| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Using webkit to compute animations is still slow in some cases.
When animating large elements, we seems to sometimes bogs the GPU,
which then makes the UI takes longer to render a frame. This in turn
slow the rate at which we can call into webkit (to update the position
of the animated layers), which results in perceived stuttering.
We previously had an implementation of CSS animations that could run
fully on the UI thread, without having to call back into webkit.
We turned it off because there was still some glitches, and calling
into webkit seemed to work well enough -- but as we can see, even
if that's the case in general, edge cases still benefit from running
the animations outside of webkit.
The CL fixes the remaining glitches we had (mostly, it was the
non support of fillMode) and re-enable our CSS animations implementation.
bug:5297559
Change-Id: I1f00bc060a76c9dfd55bd6d8ae85d5d6da68ddb5
|
|
|
|
|
|
|
|
| |
This also improves performances a bit when animating small layers.
bug:5472082 bug:5297559
Change-Id: I0b9fa0a9abf98ab53928f97f53c3134056b15889
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
bug:5461107
Tiles were being destroyed, and subsequently dereferenced in TransferQueue
Change-Id: I4fea289e5fda03a69f07554f57120c4c5bf7b016
|
|/
|
|
|
|
|
|
| |
bug:5320471
Also, reduce prefetch distance
Change-Id: I9e677c16b5d0cdc16b87bd820591088ab10ab44b
|
|\ |
|
| |
| |
| |
| |
| |
| | |
Bug: 5282993
Change-Id: Ia47c420d867d3aaed8a8ea6f91285cb3b634d970
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
bug:5452786
Fixed layer initialization to reference its current layer's SkPicture when it's
created
Change-Id: I4ac33ad6874a982d375859445e2a014ff35d7e12
|
| |/
|/|
| |
| |
| |
| |
| |
| |
| |
| | |
- no layout / bgd color check if no inval on the base layer
- throttle the number of ImageTexture uploaded per draw call
- disable fast layer position update for now
bug:5297559 bug:5421309 bug:5218173
Change-Id: I9ff9867dd3cfc2e0805e378d75ea75667fef8673
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This is part 2 of 2 of the fix.
See http://trac.webkit.org/changeset/89836
Bug: 5249309
Change-Id: If23a41a8e75e5fb6ba9e84e481ca44a319f678ef
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This is part 1 of 2 of the fix.
See http://trac.webkit.org/changeset/89165
Bug: 5249309
Change-Id: I1159aab70e1fa6d3025761423c4164682dc97b64
|
|\ \ |
|
| |/
| |
| |
| |
| | |
bug:5452814
Change-Id: Ib13a0f4e5189af0e4928bc376b29610e20eeb4dc
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| |
| | |
Bug: 5453034
Change-Id: I426aa4ca1b1c253f99c32aaf8cad597c09144af0
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
bug:5453156
simplified m_owner management - ONLY modified on UI thread
Previously deferring clear of m_owner, now set immediately since texture thread
doesn't use need stale owner info
Change-Id: I9b6f7bfd27af44a818d378ff750c0170d8122cfd
|
|\ \
| |/
|/|
| | |
into ics-mr0
|
| |
| |
| |
| |
| | |
bug: 5454889
Change-Id: I34e125d4b6aa5b9c0da7017a975e80761b6183df
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
We add a separate Images Manager, working
both in GL and software rendering.
bug:5425148
Change-Id: Ib5bc88ad94d49c7f1eb969a716691ac98dc10b77
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously the frame inversion state was passed to the UI thread
when the layer tree was synced. This CL removes that restriction
and updates it immediately by storing the state in a cross thread
refcounted object.
bug: 5283034
Change-Id: I58a7ea0f2ed261008b945241bc783cfee4f33466
|
|\ \
| |/
|/| |
|
| |
| |
| |
| |
| | |
bug:5367327
Change-Id: I94589e007f9669fa1b43ab5cee42bfa44d9e503c
|
|\ \ |
|
| |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
bug:5349958
Clear a flag when a layer tile can't allocate a texture, and only try and redraw
from tile dirtiness if that flag is set.
Also, don't ask for redraw if offending tiles are offscreen.
Change-Id: Iadb0cb267a9c1f308e5b42a6e0e3b4bc71d18ece
|
|/
|
|
|
|
|
|
|
|
|
|
| |
bug:5409902
Tiles were being discarded from the queue (and simply unsuccessfully
added). This caused them to get stuck in the 'ValidatedUntransferred' state.
Now if a tile isn't added successfully, or if it's discarded, it removes its
painting texture and will have to repaint from scratch.
Change-Id: I551e00fb8a6be3b0f3cabeabaa91e8b8b30019d5
|
|\ |
|
| |
| |
| |
| |
| | |
bug: 5436630
Change-Id: I058a0d750f185c56c0fe70ac0bff7c9781c35dc8
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Bug 5383517
Data URLs were being stripped, preventing long click from doing anything.
Change-Id: I5eba97725375a155d39f2489f97714a2b8bc8189
|
|\ \ \
| |_|/
|/| | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
bug:5309934
Defer picture changes and invalidations when:
1) a tiled texture hasn't completed painting
2) the user is scrolling (done in PaintedSurface)
Remaining issues:
Should block updates on zooming
Change-Id: I60ee722e651f96e95a4ac0e819999fc1589037c3
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This CL refactors our implementation to support multiple
non-overlapping video surfaces per plugin instance. This
will allow plugins to hardware decode simulatenous video
streams using hardware.
Additionally, this CL improves plugin startup latency by
constructing the primary plugin surface as soon as possible
in the UI thread.
bug: 5318290
Change-Id: I53ceb8b6696159ecc4765c922704727a9a75a17d
|
|\ \ \
| |/ /
|/| | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Bug: 5354191
Refactor how extras are drawn in GL
Change-Id: Ifb61c73abecd2d074903a7fb44fe4b82aacc64ac
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
With the new direct update, we are setting directly the layers position;
but for ScrollableLayer, we already set their position in the UI, so for
them their webkit counterpart is actually older/wrong.
The fix consists simply in not updating the positioning of
ScrollableLayer.
bug:5421867
Change-Id: Ia14e41aff641173a45ed898cd3b3eceab3f7bd7a
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The GPU texture upload workaround broke it, as we are not using the
Shaders when doing the copy. Adding some code to instead use a final
shader that can be either normal or inverted, on demand.
bug:5424589
Change-Id: If845233e12c65738925fbc81b21f1dc0fd38546e
|
|/ /
| |
| |
| |
| |
| | |
bug:5425608 bug:5218173
Change-Id: Ib14a5e53466b05781a842e24443e89a4779dba68
|
| |
| |
| |
| |
| | |
bug:5297412
Change-Id: I6b483e174339bc854e5d301c5c1f2e5851e57772
|
|\ \
| | |
| | |
| | | |
transform and position. This makes updates faster and less dependent on other webkit work."
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Directly update the layers transform and position.
This makes updates faster and less dependent on other webkit work.
counterpart java CL: https://android-git.corp.google.com/g/#/c/139853/
bug:5218173
Change-Id: I03a76ab853e81f0f12177fb785707ffb8dace330
|
|\ \ \
| |_|/
|/| | |
|
| | |
| | |
| | |
| | |
| | |
| | | |
Bug: 5410522
Change-Id: Id8457715665dbd51761900011beba57b7f634fff
|
|/ /
| |
| |
| |
| |
| | |
Bug: 5406350
Change-Id: I95507ee163a2cb41cc605efaabf32b7cad814533
|
| |
| |
| |
| |
| |
| | |
bug:5382635
Change-Id: I0da21245ec7808da452941a45ae5008e278f3007
|
|/
|
|
|
|
| |
Bug: 5353510
Change-Id: I1bb306381864357dab4c38fd5c182fda2dd08d8c
|
|\ |
|
| |
| |
| |
| |
| | |
bug:5395019
Change-Id: Ifc16eb59fec62e5f4b95ad4e35f140969226506d
|
|\ \ |
|