linux-kernel - Re: [PATCH v3 1/3] lockdep: Make LOCKDEP_CROSSRELEASE configs all part of PROVE

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20170829155205.GA17290@redhat.com>
Date:   Tue, 29 Aug 2017 17:52:05 +0200
From:   Oleg Nesterov <oleg@...hat.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Byungchul Park <byungchul.park@....com>, mingo@...nel.org,
        linux-kernel@...r.kernel.org, kernel-team@....com,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Dave Chinner <david@...morbit.com>, Tejun Heo <tj@...nel.org>,
        johannes@...solutions.net
Subject: Re: [PATCH v3 1/3] lockdep: Make LOCKDEP_CROSSRELEASE configs all
 part of PROVE_LOCKING

Peter, sorry for delay, didn't have a chance to return to this discussion...

On 08/23, Peter Zijlstra wrote:
>
> > > It was added by Oleg in commit:
> > >
> > >   a67da70dc095 ("workqueues: lockdep annotations for flush_work()")
> >
> > No, these annotations were moved later into start_flush, iiuc...
> >
> > This
> >
> > 	lock_map_acquire(&work->lockdep_map);
> > 	lock_map_release(&work->lockdep_map);
> >
> > was added by another commit 0976dfc1d0cd80a4e9dfaf87bd8744612bde475a
> > "workqueue: Catch more locking problems with flush_work()", and at
> > first glance it is fine.
>
> Those are fine and are indeed the flush_work() vs work inversion.
>
> The two straight forward annotations are:
>
> flush_work(work)	process_one_work(wq, work)
>   A(work)		  A(work)
>   R(work)		  work->func(work);
> 			  R(work)
>
> Which catches:
>
> Task-1:			work:
>
>   mutex_lock(&A);	mutex_lock(&A);
>   flush_work(work);

Yes, yes, this is clear.

But if we ignore the multithreaded workqueues, in this particular case
we could rely on A(wq)/R(wq) in start_flush() and process_one_work().

The problem is that start_flush_work() does not do acquire/release
unconditionally, it does this only if it is going to wait, and I am not
sure this is right...

Plus process_one_work() does lock_map_acquire_read(), I don't really
understand this too.

> And the analogous:
>
> flush_workqueue(wq)	process_one_work(wq, work)
>   A(wq)			  A(wq)
>   R(wq)			  work->func(work);
> 			  R(wq)
>
>
> The thing I puzzled over was flush_work() (really start_flush_work())
> doing:
>
>         if (pwq->wq->saved_max_active == 1 || pwq->wq->rescuer)
>                 lock_map_acquire(&pwq->wq->lockdep_map);
>         else
>                 lock_map_acquire_read(&pwq->wq->lockdep_map);
>         lock_map_release(&pwq->wq->lockdep_map);
>
> Why does flush_work() care about the wq->lockdep_map?
>
> The answer is because, for single-threaded workqueues, doing
> flush_work() from a work is a potential deadlock:

Yes, but the simple answer is that flush_work() doesn't really differ
from flush_workqueue() in this respect?

If nothing else, if some WORK is the last queued work on WQ, then
flush_work(WORK) is the same thing as flush_workqueuw(WQ), more or less.
Again, I am talking about single-threaded workqueues.

> workqueue-thread:
>
> 	work-n:
> 	  flush_work(work-n+1);
>
> 	work-n+1:
>
>
> Will not be going anywhere fast..

Or another example,

	lock(LOCK);
	flush_work(WORK);
	unlock(LOCK);

	workqueue-thread:
		another_pending_work:
			LOCK(LOCK);
			UNLOCK(LOCK);

		WORK:

In this case we do not care about WORK->lockdep_map, but
taking the wq->lockdep_map from flush_work() (if single-threaded) allows
to report the deadlock.

Again, this is just like flush_workqueue().

Oleg.