Control: tag -1 patch moreinfo On Tue, 2016-10-18 at 11:19 +0200, Lucas Nussbaum wrote: > Package: src:linux > Version: 3.16.36-1+deb8u1 > Severity: normal > > Hi, > > When booting, on about 5% of boots, the system hangs for several minutes > while waiting for systemd-udev-settle to complete. (systemd-udev-settle > is triggered by lvm2) [...] > It happens on various machines, of various models (all Dell, but I'm not sure > this is relevant as all our recent machines are Dell machines). A hardware > issue is unlikely.
I've never seen this and I don't recall any similar reports, so I suspect it may be triggered by particular drivers or configurations. Anyway: > It is fixed in stretch and unstable. > > I bisected it, and found that commit 6f942a1f264e875c5f3ad6f505d7b500a3e7fa82 > fixed it. That commit is: > > commit 6f942a1f264e875c5f3ad6f505d7b500a3e7fa82 > > Author: Peter Zijlstra <pet...@infradead.org> > Date: Wed Sep 24 10:18:46 2014 +0200 > > locking/mutex: Don't assume TASK_RUNNING [...] > Unfortunately, the code around this was changed after 3.16, making a backport > non-trivial. [...] It seems simple enough... can you try the attached patch? Ben. -- Ben Hutchings No political challenge can be met by shopping. - George Monbiot
From: Peter Zijlstra <pet...@infradead.org> Date: Wed, 24 Sep 2014 10:18:46 +0200 Subject: locking/mutex: Don't assume TASK_RUNNING Origin: https://git.kernel.org/linus/6f942a1f264e875c5f3ad6f505d7b500a3e7fa82 Bug-Debian: https://bugs.debian.org/841171 We're going to make might_sleep() test for TASK_RUNNING, because blocking without TASK_RUNNING will destroy the task state by setting it to TASK_RUNNING. There are a few occasions where its 'valid' to call blocking primitives (and mutex_lock in particular) and not have TASK_RUNNING, typically such cases are right before we set TASK_RUNNING anyhow. Robustify the code by not assuming this; this has the beneficial side effect of allowing optional code emission for fixing the above might_sleep() false positives. Signed-off-by: Peter Zijlstra (Intel) <pet...@infradead.org> Cc: t...@linutronix.de Cc: ilya.dryo...@inktank.com Cc: umgwanakikb...@gmail.com Cc: Oleg Nesterov <o...@redhat.com> Cc: Linus Torvalds <torva...@linux-foundation.org> Link: http://lkml.kernel.org/r/20140924082241.988560...@infradead.org Signed-off-by: Ingo Molnar <mi...@kernel.org> [bwh: Backported to 3.16: adjust context] --- kernel/locking/mutex.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) --- a/kernel/locking/mutex.c +++ b/kernel/locking/mutex.c @@ -480,8 +480,14 @@ static bool mutex_optimistic_spin(struct mutex *lock, * reschedule now, before we try-lock the mutex. This avoids getting * scheduled out right after we obtained the mutex. */ - if (need_resched()) + if (need_resched()) { + /* + * We _should_ have TASK_RUNNING here, but just in case + * we do not, make it so, otherwise we might get stuck. + */ + __set_current_state(TASK_RUNNING); schedule_preempt_disabled(); + } #endif spin_lock_mutex(&lock->wait_lock, flags);
signature.asc
Description: This is a digitally signed message part
