mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Paul E. McKenney" <paulmck@kernel.org>
To: Rik van Riel <riel@surriel.com>
Cc: linux-kernel@vger.kernel.org, Tejun Heo <tj@kernel.org>,
	Lai Jiangshan <jiangshanlai@gmail.com>,
	Breno Leitao <leitao@debian.org>,
	Anhad Jai Singh <ffledgling@meta.com>,
	Oleg Nesterov <oleg@redhat.com>, Jens Axboe <axboe@kernel.dk>,
	Christian Brauner <brauner@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	"Matthew Wilcox (Oracle)" <willy@infradead.org>,
	Chris Mason <clm@fb.com>
Subject: Re: [PATCH misc 1/2] workqueue: Add check for clocks going backwards to wq_worker_tick()
Date: Fri, 2 Aug 2024 09:39:19 -0700	[thread overview]
Message-ID: <1452aae8-c8d7-43a2-979a-5b3878ddc2fa@paulmck-laptop> (raw)
In-Reply-To: <bc6cf2940cee5bf3414616051c821c0b17dd61b1.camel@surriel.com>

On Fri, Aug 02, 2024 at 12:06:06PM -0400, Rik van Riel wrote:
> On Thu, 2024-08-01 at 17:30 -0700, Paul E. McKenney wrote:
> > 
> > +++ b/kernel/workqueue.c
> > @@ -1482,6 +1482,7 @@ void wq_worker_tick(struct task_struct *task)
> >  	 * If the current worker is concurrency managed and hogged
> > the CPU for
> >  	 * longer than wq_cpu_intensive_thresh_us, it's
> > automatically marked
> >  	 * CPU_INTENSIVE to avoid stalling other concurrency-managed
> > work items.
> > +	 * If the time is negative, ignore, assuming a backwards
> > clock.
> >  	 *
> >  	 * Set @worker->sleeping means that @worker is in the
> > process of
> >  	 * switching out voluntarily and won't be contributing to
> > @@ -1491,6 +1492,7 @@ void wq_worker_tick(struct task_struct *task)
> >  	 * We probably want to make this prettier in the future.
> >  	 */
> >  	if ((worker->flags & WORKER_NOT_RUNNING) ||
> > READ_ONCE(worker->sleeping) ||
> > +	    WARN_ON_ONCE((s64)(worker->task->se.sum_exec_runtime -
> > worker->current_at) < 0) ||
> >  	    worker->task->se.sum_exec_runtime - worker->current_at <
> >  	    wq_cpu_intensive_thresh_us * NSEC_PER_USEC)
> >  		return;
> 
> What is the code path by which sum_exec_runtime could go backward
> in time, if the TSC and sched_clock() jump backward?
> 
> Might it make sense to check in the place where sum_exec_runtime is
> updated, instead, and catch a wider net?
> 
> On the flip side, the run time increments are "fairly large" in
> number of TSC cycles, while most of the negative TSC jumps we 
> have seen are quite small, so even that wider net might not catch
> much because of how coarse these updates typically are...

Good points!  Even more telling, this patch didn't catch anything during
Breno's tests.  I will drop it with a workqueue.2024.08.01 branch in
-rcu in case someone needs it later.

							Thanx, Paul

> All Rights Reversed.

  reply	other threads:[~2024-08-02 16:39 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-08-02  0:30 [PATCH 0/2] Miscellaneous updates for v6.12' Paul E. McKenney
2024-08-02  0:30 ` [PATCH misc 1/2] workqueue: Add check for clocks going backwards to wq_worker_tick() Paul E. McKenney
2024-08-02 16:06   ` Rik van Riel
2024-08-02 16:39     ` Paul E. McKenney [this message]
2024-08-02  0:30 ` [PATCH misc 2/2] exit: Sleep at TASK_IDLE when waiting for application core dump 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=1452aae8-c8d7-43a2-979a-5b3878ddc2fa@paulmck-laptop \
    --to=paulmck@kernel.org \
    --cc=akpm@linux-foundation.org \
    --cc=axboe@kernel.dk \
    --cc=brauner@kernel.org \
    --cc=clm@fb.com \
    --cc=ffledgling@meta.com \
    --cc=jiangshanlai@gmail.com \
    --cc=leitao@debian.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=oleg@redhat.com \
    --cc=riel@surriel.com \
    --cc=tj@kernel.org \
    --cc=willy@infradead.org \
    /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

all inboxes | Powered by JetHome®