Re: wlcore getting stuck on hikey after the runtime PM autosuspend support change

From: Ricardo Salveti <rsalveti@rsalveti.net>
To: Tony Lindgren <tony@atomide.com>
Cc: John Stultz <john.stultz@linaro.org>,
	linux-wireless@vger.kernel.org, anders.roxell@linaro.org
Subject: Re: wlcore getting stuck on hikey after the runtime PM autosuspend support change
Date: Wed, 12 Dec 2018 17:20:04 -0200	[thread overview]
Message-ID: <CAHYQr0pqUpTCmN0f+DFA9KkEKkmtuXcAvf7tn7LqsHHa3QHSyg@mail.gmail.com> (raw)
In-Reply-To: <20181212012539.GA39861@atomide.com>

On Tue, Dec 11, 2018 at 11:25 PM Tony Lindgren <tony@atomide.com> wrote:
>
> * Ricardo Salveti <rsalveti@rsalveti.net> [181211 21:51]:
> > Tried to change back wl12xx_sdio_power_off to use pm_runtime_put_sync
> > as a test, and noticed I always get -EBUSY when reproducing the hang,
> > so it looks like this could be a race between pm_runtime_put/get when
> > doing if down/up (and the side effect on the mmc controller).
>
> Hmm so -EBUSY for pm_runtime means transition is already happening
> and in many cases the test should be:
>
> error = pm_runtime_get(dev);
> if (error < 0 && error != -EBUSY) {
>         ...
> }
> ...
> pm_runtime_put(dev);
>
> Where are you getting it?

What I noticed is that with autosuspend support I always get -EBUSY in
wl12xx_sdio_power_off -> pm_runtime_put, and I would guess that the
additional delay here creates the issue later when bringing the
interface up again.

Without autosuspend I never get -EBUSY in wl12xx_sdio_power_off.

Cheers,
-- 
Ricardo Salveti de Araujo