This gives us a snapshot of what the system process was doing after 30 seconds
of apparent inactivity as well as after 1 minute, to help distinguishing actual
deadlocks from too-slow progress, livelock, etc.
Change-Id: I19758861d1b25f298e88788e8f1c7ec7bf828823
They're designed to handle one time upgrade code, which may
take a significant amount of time. Third party code cannot
receive this broadcast so the timing should be able to be
kept under control.
Bug: 2713849
Change-Id: I3733b6530198bebd9b8c2ba6f9f27a7999336b8d
It looks like there was a subtle bug where Process.setOomAdj() could
return false just because the given process doesn't exist, even though
it is documented to only return false if OOM killing is not supported
at all. This would cause the activity manager to fall into its code
path of trying to clean up processes itself, which it does a much
poorer problem at. I am thinking we may be seeing this problem more
now that the activity manager is killing background processes itself
when there are too many of them.
In addition, this change cleans up and reduces some of the logging
around killing processes.
Finally, try to improve process LRU management a bit by taking
into account process dependencies. Any dependent processes are
pulled up in the LRU list with the processes that is actually
moving. Also, we bring a process up if someone accesses its content
provider.
Change-Id: I34ea161f839679345578ffe681e8d9c5d26ab948
With the .apk file names now changing during an update, we need
to make sure to flush all caches related to a package when the
package is removed. Otherwise we can continue to use the old
package, since its old file may still exist if we try to load it
too soon.
Change-Id: I15f08dffca3feac999dbca4f24bef12a30ca0a66
When a service transitions from foreground to background, we now push it
to the top of the LRU list. Also fix the activity manager to take care
of killing processes if we go beyond a reasonable number of background
process to keep around.
Change-Id: Ic9f44c02af7a111ee6f1d06142386b301948bafe
The value returned is purely a function of what's passed in, all
following final members.
BUG=2606839
Change-Id: I506fb9b1f3be965576c2b3317a30ff1f990ccda3
Make sure, if an activity is started with clear task when reset, if that
activity is finished in the middle of a stack that the behavior is retained
by propagating it to the next activity.
Change-Id: Ie31b8f968558b0e64e0ef7efa55950a722c6afa5
...ordered broadcast for ACTION_EXTERNAL_APPLICATIONS_UNAVAILABLE
Turns out this was because the broadcast receiver for ContextImpl was
not correctly being created, so when it received an ordered broadcast
it would not tell the activity manager when it was done.
This is now fixed, along with a ton of superficial changes to debug
output to help track this down and a little cleanup of dealing with
error cases in dispatching broadcasts. Also a fix for a NPE when
dumping the broadcast state.
Finally, a little fiddling with package manager to get rid of a lot
of the noise when removing and re-adding packages on the SD card.
Change-Id: I961c14836dc613d3ea8122b6e910ef866e7fcb25
...SD card causes reboot.
Add option to not include temporarily unavailable recent tasks,
for the recent tasks UI to not display them.
Change-Id: I1495ca217b4292fd56f537459b44c0624262c292
Also a little tweak to the activity manager to behave better when
an application crash, to hopefully mostly avoid situations where
you get into a crash loop.
Change-Id: I627cc1da3a0f16a180957f02bfbe5c81ecd31758
Don't hold the activity manager lock the entire time we are processing an ANR,
which often causes us to cause a second ANR as someone else calls into the
activity manager and gets blocked.
Change-Id: Ife4db82b12d6f4378ac2705bd6f60beb1244a1e6
There was a really dumb bug that was causing us to not always apply
the new configuration. As a result of fixing this, there were new
glithes in the transition between car and regular mode, so further
work here to fix that. And since I was actually working during the
night and seeing night mode, I noticed how obnoxiously bright the
status bar is compared to the car home at night, so it now nicely
dims itself when we switch to the night config. Oh and in doing
that I also found and fixed a bug in dispatching config changes to
a window (where they wouldn't get dispatched if the window didn't
resize).
FINALLY... tweak the wallpaper enter/exit animations a bit to
make them a little smoother.
Change-Id: I4e062093cbcfbc919307799a794846db2920216f
Use new method in UsbStorageActivity.
Fix moving dex files.
moveDex should be suffixed with LI since it uses Installer
Change-Id: Id5ef0254578e84b9aae2c2ac44f722eb5a0fda1c
Fiddle with how we go into car mode to try to ensure we get a clean
transition. Also have the system take care of remembering the night
mode setting so it will stay at what you want.
Change-Id: Icb94fdd961c7a192f7707ec71544485a1ea12455
PackageManager invokes this call back when its done handling
the media status update.
Add new uid check for updateExternalMediaStatus
Change killPids method in ActivityManager.
Remove mountsd command in Pm.java We cannot arbitrarily enable/disable
packages in PackageManager now.
Change-Id: I28dcba4afd2b4486f68abdaa1628a31b66544c91
Decouples the input dispatch thread from the battery-stats object lock regime,
to avoid the possibility of ever blocking the input dispatch thread on its
behalf. The stats object is widely used and can sometimes be locked for a
very long time (on the order of seconds) during certain extensive dump
operations.
This change does not alter the data format of the battery stats' externalized
representations.
Fixes bug #2530346
Change-Id: Iee288be3bf4936641b532dceecb8f6de8f552bf0
No steps to repro, but makes the code more robust by using the standard
JournaledFile class and doing sanity checks on the input it reads.
This required moving the JournaledFile class in to the framework (and
we really should get rid of either it or AtomicFile, but they have
different recovery semantics so that is tough). Also went through and
cleaned up the file management in various places.
Change-Id: Ieb7268d8435e77dff66b6e67bb63b62e5dea572e
This allows us to keep the assigned permissions when apps are temporarily
removed due to the SD card being unmounted, and also if you use the
facility to uninstall an app but keep its data.
Also fixes issue #2515189: Potential permission spoofing attack in
Android (external bug 7166)
Change-Id: I2a120ec938552028c989f9e0e890c32773957738
Fixes issue #2437252: Starting activity by means of startActivityForResult
causes 5 seconds delay if "android:windowIsTranslucent" is true
The optimization to avoid showing an activity window when a new
activity is being started was a little too aggressive. Now it
avoids doing this if there is not actually a fullscreen activity
on top to cover it.
Change-Id: I630e37a1f1d3b874b5a25572cbf887cebc2e3e91
Shelling out to logcat from the system server makes me queasy,
so this is turned off by default -- it must be enabled individually
for each error type (system_app_anr, etc) via a secure settings
value (which I plan to poke into from gservices for internal use).
Even when enabled, it happens in a side thread, unless the system
server is about to die anyway (system server restart).
Change-Id: Id6d88bcd78d3625f0364a5fe9c771046601a5a14
ShutdownThread.reboot() does return so we need to block after calling it
to prevent PowerManager.reboot() from returning.
Since PowerManager.reboot() can now take significantly longer than before,
we now ignore ANRs during shutdown.
Change-Id: Ibceeb265ae382567215f6a399108d8be3a7bbc95
Signed-off-by: Mike Lockwood <lockwood@android.com>
Fix issue #2493497: Stuck in the Emergency dialer - Home/Back keys doesn't work
This was another case of not updating the window focus when needed, this time
when the lock screen was hidden.
Also re-arrange the layout/animate flow to address issues where you would see
a flicker of whatever was behind the lock screen when showing a new activity that
hides the lock screen. This was because we were deciding to hide the lock screen
during the layout phase, which meant we had to do it without considering whether
it had drawn. So we could hide the lock screen before the window is shown for the
first time after being drawn. Now we can do this in the policy during animate, so
we can wait until the window is drawn and actually being shown.
The flow in perform layout is thus significantly changed, where the layout and
animate loops are both under the same repeating loop. The actual flow from this
should be the same, but it now allows the policy to request a new layout after
the animation loop is done. This actually cleans up a number of things in this
code as the complexity has increased.
Finally this includes a change to the ui mode manager when switching modes, to do
the resource configuration switch at a different time. This makes transitions
between modes much cleaner (though not yet perfect).
Change-Id: I5d9e75c1e79df1106108dd522f8ffed6058ef82b
Several functions operate on variables to which access needs to be
synchronized. However, it happens that the functions in question
are only ever called from places which have already synchronized.
Therefore, nothing is really wrong, but the functions ought to
have 'Locked' appended to their names, to indicate that it is the
caller's responsibility to synchronize before calling them.
Change-Id: I44e7dc0dff6da9436677cb10908dce41ffeba195
Capture stack traces from the system process using the same
mechanism as ANRs (which will initialize traces.txt, etc).
Also record the watchdog reset in the dropbox for uploading.
Bug: 2475557
Bug #2376231: Apps lose window focus (and back key causes ANR) if the
lock screen is dismissed while the phone is in landscape mode
This is another case where we weren't recomputing the focused window
after changing the visibility policy.
bug #2479958: Investigate source of "Resources don't contain package
for resource number 0x7f0a0000"
Um, okay, so it turns out there were bugs all over the place where
we would load an XML resource from a another application, but not
use the Resources for that application to retrieve its resources...!
I think the only reason any of this stuff was working at all was
because it typically only cared about retrieving the resource
identifiers of the items (it would look up the values later).
Bug #2401082: Passion ERE26 monkey crash - InputMethodManagerService
Add some null checks.
There exists a race condition when starting a process that recently has died.
If the ActivityManager receives the death notification for the died process
after the new process has been started but before an application thread has
been attached to the new process will the newly created process be removed
during the cleanup of the died process. If this happens when sending a broadcast
could it result in an ANR.
This is solved by doing the clean up before starting a new process that uses
the same process record.
Add null checks to a few places to avoid crashes when dumping
debug data.
Also add some sanity checks for accessing content providers in
the activity manager.
2185256: After open &close of device keyboard shortcut does not added to Home desktop.
ActivityThread was losing the last saved state when restarting or launching into
a paused state.
2366174: defaults not cleared when they should be
PackageManagerService now removes any preferred activity records for a package
when it is uninstalled.
2154556: Battery stats can have an unbounded number of wake locks
We now start combining wake locks into one shared record when we hit a
maximum limit (currently 20).
2442519: Foreground services can have no notification by providing a bogus one.
If the notification manager rejects our notification, the service is forced to
no longer be in the foreground.
2442383: Finalization issues in com.android.server.am.PendingIntentRecord.java
Cleaned up finalization to call super class and avoid the big activity manager
lock (we still need to use the locks inside of the message system, but these
are much less likely to be a problem).
2284190: Cannot call a phone number using adb
We weren't getting the calling uid/pid in startActivity() if the caller did not
supply an application record.
I am getting tired of writing package monitor code, realized this is missing in
a number of places, and at this point it has gotten complicated enough that I
don't think anyone actually does it 100% right so:
Introducing PackageMonitor.
Yes there are no Java docs. I am still playing around with just what this
thing is to figure out what makes sense and how people will use it. It is
being used to fix this bug for monitoring voice recognizers (integrating the
code from the settings provider for setting an initial value), to replace
the existing code for monitoring input methods (and fix the bug where we
wouldn't remove an input method from the enabled list when it got
uninstalled), to now monitor live wallpaper package changes (now allowing
us to avoid reverting back to the default live wallpaper when the current
one is updated!), and to monitor device admin changes.
Also includes a fix so you can't uninstall an .apk that is currently enabled
as a device admin.
Also includes a fix where the default time zone was not initialized early
enough which should fix issue #2455507 (Observed Google services frame work crash).
In addition, this finally introduces a mechanism to determine if the
"force stop" button should be enabled, with convenience in PackageMonitor
for system services to handle it. All services have been updated to support
this. There is also new infrastructure for reporting battery usage as an
applicatin error report.
This is a bunch of reworking of how configuration changes are handled:
- When orientation is changing (for whatever reason), the window manager no
longer tries to pre-emptively compute a new configuration. Instead, it
just determines change is happening and tells the window manager.
- The activity manager is now responsible for giving the window manager the
final configuration it is using. This is both so it knows whem the
activity manager is done with its configuration updates, and so the window
manager can use the "real" configuration.
- When an orientation or other configuration change is happening, freeze the
screen and keep it frozen until the activity manager has given us the
final configuration.
- The window manager can now send new configurations to its clients during
its layout pass, as part of a resize, if it has determined that it has
changed. This allows for a new View.onConfigurationChanged() API for any
view to easily find out when the configuration has changed.
- ViewRoot now also works with the activity thread to make sure the process's
current resources are updated to the new configuration when it receives one
from a window. This ensures that at the time onConfigurationChanged() and
other view callbacks are happening, the correct configuration is in force.
- There is now a sequence number associated with Configuration, which
ActivityThread uses to avoid using stale configurations. This is needed now
that it can receive configurations asynchronously from both the window
manager and activity manager.
- The hack for keeping the locale has been removed, and underlying problem
fixed by having Configuration initialize its locale to "unknown" instead of
a valid default value.
Merge commit '6371a2d31f30b473e91be9cced64a2527c086af1'
* commit '6371a2d31f30b473e91be9cced64a2527c086af1':
Don't crash the system process when apps give us a bad foreground service notification.
Merge commit 'ee3bbefd34fd5330ebbc59175a328197ab7526af' into eclair-plus-aosp
* commit 'ee3bbefd34fd5330ebbc59175a328197ab7526af':
Don't crash the system process when apps give us a bad foreground service notification.
For the system server process, do the disableJitCompilation/startJitCompilation
callbacks depending on whether the system is in safe mode or not.
In addition, if the system is found to be in safe mode, a flag will be set in
the Zygote class which will be used to launch subsequent apps in VM safe mode.
Bug: 2267590
Refactoring of the dumpsys framework introduced a comparison that
always made the action dump all the services.
Change-Id: I1f3ac2bd903e39e2a6e8632e8cbb23a982dfe0c6
Include a proper name for processes that are in the background
with running services that have been running for a long time (instead
of showing them as bg-empty).
Batch together multiple processes into the same background bin when
there are significantly more processes than bins. Arguably this
should be smarter and base the binning on the number of background/empty
processes instead of the total count, but this should be good
enough for now.
