[Intel-gfx] [PATCH] drm/i915: Don't try to handle GuC when GuC is not supported.

[Vivi, Rodrigo]

On Thu, 2016-10-06 at 09:36 -0300, Paulo Zanoni wrote:
> Em Qua, 2016-10-05 às 23:37 +0000, Vivi, Rodrigo escreveu:
> > Hi Daniel,
> > 
> > So, can we close https://bugs.freedesktop.org/show_bug.cgi?id=97573
> > with
> > wontfix or notabug?
> > 
> > I don't have a strong side on that actually, but Jani was against it
> > it
> > seems.
> 
> Just my opinion:
> 
> Considering that we already identified the problem and the fix is
> simple, I really think it's better to just fix it. In fact I thought
> you were going to submit V2 and I was planning to do the review.
> 
> Fixing this may help reducing future bug triaging time, because if we
> keep the problem unfixed we may get the same bug report again and again
> and again. It's easy to say "users get to keep all the pieces of the
> broken kernel", but we usually have to triage the problems regardless,
> discuss what to do, etc.

Here it is:
https://patchwork.freedesktop.org/patch/113922/

> 
> > 
> > Thanks,
> > Rodrigo.
> > 
> > On Wed, 2016-10-05 at 15:50 +0200, Daniel Vetter wrote:
> > > 
> > > On Thu, Sep 22, 2016 at 04:55:07PM +0000, Vivi, Rodrigo wrote:
> > > > 
> > > > On Wed, 2016-09-21 at 18:00 -0300, Paulo Zanoni wrote:
> > > > > 
> > > > > Em Qua, 2016-09-21 às 11:22 -0700, Rodrigo Vivi escreveu:
> > > > > > 
> > > > > > Avoid any kind of GuC handling if GuC is not supported
> > > > > > on a giving platform.
> > > > > > 
> > > > > > Besides being useless handling, our driver needs
> > > > > > to be smarter than the user trying to use an invalid
> > > > > > paramenter.
> > > > > 
> > > > > So the problem is when a platform doesn't support guc and the
> > > > > user
> > > > > passes i915.enable_guc_something=1, right?
> > > > 
> > > > 1 is not a problem actually since it means "use if available".
> > > > There is
> > > > not firmware and execution continues.
> > > > 
> > > > 2 is the problem because it means "use guc or fail if not
> > > > available".
> > > > But platforms that don't have guc can't fail. driver needs to be
> > > > smarter
> > > > than that.
> > > 
> > > Not sure it needs to be smarter than that really, since all these
> > > debug
> > > options auto-taint the kernel if you touch them. As in: You get to
> > > keep
> > > all the pieces.
> > > 
> > > We can still do some auto-cleanup of modoptions ofc if there's a
> > > good need
> > > for them.
> > > -Daniel
> > > 
> > > > 
> > > > 
> > > > > 
> > > > > 
> > > > > > 
> > > > > > 
> > > > > > Cc: Jani Nikula <jani.nikula at intel.com>
> > > > > > Cc: Anusha Srivatsa <anusha.srivatsa at intel.com>
> > > > > > Cc: Christophe Prigent <christophe.prigent at intel.com>
> > > > > > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=97573
> > > > > > Signed-off-by: Rodrigo Vivi <rodrigo.vivi at intel.com>
> > > > > > ---
> > > > > >  drivers/gpu/drm/i915/intel_guc_loader.c | 7 +++++++
> > > > > >  1 file changed, 7 insertions(+)
> > > > > > 
> > > > > > diff --git a/drivers/gpu/drm/i915/intel_guc_loader.c
> > > > > > b/drivers/gpu/drm/i915/intel_guc_loader.c
> > > > > > index 6fd39ef..da0f5ed 100644
> > > > > > --- a/drivers/gpu/drm/i915/intel_guc_loader.c
> > > > > > +++ b/drivers/gpu/drm/i915/intel_guc_loader.c
> > > > > > @@ -720,6 +720,13 @@ void intel_guc_init(struct drm_device
> > > > > > *dev)
> > > > > >  	struct intel_guc_fw *guc_fw = &dev_priv->guc.guc_fw;
> > > > > >  	const char *fw_path;
> > > > > >  
> > > > > > +	if (!HAS_GUC(dev)) {
> > > > > > +		i915.enable_guc_loading = 0;
> > > > > > +		i915.enable_guc_submission = 0;
> > > > > > +		fw_path = NULL;
> > > > > > +		return;
> > > > > > +	}
> > > > > 
> > > > > Instead of this, how about we just patch the code below with:
> > > > > 
> > > > > if (!HAS_GUC(dev_priv)) {
> > > > > 	i915.enable_guc_loading = 0;
> > > > > 	i915.enable_guc_submission = 0;
> > > > > } else {
> > > > > 	/* A negative value means "use platform default" */
> > > > > 	if (i915.enable_guc_loading < 0)
> > > > > 		i915.enable_guc_loading =
> > > > > HAS_GUC_UCODE(dev_priv);
> > > > > 	if (i915.enable_guc_submission < 0)
> > > > > 		i915.enable_guc_submission =
> > > > > HAS_GUC_SCHED(dev_priv);
> > > > > }
> > > > 
> > > > yeap, this works as well. I just went for the simplest option
> > > > that
> > > > minimized at most any interactions for platforms where GuC simply
> > > > doesn't exist.
> > > > 
> > > > > 
> > > > > 
> > > > > Or we could even go with our current "design pattern" and
> > > > > create
> > > > > intel_sanitize_guc_options().
> > > > 
> > > > This is indeed a very good idea.
> > > > 
> > > > > 
> > > > > 
> > > > > This way we'll be able to avoid adding a second failure code
> > > > > path,
> > > > > since we already have one for platforms with guc but options
> > > > > disabled.
> > > > > 
> > > > > 
> > > > > > 
> > > > > > +
> > > > > >  	/* A negative value means "use platform default" */
> > > > > >  	if (i915.enable_guc_loading < 0)
> > > > > >  		i915.enable_guc_loading =
> > > > > > HAS_GUC_UCODE(dev);
> > > > 
> > > > _______________________________________________
> > > > Intel-gfx mailing list
> > > > Intel-gfx at lists.freedesktop.org
> > > > https://lists.freedesktop.org/mailman/listinfo/intel-gfx
> > > 
> >