These exceptions are normal when an app has cancelled an outstanding
thumbnail request.
Bug: 11385378
Change-Id: I8aed5721b447cda5baf447ac7afd627aa1062863
Target the two biggest offenders:
- Coalesce keyguard setHidden(false) calls during unlock.
- Make sysui->WM call async.
Found during investigation into b/11221659.
Bug: 11221659
Change-Id: Icab48376bc356a933e0f9940bc2f924e2e77ab22
The ActivityChooserModel keeps a history of the last fifty
share targets and based on past usage orders the targets in
the UI. The soring implementation is using a map for improving
performance. However, the activities in this map were keyed
on the package name but there maybe more that one share
target per package. Thus, the sorting was generating bad
results. Now the unique component name is used.
bug:11195578
Change-Id: I8c7018fea168b7253ddbe57b477028368726e75e
This changes filters out share targets that we cannot start
because the target does not properly implement the SEND protocol
and has either share target activity hidden or requires a
permission to launch it. Also the code that launches the share
target activity catches the runtime exception and shows an error
message. Note that being able to launch an activity in a moment
of time is not a guarantee that one can do that latter. Hence,
being able to launch an activity while building the share UI
does not guarantee that one can launch it when selecting the
share target.
bug:11402139
Change-Id: Id35732510755b2eeb9eccacc046d289c2f2ee856
1. Implemented the advanced printer options integration. Now a print service
may declare an advanced print options activity which may be launched by
the user if the current printer supports advanced print options. These options
are visible only to the print service that added them and it is the only party
that will interpret the options.
2. Fixed a couple of bugs in the saved print jobs parsing. One was that if there
are more than one page range, a half of the print job properties was not
properly parsed. The other was that the media size constructor was using
incorrect argument order, thus creating a media size with wring width.
3. Fixed and edge case where old print jobs and their docs can get stuck in
the spooler. If the app did not write the requested pages we were not showing
an error message, rather just finish the activity without canceling the print
job and this print job is stuck in the spooler. Now we show an error message
and the user may retry, cancel. If the user cancels the print job is also
cancelled, thus no leftover in the spooler.
4. Fixed the background color of the print dialog to meet UX spec.
bug:11241800
Change-Id: I352440bc86aec824a805883fc9579d96a06d11e6
With the new tuned vsync offset, vsyncs are likely to occur shortly
after the input is received, meaning we will empty the input queue,
and thus won't schedule input consumption until more input is
received. If an application then speculatively posts draw commands to
the main looper faster than 60 hz, it will eventually end up blocking
in eglSwapBuffers. Since we're blocking in eglSwapBuffers, we won't
even schedule consumption until after the current frame (8-16ms), and
it's entirely likely we won't actually get around to consuming input
until after the next frame (another 16 ms of latency). This means we
can often go 16-32ms without processing any input events, causing
very noticeable amounts of jank.
Rather than waiting for the next input event to schedule input
consumption, speculatively schedule it every frame as long as we've
consumed some motion batch during this frame.
Bug: 11398045
Change-Id: I25e46308e00e9f9de00a1d8906f6b0e0f2e845b4
Also always lower-case extension to extract MIME type correctly,
we may want to fix in MimeUtils at some point.
Bug: 11354402
Change-Id: I77d0c06a663cd0c4320c41553ff1980b6f9a7778
If the remote side of PFD has already written a status message, then
they've also closed their end of the comm FD, and we're going to
EPIPE if we try sending our own status. So, skip writing status if
a remote status is present.
Only one end of the openFile() comm socket needs to be blocking,
otherwise detachFd() would end up blocking forever.
Bug: 11385467
Change-Id: I346d40cc1ca4a6683cec4c2d2b7db2b32ac94a55
This prevents Keep from being selected automatically when shown in the Share dialog
if Keep was previously chosen by the user as Always use for a send action.
This was a regression introduced by the new intent disambig behavior.
Bug: 11294904
Change-Id: I6745060a8ee0a6d680e657b55ee46aaec27bbacb
The current tracking of scene objects in a static ThreadLocal is
problematic as it leaks the Context associated with the SceneRoot and
returns the wrong Scene object if the same layout ID is used across
different scene roots.
Track Scene objects on the scene root view instead to avoid these
issues.
Change-Id: I891986897f757f2666897c937b5ebb0ed1d531c1
The android package is now a special case, not being added to the package list
when creating a multi-process component. There is no need, since this package
is actually the framework itself which must be loaded in every process.
Also cleaned up some of the procstats dump output to help see what is going
on here.
Change-Id: If65d35ecd562f3154bdebfded69c454af6ce8c96
Transitions, when started, add an OnPreDrawListener to the current
ViewTreeObserver (which is global to the view hierarchy). This listener
is removed when the listener is called.
It is possible to add this listener and then remove the view from
the hierarchy before the listener is called. This could result in
either the listener not getting called at all (since there was no
drawing event) or (in the case of this bug) the listener getting called
when the sceneRoot had no AttachInfo (which is the case when that
root has been removed from the hierarchy). This results in the listener
trying to remove itself from a *different* ViewTreeObserver than the one
it added itself to, leaving the actual listener still sitting on a list
of listeners in that original VTO. This can result in a growing list of
listeners and a growing amount of work that gets done on every frame.
It can also lead to a serious memory leak, since the objects referred to
by the transition may be non-trivial (as in the case of this bug).
The fix is to add another mechanism for the listener to get removed.
Specifically, we now listen for detach events on the sceneRoot. If that
view gets detached before the listener is called, then we have a chance to
remove it from the correct VTO before the AttachInfo becomes null.
Issue #11307391 keyguard is slow after updating to krt16c and playing music
Change-Id: I108413ea2f18f5351df0a11d4ae56fec0b4aa154
The APIs are createBond, setPin, setPairingConfirmation
The intent is ACTION_PAIRING_REQUEST
bug 11101076
Change-Id: I3a314efd973b3ce078ab5347159c336f222d9f15
Bug 11291911
These deleted classes were previously public APIs and so need to remain
in the build (but hidden) in order to keep existing apps working.
(Partially reverts Change-Id: I02549a71104b35d86d99058c71f43e054730ec7d)
Change-Id: I28e53b056f41e66645136f5e18fba2ff55a65fe5
Although the IME windows are now allowed to extend into
the nav bar, some IMEs were making assumptions about
computed insets based on the height of the content view.
So our navigation bar view (opaque view blocking the nav bar
area to avoid the island effect when transparent) needs to live
above the content view in the hierarchy, making the content view
the same height as it was before.
A surgical spot to put the guard view is up at the root view
(PhoneWindow.DecorView). fitSystemWindows is always called since
this view is not recreated, and the layout is stable: waiting until
the IME is attached to the window is too late to add a guard view.
This is above the screen_* layouts, so will work without having to
touch all of them. And it only affects windows of TYPE_INPUT_METHOD.
Bug:11237795
Change-Id: I6a93f30aec83f1cecfb854073046cbc87ab4aa66
1. For an app to print it creates a PrintDocumentAdapter implementation
which is passed to the print dialog activity. If the activity that
created the adapter is destroyed then the adapter, which may rely on
the activity state, may be in an invalid state. For example, an app
creates an adapter and calls print resuting in the app activity and
the print dialog activity being stacked. Now the user rotates the
device which triggers the recreating of the activity stack (assume the
app does not handle rotation). The recreated print dialog activity
receives the intent that originally created it with containing the
adapter that was constructed in the context of the old, now destroyed,
app activity instance.
To handle this we are limiting an app to be able to print only from
and activity and when this activity is destroyed we mark the adapter
as invalid which will result in hiding the print dialog activity. Note
that if the app process is killed we already handle this in the print
dialog activiy by registering a death recipient on the adapter binder.
2. In the PrintManager.PrintDocumentAdapterDelegate some of the state is
accessed only on the main thread and some from miltiple threads. The
code was trying to avoid locking for state that is not accessed by
multiple threads but this is error prone and the benefit does not
justify the complexity and added fragility. Now grabbing a lock all
the time.
3. The PrintJobConfigActivity waits for it to bind to the print spooler
service before instantiating its print controller and editor. However,
these can be accessed by invoking some of the activity cycle callbacks.
This change is adding null checks for the case where the activity
callbacks are called before the binding to the spooler is completed.
bug:11242661
Change-Id: Id906b3170e4f0a0553772dfa62686f06fdca0eaf
