am skip reason: Change-Id Ic502acc8831fe4da0b32a826171d10e9c0f9876d with SHA-1 167c3db934 is in history
Change-Id: I39493ac64ca88d768c8b6a8a0540124d0694ce8c
Define the atom protos for
UpdateEngineUpdateAttemptReported
UpdateEngineSuccessfulUpdateReported
The fields for both atoms are copied from the already reported Tron
counters, except for the source fingerprint. By defining our own proto,
we can have a better insight about each update events. It also makes the
query easier without the server side work to correlate each counters.
Sample output from testdrive shows:
data {
elapsed_timestamp_nanos: 64932607816092
atom {
update_engine_update_attempt_reported {
attempt_number: 1
payload_type: FULL
duration_boottime_in_minutes: 0
duration_monotonic_in_minutes: 0
payload_size_mib: 510
attempt_result: METADATA_VERIFICATION_FAILED
error_code: DOWNLOAD_METADATA_SIGNATURE_MISMATCH
fingerprint: "google/walleye/walleye:R/MASTER/eng.xuncha.20190731.151212:userdebug/dev-keys"
}
}
}
Bug: 138253582
Bug: 137682371
Test: run statsd_testdrive and check events
Merged-In: Ic502acc8831fe4da0b32a826171d10e9c0f9876d
(cherry picked from commit 167c3db934)
Change-Id: Ifa57f23135b4cd63888df2b8f25650be133bb1de
(cherry picked from commit f00d7fb3bd)
IonMemoryUtil and ProcFsMemoryUtil were package-private classes loaded
in the system server class loader and had the same package as
StatsCompanionService, which is loaded in a separate class loader for
statsd. This caused a crash when statsd tried to access either. This cl
moves IonMemoryUtil and ProcFsMemoryUtil along with StatsPullAtomService
to a separate package so that no classes are shared in the same package.
Bug: 147792532
Test: pulled all relevant atoms and make sure they worked
Test: adb shell cmd stats pull-source 10064
Test: adb shell cmd stats pull-source 10061
Test: adb shell cmd stats pull-source 10042
Test: adb shell cmd stats pull-source 10056
Change-Id: I5107aa47045321e84549a7f2d55d0ee27f0d080e
Bug: 135998869
Test: Compiles
Test: Number of compile errors when compiled against system_current
reduces.
Change-Id: I67d2a435e8fd8c823b6cb1bf4becb01362dcc5e0
There might be several Views with the same ID in View hierarchy,
and if we search the user specified next focused view by comparing
the View ID, it may return the wrong View with the same ID. To fix
that, we should compare the View itself.
Fixes: 147829061
Test: manual
Change-Id: I3eeb68ed3ea30ce9461165eea56227cb81b21929
Vold calls into StorageManagerService which in turn calls into the
FUSE daemon to start the session. If the system_server is very busy
e.g right after an OTA, the session may not start quickly enough and
we will timeout, causing vold to fail the mount.
Previously, the only solution would be for the user to reboot.
Now, we naively retry every 10s.
Bug: 147610762
Test: Causing an artificial timeout when starting a session for the
firs time, verified we recover after a minute
Change-Id: Ifdb2384696e85b6774a12394798fca8eb75a584e
