cc: update scaled transform to occlusion tracker during append quad.

cc: update scaled transform to occlusion tracker during append quad.

We were missing sending scaled transform to occlusion tracker
while retrieving UnoccludedContentRect in during append quad,
hence the visible_geometry_rect was going wrong.

BUG=384967
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=279794

[sohanjg, 2014-06-24 16:18:02]

Please take a look, this is not final soln, but something in the direction to
solve the issue.

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
File cc/layers/picture_layer_impl.cc (right):

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
cc/layers/picture_layer_impl.cc:142: float max_contents_scale =
MaximumTilingContentsScale();
If we avoid scaling to max scale here, the black patch issue is not observed.

The only way i could fix it keeping the scale is setting the draw properties
again before setting up SQS, i am not sure maybe LTH still needs the updated
transform/bounds/rects.

But, this causes some flicker in gmail, maybe im still missing something or
there approach is wrong :(

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
cc/layers/picture_layer_impl.cc:148: gfx::ToCeiledSize(gfx::ScaleSize(bounds(),
max_contents_scale));
this is keeping same behavior as old code, where we set scale the layer bounds 
and set it as content bounds in CalcContentScale, this reduces the flcker.

[enne (OOO), 2014-06-24 18:06:11]

What is the underlying problem?

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
File cc/layers/picture_layer_impl.cc (right):

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
cc/layers/picture_layer_impl.cc:148: gfx::ToCeiledSize(gfx::ScaleSize(bounds(),
max_contents_scale));
On 2014/06/24 16:18:01, sohanjg wrote:
> this is keeping same behavior as old code, where we set scale the layer bounds

> and set it as content bounds in CalcContentScale, this reduces the flcker.

This does not seem right to me.  This makes scaled_content_bounds (bounds * max
content scale) and then you intersect with visible_content_rect (in the space of
bounds * content scale [i.e. 1]).  That is incorrect.

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
cc/layers/picture_layer_impl.cc:155: draw_properties().target_space_transform =
scaled_draw_transform;
Sorry, but no.  Draw properties should only be set in UpdateDrawProperties and
should be considered const outside of that.

[sohanjg, 2014-06-25 13:26:53]

The problem here is the content_bounds and visible_content_rect which we send to
the SQS, i think on mobile gmail page as we adjust the layers bounds when we
touch the screen, the calculation here is having some issues.

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
File cc/layers/picture_layer_impl.cc (right):

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
cc/layers/picture_layer_impl.cc:148: gfx::ToCeiledSize(gfx::ScaleSize(bounds(),
max_contents_scale));
On 2014/06/24 18:06:11, enne wrote:
> On 2014/06/24 16:18:01, sohanjg wrote:
> > this is keeping same behavior as old code, where we set scale the layer
bounds
> 
> > and set it as content bounds in CalcContentScale, this reduces the flcker.
> 
> This does not seem right to me.  This makes scaled_content_bounds (bounds *
max
> content scale) and then you intersect with visible_content_rect (in the space
of
> bounds * content scale [i.e. 1]).  That is incorrect.

Do you mean the scaled_content_bounds calculation ? or the
scaled_visible_content_rect calculations ?
What is you suggestion here ? Should we avoid the intersection ? but can we have
visible content outside content bounds ?

https://codereview.chromium.org/350183005/diff/20001/cc/layers/picture_layer_...
cc/layers/picture_layer_impl.cc:155: draw_properties().target_space_transform =
scaled_draw_transform;
On 2014/06/24 18:06:11, enne wrote:
> Sorry, but no.  Draw properties should only be set in UpdateDrawProperties and
> should be considered const outside of that.

Yes, this was just to try and fix the black patch issue.

[sohanjg, 2014-06-25 14:36:54]

This should fix it, we were missing sending scaled transform to occlusion
tracker while retrieving UnoccludedContentRect, hence the visible_geometry_rect
was going wrong.
Please take a look, Thanks.

[enne (OOO), 2014-06-25 16:55:15]

lgtm, thanks!

[danakj, 2014-06-25 17:12:46]

Oh awesome, LGTM2

[danakj, 2014-06-25 17:12:48]

The CQ bit was checked by danakj@chromium.org

[commit-bot: I haz the power, 2014-06-25 17:15:08]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/sohan.jyoti@samsung.com/350183005/100001

[commit-bot: I haz the power, 2014-06-25 20:01:56]

Change committed as 279794

[sohanjg, 2015-11-18 09:38:02]

Description was changed from

==========
cc: update scaled transform to occlusion tracker during append quad.

We were missing sending scaled transform to occlusion tracker
while retrieving UnoccludedContentRect in during append quad,
hence the visible_geometry_rect was going wrong.

BUG=384967

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=279794
==========

to

==========
cc: update scaled transform to occlusion tracker during append quad.

We were missing sending scaled transform to occlusion tracker
while retrieving UnoccludedContentRect in during append quad,
hence the visible_geometry_rect was going wrong.

BUG=384967

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=279794
==========