From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>,
Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@elte.hu>,
"H. Peter Anvin" <hpa@zytor.com>,
Linux-Next Mailing List <linux-next@vger.kernel.org>,
Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: linux-next: build failure after merge of the rcu tree
Date: Fri, 11 Aug 2017 13:12:47 -0700 [thread overview]
Message-ID: <20170811201247.GH3730@linux.vnet.ibm.com> (raw)
In-Reply-To: <20170811144150.26gowhxte7ri5fpk@hirez.programming.kicks-ass.net>
On Fri, Aug 11, 2017 at 04:41:50PM +0200, Peter Zijlstra wrote:
> On Fri, Aug 11, 2017 at 11:14:34AM +0200, Peter Zijlstra wrote:
> > On Thu, Aug 10, 2017 at 09:54:53PM -0700, Paul E. McKenney wrote:
> > > On Fri, Aug 11, 2017 at 02:43:52PM +1000, Stephen Rothwell wrote:
> > >
> > > Looks like I need to rebase my patch on top of a9668cd6ee28, and
> > > than put an smp_mb__after_spinlock() between the lock and the unlock.
> > >
> > > Peter, any objections to that approach? Other suggestions?
> >
> > Hurm.. I'll have to try and understand that comment there again it
> > seems.
>
> OK, so per commit b5740f4b2cb3 ("sched: Fix ancient race in do_exit()")
> the race is with try_to_wake_up():
>
> down_read()
> p->state = TASK_UNINTERRUPTIBLE;
>
> try_to_wake_up(p)
> spin_lock(p->pi_lock);
> /* sees TASK_UNINTERRUPTIBLE */
> ttwu_remote()
> /* check stuff, no need to schedule() */
> p->state = TASK_RUNNING
>
>
> p->state = TASK_DEAD
>
> p->state = TASK_RUNNING /* whoops! */
> spin_unlock(p->pi_lock);
>
> __schedule(false);
> BUG();
>
>
>
>
> So given that, I think that:
>
> spin_lock(¤t->pi_lock);
> spin_unlock(¤t->pi_lock);
>
> current->state = TASK_DEAD;
>
> is sufficient. I don't see a need for an additional smp_mb here.
>
> Either the concurrent ttwu is finished and we must observe its RUNNING
> store, or it will observe our RUNNING store.
Makes sense to me! Please see below for the updated commit.
Thanx, Paul
------------------------------------------------------------------------
commit 23a9b748a3d27f67cdb078fcb891a920285e75d9
Author: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Date: Thu Jun 29 12:08:26 2017 -0700
sched: Replace spin_unlock_wait() with lock/unlock pair
There is no agreed-upon definition of spin_unlock_wait()'s semantics,
and it appears that all callers could do just as well with a lock/unlock
pair. This commit therefore replaces the spin_unlock_wait() call in
do_task_dead() with spin_lock() followed immediately by spin_unlock().
This should be safe from a performance perspective because the lock is
this tasks ->pi_lock, and this is called only after the task exits.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Andrea Parri <parri.andrea@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
[ paulmck: Drop smp_mb() based on Peter Zijlstra's analysis:
http://lkml.kernel.org/r/20170811144150.26gowhxte7ri5fpk@hirez.programming.kicks-ass.net ]
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 17c667b427b4..5d22323ae099 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3352,8 +3352,8 @@ void __noreturn do_task_dead(void)
* To avoid it, we have to wait for releasing tsk->pi_lock which
* is held by try_to_wake_up()
*/
- smp_mb();
- raw_spin_unlock_wait(¤t->pi_lock);
+ raw_spin_lock_irq(¤t->pi_lock);
+ raw_spin_unlock_irq(¤t->pi_lock);
/* Causes final put_task_struct in finish_task_switch(): */
__set_current_state(TASK_DEAD);
next prev parent reply other threads:[~2017-08-11 20:12 UTC|newest]
Thread overview: 152+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-11 4:43 linux-next: build failure after merge of the rcu tree Stephen Rothwell
2017-08-11 4:54 ` Paul E. McKenney
2017-08-11 9:14 ` Peter Zijlstra
2017-08-11 14:39 ` Paul E. McKenney
2017-08-11 14:45 ` Peter Zijlstra
2017-08-11 14:41 ` Peter Zijlstra
2017-08-11 20:12 ` Paul E. McKenney [this message]
-- strict thread matches above, loose matches on Subject: below --
2024-01-24 4:17 Stephen Rothwell
2024-01-24 9:49 ` Jiri Wiesner
2024-01-24 12:12 ` Paul E. McKenney
2024-01-24 13:31 ` Jiri Wiesner
2024-01-24 14:20 ` Paul E. McKenney
2023-07-27 4:19 Stephen Rothwell
2023-07-27 14:08 ` Paul E. McKenney
2023-05-19 0:59 Stephen Rothwell
2023-05-19 2:12 ` Paul E. McKenney
2023-05-22 1:45 ` Stephen Rothwell
2023-05-22 14:57 ` Paul E. McKenney
2023-03-14 1:29 Stephen Rothwell
2023-03-14 4:43 ` Paul E. McKenney
2022-10-17 23:26 Stephen Rothwell
2022-10-18 10:43 ` Frederic Weisbecker
2022-10-18 14:57 ` Paul E. McKenney
2022-04-19 2:36 Stephen Rothwell
2022-04-19 3:31 ` Paul E. McKenney
2021-05-03 0:11 Stephen Rothwell
2021-05-03 16:25 ` Paul E. McKenney
2021-04-22 4:10 Stephen Rothwell
2021-04-22 16:36 ` Paul E. McKenney
2021-03-17 5:36 Stephen Rothwell
2021-03-17 14:23 ` Paul E. McKenney
2021-01-04 0:37 Stephen Rothwell
2021-01-04 12:56 ` Frederic Weisbecker
2020-12-04 8:25 Stephen Rothwell
2020-12-04 19:20 ` Paul E. McKenney
2020-12-06 21:39 ` Stephen Rothwell
2020-12-07 4:48 ` Paul E. McKenney
2020-12-07 8:59 ` Stephen Rothwell
2020-09-17 5:19 Stephen Rothwell
2020-09-17 22:01 ` Paul E. McKenney
2020-09-18 0:00 ` Stephen Rothwell
2020-09-08 5:38 Stephen Rothwell
2020-09-08 13:54 ` Paul E. McKenney
2020-08-18 1:43 Stephen Rothwell
2020-08-18 14:08 ` Paul E. McKenney
2020-06-25 2:57 Stephen Rothwell
2020-06-25 3:45 ` Paul E. McKenney
2020-05-28 9:05 Stephen Rothwell
2020-05-28 16:33 ` Paul E. McKenney
2020-05-28 21:03 ` Paul E. McKenney
2020-04-05 1:49 Stephen Rothwell
2020-04-05 3:10 ` Paul E. McKenney
2020-01-17 3:07 Stephen Rothwell
2019-12-12 2:45 Stephen Rothwell
2019-12-12 4:07 ` Paul E. McKenney
2019-12-12 4:26 ` Stephen Rothwell
2019-12-12 4:41 ` Paul E. McKenney
2020-01-17 3:09 ` Stephen Rothwell
2019-08-13 7:57 Stephen Rothwell
2019-08-13 15:31 ` Paul E. McKenney
2019-08-12 6:12 Stephen Rothwell
2019-08-12 16:19 ` Paul E. McKenney
2019-08-13 5:25 ` Stephen Rothwell
2019-08-13 14:38 ` Paul E. McKenney
2017-09-04 4:50 Stephen Rothwell
2017-09-04 16:39 ` Paul E. McKenney
2017-08-28 4:25 Stephen Rothwell
2017-08-28 17:50 ` Paul E. McKenney
2017-05-29 6:02 Stephen Rothwell
2017-05-29 21:15 ` Paul E. McKenney
2017-05-30 1:40 ` Stephen Rothwell
2017-05-30 1:54 ` Joe Perches
2017-05-30 2:14 ` Paul E. McKenney
2017-05-30 2:20 ` Joe Perches
2017-05-30 3:13 ` Stephen Rothwell
2017-05-30 4:10 ` Michael Ellerman
2017-06-02 17:51 ` Paul E. McKenney
2017-04-20 5:36 Stephen Rothwell
2017-04-20 14:23 ` Paul E. McKenney
2017-04-19 3:50 Stephen Rothwell
2017-04-19 4:06 ` Paul E. McKenney
2017-04-19 5:45 ` Stephen Rothwell
2017-03-08 1:16 Stephen Rothwell
2017-03-08 10:13 ` Daniel Vetter
2017-03-08 17:40 ` Paul E. McKenney
2017-01-19 3:34 Stephen Rothwell
2017-01-19 21:54 ` Paul McKenney
2017-02-13 2:21 ` Stephen Rothwell
2017-02-13 4:37 ` Paul E. McKenney
2017-02-13 6:43 ` Stephen Rothwell
2017-03-08 1:16 ` Stephen Rothwell
2017-03-08 1:37 ` Paul E. McKenney
2017-03-08 18:05 ` Paul E. McKenney
2016-05-02 4:37 Stephen Rothwell
2016-05-02 11:06 ` Paul E. McKenney
2016-02-01 2:55 Stephen Rothwell
2016-02-01 9:53 ` Paul E. McKenney
2016-01-07 8:57 Stephen Rothwell
2016-01-07 18:02 ` Paul E. McKenney
2016-01-07 20:19 ` Stephen Rothwell
2016-01-07 20:52 ` Paul E. McKenney
2016-01-08 1:37 ` Boqun Feng
2016-01-08 3:41 ` Paul E. McKenney
2016-01-08 4:08 ` Stephen Rothwell
2016-01-08 4:48 ` Paul E. McKenney
2016-01-08 4:54 ` Boqun Feng
2016-01-08 15:53 ` Paul E. McKenney
2016-01-08 15:57 ` Tejun Heo
2016-01-08 16:18 ` Paul E. McKenney
2016-01-08 15:58 ` Boqun Feng
2016-01-08 4:10 ` Stephen Rothwell
2015-09-01 3:50 Stephen Rothwell
2015-09-01 7:49 ` Paul E. McKenney
2015-09-02 3:58 ` Stephen Rothwell
2015-09-02 5:26 ` Paul E. McKenney
2015-09-02 6:40 ` Davidlohr Bueso
2015-09-02 7:14 ` Paul E. McKenney
2015-09-02 7:29 ` Ingo Molnar
2015-09-02 8:34 ` Paul E. McKenney
2015-07-16 3:14 Stephen Rothwell
2015-07-16 3:51 ` Paul E. McKenney
2015-07-16 5:50 ` Stephen Rothwell
2015-07-17 11:40 ` Ingo Molnar
2015-07-17 17:35 ` Paul E. McKenney
2015-07-17 18:53 ` Paul E. McKenney
2015-07-17 19:51 ` Ingo Molnar
2015-07-17 21:33 ` Paul E. McKenney
2015-07-18 2:40 ` Ingo Molnar
2015-04-13 10:39 Stephen Rothwell
2015-04-13 11:06 ` Borislav Petkov
2015-04-13 11:34 ` Ingo Molnar
2015-04-13 12:40 ` Paul E. McKenney
2015-02-27 2:18 Stephen Rothwell
2015-02-27 5:59 ` Paul E. McKenney
2014-12-26 7:51 Stephen Rothwell
2014-12-26 16:54 ` Paul E. McKenney
2014-12-27 16:24 ` Pranith Kumar
2014-12-27 17:20 ` Pranith Kumar
2014-12-31 1:45 ` Paul E. McKenney
2014-12-12 6:12 Stephen Rothwell
2014-12-12 17:23 ` Paul E. McKenney
2014-12-10 8:09 Stephen Rothwell
2014-12-10 15:03 ` Pranith Kumar
2014-12-10 15:18 ` Paul E. McKenney
2014-12-09 11:42 Stephen Rothwell
2014-12-09 14:07 ` Paul E. McKenney
2012-04-16 4:11 Stephen Rothwell
2012-04-16 17:02 ` Paul E. McKenney
2010-09-17 2:42 Stephen Rothwell
2010-09-17 4:39 ` David Miller
2010-09-17 5:34 ` Eric Dumazet
2010-09-17 23:17 ` Paul E. McKenney
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=20170811201247.GH3730@linux.vnet.ibm.com \
--to=paulmck@linux.vnet.ibm.com \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-next@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=peterz@infradead.org \
--cc=sfr@canb.auug.org.au \
--cc=tglx@linutronix.de \
/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).