Delete libplatformprotos since its not used.
Test: m -j
Bug: 147599928
Merged-In: I80203756d714bf123978d26bbe81351a1abdc967
Change-Id: I80203756d714bf123978d26bbe81351a1abdc967
With the goal of reducing log spam, print only a summary 'Found stale
backup journal' messages instead of logging within the inner loop.
Previously, over 12k messages could be printed at a time from this
function.
Before this CL:
- a backup was scheduled for each packageName from each stale
journal
- one (or two, if MORE_DEBUG) message was logged for each
packageName in each journal file.
After this CL:
- packageNames are de-duplicated before scheduling backups or logging
(it's not clear to me whether duplicate packageNames previously
occurred, in practice).
- one message is logged for the number (if > 0) of stale journals.
- one message is logged for the number (including their names, if
MORE_DEBUG) of packages.
Bug: 161940947
Test: fewer 'Found state backup journal' messages printed
Merged-In: Ia1343e4cea31feb1eba9da561d20736eb5df0a14
Change-Id: Ia1343e4cea31feb1eba9da561d20736eb5df0a14
With the goal of reducing log spam, print only a summary 'Found stale
backup journal' messages instead of logging within the inner loop.
Previously, over 12k messages could be printed at a time from this
function.
Before this CL:
- a backup was scheduled for each packageName from each stale
journal
- one (or two, if MORE_DEBUG) message was logged for each
packageName in each journal file.
After this CL:
- packageNames are de-duplicated before scheduling backups or logging
(it's not clear to me whether duplicate packageNames previously
occurred, in practice).
- one message is logged for the number (if > 0) of stale journals.
- one message is logged for the number (including their names, if
MORE_DEBUG) of packages.
Bug: 161940947
Test: fewer 'Found state backup journal' messages printed
Merged-In: Ia1343e4cea31feb1eba9da561d20736eb5df0a14
Change-Id: Ia1343e4cea31feb1eba9da561d20736eb5df0a14
(cherry picked from commit 112e3c2d049d304b0aa57750fcaa6415ea2b6fef)
For backwards compatibility, we allow phone, system UID to
access cell identity through getServiceState. Cell broadcast
service, which is used to be part of telephony, now is running
under network stack UID. We should consider network stack UID as
trustable components like system/telephony.
Fix: 157618105
Test: Manual
Change-Id: I9b2dfc778752e130f5c7e44c4b2e45b63ac6f918
When the preferred handler of an action is saved with "always", the set
of components displayed as options to the user are saved with it. On
subsequent queries, if that set of components change, the preference is
cleared to allow the user to consider the new set. With the introduction
of app visibility filtering in R (and to a lesser extent, Instant Apps
in O), the set of apps that an app can see may be less than the set that
the system can. As a result, if the caller calls resolveIntent and
cannot see all of the apps the system can, it may reset the user
preference unexpectedly.
This change ensures that before we clear the user preference that we're
comparing against the full set of apps. As such, if we encounter a call
to the method from an app that cannot see all apps, we avoid any
mutations and instead just return the set of apps that the user can see
or the preferred app if it's part of the set they can see. If the system
calls or if the call is part of an activity start, we'll allow reset of
the preference.
Test: Manual - Install 3 cameras, verify preference only reset on start
Fixes: 157548344
Change-Id: Icb50a66500edba26c4c6c3e80a6a9c544d4d63d6
Currently a native APEX can get and use a VINTF AIDL. However, this
can't be passed over JNI to be used by Java code.
This does not open VINTF AIDLs to any type of apps, where AIDL as an API
is completely disallowed. Also, no Java ServiceManager APIs are
available, and they couldn't be until b/139325468 is fixed.
Bug: 161501127
Test: atest binderVendorDoubleLoadTest
Change-Id: I1ab977248b1c39a7d08e0ca34389a7ba168c05b0
Currently a native APEX can get and use a VINTF AIDL. However, this
can't be passed over JNI to be used by Java code.
This does not open VINTF AIDLs to any type of apps, where AIDL as an API
is completely disallowed. Also, no Java ServiceManager APIs are
available, and they couldn't be until b/139325468 is fixed.
Bug: 161501127
Test: atest binderVendorDoubleLoadTest
Change-Id: I1ab977248b1c39a7d08e0ca34389a7ba168c05b0
Merged-In: I1ab977248b1c39a7d08e0ca34389a7ba168c05b0
