Haven't found the underlying cause, but this will give us more
information when we get into the bad state.
Change-Id: I9aebd3a025a7c0d931f43098461b64ee3c220746
We now have the activity manager kill long-running processes
during idle maintanence.
This involved adding some more information to the activity manager
about the current memory state, so that it could know if it really
should bother killing anything. While doing this, I also improved
how we determine when memory is getting low by better ignoring cases
where processes are going away for other reasons (such as now idle
maintenance). We now won't raise our memory state if either a process
is going away because we wanted it gone for another reason or the
total number of processes is not decreasing.
The idle maintanence killing also uses new per-process information
about whether the process has ever gone into the cached state since
the last idle maintenance, and the initial pss and current pss size
over its run time.
Change-Id: Iceaa7ffb2ad2015c33a64133a72a272b56dbad53
In the case where the top task is finishing and another task is
launching make sure that the next task will be launched once the
top task actually completes pausing.
In the case of b/10550460 the top task, Dialtacts, was finishing
but had not yet completed pausing. It was configured to return to
the home screen (mLaunchHomeTaskNext true) but because its finishing
flag was set all the tests we have thought that the InCallActivity
task was the top task. When it finally did complete the
mLaunchHomeTaskNext flag caused the home activity to be started
instead of the InCallActivity.
If the InCallActivity task had been moved above the Dialtacts task
at the time it was judged to be the top task the home activity
would not have been launched when Dialtacts completed. This fix
moves the judged top task over the finishing top task.
Fixes bug 10550460.
Change-Id: I14052ad2282008679b560dd7fb16b216664ec24d
Dumb typo was clearing the wrong service array, causing
us to sometimes forget we were launching a service.
Change-Id: Ie1aba0e07d19e85a104a5985e3cead5f28a0556a
This significantly reworks the logging we do when
all cached processes are killed:
- We now collect the list of processes in-place so we
have a snapshot of exactly when the low memory situation
happened.
- In that snapshot we include the key process state: oom
adj, proc state, adj reasons.
- The report then asynchronously collects pss information
for those processes.
- The ultimate data printed to the log looks like a mix
between the "dumpsys meminfo" and "dumpsys activity"
output. This code no longer uses "dumpsys meminfo"
itself, so some of that data is no longer included,
in particular pss organized by allocation type.
In doing this, I realized that the existing code that is
supposed to run "procstats" is not currently working. And
at that point I realized, really, when we are collecting
this pss data we'd really like to include all those native
processes using ghod-only-knows how much RAM. And guess
what, we have a list of processes available in
ProcessCpuTracker.
So we now also collect and print information for native
processes, and we also do this for "dumpsys meminfo" which
really seems like a good thing when we are printing summaries
of all pss and such.
I also improved the code for reading /proc/meminfo to be
able to load all the interesting fields from there, and
am now printing that as well.
Change-Id: I9e7d13e9c07a8249c7a7e12e5433973b2c0fdc11
Because recents sits on the same stack as launcher it can sometimes be
above launcher. When we were launching home activity because the flag
told us to we would sometimes launch recents instead. This fix makes
sure that the home activity is on the top when it is supposed to be
launched next.
Previously this was fixed by having recents move itself to the back
of the stack after it launched an activity (b/9750207 and ag/336019).
But that solution caused the AppTransition to be set to
TRANSIT_TASK_TO_BACK which left the SOFT_INPUT_IS_FORWARD_NAVIGATION
flag unset. This in turn caused IMEs to remain unlaunched when
returning from recents (b/10240567).
Fixes bug 10240567.
Change-Id: I35c6619af0e68d0e6d9ab87cad06ea7c27e11e27
Added some code to the activity manager to keep track of
services that are launching and limit the number that can
be launched concurrently. This only comes into play under
specific circumstances: when the service launch is a background
request (so timing is not important) and its process is not
already running at a high priority.
In this case, we have a list of services that are currently
launching and when that gets too big we start delaying the
launch of future services until currently launching ones are
finished.
There are some important tuning parameters for this: how many
background services we allow to launch concurrently (currently
1 on low-ram devices, 3 on other devices), and how long we
wait for a background service to run before consider it to be
a more long-running service and go on to the next pending
launch (currently set to 15 seconds).
Also while in here, did some cleanup of the service code:
- A little refactoring to make per-user data cleaner.
- Switch to ArrayMap.
Change-Id: I09f372eb5e0f81a8de7c64f8320af41e84b90aa3
Before:
4539 does not have permission:android.permission.CLEAR_APP_USER_DATA to clear datafor process:com.android.chrome
After:
PID 4539 does not have permission android.permission.CLEAR_APP_USER_DATA to clear data of package com.android.chrome
Change-Id: Ic466decb050e4fa7f3fee4098c4f2abdc6cedf5c
Reverts extension to assist context API to query
foreground services for assist context data.
Also hides Intent.ACTION_VOICE_ASSIST because
nobody's actually using it yet.
Bug: 10461702
Change-Id: Idf6836adc659b434e11ebb2b98e8b814c94a7227
Make it a little easier to diagnose input dispatch timeouts by
providing the detailed reason as the ANR annotation in the log.
Bug: 10689184
Change-Id: Ie18fd9ad066b0673d1f57c030e027ad0085f4650
In cases where the client is waiting for an activity to launch
(startActivityMayWait()) it is a bad idea to clear
ActivityRecord.displayStartTime when going into the pause state. If
displayStartTime is cleared before the activity is displayed,
the client will never be released.
This fix keeps pause from clearing displayStartTime if any client
is waiting for the activity to be displayed.
Fixes bug 10095558. But not a permanent fix, startActivityMayWait()
should not be called by any production code.
Change-Id: I7cbdcb04256f4a26233867c52aedd3bc4151adc3
Changes for translucent activity were causing activities to be
launched twice due to a recursive call into resumeTopActivity.
Putting the translucent action onto a handler removes the recursivity
and fixes the multiple launch problem.
Fixes bug 10556969.
Change-Id: I2bb53cd555b0aaf093ab35db2859acb10b58211e
We now keep track of which process and service states are actively
in use, and remove any that are not in use during a commit. The
activity manager needed to be tweaked to report this data, and ensure
it does not try to operate on one of these structures when not in
use.
Also some other fixes:
- We now keep track of process names associated with services, for
display in the UI.
- Keep track of total run time for each service, also for UI.
- The parceled format is more efficient, not storing duplicates of
process/package names, and writing times as ints when possible.
- Reduced commit period from 1 day to 12 hours, so that our UI can
be a little closer at its attempt to display the stats over 1 day.
Change-Id: Ifeda0ffe963a7b49d8eb2a3f6923f3a5e71a4e43
http://ag/323631 added the provideAssistData
attribute to the Service manifest tag.
It did not, however, add that attribute to
public.xml - making it impossible to actually
build a service that defines said tag.
So, add it.
Now that the attribute is where it should be,
restore @link notation to ServiceInfo.java
without breaking the offline docs build
( see http://ag//340279 ).
Also, make some log warnings related to
providing assist data slightly more verbose.
Bug: 10573008
Change-Id: Ie2bcb411c182d69738a2fa4a74de3171b9b9c455
There was a bug in counting the number of starting services
in a process that would cause it to count too many (it would
increment at any state change while the service is started, not
just when starting/stopping).
Also reduce dumpsys output -- only print summaries for old
data. There is probably no utility in printing the long details
of all that data.
Change-Id: I1c1e901b0214c01eb7d071f23166fc6f3702ca67
- There is an assumption that moving a task to the back moves it
behind the home activity. This fix ensures that the home activity
will be launched by some activity before the task that has been
moved to the back.
- Convert "private final" declarations to private only.
- Convert container tests for size()==0 to isEmpty().
- Remove unused variable.
Fixes bug 10228536.
Change-Id: I51741ee7a09652e27afffcc50c14953bf0aa27ea
When relaunching an activity re-evaluate the flag that determines
whether to return to the next activity on the stack or to the home
activity upon return.
Fixes bug 10545580.
Change-Id: I5868ce1b494afffcc5afc22abe3b65aa339827bb
Since ContentProvider file operations can end up doing substantial
network I/O before returning the file, allow clients to cancel their
file requests with CancellationSignal.
Ideally this would only be needed for openFile(), but ContentResolver
heavily relies on openAssetFile() and openTypedAssetFile() for common
cases.
Also improve documentation to mention reliable ParcelFileDescriptors
and encourage developers to move away from "rw" combination modes,
since they restrict provider flexibility. Mention more about places
where pipes or socket pairs could be returned.
Improve DocumentsContract documentation.
Bug: 10329944
Change-Id: I49b2825ea433eb051624c4da3b77612fe3ffc99c
When Activity.convert{To|From}Translucent() is called the
ViewRootImpl is now notified when it is safe to convert the Canvas
from translucent to opaque and back to translucent. This will make
it possible to save resources when compositing opaque layers.
Fixes bug 10349536.
Change-Id: I7282aee1d54601fb00611d20be204bf164d873f6
Having a single mHomeProcess in ActivityManagerService ignores the
possibility of multiple processes serving as home. Particularly when
we have a home activity that only serves to launch the true home
activity.
Fixes bug b10296766
Change-Id: I9e5c9c955c64f4b3df856b6bc66d6a0d93f85aab
The warning about an ActivityRecord not being assigned a task was
incorrect in the location modified by this fix. In this case an
existing ActivityRecord is relaunched so it is not necessary to
assign the task to the passed ActivityRecord.
Addresses the warning associated with bug 10181389.
Change-Id: I76d5066c320bf9da2663bc34bcaca801ad4953bc
Now that granted Uri permissions can be persisted across reboots,
offer APIs to discover them. Returns incoming or outgoing grants
matching the requested flags and mask. Add helper method to discover
"open" documents using this new API and filtering for non-documents.
Require that callers own at least of the filtering packages to avoid
exposing all grants. Switch internal grant tracking to use ArrayMap.
Change-Id: I0a755f221d0d160b411f8d3cfc48279b64345733
