Fixed the vendor data/network service non-responsive
issue. The callback binder passed from frameworks
might be GC'd so that vendor data/network service skipped
calling the callback. This eventually caused data
connection state machine messed up.
Fixed by turning the weak refernce to the callback into
a strong reference. This ensure the binder alive when
vendor service needs to invoke the callback.
Test: Telephony sanity tests
Bug: 139076980
Change-Id: Ica0b7b810ffd5416ffd1b2b61f7ebc4af0dcb8ce
Added debug messages and watchdog to catch
unresponsive data service issue.
Bug: 139076980
Test: Manual
Change-Id: I52dd126fddc1b1e8b18d5388a92aabaa3e2e48d5
The ACTIVITY_RECOGNITION permission applies to TYPE_STEP_DETECTOR, not
TYPE_GEOMAGNETIC_ROTATION_VECTOR.
Bug: 139363621
Test: n/a, comment update only
Change-Id: Ibce74be1e704ded04ae710988d0e03df43ab1cf9
As of Android Q, an app is included in the return value of
LauncherApps#getActivityList() unless it fulfills specific criteria.
This CL describes the conditions necessary for an app's activities
to *not* appear in the list.
Test: make ds-docs -j32
Bug: 130352392
Change-Id: Iad888a7a7f47d090da1b143ac152705577144d20
Code duplication was causing the wrong sensor to be re-registered,
also, binned brightness isn't a drop-in replacement for proximity
as expected.
Brightness can be low when prox is not covered, the correct behavior
is to verify if brightness is not 0
Test: manual, covering sensor
Test: atest DozeSensorsTest
Fixes: 139298207
Change-Id: I8673e7d9bcb56d7a34741cfad35cf397d976c18d
Previously we were using a fixed duration and as a result the
scrims duration was much shorter right after boot when the
fading away duration takes 2 seconds. We're now always using
the fading away duration which simplifies also the bypass
transition.
Fixes: 138305791
Test: atest SystemUITests
Change-Id: I2eba68c75c9777ab1ba6ffadadafed3836eb74e3
This key is introduced for AT&T
This will also require a carrier-config side change
to set show_4g_for_3g_data_icon_bool to true.
This will replace H/H+ that is shown by AT&T today.
Bug: 138655095
Test: manually
Change-Id: I480e14be41f8edbe607472351efd39d6d91b81b0
am: 68f3bb1305 -s ours
am skip reason: change_id I615b2f7fddca30dae67dbaab0e5d54a824a4c441 with SHA1 2d35980e72 is in history
Change-Id: I37cdb13b72ac7c4da5df1db6ee07209ed20353db
We see cases where bootanim is IO sleeping on boot and its causing
jank. Change the ioprio to the highest in an attempt to reduce this.
Bug: 138459662
Test: manual - flash and factory reset, looking for artifacts in the boot animation
Change-Id: I9734edea37c92e365a2b359a5180303e9b9284f7
My descendants will vilify this CL for generations to come. We'll
clean it up for R, but this is our last, best hope for fixing things
in Q.
Bug: 138775282
Test: manual
Change-Id: I615b2f7fddca30dae67dbaab0e5d54a824a4c441
Merged-In: I615b2f7fddca30dae67dbaab0e5d54a824a4c441
(cherry picked from commit 2d35980e72)
The getInterfaceVersion method was missed when freezing the AIDL
interface version.
This causes Q devices to return 0 instead of 3 on this interface.
Fixes: 139110717
Test: toggle tethering, dumpsys network_stack version
Merged-In: I782f479de2d7d4c6579f206645a580c905009d85
(cherry picked from commit c6c1c69378)
Change-Id: I16b08e77351399d937b037c700b6e454b09f6d16
