep.
> >
> > So what's happening here?
>
> Hrm, I thought the softlockup detector also complains about tasks stuck
> in uninterruptible sleep, but you're right, it *does* schedule() so the
> softlockup detector won't complain about it.
>
> I think. Let me go try to prove that empirically. :)
Hrm. If I change the bottom of the function to:
wait_event(fc->blocked_waitq, <some false expression>);
Then I get softlockup warnings because the process state gets set to
UNINTERRUPTIBLE, schedule() is called to pick another process, and the
umount process never reaches runnable state ever again.
If instead I change it to:
while (wait_event_timeout(fc->blocked_waitq, <false expr>, HZ) == 0) {
/* empty */
}
then I do not get softlockup warnings, because the umount process
actually does get scheduled off and on the system, repeatedly.
Conclusion: The loop is necessary to avoid softlockup warnings while the
fuse requests are processed by the server, but it is not necessary to
touch the watchdog in the loop body.
Thanks for challenging me, now I've learned something useful. :)
--D
next prev parent reply other threads:[~2025-09-23 22:34 UTC|newest]
Thread overview: 126+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20250916000759.GA8080@frogsfrogsfrogs>
2025-09-16 0:18 ` [PATCHSET RFC v5 1/8] fuse: general bug fixes Darrick J. Wong
2025-09-16 0:24 ` [PATCH 1/8] fuse: fix livelock in synchronous file put from fuseblk workers Darrick J. Wong
2025-09-23 10:57 ` Miklos Szeredi
2025-09-16 0:24 ` [PATCH 2/8] fuse: flush pending fuse events before aborting the connection Darrick J. Wong
2025-09-23 11:11 ` Miklos Szeredi
2025-09-23 14:54 ` Darrick J. Wong
2025-09-23 18:56 ` Miklos Szeredi
2025-09-23 20:59 ` Darrick J. Wong
2025-09-23 22:34 ` Darrick J. Wong [this message]
2025-09-24 12:04 ` Miklos Szeredi
2025-09-24 17:50 ` Darrick J. Wong
2025-09-24 18:19 ` Miklos Szeredi
2025-09-24 20:54 ` Darrick J. Wong
2025-09-30 10:29 ` Miklos Szeredi
2025-09-30 17:56 ` Darrick J. Wong
2025-09-16 0:24 ` [PATCH 3/8] fuse: capture the unique id of fuse commands being sent Darrick J. Wong
2025-09-23 10:58 ` Miklos Szeredi
2025-09-16 0:25 ` [PATCH 4/8] fuse: signal that a fuse filesystem should exhibit local fs behaviors Darrick J. Wong
2025-09-17 17:18 ` Joanne Koong
2025-09-18 16:52 ` Darrick J. Wong
2025-09-19 9:24 ` Miklos Szeredi
2025-09-19 17:50 ` Darrick J. Wong
2025-09-23 14:57 ` Miklos Szeredi
2025-09-23 20:51 ` Darrick J. Wong
2025-09-24 13:55 ` Miklos Szeredi
2025-09-24 17:31 ` Darrick J. Wong
2025-09-25 19:17 ` Darrick J. Wong
2025-09-16 0:25 ` [PATCH 5/8] fuse: implement file attributes mask for statx Darrick J. Wong
2025-09-16 0:25 ` [PATCH 6/8] fuse: update file mode when updating acls Darrick J. Wong
2025-09-16 0:25 ` [PATCH 7/8] fuse: propagate default and file acls on creation Darrick J. Wong
2025-09-16 6:41 ` Chen Linxuan
2025-09-16 14:48 ` Darrick J. Wong
2025-09-16 0:26 ` [PATCH 8/8] fuse: enable FUSE_SYNCFS for all fuseblk servers Darrick J. Wong
2025-09-23 10:58 ` Miklos Szeredi
2025-09-16 0:18 ` [PATCHSET RFC v5 2/8] iomap: cleanups ahead of adding fuse support Darrick J. Wong
2025-09-16 0:26 ` [PATCH 1/2] iomap: trace iomap_zero_iter zeroing activities Darrick J. Wong
2025-09-16 13:49 ` Christoph Hellwig
2025-09-16 14:49 ` Darrick J. Wong
2025-09-16 0:26 ` [PATCH 2/2] iomap: error out on file IO when there is no inline_data buffer Darrick J. Wong
2025-09-16 13:50 ` Christoph Hellwig
2025-09-16 14:50 ` Darrick J. Wong
2025-09-16 0:18 ` [PATCHSET RFC v5 3/8] fuse: cleanups ahead of adding fuse support Darrick J. Wong
2025-09-16 0:26 ` [PATCH 1/5] fuse: allow synchronous FUSE_INIT Darrick J. Wong
2025-09-17 17:22 ` Joanne Koong
2025-09-18 18:04 ` Darrick J. Wong
2025-09-16 0:27 ` [PATCH 2/5] fuse: move the backing file idr and code into a new source file Darrick J. Wong
2025-09-25 14:11 ` Miklos Szeredi
2025-09-16 0:27 ` [PATCH 3/5] fuse: move the passthrough-specific code back to passthrough.c Darrick J. Wong
2025-09-17 2:47 ` Amir Goldstein
2025-09-18 18:02 ` Darrick J. Wong
2025-09-19 7:34 ` Miklos Szeredi
2025-09-19 9:36 ` Amir Goldstein
2025-09-19 17:43 ` Darrick J. Wong
2025-09-16 0:27 ` [PATCH 4/5] fuse_trace: " Darrick J. Wong
2025-09-16 0:27 ` [PATCH 5/5] fuse: move CREATE_TRACE_POINTS to a separate file Darrick J. Wong
2025-09-25 14:25 ` Miklos Szeredi
2025-09-16 0:19 ` [PATCHSET RFC v5 4/8] fuse: allow servers to use iomap for better file IO performance Darrick J. Wong
2025-09-16 0:28 ` [PATCH 01/28] fuse: implement the basic iomap mechanisms Darrick J. Wong
2025-09-19 22:36 ` Joanne Koong
2025-09-23 20:32 ` Darrick J. Wong
2025-09-23 21:24 ` Joanne Koong
2025-09-23 22:10 ` Darrick J. Wong
2025-09-23 23:08 ` Darrick J. Wong
2025-09-16 0:28 ` [PATCH 02/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:28 ` [PATCH 03/28] fuse: make debugging configurable at runtime Darrick J. Wong
2025-09-16 0:29 ` [PATCH 04/28] fuse: adapt FUSE_DEV_IOC_BACKING_{OPEN,CLOSE} to add new iomap devices Darrick J. Wong
2025-09-17 3:09 ` Amir Goldstein
2025-09-18 18:17 ` Darrick J. Wong
2025-09-18 18:42 ` Amir Goldstein
2025-09-18 19:03 ` Darrick J. Wong
2025-09-19 7:13 ` Miklos Szeredi
2025-09-19 9:54 ` Amir Goldstein
2025-09-19 17:42 ` Darrick J. Wong
2025-09-23 7:10 ` Miklos Szeredi
2025-09-16 0:29 ` [PATCH 05/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:29 ` [PATCH 06/28] fuse: flush events and send FUSE_SYNCFS and FUSE_DESTROY on unmount Darrick J. Wong
2025-09-16 0:29 ` [PATCH 07/28] fuse: create a per-inode flag for toggling iomap Darrick J. Wong
2025-09-16 0:30 ` [PATCH 08/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:30 ` [PATCH 09/28] fuse: isolate the other regular file IO paths from iomap Darrick J. Wong
2025-09-16 0:30 ` [PATCH 10/28] fuse: implement basic iomap reporting such as FIEMAP and SEEK_{DATA,HOLE} Darrick J. Wong
2025-09-16 0:30 ` [PATCH 11/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:31 ` [PATCH 12/28] fuse: implement direct IO with iomap Darrick J. Wong
2025-09-16 0:31 ` [PATCH 13/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:31 ` [PATCH 14/28] fuse: implement buffered " Darrick J. Wong
2025-09-16 0:31 ` [PATCH 15/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:32 ` [PATCH 16/28] fuse: implement large folios for iomap pagecache files Darrick J. Wong
2025-09-16 0:32 ` [PATCH 17/28] fuse: use an unrestricted backing device with iomap pagecache io Darrick J. Wong
2025-09-16 0:32 ` [PATCH 18/28] fuse: advertise support for iomap Darrick J. Wong
2025-09-16 0:32 ` [PATCH 19/28] fuse: query filesystem geometry when using iomap Darrick J. Wong
2025-09-16 0:33 ` [PATCH 20/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:33 ` [PATCH 21/28] fuse: implement fadvise for iomap files Darrick J. Wong
2025-09-16 0:33 ` [PATCH 22/28] fuse: invalidate ranges of block devices being used for iomap Darrick J. Wong
2025-09-16 0:33 ` [PATCH 23/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:34 ` [PATCH 24/28] fuse: implement inline data file IO via iomap Darrick J. Wong
2025-09-16 0:34 ` [PATCH 25/28] fuse_trace: " Darrick J. Wong
2025-09-16 0:34 ` [PATCH 26/28] fuse: allow more statx fields Darrick J. Wong
2025-09-16 0:35 ` [PATCH 27/28] fuse: support atomic writes with iomap Darrick J. Wong
2025-09-16 0:35 ` [PATCH 28/28] fuse: disable direct reclaim for any fuse server that uses iomap Darrick J. Wong
2025-09-16 0:19 ` [PATCHSET RFC v5 5/8] fuse: allow servers to specify root node id Darrick J. Wong
2025-09-16 0:35 ` [PATCH 1/3] fuse: make the root nodeid dynamic Darrick J. Wong
2025-09-16 0:35 ` [PATCH 2/3] fuse_trace: " Darrick J. Wong
2025-09-16 0:36 ` [PATCH 3/3] fuse: allow setting of root nodeid Darrick J. Wong
2025-09-16 0:19 ` [PATCHSET RFC v5 6/8] fuse: handle timestamps and ACLs correctly when iomap is enabled Darrick J. Wong
2025-09-16 0:36 ` [PATCH 1/9] fuse: enable caching of timestamps Darrick J. Wong
2025-09-16 0:36 ` [PATCH 2/9] fuse: force a ctime update after a fileattr_set call when in iomap mode Darrick J. Wong
2025-09-16 0:36 ` [PATCH 3/9] fuse: allow local filesystems to set some VFS iflags Darrick J. Wong
2025-09-16 0:37 ` [PATCH 4/9] fuse_trace: " Darrick J. Wong
2025-09-16 0:37 ` [PATCH 5/9] fuse: cache atime when in iomap mode Darrick J. Wong
2025-09-16 0:37 ` [PATCH 6/9] fuse: let the kernel handle KILL_SUID/KILL_SGID for iomap filesystems Darrick J. Wong
2025-09-16 0:37 ` [PATCH 7/9] fuse_trace: " Darrick J. Wong
2025-09-16 0:38 ` [PATCH 8/9] fuse: update ctime when updating acls on an iomap inode Darrick J. Wong
2025-09-16 0:38 ` [PATCH 9/9] fuse: always cache ACLs when using iomap Darrick J. Wong
2025-09-16 0:19 ` [PATCHSET RFC v5 7/8] fuse: cache iomap mappings for even better file IO performance Darrick J. Wong
2025-09-16 0:38 ` [PATCH 01/10] fuse: cache iomaps Darrick J. Wong
2025-09-16 0:38 ` [PATCH 02/10] fuse_trace: " Darrick J. Wong
2025-09-16 0:39 ` [PATCH 03/10] fuse: use the iomap cache for iomap_begin Darrick J. Wong
2025-09-16 0:39 ` [PATCH 04/10] fuse_trace: " Darrick J. Wong
2025-09-16 0:39 ` [PATCH 05/10] fuse: invalidate iomap cache after file updates Darrick J. Wong
2025-09-16 0:39 ` [PATCH 06/10] fuse_trace: " Darrick J. Wong
2025-09-16 0:40 ` [PATCH 07/10] fuse: enable iomap cache management Darrick J. Wong
2025-09-16 0:40 ` [PATCH 08/10] fuse_trace: " Darrick J. Wong
2025-09-16 0:40 ` [PATCH 09/10] fuse: overlay iomap inode info in struct fuse_inode Darrick J. Wong
2025-09-16 0:41 ` [PATCH 10/10] fuse: enable iomap Darrick J. Wong
2025-09-16 0:20 ` [PATCHSET RFC v5 8/8] fuse: run fuse servers as a contained service Darrick J. Wong
2025-09-16 0:41 ` [PATCH 1/2] fuse: allow privileged mount helpers to pre-approve iomap usage Darrick J. Wong
2025-09-16 0:41 ` [PATCH 2/2] fuse: set iomap backing device block size 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=20250923223447.GJ1587915@frogsfrogsfrogs \
--to=djwong@kernel.org \
--cc=John@groves.net \
--cc=bernd@bsbernd.com \
--cc=joannelkoong@gmail.com \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-xfs@vger.kernel.org \
--cc=miklos@szeredi.hu \
--cc=neal@gompa.dev \
/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