Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Thu, 22 Oct 2020 15:24:12 +0200
From:   Maxime Ripard <maxime@cerno.tech>
To:     Takashi Iwai <tiwai@suse.de>
Cc:     Jaroslav Kysela <perex@perex.cz>, Takashi Iwai <tiwai@suse.com>,
        Liam Girdwood <lgirdwood@gmail.com>,
        Mark Brown <broonie@kernel.org>, alsa-devel@alsa-project.org,
        linux-kernel@vger.kernel.org, Dom Cobley <dom@raspberrypi.com>,
        Dave Stevenson <dave.stevenson@raspberrypi.com>,
        Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Subject: Re: Context expectations in ALSA
Message-ID: <20201022132412.vntap4kfb2aj24qy@gilmour.lan>
References: <20201022095041.44jytaelnlako54w@gilmour.lan>
 <30226f94-72e9-34d2-17d0-11d2501053f0@perex.cz>
 <20201022125741.xxibhwgcr2mhxehe@gilmour.lan>
 <s5ha6webdn2.wl-tiwai@suse.de>
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha256;
        protocol="application/pgp-signature"; boundary="rwqqljnlxayusdb4"
Content-Disposition: inline
In-Reply-To: <s5ha6webdn2.wl-tiwai@suse.de>
Precedence: bulk


--rwqqljnlxayusdb4
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Hi Takashi,

On Thu, Oct 22, 2020 at 03:20:49PM +0200, Takashi Iwai wrote:
> On Thu, 22 Oct 2020 14:57:41 +0200,
> Maxime Ripard wrote:
> >=20
> > On Thu, Oct 22, 2020 at 12:03:19PM +0200, Jaroslav Kysela wrote:
> > > Dne 22. 10. 20 v 11:50 Maxime Ripard napsal(a):
> > >=20
> > > > So, I'm not really sure what I'm supposed to do here. The drivers
> > > > involved don't appear to be doing anything extraordinary, but the i=
ssues
> > > > lockdep report are definitely valid too. What are the expectations =
in
> > > > terms of context from ALSA when running the callbacks, and how can =
we
> > > > fix it?
> > >=20
> > > I think that you should set the non-atomic flag and wake up the workq=
ueue or
> > > so from interrupt handler in this case. Call snd_pcm_period_elapsed()=
 from the
> > > workqueue not the interrupt handler context.
> >=20
> > Yeah, that was my first guess too. However, the DMA driver uses some
> > kind of generic helpers using a tasklet, so getting rid of it would take
> > some work and would very likely not be eligible for stable.
>=20
> Who sets the nonatomic flag for vc4?  I couldn't find the relevant
> code in the latest upstream.

Sorry if this wasn't clear enough, it's not there at the moment, ALSA
takes a spinlock and lockdep complains that we're sleeping in an atomic
context.

I tried to add the nonatomic flag in my tree to see if it was fixing the
issue, but ran into another lockdep complain now with ALSA taking a
mutex in a tasklet.

> Ideally dmaengine PCM helper should support the nonatomic mode, but
> until then, the other side needs to drop the nonatomic flag, I
> suppose.

In this case, I'm not sure the blame is in the PCM helper but if there's
any blame, I guess it's the virt-chan layer inside dmaengine (so for
providers) that use a tasklet instead of something that allows sleeping

Maxime

--rwqqljnlxayusdb4
Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----

iHUEABYIAB0WIQRcEzekXsqa64kGDp7j7w1vZxhRxQUCX5GH/AAKCRDj7w1vZxhR
xepDAP96aJ4MTdcE8LJZ10+kDyzH1UtQlV//DYZWfptlugY+vgEA2+KK5p2ApO4B
kDALlv1Yd5PHZH7QHStcrelNXI69ygs=
=e2G2
-----END PGP SIGNATURE-----

--rwqqljnlxayusdb4--