[alsa-devel] [PATCH] vga_switcheroo: Fix missing gpu_bound call at audio client registration

[Takashi Iwai]

On Mon, 05 Nov 2018 12:55:17 +0100,
Lukas Wunner wrote:
> 
> On Mon, Nov 05, 2018 at 12:02:53PM +0100, Takashi Iwai wrote:
> > The commit 37a3a98ef601 ("ALSA: hda - Enable runtime PM only for
> > discrete GPU") added a new ops gpu_bound to be called when GPU gets
> > bound.  The patch overlooked, however, that vga_switcheroo_enable() is
> > called only once at GPU is bound.  When an audio client is registered
> > after that point, it would miss the gpu_bound call.  This leads to the
> > unexpected lack of runtime PM in HD-audio side.
> > 
> > For addressing that regression, just call gpu_bound callback manually
> > at vga_switcheroo_register_audio_client() when the GPU was already
> > bound.
> > 
> > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201615
> > Fixes: 37a3a98ef601 ("ALSA: hda - Enable runtime PM only for discrete GPU")
> > Cc: <stable at vger.kernel.org>
> > Signed-off-by: Takashi Iwai <tiwai at suse.de>
> > ---
> >  drivers/gpu/vga/vga_switcheroo.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/drivers/gpu/vga/vga_switcheroo.c b/drivers/gpu/vga/vga_switcheroo.c
> > index cf2a18571d48..a132c37d7334 100644
> > --- a/drivers/gpu/vga/vga_switcheroo.c
> > +++ b/drivers/gpu/vga/vga_switcheroo.c
> > @@ -380,6 +380,9 @@ int vga_switcheroo_register_audio_client(struct pci_dev *pdev,
> >  			mutex_unlock(&vgasr_mutex);
> >  			return -EINVAL;
> >  		}
> > +		/* notify if GPU has been already bound */
> > +		if (ops->gpu_bound)
> > +			ops->gpu_bound(pdev, id);
> >  	}
> >  	mutex_unlock(&vgasr_mutex);
> 
> If the audio client registers before vga_switcheroo becomes enabled,
> ->gpu_bound will be executed twice.

Is it?  The addition is in the if-block of vgasr_priv.active, so this
is executed only when vga_switcheroo_enable() was already called
beforehand.

> AFAICS this only causes a minor
> cosmetic issue, namely that dev_info() is called twice for blacklisted
> devices in set_default_power_save().  This could be eliminated by
> amending the if-condition with "&& vgasr_priv.active".  Since this
> is only minor, up to you if you want to respin.
> 
> Either way,
> Reviewed-by: Lukas Wunner <lukas at wunner.de>


Thanks!

Takashi