Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1526753

Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer wakeups

From Oleg Nesterov <oleg@redhat.com>
Newsgroups linux.kernel
Subject Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer wakeups
Date 2016-11-21 16:10 +0100
Message-ID <sFYoh-2MJ-39@gated-at.bofh.it> (permalink)
References <sEWye-2tW-9@gated-at.bofh.it> <sEWyf-2tW-55@gated-at.bofh.it> <sFVTr-18L-7@gated-at.bofh.it> <sFVTr-18L-5@gated-at.bofh.it> <sFWcN-1f8-9@gated-at.bofh.it>
Organization linux.* mail to news gateway

Show all headers | View raw


On 11/21, Oleg Nesterov wrote:
>
> No, no, I meant that afaics both readers can see per_cpu_sum() != 0 and
> thus the writer won't be woken up. Till the next down_read/up_read.
>
> Suppose that we have 2 CPU's, both counters == 1, both readers decrement.
> its counter at the same time.
>
> 	READER_ON_CPU_0			READER_ON_CPU_1
>
> 	--ctr_0;			--ctr_1;
>
> 	if (ctr_0 + ctr_1)		if (ctr_0 + ctr_1)
> 		wakeup();			wakeup();
>
> Why we can't miss a wakeup?
>
> This patch doesn't even add a barrier, but I think wmb() won't be enough
> anyway.

And in fact I am not sure this optimization makes sense... But it would be
nice to avoid wake_up() when the writer sleeps in rcu_sync_enter(). Or this
is the "slow mode" sem (cgroup_threadgroup_rwsem).

I need to re-check, but what do you think about the change below?

Oleg.

--- x/kernel/locking/percpu-rwsem.c
+++ x/kernel/locking/percpu-rwsem.c
@@ -103,7 +103,9 @@ void __percpu_up_read(struct percpu_rw_s
 	__this_cpu_dec(*sem->read_count);
 
 	/* Prod writer to recheck readers_active */
-	wake_up(&sem->writer);
+	smp_mb();
+	if (sem->readers_block)
+		wake_up(&sem->writer);
 }
 EXPORT_SYMBOL_GPL(__percpu_up_read);
 

Back to linux.kernel | Previous | NextPrevious in thread | Next in thread | Find similar | Unroll thread


Thread

[PATCH -tip 0/3] locking/percpu-rwsem: writer-side optimizations Davidlohr Bueso <dave@stgolabs.net> - 2016-11-18 20:00 +0100
  [PATCH 1/3] locking/percpu-rwsem: Move text file into Documentation/locking/ Davidlohr Bueso <dave@stgolabs.net> - 2016-11-18 20:00 +0100
  [PATCH 2/3] locking/percpu-rwsem: Replace bulky wait-queues with swait Davidlohr Bueso <dave@stgolabs.net> - 2016-11-18 20:00 +0100
    Re: [PATCH 2/3] locking/percpu-rwsem: Replace bulky wait-queues         with swait Oleg Nesterov <oleg@redhat.com> - 2016-11-21 14:00 +0100
      Re: [PATCH 2/3] locking/percpu-rwsem: Replace bulky wait-queues with  swait Davidlohr Bueso <dave@stgolabs.net> - 2016-11-21 18:30 +0100
  [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer wakeups Davidlohr Bueso <dave@stgolabs.net> - 2016-11-18 20:00 +0100
    Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer  wakeups Peter Zijlstra <peterz@infradead.org> - 2016-11-21 13:30 +0100
      Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer         wakeups Oleg Nesterov <oleg@redhat.com> - 2016-11-21 13:50 +0100
        Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer         wakeups Oleg Nesterov <oleg@redhat.com> - 2016-11-21 16:10 +0100
          Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer  wakeups Davidlohr Bueso <dave@stgolabs.net> - 2016-11-22 05:00 +0100
            Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer         wakeups Oleg Nesterov <oleg@redhat.com> - 2016-11-23 15:50 +0100
    Re: [PATCH 3/3] locking/percpu-rwsem: Avoid unnecessary writer         wakeups Oleg Nesterov <oleg@redhat.com> - 2016-11-21 13:30 +0100

csiph-web