Cyclic references can occur between a Service object held by an
application and a ServiceRecord object held by the system server.
A part of the problem is that binders are leaked and since many binders
are implemented by inner classes of services these services are also leaked.
This causes low memory problems. The solution is: When a Service is beeing
destroyed, go through the ServiceRecord's all IntentBindRecord and set its
binder references to null. This allows the binder and the service object to
be garbage collected.
Change-Id: I5a257521964851f34c08ffb3908feaad96b1bafe
When using sendOrderedBroadcast(..) with a BroadcastReceiver the
BroadcastReceiver instance was not released. The reason for this was that
the resultTo field in the BroadcastRecord kept a reference until it was pushed
out of the mBroadcastHistory. This reference in turn kept a reference to the
process side IIntentReceiver (implemented in ReceiverDispatcher$InnerReceiver).
This in turn had a strong reference (through mStrongRef) to the Context.
In order to keep the debug output the resultTo is also kept as a String in the
new resultToString variable.
Change-Id: I4382a22a541c27b3694fb2b78a04ee820b235f8f
Cyclic references can occur between a Service object held by an
application and a ServiceRecord object held by the system server.
A part of the problem is that binders are leaked and since many binders
are implemented by inner classes of services these services are also leaked.
This causes low memory problems. The solution is: When a Service is beeing
destroyed, go through the ServiceRecord's all IntentBindRecord and set its
binder references to null. This allows the binder and the service object to
be garbage collected.
Change-Id: I5a257521964851f34c08ffb3908feaad96b1bafe
When using sendOrderedBroadcast(..) with a BroadcastReceiver the
BroadcastReceiver instance was not released. The reason for this was that
the resultTo field in the BroadcastRecord kept a reference until it was pushed
out of the mBroadcastHistory. This reference in turn kept a reference to the
process side IIntentReceiver (implemented in ReceiverDispatcher$InnerReceiver).
This in turn had a strong reference (through mStrongRef) to the Context.
In order to keep the debug output the resultTo is also kept as a String in the
new resultToString variable.
Change-Id: I4382a22a541c27b3694fb2b78a04ee820b235f8f
ServiceRecord's bindings is a hashmap to keep track of all active
bindings to the service. This is not cleared when the service is
brought down by activity manager. This adds up the references to
IntentBindRecords and its references to ServiceRecord. Fix is to
clear the bindings.
ServiceRecord's restarter is a reference to the service and is not
cleared when the service is brought down by activity manager. This
adds up the references to ServiceRecord. Fix is to set the reference
to null when the service is brought down by activity manager.
Change-Id: Ica448cd5f60192c8adb23209b5d0e2cf0c04e446
The service connection to the previous live wallpaper is unbound when
a new wallpaper is effective. Although the service connection is
unbound it is not disconnected and its reference to wallpaper's
service and engine is still effective. This adds up to the total
JNI references and causes dalvik (hosting system_server) to abort.
Fix is to release the references in clearWallpaperComponentLocked.
Change-Id: Idd2bab83a56d2e6c6dd7ab9be08d5e14887aa384
ServiceRecord's bindings is a hashmap to keep track of all active
bindings to the service. This is not cleared when the service is
brought down by activity manager. This adds up the references to
IntentBindRecords and its references to ServiceRecord. Fix is to
clear the bindings.
ServiceRecord's restarter is a reference to the service and is not
cleared when the service is brought down by activity manager. This
adds up the references to ServiceRecord. Fix is to set the reference
to null when the service is brought down by activity manager.
Change-Id: Ica448cd5f60192c8adb23209b5d0e2cf0c04e446
...current process has android.permission.WAKE_LOCK
When updating a system app, we would actually uninstall the package
of the system app, which also meant removing its uid...! It was just
luck that we would get the same uid when installing the update after
that. During that time, if anyone tried to do anything related to
that uid, it would be unknown.
This change tweaks how we go about replacing system apps by making
it more like normal apps -- to make this work, if we need to disable
the system app, we generate a new PackageSetting from the current
system app and replace it into our data structures, so we can update
that without trashing the current correct information about the (still
actually there) system app.
Also fixed a problem where we were not killing the currently running
app before installing, like we do when updating a normal application.
And fixed a problem where we were not deleting the /data .apk when
uninstalling a system app update.
And added a new option to the "pm" command to clear the data associated
with an app.
Change-Id: I0e879677849aa42950a3c360bf78ad820e87674b
Rewrote interceptKeyBeforeQueueing to make the handling more systematic.
Behavior should be identical except:
- We never pass keys to applications when the screen is off and the keyguard
is not showing (the proximity sensor turned off the screen).
Previously we passed all non-wake keys through in this case which
caused a bug on Crespo where the screen would come back on if a soft key
was held at the time of power off because the resulting key up event
would sneak in just before the keyguard was shown. It would then be
passed through to the dispatcher which would poke user activity and
wake up the screen.
- We propagate the key flags when broadcasting media keys which
ensures that recipients can tell when the key is canceled.
- We ignore endcall or power if canceled (shouldn't happen anyways).
Changed the input dispatcher to not poke user activity for canceled
events since they are synthetic and should not wake the device.
Changed the lock screen so that it does not poke the wake lock when the
grab handle is released. This fixes a bug where the screen would come
back on immediately if the power went off while the user was holding
one of the grab handles because the sliding tab would receive an up
event after screen turned off and release the grab handles.
Bug: 3144874
Change-Id: Iebb91e10592b4ef2de4b1dd3a2e1e4254aacb697
This fix improves the performance by caching the string that should
be returned, and reuse it next time if possible.
This will make it faster to switch between activities, approximately
half the time to create the new view when changing from landscape to
portrait. Also, the time for starting a new application is be reduced
as WindowState.toString is being called thousands of times in this
case.
Change-Id: I2b8b9bc1e251d1af43b6c85f049c01452f2573a2
Reports that we sometimes didn't report NO_CONNECTIVITY led to this suggested change.
Could not repro the problem, but the change looks ok anyway. Better safe than sorry.
bug:3276408
Change-Id: I0cdb48a05a5c9dfcf3a0b468a6eae43d461023b1
This fix improves the performance by caching the string that should
be returned, and reuse it next time if possible.
This will make it faster to switch between activities, approximately
half the time to create the new view when changing from landscape to
portrait. Also, the time for starting a new application is be reduced
as WindowState.toString is being called thousands of times in this
case.
Change-Id: I2b8b9bc1e251d1af43b6c85f049c01452f2573a2
Includes some other small fixes to battery collection and a few
other things.
Output of package info looks like this:
5,0,i,uid,1000,com.android.settings
5,0,i,uid,1000,com.android.providers.subscribedfeeds
5,0,i,uid,1000,com.android.providers.settings
5,0,i,uid,1000,com.android.server.vpn
5,0,i,uid,1000,android
5,0,i,uid,1000,com.android.systemui
5,0,i,uid,1000,com.google.android.backup
5,0,i,uid,1001,com.android.phone
5,0,i,uid,1001,com.android.providers.telephony
5,0,i,uid,1022,com.android.nfc
5,0,i,uid,10021,com.google.android.location
5,0,i,uid,10021,com.google.android.syncadapters.calendar
5,0,i,uid,10021,com.google.android.gsf
5,0,i,uid,10021,com.google.android.syncadapters.contacts
5,0,i,uid,10026,com.android.providers.downloads.ui
5,0,i,uid,10026,com.android.providers.media
5,0,i,uid,10026,com.android.providers.drm
5,0,i,uid,10026,com.android.providers.downloads
5,0,i,uid,10032,com.android.launcher
5,0,i,uid,10039,com.google.android.gm
5,0,i,uid,10041,com.google.android.gallery3d
5,0,i,uid,10049,com.android.providers.calendar
Change-Id: I9e38f254eef146339113ad270f5c6e8b60fb7a1d
To monitor the dropbox an application have to either poll the dropbox
and keep track of all entries or observ the /data/system/dropbox
directory. The later requires that the application runs as system-user.
This commit adds that a broadcast intent is sent when something is written
to the dropbox and an application can just listen on this intent and
then reads the entry with help of the DropboxManager class.
The application have to hold the permission android.permission.READ_LOGS
to get the intent.
Change-Id: I1f77f206a243df69f4ed5306078c47f7bf6181ec
* commit 'a2977c383d363e1e88a5b36230b1fa4c312807d2':
[3171580] SurfaceFlinger Bypass mode. (DO NOT MERGE)
[3171580] Add transform field to native buffers. (DO NOT MERGE)
* commit '05813b0eb92cb1bc79607ee402f14ca1e4b43f6d':
[3253328, 3171580] Treat GONE and INVISIBLE views the same when calculating transparent regions
[3171580] Fix two typos related to fixed-size buffers
This is a poor's man precursor to the h/w composer HAL.
Basically we detect when a window is full screen and in
that case we bypass surfaceflinger's composition step, which
yields to much improved performance.
Change-Id: Ie03796ae81a1c951949b771c9323044b980cb347
mFixedSize was never set, this bug was introduced during some "cleanup", in
practice this could cause some issues when a fixed-size buffer was used and
the window was resized.
Layer::drawForSreenShot() had a typo that had no effect.
mFixedSize was used to determine if filtering was needed, which was a bit too
conservative and created a dependency between filtering and "fixed size" states
which should exist.
Now we enable filtering based on the size of the buffer vs. the size of the layer.
Change-Id: I32044e91b0c944c1b137efdceb3f01dfaa78119d
some of these failures are not fatal and even expected in some cases
so they should not emit a dump in the log in those cases.
Change-Id: Idcfa252e3bfa9d74e27fe4ad8f8623aa01aa9c5e
This patch allows users of WindowManagerService.injectKeyEvent() to
set flags on the key event being injected.
In particular this allows long presses (FLAG_LONG_PRESS) to be
injected into the window manager.
The problem is that the code in AudioPolicyManagerBase::checkAndSetVolume() that forces
voice volume to max when setting bluetooth SCO volume is not called if the bluetooth stream
volume did not actually change. So even if we re apply volumes when switching to bluetooth
device, the volume voice volume is not changed and remains what it was when routed to earpiece
What makes things worse on Passion is that stream volumes are limited when connected to bluetooth
and their actual value does not change as soon as they exceed the limit threshold.
Change-Id: Id7c317db45b392a1c20dca2859678e3c64a371ed
There is a bug in the way audio policy manager handles A2DP interface suspend/restore
when SCO is used. This bug is not new but has been triggered by a change in the timing
of the events received by audio policy manager when a call is setup and torn down
introduced by commit 164a8f86c7e48992691368c4895709c3bdb835a4.
The fix consists in grouping the control of A2DP suspended state in a single function
that is called systematically when conditions affecting this state are changed:
- call state change
- device connection/disconnection
- change in forced usage.
Change-Id: I46ee2399ee5547b60511fc6cfd32e2720091b0f8
