Fixes an issue with non-fullscreen modal windows (such as dialogs), where the touch exclusion of
the window behind was still being applied, even though the window behind did not actually
receive the touches in the exclusion, because they would go to the modal window in front.
Bug: 135522625
Test: atest testCalculateSystemGestureExclusion_modal
Change-Id: Ia99f4f601e780715abaf966f6f297fd9d555fd0b
am: c0e4ee7332 -s ours
am skip reason: change_id I2f78f0dbbace212309b87f779efad020cf255196 with SHA1 40431ae1bc is in history
Change-Id: I2f6d0b75851879f84167c689802a98645d96d3c1
Users are incorrectly being directed to the settings app when
selecting 'Always', making them unable to actually open an app through
the resolver.
Bug: 132071949
Test: Steps are reported in b/136451610
Change-Id: I11775d7b0b780195f08273ad2b90b8ecd2ee1db4
Make sure the client post execution lifecycle state to RESUMED
only if the activity was RESUMED while delivering new intent.
Bug: 135715788
Test: making skype calls
Test: atest ActivityThreadTest
Change-Id: I1e3054e1d1611aecf6ddf6d482abf2cb3ebdf9a4
Fixes: 136257475
Test: switch sms holder in settings, and ensure ceceived broadcasts for both
becoming and leaving the sms app status
Change-Id: I10dda3fd8ebd38ab12bdb8192ed4f3863f4238b8
Interface stats on stacked interfaces is usually clatd. For
xt_qtaguid supported device, the stats is already accounted
against its final egress interface by the kernel.
Framework side does not need to handle stats on statcked
interface at all.
However, on devices that support BPF offload, xlat464
packets are seen by the iptables rules as arriving on stack
interface only. Thus, add stack interface into accounting is
needed.
Bug: 136193260
Test: 1. atest FrameworksNetTests
2. atest android.app.usage.cts.NetworkUsageStatsTest
3. manual test on ipv6-only wifi network
Merged-In: I8ebbefbe4df00e40f4896a17fa52c8438d41286e
Change-Id: I8ebbefbe4df00e40f4896a17fa52c8438d41286e
(cherry-pick from aosp/1009306)
(cherry picked from commit feb79bd0d7)
The file permissions of the idmap2 binary are currently not set
correctly when the system forks and execs the idmap binary during
zygote. This chnages sets the uid and gid after forking to the same uid
and gid of the parent process.
Bug: 134897503
Test: device boots and generates idmap
Change-Id: Ic7fac49e5982f3c47713603b905c3a6be117a05b
When mutiple apps share shortcuts and are given to the UI to display,
the ranked order from both the App Prediction Service and the legacy
ranker is inadvertently being discarded. This means that targets the
user almost never uses will now be shown as high priority, causing
confusion and pretty horrible user experience.
Address the issue by adding in a score for shortcuts, based on the
index in the ordered list.
Bug: 136201796
Test: atest ChooserActivityTest and manually compare results from
AppPredictionService to verify order is consistent
Change-Id: I1fe2c2d2e13195f6f6c3a5818fade4844c15488e
Fixes: 136120151
Test: atest android.autofillservice.cts.SimpleSaveActivityTest#testDelayedSave
Change-Id: Ie0fd8ef0fbb8457fafc7cb5bed30813abaa0b893
Exempt-From-Owner-Approval: Felipe is no longer with the team
