[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <alpine.DEB.2.20.1702171255150.3536@nanos>
Date: Fri, 17 Feb 2017 12:55:56 +0100 (CET)
From: Thomas Gleixner <tglx@...utronix.de>
To: Chris Wilson <chris@...is-wilson.co.uk>
cc: linux-kernel@...r.kernel.org, intel-gfx@...ts.freedesktop.org,
Tvrtko Ursulin <tvrtko.ursulin@...el.com>,
Hannes Reinecke <hare@...e.com>, Jens Axboe <axboe@...nel.dk>,
Bjorn Helgaas <bhelgaas@...gle.com>,
Alexander Potapenko <glider@...gle.com>,
Chen Fan <chen.fan.fnst@...fujitsu.com>,
Ingo Molnar <mingo@...nel.org>,
"Peter Zijlstra (Intel)" <peterz@...radead.org>,
Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
Johannes Thumshirn <jthumshirn@...e.de>,
Emese Revfy <re.emese@...il.com>,
Sagi Grimberg <sagi@...mberg.me>,
Eric Dumazet <edumazet@...gle.com>,
Tom Herbert <therbert@...gle.com>,
Ben Hutchings <bhutchings@...arflare.com>
Subject: Re: [PATCH v3] softirq: Prevent looping on disabled tasklets
On Sun, 12 Feb 2017, Chris Wilson wrote:
> On Sun, Feb 12, 2017 at 03:46:09PM +0000, Chris Wilson wrote:
> > +void tasklet_enable(struct tasklet_struct *t)
> > +{
> > + if (!atomic_dec_and_test(&t->count))
> > + return;
> > +
> > + if (test_bit(TASKLET_STATE_SCHED, &t->state))
> > + raise_softirq(HI_SOFTIRQ | TASKLET_SOFTIRQ);
>
> And of course this can't work as raise_softirq() is local to the cpu.
Indeed. tasklets are a horror by design. We rather should make them go away.
Thanks,
tglx
Powered by blists - more mailing lists