Re: [PATCH 4/4] drm/i915: Clear crtc atomic flags at beginning of transaction

[Matt Roper <matthew.d.roper@intel.com>]

On Thu, Apr 09, 2015 at 02:48:43PM +0200, Daniel Vetter wrote:
> On Wed, Apr 08, 2015 at 06:56:54PM -0700, Matt Roper wrote:
> > Once we have full atomic modeset, these kind of flags should be in a
> > real intel_crtc_state that's tracked properly.  In the meantime, make
> > sure we clear out any old flags at the beginning of a transaction so
> > that we don't wind up seeing leftover flags from old transactions that
> > were checked, but never went to the commit step.
> > 
> > A simple memset would have done here, but I expect there to be a few
> > more things that we *don't* want to clear that get added into this
> > structure before we're ready to kill off and roll everything into the
> > CRTC state.
> > 
> > Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
> 
> Not sure whether this helps a lot with moving intel_crtc->atomic into
> intel_crtc_state. We probably want to just move things one-by-one to
> reduce the overall churn, and then we can add them one-by-one to the intel
> crtc_state_duplicate function.
> 
> Or do you have some bigger plans here?
> -Daniel

My in-progress WM work had to stick some other things in
intel_crtc->atomic, but maybe with Ander's latest patch sets we're close
enough that I can rewrite those to use crtc_state instead; I'll have to
go back and check.

Regardless, I think we do still have a bug today where an uncommitted
transaction (e.g., because check or prepare fail) will leave stale flags
in intel_crtc->atomic that the next transaction doesn't realize it needs
to clear.  We either need to clear those out (as I'm doing here,
although a simple memset would work too), or we need to transition all
those flags over to CRTC state and kill off the hacky intel_crtc->atomic
structure.


Matt

> 
> > ---
> >  drivers/gpu/drm/i915/intel_atomic.c  | 18 ++++++++++++++++++
> >  drivers/gpu/drm/i915/intel_display.c |  2 +-
> >  drivers/gpu/drm/i915/intel_drv.h     |  1 +
> >  3 files changed, 20 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/intel_atomic.c b/drivers/gpu/drm/i915/intel_atomic.c
> > index 3903b90..542230d 100644
> > --- a/drivers/gpu/drm/i915/intel_atomic.c
> > +++ b/drivers/gpu/drm/i915/intel_atomic.c
> > @@ -35,6 +35,22 @@
> >  #include <drm/drm_plane_helper.h>
> >  #include "intel_drv.h"
> >  
> > +/**
> > + * intel_clear_atomic_crtc_flags
> > + * @crtc: CRTC to clear flags for
> > + *
> > + * Until we have proper atomic handling of CRTC states, we dump some atomic
> > + * task flags in intel_crtc->atomic.  Those flags need to be cleared at
> > + * the beginning of each transaction so that we don't carry over stale flags
> > + * from previous transactions.
> > + *
> > + * Note that the whole intel_crtc->atomic structure is a temporary hack and
> > + * should transition into the CRTC state eventually.
> > + */
> > +void intel_clear_atomic_crtc_flags(struct intel_crtc *crtc)
> > +{
> > +	memset(&crtc->atomic, 0, sizeof(crtc->atomic));
> > +}
> >  
> >  /**
> >   * intel_atomic_check - validate state object
> > @@ -76,6 +92,8 @@ int intel_atomic_check(struct drm_device *dev,
> >  	state->allow_modeset = false;
> >  	for (i = 0; i < ncrtcs; i++) {
> >  		struct intel_crtc *crtc = to_intel_crtc(state->crtcs[i]);
> > +		if (crtc)
> > +			intel_clear_atomic_crtc_flags(crtc);
> >  		if (crtc && crtc->pipe != nuclear_pipe)
> >  			not_nuclear = true;
> >  	}
> > diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> > index 3a74923..bb8d345 100644
> > --- a/drivers/gpu/drm/i915/intel_display.c
> > +++ b/drivers/gpu/drm/i915/intel_display.c
> > @@ -12812,7 +12812,7 @@ static void intel_finish_crtc_commit(struct drm_crtc *crtc)
> >  			intel_update_sprite_watermarks(p, crtc, 0, 0, 0,
> >  						       false, false);
> >  
> > -	memset(&intel_crtc->atomic, 0, sizeof(intel_crtc->atomic));
> > +	intel_clear_atomic_crtc_flags(intel_crtc);
> >  }
> >  
> >  /**
> > diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
> > index d5ea24f..28d838e 100644
> > --- a/drivers/gpu/drm/i915/intel_drv.h
> > +++ b/drivers/gpu/drm/i915/intel_drv.h
> > @@ -1303,6 +1303,7 @@ void intel_pre_disable_primary(struct drm_crtc *crtc);
> >  void intel_tv_init(struct drm_device *dev);
> >  
> >  /* intel_atomic.c */
> > +void intel_clear_atomic_crtc_flags(struct intel_crtc *crtc);
> >  int intel_atomic_check(struct drm_device *dev,
> >  		       struct drm_atomic_state *state);
> >  int intel_atomic_commit(struct drm_device *dev,
> > -- 
> > 1.8.5.1
> > 
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx@lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> 
> -- 
> Daniel Vetter
> Software Engineer, Intel Corporation
> http://blog.ffwll.ch

-- 
Matt Roper
Graphics Software Engineer
IoTG Platform Enabling & Development
Intel Corporation
(916) 356-2795
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx