From: Peter Zijlstra <peterz@infradead.org>
To: Oleg Nesterov <oleg@redhat.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
linux-rt-users <linux-rt-users@vger.kernel.org>,
LKML <linux-kernel@vger.kernel.org>,
Thomas Gleixner <tglx@linutronix.de>,
Daniel Bristot de Oliveira <bristot@redhat.com>,
Clark Williams <williams@redhat.com>,
Juri Lelli <juri.lelli@redhat.com>,
jack@suse.com, Waiman Long <longman@redhat.com>,
Davidlohr Bueso <dave@stgolabs.net>
Subject: Re: [RT WARNING] DEBUG_LOCKS_WARN_ON(rt_mutex_owner(lock) != current) with fsfreeze (4.19.25-rt16)
Date: Fri, 3 May 2019 17:25:13 +0200 [thread overview]
Message-ID: <20190503152513.GE2650@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <20190503145059.GC2606@hirez.programming.kicks-ass.net>
On Fri, May 03, 2019 at 04:50:59PM +0200, Peter Zijlstra wrote:
> So how about something like so then?
> --- a/kernel/locking/percpu-rwsem.c
> +++ b/kernel/locking/percpu-rwsem.c
> @@ -63,7 +66,7 @@ int __percpu_down_read(struct percpu_rw_
> * If !readers_block the critical section starts here, matched by the
> * release in percpu_up_write().
> */
> - if (likely(!smp_load_acquire(&sem->readers_block)))
> + if (likely(!atomic_read_acquire(&sem->block)))
> return 1;
>
> /*
> @@ -80,14 +83,8 @@ int __percpu_down_read(struct percpu_rw_
> * and reschedule on the preempt_enable() in percpu_down_read().
> */
> preempt_enable_no_resched();
> -
> - /*
> - * Avoid lockdep for the down/up_read() we already have them.
> - */
> - __down_read(&sem->rw_sem);
> + wait_event(sem->waiters, !atomic_read(&sem->block));
That should be:
wait_event(sem->waiters, !atomic_read_acquire(&sem->block));
I suppose.
> this_cpu_inc(*sem->read_count);
> - __up_read(&sem->rw_sem);
> -
> preempt_disable();
> return 1;
> }
> @@ -104,7 +101,7 @@ void __percpu_up_read(struct percpu_rw_s
> __this_cpu_dec(*sem->read_count);
>
> /* Prod writer to recheck readers_active */
> - rcuwait_wake_up(&sem->writer);
> + wake_up(&sem->waiters);
> }
> EXPORT_SYMBOL_GPL(__percpu_up_read);
>
> @@ -139,18 +136,22 @@ static bool readers_active_check(struct
> return true;
> }
>
> +static inline bool acquire_block(struct percpu_rw_semaphore *sem)
> +{
> + if (atomic_read(&sem->block))
> + return false;
> +
> + return atomic_xchg(&sem->block, 1) == 0;
> +}
> +
> void percpu_down_write(struct percpu_rw_semaphore *sem)
> {
> + rwsem_acquire(&sem->dep_map, 0, 0, _RET_IP_);
> +
> /* Notify readers to take the slow path. */
> rcu_sync_enter(&sem->rss);
>
> - down_write(&sem->rw_sem);
> -
> - /*
> - * Notify new readers to block; up until now, and thus throughout the
> - * longish rcu_sync_enter() above, new readers could still come in.
> - */
> - WRITE_ONCE(sem->readers_block, 1);
> + wait_event_exclusive(sem->waiters, acquire_block(sem));
>
> smp_mb(); /* D matches A */
And we can remove that smp_mb() and rely on the atomic_xchg() from
acquire_block().
>
> @@ -161,7 +162,7 @@ void percpu_down_write(struct percpu_rw_
> */
>
> /* Wait for all now active readers to complete. */
> - rcuwait_wait_event(&sem->writer, readers_active_check(sem));
> + wait_event(sem->waiters, readers_active_check(sem));
> }
> EXPORT_SYMBOL_GPL(percpu_down_write);
>
> @@ -177,12 +178,8 @@ void percpu_up_write(struct percpu_rw_se
> * Therefore we force it through the slow path which guarantees an
> * acquire and thereby guarantees the critical section's consistency.
> */
> - smp_store_release(&sem->readers_block, 0);
> -
> - /*
> - * Release the write lock, this will allow readers back in the game.
> - */
> - up_write(&sem->rw_sem);
> + atomic_set_release(&sem->block, 0);
> + wake_up(&sem->waiters);
>
> /*
> * Once this completes (at least one RCU-sched grace period hence) the
> @@ -190,5 +187,21 @@ void percpu_up_write(struct percpu_rw_se
> * exclusive write lock because its counting.
> */
> rcu_sync_exit(&sem->rss);
> +
> + rwsem_release(&sem->dep_map, 1, _RET_IP_);
> }
> EXPORT_SYMBOL_GPL(percpu_up_write);
next prev parent reply other threads:[~2019-05-03 15:25 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-03-26 9:34 [RT WARNING] DEBUG_LOCKS_WARN_ON(rt_mutex_owner(lock) != current) with fsfreeze (4.19.25-rt16) Juri Lelli
2019-03-28 10:17 ` Sebastian Andrzej Siewior
2019-04-19 8:56 ` Juri Lelli
2019-04-30 12:51 ` Sebastian Andrzej Siewior
2019-04-30 13:28 ` Peter Zijlstra
2019-04-30 13:45 ` Sebastian Andrzej Siewior
2019-04-30 14:01 ` Peter Zijlstra
2019-04-30 14:15 ` Oleg Nesterov
2019-04-30 14:29 ` Peter Zijlstra
2019-04-30 14:42 ` Oleg Nesterov
2019-04-30 14:44 ` Peter Zijlstra
2019-04-30 14:53 ` Oleg Nesterov
2019-05-01 17:09 ` Peter Zijlstra
2019-05-01 17:26 ` Waiman Long
2019-05-01 18:54 ` Peter Zijlstra
2019-05-01 19:22 ` Davidlohr Bueso
2019-05-01 19:25 ` Peter Zijlstra
2019-05-02 10:09 ` Oleg Nesterov
2019-05-02 11:42 ` Oleg Nesterov
2019-05-03 14:50 ` Peter Zijlstra
2019-05-03 15:25 ` Peter Zijlstra [this message]
2019-05-06 16:50 ` Oleg Nesterov
2019-06-19 9:50 ` Peter Zijlstra
2019-05-03 14:16 ` Peter Zijlstra
2019-05-03 15:37 ` Oleg Nesterov
2019-05-03 15:46 ` Peter Zijlstra
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=20190503152513.GE2650@hirez.programming.kicks-ass.net \
--to=peterz@infradead.org \
--cc=bigeasy@linutronix.de \
--cc=bristot@redhat.com \
--cc=dave@stgolabs.net \
--cc=jack@suse.com \
--cc=juri.lelli@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rt-users@vger.kernel.org \
--cc=longman@redhat.com \
--cc=oleg@redhat.com \
--cc=tglx@linutronix.de \
--cc=williams@redhat.com \
/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).