A job's NetworkRequest should have the source app's UID as the UID. The
previous code attempted to set the UID, but didn't modify the stored
job, leading to potentially unexpected behavior of proxied jobs.
Rearranging code to ensure we properly set the NetworkRequest UID.

Bug: 311727153
Test: N/A
Change-Id: I8fb49b74ff861ea82d89ab8d2791e4b917fd6ef2

Switch to using static_libs in one location to avoid loading
the same class into the boot path multiple times.

Remove redundant links.

Bug: 309674488
Test: atest android.compat.sjp.cts.StrictJavaPackagesTest
Test: atest frameworks/base/services/tests/mockingservicestests/src/com/android/server/job
Test: atest frameworks/base/services/tests/servicestests/src/com/android/server/job
Test: atest CtsAppTestCases
Test: atest CtsJobSchedulerTestCases:JobSchedulingTest
Test: atest CtsNotificationTestCases:NotificationManagerZenTest
Test: atest CtsNotificationTestCases:AutomaticZenRuleTest
Test: atest CtsOsTestCases:PowerManagerTest
Test: atest FrameworksUiServicesTests:ZenModeConfigTest
Test: atest PowerServiceTests:PowerManagerServiceTest
Change-Id: I917cc4ae662c55a030fbf266133fba04fec496f2

This reverts commit efba4176.

Reason for revert: DroidMonitor: Potential culprit for Bug b/310080905 - verifying through ABTD before revert submission. This is part of the standard investigation process, and does not mean your CL will be reverted.

Bug: 310080905
Change-Id: Id7d6c1fbc6e58b08c14ad52b74a10adbff8abc4e

Switch to using libs to avoid loading the same class into the
boot path multiple times.

Bug: 309674488
Test: atest android.compat.sjp.cts.StrictJavaPackagesTest
Test: atest frameworks/base/services/tests/mockingservicestests/src/com/android/server/job
Test: atest frameworks/base/services/tests/servicestests/src/com/android/server/job
Test: atest CtsJobSchedulerTestCases:JobSchedulingTest
Test: atest CtsOsTestCases:PowerManagerTest
Test: atest PowerServiceTests:PowerManagerServiceTest
Change-Id: I4a00922aec9ae295da4480784088d18d0c6bf806

Revert submission 24499968-appops-deviceid

Reason for revert: Causing perf regression b/308786093

Reverted changes: /q/submissionid:24499968-appops-deviceid

Change-Id: I211edc6b9bb9f59eb6259be54938000f9ad65c45

Bug: 244264995
Test: Compiles
(cherry picked from https://android-review.googlesource.com/q/commit:ad387f90be5f17181f9b7b03b54bac570bafe2b1)
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:7c67bf850f1a1d59a070663248a52a416591a096)
Merged-In: Ifabc201418c54c2f910ff4b95b5240a58a683001
Change-Id: Ifabc201418c54c2f910ff4b95b5240a58a683001

Bug: 244264995
Test: Compiles
Merged-In: Ifabc201418c54c2f910ff4b95b5240a58a683001
Change-Id: Ifabc201418c54c2f910ff4b95b5240a58a683001

Bug: 230386578
Test: manual
Change-Id: Id1aaa54ccd314108aff02e836ea08ce1fbe80539

Bug: 244264995
Test: Compiles
(cherry picked from https://android-review.googlesource.com/q/commit:ad387f90be5f17181f9b7b03b54bac570bafe2b1)
Merged-In: Ifabc201418c54c2f910ff4b95b5240a58a683001
Change-Id: Ifabc201418c54c2f910ff4b95b5240a58a683001

Switch the flex network behavior from unmetered vs metered to looking at
network transports. Transport affinities indicate which transports are
preferred for running jobs, and which ones the system should try to
avoid, if possible. By default, when flex scheduling is enabled, the
system will avoid running jobs on the cellular network and prefer
running jobs on wifi and/or ethernet networks.

Watches tend to send most traffic over the bluetooth network. However,
there are plans to modify job network traffic in other ways for watches.
For now, watches will be excluded from network flex scheduling until
those plans are resolved.

Bug: 236261941
Bug: 299329948
Bug: 299346198
Test: atest CtsJobSchedulerTestCases:ConnectivityConstraintTest
Test: atest CtsJobSchedulerTestCases:FlexibilityConstraintTest
Test: atest frameworks/base/services/tests/mockingservicestests/src/com/android/server/job
Test: atest frameworks/base/services/tests/servicestests/src/com/android/server/job
Change-Id: I675d72b18ae6ddc43fb1503d481ff9bb56c2dcb6

Make it possible to disable wakelocks for background apps while the
device is in light device idle mode.

Bug: 299329948
Test: atest PowerServiceTests:PowerManagerServiceTest
Change-Id: Ice07f936f6a2ac329825564cb4d8950a92e65fc3

1. Avoid scheduling the idle alarm if the device is already considered
   idle.
2. Avoid scheduling the idle alarm again if the currently scheduled
   alarm is for the same time.

Bug: 309292340
Test: atest FrameworksMockingServicesTests:DeviceIdlenessTrackerTest
Change-Id: I0cb49041d15e5b2bd880ddc06cb2f1aa95e8c174

If a job doesn't say that it needs network, then there's no reason to
ensure that the app has access to network while the job is running.
Currently, the network restriction bypass flags are only given to
expedited or user-initiated jobs. Don't include the flags when the job
doesn't require connectivity to avoid accidental network usage/leakage.

Bug: 297106511
Bug: 299329948
Bug: 299346198
Test: atest CtsJobSchedulerTestCases:ConnectivityConstraintTest
Test: atest CtsJobSchedulerTestCases:ExpeditedJobTest
Change-Id: I9c632595f27655a977d6b084760839ce02c8c07b

Developers are informed that expedited jobs can be stopped earlier than
regular jobs and so shouldn't use them for long pieces of work.
The standard timeout is 3 minutes for EJs. There's little reason to let
them run longer than the normal 10 minutes, and letting them run up to
30 minutes may give developers the wrong impression and increases the
potential for accidental power issues. Lower the maximum execution limit
of EJs to 10 minutes (the base execution limit of regular jobs) to be
more in line with what EJs should be used for and prevent additional
issues.

Bug: 284512488
Bug: 297106511
Bug: 299329948
Bug: 299346198
Test: atest FrameworksMockingServicesTests:JobSchedulerServiceTest
Change-Id: Ia4950f0801f2881a34976987f735e5ab7f493ee1

Create a separate JobScheduler idle value by charging + battery-not-low
state so that we can have different values when the device is charging
compared to when it's on battery.

Bug: 236261941
Bug: 297106511
Bug: 299329948
Test: atest CtsJobSchedulerTestCases:IdleConstraintTest
Test: atest frameworks/base/services/tests/mockingservicestests/src/com/android/server/job
Test: atest frameworks/base/services/tests/servicestests/src/com/android/server/job
Change-Id: I16f41f05f7ffe5ab7fd1f660e0e08ea7544ff921

startOp, finishOp methods

This allows us to reduce the number of parameters being passed around
and also enables easy addition of new parameters such as device id.

Note that this change is unflagged and is meant to be a pure refactor.
This is preparatory CL and no new methods are being added to AppOpsManager yet, but will be needed
to allow clients to pass in device id.

Bug: 299160174
Test: atest CtsAppOpsTestCases CtsAppOps2TestCases
Change-Id: I591b1dc7d7e4af0371c2d1d80b06d858e7213a94

This reverts commit eb72459a.

Reason for revert: The plan is to apply it to jobs by default.

Bug: 236261941
Bug: 299346198
Test: atest FrameworksMockingServicesTests:ConnectivityControllerTest
Test: atest FrameworksMockingServicesTests:FlexibilityControllerTest
Test: atest FrameworksMockingServicesTests:JobStatusTest
Change-Id: Ie9a5643caf6e6b68642580b92b10c05a70bd2998

The job bias is meant to be an indication of the job type and the state
the app was in when the job was scheduled. This can only be determined
by the system and certain privileged apps. The job bias can affect some
things such as which system health factors are taken into consideration
when deciding when to run a job. Given all this, the bias should only be
set by the system or certain privileged apps. Note the required permission
on the hidden API and enforce that the calling app has the permission
when the job is scheduled.

Bug: 300477393
Test: atest CtsJobSchedulerTestCases:JobInfoTest
Test: atest CtsSyncManagerTestCases
Test: atest DownloadManagerApi28Test
Test: atest FrameworksServicesTests:BiasSchedulingTest
Change-Id: I1cb37eafc42f961dee9d2e7bc7d496a31dff31f4

during Job execution

Test: manual testing on device done by building
Bug: 255393346
Change-Id: If876aed542e8edf282ac36ff53979919bf41ce90

Don't relax the prefetch connectivity constraint requirement unless the
battery is also not low.

Bug: 299329948
Test: atest FrameworksMockingServicesTests:PrefetchControllerTest
Test: atest FrameworksMockingServicesTests:ConnectivityControllerTest
Change-Id: I4b2c0905fb15277c63ae57284a2139858998d68b

1. Remove the job ID from the wakelock tag.
2. Lazily load the tag when needed to reduce memory usage of jobs
   waiting for constraints to be satisfied.
3. Remove redundant log line.

Bug: 273758274
Bug: 302348192
Bug: 307556230
Test: atest CtsJobSchedulerTestCases
Test: Manually remove output of `adb shell dumpsys jobscheduler`
Change-Id: If8547d23ced90b17225ab8efd758ac735655687e

Use UserHandle.parseUserArg() when parsing adb command flags to enable
parsing "cur" and other special user strings.

Bug: 282063891
Test: atest CtsJobSchedulerTestCases
Test: `adb shell cmd jobscheduler run --user cur com.android.settings 2131427352`
Change-Id: Ibe06f536f1de760a5604d067b6c1894f756a17c6

Having the callbacks on RevocableFileDescriptor coming in on the main
thread of system_server can create problems:
- system_server's main thread is heavily contended
- it can cause deadlocks: callbacks come in from vold with vold's global
  lock held; this callback needs the system_server main thread to make
  progress. But if the main thread is busy with another call into vold
  (unrelated to RevocableFd), this will result in deadlock.

Bug: 300351508
Test: atest BlobStoreManagerTest
Change-Id: Ie4c3c65bdb9303f4aaab8f76b95d3f9f133b4c3e

The uid parameter to scheduleAsPackage is the calling UID. It was
incorrectly being used as the source UID in the statsd logging, causing
the end metrics to have the wrong information. This fixes the UID
logging and renames the input parameter to avoid further confusion.

Bug: 138239687
Test: N/A
Change-Id: Ifdfe2ffbc7b5f8b8fe74291a1ec54068541f740b

When loading from disk, skip files such as backup & temp files created
during the AtomicFile writing process to avoid incorrectly reading
duplicate jobs from them.

Bug: 289062813
Bug: 305169670
Test: atest FrameworksServicesTests:JobStoreTest
Change-Id: I28a180e2b24357219d34be6a22c785311be6e72e

Various issues here, but the most recurring one is referencing SystemApi
from public javadoc. Remove or rewrite such references.

Also common is trying to @link IntDefs. Usually this is not necessary,
as the presence of IntDef in the @param/@return auto-generates relevant
documentation.

Bug: 303184203
Test: m docs
Change-Id: Ie59e61881c45c73fbbe08250493bd35eca268ef6

Now that exact alarms are denied to newer app installs for apps targeting T, the deny list is obsolete and the overhead of maintaining it is not worthwhile.

Test: atest CtsAlarmManagerTestCases
Test: atest FrameworksMockingServicesTests:com.android.server.alarm

Bug: 304846238
Change-Id: I4376e5ddbb2e470f3e3d690dd0866f4b470f455b

Bug: 299329948
Change-Id: If1f6ef1a950ee93c2cb4413205c87a9454734573
Test: N/A

If the persisted files have somehow accumulated duplicate versions of
the same job (identified by uid-namespace-jobId), then drop the
duplicates and only keep one.

Bug: 289062813
Bug: 296890885
Test: atest JobStoreTest
Change-Id: I3179261309031801f0506cf905206376e59af1fe
(cherry picked from commit c8fbd937)
Merged-In: I3179261309031801f0506cf905206376e59af1fe

The system may be busy when trying to bind to the app, so don't penalize
it for slow binding. Now, we'll retain the job that was slow to bind and
stop triggering an ANR.

Bug: 292090044
Bug: 303533836
Test: atest CtsJobSchedulerTestCases:JobSchedulingTest
Change-Id: Ida52d63c5e9710d9d590388d20fceb4fa8f2e0d5
(cherry picked from commit 71fdfd48)
Merged-In: Ida52d63c5e9710d9d590388d20fceb4fa8f2e0d5

Reduce the prefetch job launch time threshold to one hour so we will
only consider running a prefetch job at most one hour before the app is
estimated to be launched.

Bug: 295046507
Test: atest FrameworksMockingServicesTests:PrefetchControllerTest
Change-Id: Ia3d4d049fb44b8f12418c987425949b60b37bd65

The system may be busy when trying to bind to the app, so don't penalize
it for slow binding. Now, we'll retain the job that was slow to bind and
stop triggering an ANR.

Bug: 292090044
Bug: 303533836
Test: atest CtsJobSchedulerTestCases:JobSchedulingTest
Change-Id: Ida52d63c5e9710d9d590388d20fceb4fa8f2e0d5

Make it possible to increase the light idle time linearly instead of
exponentially.

Bug: 299329948
Test: atest DeviceIdleTest
Test: atest FrameworksMockingServicesTests:DeviceIdleControllerTest
Change-Id: I4e7c07fa1d2653446ae5617f11fbd26ec39438ab

Only let prefetch jobs use the opportunistic quota when the device
is on charger to reduce excess battery drain. The opportunistic quota
usage behavior is optional and not required for prefetch jobs, so we can
disable it without impacting app or developer expectations.

Bug: 295046507
Bug: 299329948
Test: atest FrameworksMockingServicesTests:ConnectivityControllerTest
Test: atest FrameworksMockingServicesTests:PrefetchControllerTest
Change-Id: Idc137279fea1100f53ecf20279153fe92fdb5617

Platform_testing large test annotation is for
host side tests and restricted, device tests should use
androidx annotation.

Bug: 293837621
Test: presubmit
Change-Id: I835c4c837fa2074c937d7068f17c75467fb213dc

The statsd logging was incorrectly logging for only the source app.
Including the calling app will enable better metrics and debugging.

Bug: 138239687
Test: statsd_testdrive 8 while running CtsSyncManagerTestsCases
Test: statsd_testdrive 150 while running CtsSyncManagerTestsCases
Change-Id: I9ba3097135b1f11368df800b043d730d780f3a33

Bug: 294435720

Test: build

Change-Id: I3a788a7202827623b50d25b0a71bd6237a04d47c

Bug: 294435720

Test: atest FrameworksMockingServicesTests: com.android.server.DeviceIdleControllerTest

Change-Id: I908b0a8db05ef0feb19c5ecab47c1065295d17a4

The background thread is prone to severe delays in processing time near
boot, which causes pull timeouts and event queue overflow in statsd. The
vast majority of pullers use the DIRECT_EXECUTOR to execute pulls in the
incoming binder thread, so move these pulled atoms to do the same thing.
More context is in the bug and linked docs.

Bug: 298133385
Test: verified via boot traces that the pull durations are shorter and
that the pulls no longer happen in the background thread

Change-Id: I67a3d583cc13d9547dfe7108897e95f4a5238d26
Merged-In: I67a3d583cc13d9547dfe7108897e95f4a5238d26
(cherry picked from commit 5c85f29b)

Bug: 298968496
Test: PackageInstallerArchiveTest.java
Change-Id: Ic9669cc76e7a37455d773d00268130426b2a5165