From: Matthew Wilcox <willy@infradead.org>
To: Pavel Machek <pavel@ucw.cz>,
"Rafael J. Wysocki" <rafael@kernel.org>,
jikos@suse.cz, raven@themaw.net, akpm@linux-foundation.org,
sfr@canb.auug.org.au, linux-next@vger.kernel.org,
linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org,
"Rafael J. Wysocki" <rjw@rjwysocki.net>
Subject: Re: update-binfmts breaking suspend
Date: Tue, 10 Apr 2018 21:30:52 -0700 [thread overview]
Message-ID: <20180411043052.GA15171@bombadil.infradead.org> (raw)
In-Reply-To: <20180406144355.GA20605@bombadil.infradead.org>
Ping?
On Fri, Apr 06, 2018 at 07:43:55AM -0700, Matthew Wilcox wrote:
> Pavel Machek <pavel@ucw.cz> wrote:
> > > Failure is not a hang, as they expect, but... machine locks up, but
> > > does not suspend, and then continues running after a delay..
> > >
> > > [ 35.038766] PM: Syncing filesystems ... done.
> > > [ 35.051246] Freezing user space processes ...
> > > [ 55.060528] Freezing of tasks failed after 20.009 seconds (1 tasks
> > > refusing to freeze, wq_busy
> > > =0):
> > > [ 55.060552] update-binfmts D 0 2727 1 0x80000004
> > > [ 55.060576] Call Trace:
> > > [ 55.060600] __schedule+0x37a/0x7e0
> > > [ 55.060618] schedule+0x29/0x70
> > > [ 55.060635] autofs4_wait+0x359/0x7a0
> > > [ 55.060653] ? wait_woken+0x70/0x70
> > > [ 55.060668] autofs4_mount_wait+0x4a/0xe0
> > > [ 55.060684] ? autofs4_mount_wait+0x4a/0xe0
> > > [ 55.060699] autofs4_d_automount+0xe0/0x200
> > > [ 55.060715] ? autofs4_d_automount+0xe0/0x200
> > >
> > > Did the rework of freezing start already in -next?
> >
> > Hmm, so I did git bisect, and it pointed to:
> >
> > commit 7cb03edf112fea6ead2fcd3c5fd639756d6d114b
> > Author: Matthew Wilcox <mawilcox@microsoft.com>
> > Date: Thu Mar 29 10:15:17 2018 +1100
> >
> > autofs4: use wait_event_killable
> >
> > This playing with signals to allow only fatal signals appears to
> > predate
> > the introduction of wait_event_killable(), and I'm fairly sure
> > that
> > wait_event_killable is what was meant to happen here.
>
> Umm. I'm not familiar with the freezer. Help me out here ...
>
> I see the message coming from here:
>
> pr_err("Freezing of tasks %s after %d.%03d seconds "
> "(%d tasks refusing to freeze, wq_busy=%d):\n",
> wakeup ? "aborted" : "failed",
> elapsed_msecs / 1000, elapsed_msecs % 1000,
> todo - wq_busy, wq_busy);
>
> and then backtracking in that function, I see this:
>
> for_each_process_thread(g, p) {
> if (p == current || !freeze_task(p))
> continue;
>
> in freeze_task(), I see this:
>
> if (!(p->flags & PF_KTHREAD))
> fake_signal_wake_up(p);
> else
> wake_up_state(p, TASK_INTERRUPTIBLE);
>
> which does this:
>
> if (lock_task_sighand(p, &flags)) {
> signal_wake_up(p, 0);
> unlock_task_sighand(p, &flags);
> }
>
> which does this:
>
> static inline void signal_wake_up(struct task_struct *t, bool resume)
> {
> signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
> }
>
> which does this:
>
> void signal_wake_up_state(struct task_struct *t, unsigned int state)
> {
> set_tsk_thread_flag(t, TIF_SIGPENDING);
> /*
> * TASK_WAKEKILL also means wake it up in the stopped/traced/killable
> * case. We don't check t->state here because there is a race with it
> * executing another processor and just now entering stopped state.
> * By using wake_up_state, we ensure the process will wake up and
> * handle its death signal.
> */
> if (!wake_up_state(t, state | TASK_INTERRUPTIBLE))
> kick_process(t);
> }
>
> Now I don't know why we only wake interruptible tasks here and not killable
> tasks. I've trawled git history all the way back to 2.6.12-rc2, and the
> reasoning behind signal_wake_up() (as it originally was) is lost to pre-git
> history.
>
> So ... why do we only wake interruptible tasks on suspend? Why not wake
> uninterruptible tasks too?
>
> if (lock_task_sighand(p, &flags)) {
> - signal_wake_up(p, 0);
> + signal_wake_up_state(p, TASK_WAKEKILL);
> unlock_task_sighand(p, &flags);
> }
>
> or why do we consider tasks waiting uninterruptibly to block freezing?
> Is it because they're (probably) waiting for I/O and we want the I/O
> to complete?
next prev parent reply other threads:[~2018-04-11 4:30 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20180404165559.4cd0c12c@canb.auug.org.au>
2018-04-04 7:50 ` x32 suspend failuer in Re: linux-next: Tree for Apr 4 Pavel Machek
2018-04-04 7:58 ` Rafael J. Wysocki
2018-04-04 8:49 ` Pavel Machek
2018-04-05 12:25 ` update-binfmts breaking suspend was " Pavel Machek
2018-04-05 20:30 ` Pavel Machek
2018-04-05 22:27 ` Rafael J. Wysocki
[not found] ` <SN4PR2101MB073673B12998428D48DA62E5CBBA0@SN4PR2101MB0736.namprd21.prod.outlook.com>
[not found] ` <20180406144355.GA20605@bombadil.infradead.org>
2018-04-11 4:30 ` Matthew Wilcox [this message]
2018-04-11 6:09 ` update-binfmts breaking suspend Pavel Machek
2018-04-06 22:41 ` x32 suspend failuer in Re: linux-next: Tree for Apr 4 Pavel Machek
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=20180411043052.GA15171@bombadil.infradead.org \
--to=willy@infradead.org \
--cc=akpm@linux-foundation.org \
--cc=jikos@suse.cz \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-next@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=pavel@ucw.cz \
--cc=rafael@kernel.org \
--cc=raven@themaw.net \
--cc=rjw@rjwysocki.net \
--cc=sfr@canb.auug.org.au \
/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).