| 91b22809648a33d64c159e8496773b1b3b2ab6ca |
|
29-Jun-2016 |
Robert Carr <racarr@google.com> |
Only use one SurfaceControlWithBackground per AppToken.
In the past, if an app never renders to a SurfaceView, it will be
invisible despite having FLAG_OPAQUE. This means an app could leave a
totally empty SurfaceView (never drawing in to it) on top of a second
SurfaceView, and expect the second one to be visible. This is probably
buggy app behavior because FLAG_OPAQUE means if they ever draw anything at all
in to the top SurfaceView the bottom one will become totally invisible.
However this has worked in the past, so we have to preserve things for
apps. To accomplish this we ensure only the bottom most visible
SurfaceView for a given AppToken will receive a background. We achieve
this by synchronizing through the app token whenever visibility or
layering of a SurfaceView changes.
Bug: 29580298
Change-Id: I0023326323cb961b56404fd49093384e7b72aa54
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 938a70142ebb504ef2e079708c48315f80707b53 |
|
11-Jun-2016 |
TreeHugger Robot <treehugger-gerrit@google.com> |
Merge "Restore setCropInTransaction HiddenForCrop behavior." into nyc-dev
|
| 086d2929f739ab217c38f99b76fd70f30dcfa478 |
|
09-Jun-2016 |
Robert Carr <racarr@google.com> |
Place a black layer behind SurfaceView.
For SurfaceView's which do not implement
SurfaceRedrawNeeded we can see holes through to the
background in various scenarios. We prevent this by
placing a DimLayer behind opaque SurfaceViews. SurfaceFlinger
will remove it from the visible layers when the SurfaceView
totally obscures it.
Bug: 28763785
Change-Id: I9b4d1ba8be786f30432a6a3b42e5616662cdc090
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 4320d33a065173dff97ddb462c8a6a0a649a41e5 |
|
11-Jun-2016 |
Robert Carr <racarr@google.com> |
Restore setCropInTransaction HiddenForCrop behavior.
Prior to c/1106850 setCropInTransaction hid the surface
for crop width and height <= 0. That CL allowed
setting -1 for crop width and height which SF
determines as clearing the crop. Other portions of the code
depend on the old behavior though for negative values, so restore
the behavior of setCropInTransaction and use a new clearCrop
method for the new code.
Bug: 29276588
Change-Id: I728666009c362ff635c7ebfb3ef2e83428fb03fe
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| a9408d4a4809dd229fb7fb8f9594cb6db4b1da64 |
|
03-Jun-2016 |
Robert Carr <racarr@google.com> |
PiP animation: Move window with resize when ending animation.
At the end of the animation (when going from larger to smaller),
we are left with a scaled surface, that we want to seamlessly
resize to an unscaled surface of the new size. Because we have scaled
the shadow region of the surface, the position of the content
will differ before and after the resize applies. We use new
SurfaceFlinger API to cause position updates to apply after
resize. Because we have to switch into SCALING_MODE_FREEZE,
we could end up prematurely cropping the window, so we
switch to using screen space crop for the pinned stack.
Bug: 28899837
Change-Id: I9b762a237413e4fa3d432e67d30c7125bfef484c
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| fed100742011bbc3092925c37e9f2df4ab04b6b5 |
|
26-May-2016 |
Robert Carr <racarr@google.com> |
Ensure surfaces only resize during relayout.
For clients which use an indeterminate measure spec in their
layout params (MATCH_PARENT). We could try and change their window size
due to a stack resize, before the client calls relayout. This isn't
safe as the client never had an opportunity to pause rendering, so it
could be in the middle of producing a frame for the old size to suddenly
find the buffer size change underneath it.
Bug: 28559097
Change-Id: I3982936fdf85c22def2f9c754d5508e029e4a84d
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 3cc58ddd3c39580d0c008d1cf7138d5634be1b90 |
|
21-Apr-2016 |
Chong Zhang <chz@google.com> |
Fix dead surface/window left on screen
If the surface is explicitly hidden while we're changing resizing mode,
we need to discard the preserved surface immediately, otherwise it could
linger around on screen.
Do not mark mAnimatingExit in setTokenVisibilityLocked. This flag can only
be set after client relayouts to invisible. There we set mAnimatingExit and
mAnimating, so that when animation finishes the surface is marked to be
destroyed. If we marke it too early (and not mark mAnimating at the same
time), the surface may get stuck in exiting state.
bug: 28236574
Change-Id: I5cc916dfa698113e1d783a66547e1878469a9879
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 1ca6a33f36357281b3445e85d9e67cacd1a12ede |
|
12-Apr-2016 |
Robert Carr <racarr@google.com> |
Force windows to be scalable during pinned animation.
We resize windows at the beginning of the pinned stack
animation when animating to a larger size, and so for some
duration a resize will be pending. We need to force the window
out of SCALING_MODE_FREEZE so we can animate during this period.
Bug: 27891386
Change-Id: I5cff599ed67f2c179e938662b6f0d99bd790aaba
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 6a7c90a12b5e5250e0350d35ca6547b26630653f |
|
11-Mar-2016 |
Jorim Jaggi <jjaggi@google.com> |
Implement clip modes during animation
Introduce modes how surfaces are clipped to the stack bounds
during transitions.
Use setFinalCrop to implement STACK_CLIP_AFTER_ANIM.
Add logic to determine which stack clip mode to use.
Bug: 26559810
Change-Id: I8edc47de3aaf1ef12055cefd8ceb8df536c5109a
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 628e27f1053c3e2f94f59656b82b9cefd75c6e2e |
|
18-Mar-2016 |
Rob Carr <racarr@google.com> |
Merge "Include more methods in SurfaceTrace." into nyc-dev
|
| ae35fef616c14f721cf12d6a0f78bef9fef5df9d |
|
16-Mar-2016 |
Chong Zhang <chz@google.com> |
Apply temp inset bounds to child window if it's not empty
bug: 27676101
bug: 27687126
Change-Id: Iab1129cc0224194e7a9d11d0454bc7af6897a6e8
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 6a19b4ff88913ed8db8432d9cc89f919e655f2d4 |
|
15-Mar-2016 |
Robert Carr <racarr@google.com> |
Include more methods in SurfaceTrace.
SurfaceTrace was missing setTransparentRegionHint.
Change-Id: I3fc946493cc0daa0ab5ad613feef3e99a11ac8f8
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 2728bf4bf9c4bc73837f41d3bf0251a7de6c9e16 |
|
03-Mar-2016 |
Wale Ogunwale <ogunwale@google.com> |
Added more log points for add/remove in window manager.
Bug: 27286867
Change-Id: Iecb522a1ff7e093a8feef27fdd68c50b9a80d553
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 47e36a3e270ff3e94750d730ac2a9f0bdfe96c04 |
|
01-Mar-2016 |
Chong Zhang <chz@google.com> |
Force disconnect when the surface is about to be saved.
Some client will not disconnect, and if we're saving the surface (instead
of destroying it), we need to make sure the surface is disconnected.
Otherwise the client won't be able to reconnect to the same surface.
bug: 27295820
Change-Id: I471b8fbe8f590c900e17a017167466fc8a70b87a
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 1eb39b009ef6cfa133cbd5d6ea8caaac64ee1559 |
|
25-Feb-2016 |
Jorim Jaggi <jjaggi@google.com> |
Start with mHiddenForOtherReasons=true
Initially a surface is hidden after creating, so reflect that fact
in mHiddenForOtherReasons, so it doesn't get shown before the correct
transformation is applied.
Bug: 27350530
Change-Id: Idfe8723195dd1488894da066fafee1f2c2afc12d
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 5e6968db630d26872306c4b23aa2c600d83ed454 |
|
20-Feb-2016 |
Jorim Jaggi <jjaggi@google.com> |
Fix disappearing windows after moving divider to side
Because we only hide the surface when the clip rect got empty
but never showed it again if it got non-empty, app windows
disappeared after moving the docked stack divider to the
edge of the screen. Now we reshow the surfaces if the clip-rect
gets non-empty.
However, this introduces another bug while dismissing the docked
stack: Because we move all windows to the fullscreen stack, we resize
them but until the app transition starts, it can take a while and
during this time the app surface would be visible with the wrong
bounds. To fix this, we notify the windows that we are repositioning
ourselves in our stack. When applying the clip-rect, we detect that
situation and then we set the clip rect to empty if it was just empty
before and we just moved in the stack, to fix this very specific
issue.
I'm really not proud of this solution but at this point we can't
revisit how app transitions are executed in terms of timing and
ordering, so I thought this little hack is the best solution at
this point.
Bug: 26588506
Change-Id: I78305f7f7ef6c3da3c126a58d751117fcee23ca9
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 722ff89fd1684179c342c3e9b6014c5499f90b90 |
|
18-Feb-2016 |
Wale Ogunwale <ogunwale@google.com> |
Protect against surfaceController and hasSurface getting out of sync.
WindowStateAnimator.mSurfaceController is set to null whenever a
surface is destroyed and WindowState.mHasSurface is set to false
shortly after that. However, it is possible for them to get out
of sync in a couple of places due to exceptions or duplicate destroy.
Consolidated the call to set WindowState.mHasSurface inside a finally
block in WindowStateAnimator.destroySurface
Also, cleaned up the code a little to that it is more obvious what is
going on.
Bug: 27235356
Change-Id: I7e6d0c1fb015531c393ac86dcaebebd134fad612
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 0464a9345e7e3e666a84a3489cfffd5e45a21353 |
|
03-Feb-2016 |
Chong Zhang <chz@google.com> |
Disable WM loggings
bug: 26819496
Change-Id: Ib5258391101293c9aaca3b865e1f1580234bcbe3
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| e1034cc39a574c3e8dc481810fb8f24571c41d70 |
|
01-Feb-2016 |
Robert Carr <racarr@google.com> |
Ensure Surface matrix set even when size unchanged.
Simple error preventing transform matrices from being
updated when the size isn't changed.
Bug: 26454664
Bug: 25287371
Change-Id: I0e1a71ceda725e26d49786593665cf0865213c91
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 198dcbf5231761b7b644d9d7fbceb23e1f0f9aec |
|
18-Jan-2016 |
Filip Gruszczynski <gruszczy@google.com> |
Fix several small logging issues.
This includes:
1) invert HIDE_STACK_CRAWLS to SHOW_STACK_CRAWLS so it's immediately
clear from the config file that something is enabled (if anything is
true).
2) Merge stack collection code into a method, so we can remove the
repeated code.
3) Remove copying of some constants in AppTransition and just import
them directly.
Change-Id: I3190ee0a5963720ac6285b4f48b2705e84f04ab5
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| ce4f2f18b720b2abe90f87e2066c538f51ee544b |
|
15-Dec-2015 |
Robert Carr <racarr@google.com> |
Temporary enable logging for monkey testing
Temporary enable logging for tracing 26157153 in monkey testing.
Bug: 26157153
Change-Id: I85aef42c426ceaa831b7919a37dd35ef78e9c779
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 0bd180d8880b3d1b9677f154c034a2af840b4796 |
|
08-Dec-2015 |
Filip Gruszczynski <gruszczy@google.com> |
Improve debugging setup for window manager package.
This moves debug flags to a dedicated class and also provides a
mechanism to either have fine grained tags or one common tag for easier
grepping the logcat. This mimicks the approach in activity manager
package.
Change-Id: Ie8c08b9a3d8b182335ee5547ee05d21b5933db6b
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 19723a4a2bca0660f7ee7c29926af285d94ab5a2 |
|
26-Nov-2015 |
Filip Gruszczynski <gruszczy@google.com> |
Revert "Destroy docked divider surface when it's hidden."
This reverts commit cb5f57bc580d7f73b93c8f504daa4dcd74cdce72.
Change-Id: I1f77e1ccd5382ed57b8e4165afd79db5223f51c1
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| cb5f57bc580d7f73b93c8f504daa4dcd74cdce72 |
|
24-Nov-2015 |
Filip Gruszczynski <gruszczy@google.com> |
Destroy docked divider surface when it's hidden.
Also includes bunch of small refactorings:
* destroying surfaces is now fully contained within
WindowManagerServices and mDestroySurface can be privatized;
* WMS.isDockedStackResizingLocked can be removed;
* mScreenCaptureDisabled changes from being SparseArray<Boolean> to
SparseBooleanArray, which not only avoids boxing but also makes code
simpler (no need to check for null)
Bug: 25844096
Change-Id: I0e5462760ffbc947ce6dc52ef429fa270ffc6786
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 78a08ee876794586e1d429e67d4b94209415ea5a |
|
09-Nov-2015 |
Filip Gruszczynski <gruszczy@google.com> |
Fix blink when docking a window.
We need to be very precise about removing the old window when the new
one is shown. The right moment is immediately after the first frame
of the new window entrance animation gets commited. This requires
more infrastructure and flags, rather than depending on the old ones
and hoping that they will match our needs.
Bug: 25075151
Change-Id: I0fdbf38d0915ded3300d75fc7268e619b574bcf5
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 10a80e0b5b2efb3a65b66dfa8b6fee6d0904ea42 |
|
06-Nov-2015 |
Filip Gruszczynski <gruszczy@google.com> |
Fix windows disappearing when resizing freeform or docked.
Also includes some code clarity improvements: mHasSurface is set using a
setter, some fields get private.
Change-Id: I2f834880493c008fdccf07ff6ebfebd2e26690a9
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| 63a35e2343468a04e360f0514c6c9dc03068c185 |
|
06-Nov-2015 |
Filip Gruszczynski <gruszczy@google.com> |
Fix minor problems when resizing/maximizing docked window.
When maximizing the transition should originate from visible bounds, so
the first frame matches what is visible to the user. When switching to
the big surface, we only need to increase the layer by one, instead of
having artificially large value. If we use the large value, it will
cause a flicker over system windows.
Also includes some cleanup, like static imports and necessary logging.
Bug: 24913915
Change-Id: I84d7594622aa639e2008c662f941edf9c20b3202
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|
| e6a8351bc715999d1e42dcc1003a6eda6c318dd9 |
|
04-Nov-2015 |
Robert Carr <racarr@google.com> |
Extract application window usage of SurfaceController.
Abstract the usage of SurfaceController from wm w.r.t
application windows in to a new WindowSurfaceController
class. This class tracks boring book keeping, logging,
errors, etc...to lend clarity to difficult policy
code in WindowStateAnimator et al.
Change-Id: Ifcd5d48a51e68564f49e799ae793b320cac88645
/frameworks/base/services/core/java/com/android/server/wm/WindowSurfaceController.java
|