Fixes a particularly bad transition in the case where there
is one incoming message to a group.
Change-Id: Ieddece4a496292a69e14cdcd74d94986938d8223
Fixes: 29043489
- Fixes a bug where we would not update the
overlay restrictions if the VR listener we are bound to
changes bug the VR enabled state doesn't.
- Also fixed a case where the notification listener access
and location permission were not granted/revoked for the
correct user. For example, if a vr activity in one user calls
a VR activity in another (possible for cross profiles) we end
up not revoking the special access for the vr service in the
first user. The notification listener setting was also updated
for the system user instead of the user for which we grant/revoke
specal access.
- Properly remove the overlay restrictions for the old user
if we transition from a vr actiivty in one user to a vr activity
in a profile of this user.
Bug: 29273985
Change-Id: Ib1de6f2f5445001ac61edca5c77ea3a066544307
The infrastructure is here but not yet fully plumbed, so gracefully
provide the historical behavior until the rest is in place.
Bug 29220492
Change-Id: I36feaf68099d85e4ae11a5262002ee85346703ed
The ContextThemeWrapper's configuration override must also
be applied to ResourcesManager during a configuration change.
Bug:27644297
Change-Id: Ic126308b753979c18801d422a8243aa4daf45147
Referrer information is passed as an intent extra to each Activity,
therefore if that intent contains a custom parcelable unmarshaling
the intent extras will fail in a call to getReferrer. Insulate apps
from this concern and log a warning when it happens.
Bug 28972438
Change-Id: I37dbc033bfcfe816e4b58b382c012854fbed943b
ApplicationInfo is mutable and unfortunately some apps do actually
modify the flags. Due to the lazy loading nature of the network security
config this may lead to issues. Instead cache the needed flags and
resources at application startup.
Bug: 29063413
(cherry picked from commit 276ee969be)
Change-Id: If638a716fd903b4e9dbabcbecb38bd4e26fef08c
In M we would bubble up a NameNotFoundException when failing
to load an assetmanager from another application. We switched
to throwing an IllegalArgumentException, which isn't something
anyone should handle.
This change has ResourcesManager throw Resources.NotFoundException,
which callers like ApplicationPackageManager can catch and rethrow
with the appropriate checked exception. This adds more detail to
errors than we had in M.
Bug:28876344
Change-Id: I250aeef8a4ccfd90c6abef2243edc1c5337ebde9
dumpDbInfo() uses an async task when dumping from the system thread, but
then the file descriptor is closed by the time the task is executed. As
such, the system databases are not dumped.
BUG: 29123428
Change-Id: Ibb8bbe503aa03c3ca13b2b1789f0ead0b869e9d5
am: 33d03a9e43
* commit '33d03a9e435eb474e3bdd3e958943e0057866477':
OP_ACTIVATE_VPN should not map to DISALLOW_CONFIG_VPN
Change-Id: Ib8e3b7fae441adf95b928d7139455d3ca13c1fb6
They are two different things.
OP_ACTIVATE_VPN means can we allow a vpn app to establish without user
interaction while DISALLOW_CONFIG_VPN means can the user modify vpn
config.
Testing:
1. Turn on DISALLOW_CONFIG_VPN, user cannot modify vpn config through
setting
2. Turn on DISALLOW_CONFIG_VPN, Dpm.setAlwaysOnVpn works.
3. Reboot after 2, vpn established automatically.
Bug: 29086229
Change-Id: I24899d6c7f0dd62bf441a44c6ee66fee35973a11
When ejecting a storage device, the system process needs to rapidly
release any open FDs to prevent itself from being killed by vold.
This change examines all ResourceImpls cached inside the system
process and evicts any that reference the storage device being
ejected. (ResourcesManager will gladly recreate any evicted entries
when asked again in the future.)
Also replace broken use of WeakHashMap, since we want the values to
be weak references, not the keys.
Bug: 28867548
Change-Id: Ib9cfc66497149b6d3f8d49213e9779408a331d2a
We now have a new settings key that provides all of the existing
tuning parameters, plus some newly redone ones for dealing with
different memory levels.
Changed the minimum batching for overall jobs from 2 to 1, so
we will never get in the way of immediately scheduling jobs
when the developer asks for this. We should now be able to rely
on the doze modes to do better batching of jobs for us when it
is really important.
Also work on issue #28981330: Excessive JobScheduler wakeup alarms.
Use a work source with scheduled alarms to blame them on the app
whose job they are being scheduled for, and add a check for whether
a job's timing constraint has been satisfied before considering it
a possible candidate for the next alarm. (If it is satisified,
the time is in the past, so we should not schedule an alarm for it.)
Finally clean up a bunch of the dumpsys output to make it easier
to understand.
Change-Id: I06cf2c1310448f47cf386f393e9b267335fabaeb
am: cd0aa9cda8
* commit 'cd0aa9cda8e57c224b473198a345fb008fe30b5a':
Don't bother with WeakHashMap for direct alarm bookkeeping
Change-Id: Ia86b5f27d511bb8da2bd8982ffe35d4d2906da54
