* changes:
Inline readNetworkStatsDetailInternal, make mUseBpfStats final
Remove VPN info arrays from NetworkStats(Observer|Recorder)
NetworkStatsFactory: Take VPNs into account for network/battery stats
Remove duplicate line in clat_simple test file
Remove unused lastStats parameter
Revert "Revert "Take all VPN underlying networks into account when migrating traffic for""
We are planning to use this metric to detect leaks.
This CL also decouples the actual memory sampling from AM. This means:
- Less time locking the pid list (we used to lock and then read proc)
- Less serialization / deserialization for the parcel
- Simpler to evolve (e.g. removed the HWM-specific method in AM)
Change-Id: I87a7243156dd8c88cfa85038e7e6cf4963e271e1
Test: manual, MemoryStatUtilTest, UidAtomTests
Bug: b/135418017
This change fixes detailed UID stats to ensure network and battery stats
both take VPNs into account. NetworkStatsFactory is being made aware of
VPNs enabled, and the full set of underlying networks present.
Since traffic can only be migrated over a NetworkStats delta, NSF
maintains a NetworkStats snapshot across all UIDs/ifaces/tags.
This snapshot gets updated whenever NSF records a new snapshot
(based on various hooks such as VPN updating its underlying networks,
network getting lost, etc.), or NetworkStatsService's
getDetailedUidStats() method being called.
This change widens the scope of the existing mPersistentSnapshot lock,
renaming it to mPersistentDataLock, and ensures that TUN migrations are
not done in parallel. Additionally, mVpnInfos is updated via
pointer-swapping, to reduce the scope of the mPersistentDataLock.
The safety of this change is predicated on:
1. NetworkStatsFactory lock not held, so services cannot deadlock through
the cyclical lock.
2. The broadening of the scope of the lock in NetworkStatsFactory has no
threading implications, as it is always the last (leaf node) lock held,
and therefore is impossible to have lock inversion.
Additionally, to ensure VPNs work with 464xlat, the VPN info passed to
the NetworkStatsFactory includes all underlying interfaces, instead of
only passing the first one.
This (partially) re-applies changes from:
aosp/972848: Add one more test for VPN usage stats.
aosp/972847: Addressing comments for http://ag/7700679.
aosp/885338: NetworkStatsService: Fix getDetailedUidStats to take VPNs
into account.
Co-developed with: Varun Anand <vaanand@google.com>
Bug: 113122541
Bug: 120145746
Bug: 129264869
Bug: 134244752
Test: FrameworksNetTest passing
Test: Manual tests show data usage fixes maintained.
Merged-In: I6466ec1411fc5ed6954125d27d353b6cd1be719e
Change-Id: Id45ae956ad7165be346ecc010e17d260563ac1c0
(cherry picked from commit 9fbbdebc61)
This reverts commit d8220c2050.
Reason for revert: Fix available for deadlocks.
Bug: 113122541
Bug: 134244752
Merged-In: Ib65214598837289bd39dbf040b56ab7835f893ba
Change-Id: Ia90bf2c72ef686e80800d113d03548e0efcadb66
(cherry picked from commit a84d9fa572)
Settings.Secure string for clock face may be JSON so that it can contain
the timestamp that it was set for logging purposes.
Bug: 134687399
Test: Added SettingsWrapperTest
Change-Id: I50afba479c30029428819c6616ca754db681a2b4
Using this flag when binding to a service will
allow the bound process to be held at a low
oom_adj of 250, so that it can be expunged to
reclaim memory if a more user-visible app needs
it.
Use for bindings such as job services and other
connections that the caller can easily recover
from and restart if necessary.
Adjust the lmk thresholds to use this oom_adj
as one of the levels, so they're killed before
perceptible apps (such as foreground services).
Bug: 135219821
Test: CtsAppTestCases
Manually check notification listener oom_adj
and dumpsys activity services output
Change-Id: I9f6d0891d842e4d12f7995b9b1a8f57b0903a16d
Revert API behavior changes to SurfaceView, snapping
back to Q's version.
Bug: 133378280
Test: none
Change-Id: I3a47f9bfdfab2d16707f952a9af672551736e681
Set a trim-level threshold to debug.am.run_gc_trim_level to activate it.
Bug: 135148702
Test: Manual test with "setprop debug.am.run_gc_trim_level 0", run a lot of
heavy apps and take pictures, and check logcat for the "force_gc" event log.
Test: Manual test with "setprop debug.am.run_mallopt_trim_level 0", run a lot of
heavy apps and take pictures, and check logcat for a debug log.
Change-Id: I73b4dc7374e85e9a22c98ab17da53aa6cb25a188
When the configuration changes between landscape and reverse
landscape, the app will not receive onConfigurationChanged as
orientation is not part of the public portion of the configuration.
However, when the ViewRootImpl receives such a configuration back from
relayout, it will force a layout of the client views
(see updatedConfiguration in performTraversals), this is because
Configuration#equals compares the non public part of the configuration
as well. This CL changes MSG_REPORT_RESIZED to handle the configuration
changing the same way performTraversals does, so that the app consistently
receives a configuration change.
Bug: 134643273
Test: Manual
Change-Id: If016bcd9a5b8d2a7efc5e1ab3c82a88a608caf8b
- Also save the scale with the snapshot data instead of trying to infer it
from the state
- Also recycle the reduced resolution snapshot copy after writing it
Bug: 135150767
Test: atest TaskSnapshotPersisterLoaderTest
Change-Id: I64a9ccf16f3966ad5c6d07ff0b6ce8e9dd710e40
This is still sent in an intent.
Bug: 131764329
Fixes: 131764329
Merged-In: I56c86b0c1912064d5a642991df32d2cefb6a8d5b
Change-Id: I64b9d632be97dc51e6085162371bb8c19f410258
(cherry picked from commit e546cb0bd16b7359feeb3c46ba52e64cf91ae4d3)
To determine if the CPS can get/send messages. Apparently
the IBinder can be cached in ActivityManager and onBind() is not
always called when a service is connected the second time.
Test: manual; ensure a service recieves an onsubscribe for an
active rule post requestUnbind/requestRebind
Fixes: 62584038
Change-Id: Iffe37242509f3bf26e609e6b423f3928c00156ad
(cherry picked from commit 265d093cd9)
Before this CL, the magnifier could deadlock when the following
happened:
1. the renderer is asked to draw (and a frame callback is provided)
2. a #dismiss() happens on the UI thread. This acquires mDestroyLock
(previously line 309)
3. InternalPopupWindow#destroy() is called, and this calls
mRenderer.destroy(). This attempts to destroy the renderer on the UI
thread, however the UI thread will wait until the pending frame callback
corresponding to step 1 is executed on the render thread.
4. The frame callback starts executing on the render thread, and tries
to acquire mDestroyLock (previously line 1093). However, this is held by
the UI thread, so a deadlock happens.
This CL completely removes mDestroyLock, relying on the existing
synchronization between the UI and render threads described in step 3.
Bug: 134584742
Test: manual testing
Change-Id: Ia4c75b5b997e0ed94d5a3814dd4507a8fffa124d
Annotates InputMethodSystemProperty#MULTI_CLIENT_IME_ENABLED as @TestApi
to let cts can verify if multi-client IME enabled.
Bug: 135217809
Test: build and passes
Change-Id: Id7b4dceb2dbfaf3d7ed1084957dd14b04cad0cdf
