Mac: Enable partial swap

Mac: Enable partial swap

This is pretty trivial when using the overlay path. Just create a
new layer for the damage. The partial damage layer can be created,
removed, or resized with basically no cost, because its backing is the
already-existing IOSurface.

As each partial swap comes in, union the existing overlay layer's
rect with the rect of the new partial swap. If this grows "too big",
then do a full swap (which clears the accumulated union). The
constants here are pretty arbitrary.

BUG=474299
Committed: https://crrev.com/aae371ca10e047d9b192acf869f8ff9601d3cdf7
Cr-Commit-Position: refs/heads/master@{#342771}

[ccameron, 2015-08-05 03:06:00]

ccameron@chromium.org changed reviewers:
+ andresantoso@chromium.org

[ccameron, 2015-08-05 03:06:00]

(much rejoicing)

[Andre, 2015-08-05 17:59:30]

Very nice!

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
File content/common/gpu/image_transport_surface_overlay_mac.mm (right):

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:28: -(void)
setContentsChanged;
- (void)set...

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:111: double
kMaximumFractionOfFullDamage = 0.85;
nit: const

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:112: double
fraction_of_full_damage =
nit: const

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:185:
content_bounds_rect.height())];
How about something like:
CGRect damage_rect = this_swap.dip_partial_damage_rect.ToCGRect();
[partial_damage_layer_ setFrame:damage_rect];
[partial_damage_layer_ setContentsRect:
    CGRectApplyAffineTransform(damage_rect, CGAffineTransformMakeScale(
        1. / damage_rect.size.width, 1. / damage_rect.size.height));];

Also, I feel like using a shorter name like damage_layer_ instead of
partial_damage_layer_ might make this
code more readable. WDYT?

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:199: [layer_
setContents:static_cast<id>(this_swap.io_surface.get())];
setContents:new_contents

[ccameron, 2015-08-06 18:55:17]

ccameron@chromium.org changed reviewers:
+ danakj@chromium.org

[ccameron, 2015-08-06 18:55:18]

Thanks -- updated.

danakj, could you take a look at the ui/gfx/geometry bits (copy-pasted from
rect.cc/h)?

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
File content/common/gpu/image_transport_surface_overlay_mac.mm (right):

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:28: -(void)
setContentsChanged;
On 2015/08/05 17:59:30, Andre wrote:
> - (void)set...

Done.

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:111: double
kMaximumFractionOfFullDamage = 0.85;
On 2015/08/05 17:59:30, Andre wrote:
> nit: const

Done.

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:112: double
fraction_of_full_damage =
On 2015/08/05 17:59:30, Andre wrote:
> nit: const

Done.

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:185:
content_bounds_rect.height())];
On 2015/08/05 17:59:30, Andre wrote:
> How about something like:
> CGRect damage_rect = this_swap.dip_partial_damage_rect.ToCGRect();
> [partial_damage_layer_ setFrame:damage_rect];
> [partial_damage_layer_ setContentsRect:
>     CGRectApplyAffineTransform(damage_rect, CGAffineTransformMakeScale(
>         1. / damage_rect.size.width, 1. / damage_rect.size.height));];
> 
> Also, I feel like using a shorter name like damage_layer_ instead of
> partial_damage_layer_ might make this
> code more readable. WDYT?

Actually, gfx::RectF should have a ToCGRect function, so I added that (so that
we only go to the CG types at the last minute).

https://codereview.chromium.org/1271123002/diff/1/content/common/gpu/image_tr...
content/common/gpu/image_transport_surface_overlay_mac.mm:199: [layer_
setContents:static_cast<id>(this_swap.io_surface.get())];
On 2015/08/05 17:59:30, Andre wrote:
> setContents:new_contents

Done.

[Andre, 2015-08-06 20:02:03]

LGTM

[danakj, 2015-08-06 20:30:48]

Woa, CGRect is floats? Maybe we shouldn't have those Rect conversions that they
should only be RectF then. LGTM for this anyhow.

[ccameron, 2015-08-06 20:35:07]

Thanks!

On 2015/08/06 20:30:48, danakj wrote:
> Woa, CGRect is floats?

CGRect is CGFloat, which is double!

> Maybe we shouldn't have those Rect conversions that they
> should only be RectF then. LGTM for this anyhow.

[ccameron, 2015-08-11 00:36:08]

The CQ bit was checked by ccameron@chromium.org

[ccameron, 2015-08-11 00:36:09]

The patchset sent to the CQ was uploaded after l-g-t-m from danakj@chromium.org,
andresantoso@chromium.org
Link to the patchset: https://codereview.chromium.org/1271123002/#ps40001
(title: "Rebase")

[commit-bot: I haz the power, 2015-08-11 00:36:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1271123002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1271123002/40001

[commit-bot: I haz the power, 2015-08-11 01:41:52]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-08-11 01:43:01]

Patchset 3 (id:??) landed as
https://crrev.com/aae371ca10e047d9b192acf869f8ff9601d3cdf7
Cr-Commit-Position: refs/heads/master@{#342771}