diff mbox series

drm/damage_helper: Fix handling of cursor dirty buffers

Message ID 20210817232604.160029-1-jose.souza@intel.com (mailing list archive)
State New, archived
Headers show
Series drm/damage_helper: Fix handling of cursor dirty buffers | expand

Commit Message

Souza, Jose Aug. 17, 2021, 11:26 p.m. UTC
Cursors don't have a framebuffer so the fb comparisson was always
failing and atomic state was being committed without any plane state.

So here checking if objects match when checking cursors.

Fixes: b9fc5e01d1ce ("drm: Add helper to implement legacy dirtyfb")
Cc: Daniel Vetter <daniel.vetter@intel.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Deepak Rawat <drawat@vmware.com>
Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
---
 drivers/gpu/drm/drm_damage_helper.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

Comments

Daniel Vetter Aug. 18, 2021, 9:54 a.m. UTC | #1
On Tue, Aug 17, 2021 at 04:26:04PM -0700, José Roberto de Souza wrote:
> Cursors don't have a framebuffer so the fb comparisson was always
> failing and atomic state was being committed without any plane state.
> 
> So here checking if objects match when checking cursors.

This looks extremely backwards ... what exactly is this fixing? If this
isn't based on a real world compositor usage but some igt, then I'd say
the igt here is very wrong.
-Daniel

> Fixes: b9fc5e01d1ce ("drm: Add helper to implement legacy dirtyfb")
> Cc: Daniel Vetter <daniel.vetter@intel.com>
> Cc: Rob Clark <robdclark@gmail.com>
> Cc: Deepak Rawat <drawat@vmware.com>
> Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
> Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
> ---
>  drivers/gpu/drm/drm_damage_helper.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/drm_damage_helper.c b/drivers/gpu/drm/drm_damage_helper.c
> index 8eeff0c7bdd47..595187d97c131 100644
> --- a/drivers/gpu/drm/drm_damage_helper.c
> +++ b/drivers/gpu/drm/drm_damage_helper.c
> @@ -157,12 +157,18 @@ int drm_atomic_helper_dirtyfb(struct drm_framebuffer *fb,
>  retry:
>  	drm_for_each_plane(plane, fb->dev) {
>  		struct drm_plane_state *plane_state;
> +		bool match;
>  
>  		ret = drm_modeset_lock(&plane->mutex, state->acquire_ctx);
>  		if (ret)
>  			goto out;
>  
> -		if (plane->state->fb != fb) {
> +		match = plane->state->fb == fb;
> +		/* Check if objs match to handle dirty buffers of cursors */
> +		if (plane->type == DRM_PLANE_TYPE_CURSOR && plane->state->fb)
> +			match |= fb->obj[0] == plane->state->fb->obj[0];
> +
> +		if (!match) {
>  			drm_modeset_unlock(&plane->mutex);
>  			continue;
>  		}
> -- 
> 2.32.0
>
Souza, Jose Aug. 18, 2021, 4:44 p.m. UTC | #2
On Wed, 2021-08-18 at 11:54 +0200, Daniel Vetter wrote:
> On Tue, Aug 17, 2021 at 04:26:04PM -0700, José Roberto de Souza wrote:
> > Cursors don't have a framebuffer so the fb comparisson was always
> > failing and atomic state was being committed without any plane state.
> > 
> > So here checking if objects match when checking cursors.
> 
> This looks extremely backwards ... what exactly is this fixing? If this
> isn't based on a real world compositor usage but some igt, then I'd say
> the igt here is very wrong.

Yes it is IGT.
Writing to cursor buffer current in the screen and calling drmModeDirtyFB() causes a empty atomic commit by drm_atomic_helper_dirtyfb().


> -Daniel
> 
> > Fixes: b9fc5e01d1ce ("drm: Add helper to implement legacy dirtyfb")
> > Cc: Daniel Vetter <daniel.vetter@intel.com>
> > Cc: Rob Clark <robdclark@gmail.com>
> > Cc: Deepak Rawat <drawat@vmware.com>
> > Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
> > Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
> > ---
> >  drivers/gpu/drm/drm_damage_helper.c | 8 +++++++-
> >  1 file changed, 7 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_damage_helper.c b/drivers/gpu/drm/drm_damage_helper.c
> > index 8eeff0c7bdd47..595187d97c131 100644
> > --- a/drivers/gpu/drm/drm_damage_helper.c
> > +++ b/drivers/gpu/drm/drm_damage_helper.c
> > @@ -157,12 +157,18 @@ int drm_atomic_helper_dirtyfb(struct drm_framebuffer *fb,
> >  retry:
> >  	drm_for_each_plane(plane, fb->dev) {
> >  		struct drm_plane_state *plane_state;
> > +		bool match;
> >  
> >  		ret = drm_modeset_lock(&plane->mutex, state->acquire_ctx);
> >  		if (ret)
> >  			goto out;
> >  
> > -		if (plane->state->fb != fb) {
> > +		match = plane->state->fb == fb;
> > +		/* Check if objs match to handle dirty buffers of cursors */
> > +		if (plane->type == DRM_PLANE_TYPE_CURSOR && plane->state->fb)
> > +			match |= fb->obj[0] == plane->state->fb->obj[0];
> > +
> > +		if (!match) {
> >  			drm_modeset_unlock(&plane->mutex);
> >  			continue;
> >  		}
> > -- 
> > 2.32.0
> > 
>
Daniel Vetter Aug. 19, 2021, 8:41 a.m. UTC | #3
On Wed, Aug 18, 2021 at 04:44:44PM +0000, Souza, Jose wrote:
> On Wed, 2021-08-18 at 11:54 +0200, Daniel Vetter wrote:
> > On Tue, Aug 17, 2021 at 04:26:04PM -0700, José Roberto de Souza wrote:
> > > Cursors don't have a framebuffer so the fb comparisson was always
> > > failing and atomic state was being committed without any plane state.
> > > 
> > > So here checking if objects match when checking cursors.
> > 
> > This looks extremely backwards ... what exactly is this fixing? If this
> > isn't based on a real world compositor usage but some igt, then I'd say
> > the igt here is very wrong.
> 
> Yes it is IGT.
> Writing to cursor buffer current in the screen and calling
> drmModeDirtyFB() causes a empty atomic commit by
> drm_atomic_helper_dirtyfb().

Ok if the cursor write is done through legacy cursor uapi then trying to
make that work with dirtyfb doesn't make sense.

But you've found a bug at least, namely the empty commit. I think that
should be filtered out, and our dirtyfb testcases (I hope we have some
vendor-agnostic kms_dirtyfb already) should be extended with a testcase
where we call dirtyfb on an fb which is not currently used anywhere.

Wrt the cursor: The legacy cursor ioctls get remapped onto the cursor
plane, which means they come in as full commits. So there's really not
dirtyfb required afterwards. The funny thing about the cursor ioctls is
that they never supported frontbuffer rendering. You _always_ had to call
them to upload data. So the test is invalid from a functional pov too.
-Daniel

> 
> 
> > -Daniel
> > 
> > > Fixes: b9fc5e01d1ce ("drm: Add helper to implement legacy dirtyfb")
> > > Cc: Daniel Vetter <daniel.vetter@intel.com>
> > > Cc: Rob Clark <robdclark@gmail.com>
> > > Cc: Deepak Rawat <drawat@vmware.com>
> > > Cc: Gwan-gyeong Mun <gwan-gyeong.mun@intel.com>
> > > Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
> > > ---
> > >  drivers/gpu/drm/drm_damage_helper.c | 8 +++++++-
> > >  1 file changed, 7 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/gpu/drm/drm_damage_helper.c b/drivers/gpu/drm/drm_damage_helper.c
> > > index 8eeff0c7bdd47..595187d97c131 100644
> > > --- a/drivers/gpu/drm/drm_damage_helper.c
> > > +++ b/drivers/gpu/drm/drm_damage_helper.c
> > > @@ -157,12 +157,18 @@ int drm_atomic_helper_dirtyfb(struct drm_framebuffer *fb,
> > >  retry:
> > >  	drm_for_each_plane(plane, fb->dev) {
> > >  		struct drm_plane_state *plane_state;
> > > +		bool match;
> > >  
> > >  		ret = drm_modeset_lock(&plane->mutex, state->acquire_ctx);
> > >  		if (ret)
> > >  			goto out;
> > >  
> > > -		if (plane->state->fb != fb) {
> > > +		match = plane->state->fb == fb;
> > > +		/* Check if objs match to handle dirty buffers of cursors */
> > > +		if (plane->type == DRM_PLANE_TYPE_CURSOR && plane->state->fb)
> > > +			match |= fb->obj[0] == plane->state->fb->obj[0];
> > > +
> > > +		if (!match) {
> > >  			drm_modeset_unlock(&plane->mutex);
> > >  			continue;
> > >  		}
> > > -- 
> > > 2.32.0
> > > 
> > 
>
diff mbox series

Patch

diff --git a/drivers/gpu/drm/drm_damage_helper.c b/drivers/gpu/drm/drm_damage_helper.c
index 8eeff0c7bdd47..595187d97c131 100644
--- a/drivers/gpu/drm/drm_damage_helper.c
+++ b/drivers/gpu/drm/drm_damage_helper.c
@@ -157,12 +157,18 @@  int drm_atomic_helper_dirtyfb(struct drm_framebuffer *fb,
 retry:
 	drm_for_each_plane(plane, fb->dev) {
 		struct drm_plane_state *plane_state;
+		bool match;
 
 		ret = drm_modeset_lock(&plane->mutex, state->acquire_ctx);
 		if (ret)
 			goto out;
 
-		if (plane->state->fb != fb) {
+		match = plane->state->fb == fb;
+		/* Check if objs match to handle dirty buffers of cursors */
+		if (plane->type == DRM_PLANE_TYPE_CURSOR && plane->state->fb)
+			match |= fb->obj[0] == plane->state->fb->obj[0];
+
+		if (!match) {
 			drm_modeset_unlock(&plane->mutex);
 			continue;
 		}