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
Need to clear the attached hidden flag on the window, even if it does
not yet have an actual surface.
Change-Id: I0a90f6b03cd8e347a3eae4d6bea3ab93f7128bf4
Throttle is useless when ntp and imsi are not yet ready. In the real world
the second check won't happen for 10 minutes letting people do what they want
for 10 minutes each boot. Trim this down with a quicker first check.
bug:2614247
Change-Id: I84ec33f31e8fed37b3933482fbc187b8e9e94e79
Previously, it was conceivable that a 3rd party application on a non-GED
device could publish a service that supported the (hidden) IBackupTransport
interface and spoofed the Google backup transport's package and component
name. This could allow it to secretly intercept all data moved through the
backup mechanism.
Fix by ensuring that the package in question exists and is part of the
OS itself (FLAG_SYSTEM in its ApplicationInfo description) before binding
to it.
Fixes bug #2457063
Change-Id: I3487572be45c2014fa209beacfe3ac6f8270f872
An ANR has been reported because somebody else was hogging the Netd
link. We can't pause the server thread for that, so make a new thread.
bug:2609904
Change-Id: I0430a08621da1cda0a0010ce15826cceee59a529
Cleanup of duplicate code
Caching Ntp data so we don't hit the network as often
Clearing out of MRU data when we switch to a new sim
Sending notification after we reset byte counts
bug:2605277
bug:2605272
bug:2597530
bug:2576057
Change-Id: I3df526bae6374f6ae56fed67b0defad2b4fabefd
The value returned is purely a function of what's passed in, all
following final members.
BUG=2606839
Change-Id: I506fb9b1f3be965576c2b3317a30ff1f990ccda3
Previously, this status icon reflected just the incoming
call behavior:
- gone for audible (with or without vibrate),
- wiggling phone for vibrate,
- silent (speaker + X) for no audible ring or vibrate.
Now that silent mode has changed (see VIBRATE_IN_SILENT),
you can easily get into a state where this icon is
confusing by setting the incoming call ringer to *never*
vibrate but setting the overall device silent state to allow
vibration (VIBRATE_IN_SILENT=true). In this case, rocking
the volume all the way down would show a wiggling phone
(because the device is in vibrate-only mode) but the status
bar would show speaker+X (because the incoming call would
neither ring nor vibrate).
Now these two icons are consistent. This means, however,
that if this special condition is set by the user (device
vibrates in silent mode, but incoming calls should never
vibrate) and the device is in silent mode, the status bar
will show a wiggling phone even though incoming calls will
not vibrate the device. This puts the ringer on a more even
footing with other apps on the device, which may elect not
to vibrate even if VIBRATE_IN_SILENT=true overall.
Bug: 2598014
Change-Id: Ib942f68738d77ab0d9b9da4fac718ae5957c133c
Use elapsed time not wall time for alarms so users can't play with the
system time to get around things.
Also using NTP servers to pull in an authoritative time - if we the build
is configured with an NTP server we will not advance to the next cycle
without it, but we also will not trottle - rather not throttle users
on an error.
Note that the poll alarm is just relative to the last poll time and real
time doesn't matter.
Defining the time-fetching API's as returning time in the system wallclock
range (correcting if we are using NTP time internally).
bug:2597530
Change-Id: I1c0ac0923314c2f8a04edd0b36c4845352eae99a
In lightSensorChangedLocked we ignore the light sensor if the battery is low.
But in applyButtonState() and applyKeyboardState() we were still using the previous
mLightSensorScreenBrightness value, which resulted in a race condition that could
leave the button lights on after the screen turns off.
Now we ignore the light sensor value and button brightness override if the battery
is low so the low battery behavior is consistent.
Change-Id: I4943f8904299883211a95596ee207df69d1eaea2
BUG: 2570962
Signed-off-by: Mike Lockwood <lockwood@android.com>
The NotificationPlayer sets an OnCompletionListener on the
MediaPlayer it's using to play the notification.
NotificationPlayer relies on the completion event to abandon
audio focus.
The player was started before the listener was set. The theory
about the bug is that the player finished playing before the listener
was called. Therefore the audio focus was never abandonned, and
the music didn't resume.
Change-Id: Ic3baf359ea24de0d832a655dc47e745b25bc3dec
Uses the last used data until the SIM is available (if ever). Supports data from
several SIMs for all the world travelers out there.
bug: 2576057
Change-Id: I70e34a51f1c2ccda41a480652b0233b68ff3f538
The point of this patch is to deal with the following case:
- A system application exists in the system partition. As such its
native code lies under /system/lib
- An upgrade for it is installed through Market and overrides the
version on the system partition. This places its native libraries
under /data/data/<pkgname>/lib/
- Finally, the system is upgraded through an OTA, with an even more
recent of the system application. We thus must remove the Market
package as well as the native libraries that are in <dataDir>/lib
since the new versions in /system/lib will be more recent.
The patch itself removes the files in <dataDir>/lib when such a
situation is detected. Note that this directory is created by the
PackageManager service and owned by the 'system' userID. The application
cannot create files there so simply erasing the content of the directory
is enough (and avoids reparsing the market package file to list
the installed binaries).
The patch also fixes a different issue which can be reproduced with:
- have system app installed on /system partition
- install an update (e.g. adb install -r Term.apk)
- uninstall the update (e.g. adb uninstall com.android.term)
the logcat shows that the PackageManager will unpack the native libraries
from the original /system/app/Term.apk into <dataDir>/lib, while they
already are in /system/lib. The patch fixes that too.
Change-Id: I5dfe27b91b6b31972e9cd4c196d15c3692732127
Couple of cosmetic fixes as well.
This reverts commit 0c1285fa979e8b8c170f31d0ef1b4e5d9ddcf6b1.
Conflicts:
services/java/com/android/server/PackageManagerService.java
Change-Id: Ife0f9f80525f4a41b1f8c12ffc2de62e061e7d5b
when the install location is set to internal only. Similary if install
location is set to external only(via adb), dont fall back to installing on
internal storage if sdcard is filled up.
If nothing is specified ie install location is set to let system decide,
we just check on internal storage.
Fix tests
Change-Id: I7400ccc131782d9c45284bd9ebd0dee43a3b412b
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
The byte counter for the iface goes to zero when it comes back. Catch that and
do the complex math to correct.
Also add a log entry telling us our desired throttle index.
Lastly, reset the sec/msec part of our reset time so that one calc will == another and not
be off by little bits.
bug:25494925
Change-Id: I99525b72a76316a8e601ceb457ca202fae040bae
...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
The window manager doesn't retain its keep screen on lock while
in the process of destroying and creating new activity windows during
a config change. Now it does.
Unfortunately, this doesn't help maps, since it doesn't use our
lovely KEEP_SCREEN_ON flag. :p
Change-Id: Ida4bc0fcfe132cdff2bd4eeb7ba0ce9731504d14
...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
Never got SharedPreferences working for this system service. Didn't really need it anyway.
Using a straight up file now.
bug:2576057
Change-Id: Idad926e8242a85e22718c25f3f6c03e5749badac
Don't use hard coded defaults - use build configs instead.
Iface is now ONLY set by config so gservices doesn't need to.
bug:2576057
Change-Id: I8f9e3bc51af3b1cdf8bb4290a43197d9c01e2a14
in status bar.
When unmounting sdcard, all the secure containers have to be unmounted first
before invoking the MountService call back. We send a broadcast
about disabled packages. Force a gc to clean up any stale asset references
and then unmount the containers before invoking the MountService call back.
Change-Id: I831da1ae54aeec89de3601b94a4a2fae5dc7e495
...go directly to unlock screen
Fix a problem where, when we are moving windows due to a configuration
change, our wait to have the windows get redrawn would also
inadvertently make us think their surfaces should be hidden, leading
to nasty things like the lock screen thinking it should be shown
again (and thus forcing the orientation back to portrait).
Change-Id: I79057cfbc93ff59fd3558da997d4f73796e6aabe
