Refactor the test case to support listen custom DNS/HTTP port.
Avoid binding system restricted port (<1024).

Bug: 322113686
Test: atest NetworkStackIntegrationTests:android.net.NetworkStatsIntegrationTest
Change-Id: I5d3b966cdadc9321ca2ee0d04a5b8f71ee79bd8a

Bug: 323464130
Test: presubmit
Change-Id: I67bde1506e99b42ff3806254aa8bbc6c98792e3e

For ot-daemon to perform mDNS operations, we need to implement an AIDL
service NsdPublisher. MdnsPublisher forwards the calls to NsdManager.

For now, only register & unregster a service without a custom host is supported.

Bug: 320211657
Bug: 318323473
Test: atest CtsThreadNetworkTestCases

Change-Id: Ia392a0f57046f53973ac6bd6043dec915d170666

This commit handles the ot-daemon crash by automatically restarting it
and rejoin the network if ever joined.

This also fixes the address not deleted issue in TunInterfaceContorller.

Bug: 321540728
Bug: 309792480
Change-Id: I3ed466eb70818974a1e57f01456363bdf82915d5

The comment that "The very first posted task is to initialize the
service cache, so it will be always set in other tasks running on
the handler" did not consider that if the looper was provided by the
caller, it is possible for other tasks to run on it before the
serviceCache is initialized. In particular, if the caller calls
registerListener on the looper thread right after calling the
constructor, this would be executed before the posted initialization
task.

In practice this does not happen as MdnsDiscoveryManager is created in
the NsdService constructor, and any usage of it is on messages posted to
the NsdService handler thread after the constructor has been called; so
the service cache initialization does happen before any other
MdnsDiscoveryManager task.

As this is error-prone, just initialize the service cache before first
use on the handler thread, as it is only used for
createServiceTypeClient, which runs on the handler thread.

Test: atest
Change-Id: Ie341bb71ef327dceb4f0aa6fc2d241d6e9cbb900

The check*Permission methods are a bit error-prone because people
can forget to check the return value and expect that they behave
like the enforce* methods. This was pointed out before in some bugs.

This change adds @CheckResult annotation to enforce linter warning
and rename methods accordingly.

Test: TH
Fix: 279691948
Change-Id: Ibb0df6540fe0ed40289bb190b790130daaa2f763

Instead of calling HandlerThread#getLooper immediately after starting
it, do it asynchronously in onLooperPrepared.

getLooper blocks until the looper is ready, which may take a significant
amount of time.

This involves queueing tasks until the looper is ready, so the thread
selection logic is refactored to a dedicated class.

Test: atest
Change-Id: I0444bffe8e4bb4e3f29bd0271c7b24bc46a62cf6

Test: atest NetworkStaticLibTests
atest FrameworksNetTests

Change-Id: I34a0a29eb7e78490701872229585e61f7a0c1dd3

netbpfload.mainline.rc is derived from:
  initrc-doc/bpfloader-sdk34-14-U.rc,
except that it is 'override' instead of 'updatable' and uses
the apex netbpfload instead of the platform bpfloader.

Test: TreeHugger
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: I1ecf14e3d927ece136a0669a9b209be7112e55b5

This CL is created as a best effort to migrate test targets
to the new android ownership model. If you find incorrect or unnecessary
attribution in this CL, please create a separate CL to fix that.

For more details please refer to the link below,
<add g3 doc link>

Bug: 304529413
Test: N/A
Change-Id: I243f17b3f0ad7af9ffa15ca242456e36688733f5
Merged-In: I243f17b3f0ad7af9ffa15ca242456e36688733f5

A lot of WTF logs reported were barely 1 second over the 1 hour
limit. AlarmManager has a maximum window of 1 hour so these cases
are possible and expected.
Increase the max metrics duration deviation to 2 hours to avoid
this log.

Bug: 297292877
Test: Manual test
Change-Id: I9a476e8061944c6499fe1b1d6e9ad1e19e6ce2b9

Previously only the subtypes of the last registered listener would be
included in the discovery queries.

This fixes that by combining the subtypes of all registered listeners,
to be included in the queries. By moving the queried subtypes from the
QueryTaskConfig to the QueryTask, they are now recalculated before each
send (like the existing servicesToResolve), so are always in sync with
the registered listeners.

Fixes: 322918399
Test: atest
Change-Id: I9a4c94189f997e3a9f92e46b5eb1762e9925cf4c

Address the remaining comments in aosp/2874718 to simplify the
burst time logic.

Bug: 271067818
Test: atest FrameworksNetTests NsdManagerTest
Change-Id: I619f45189b10da3e09759ee1d57a898173a9fff8

Pass buffer size to netlinkSocketForProto to allow callers to specify
buffer size. The buffer size is not set if 0 is passed in.

Following callers of netlinkSocketForProto passes SOCKET_RECV_BUFFERSIZE
to keep same behavior as before:
 * OffloadHardwareInterface.createConntrackSocket()
 * NetlinkUtils.sendOneShotKernelMessage()
 * IpSecXfrmController.Dependencies.newNetlinkSocket()
 * InetDiagMessage.getConnectionOwnerUid()
Following callers pass 0 to not update the buffer size:
 * TunInterfaceController - it uses the socket only for writing
 * NetlinkUtils.createNetLinkInetDiagSocket() same behavior as before
Added a SOCKET_DUMP_RECV_BUFFERSIZE of 128k to be used by socket
requesting dump information from kernel:
 * NetlinkUtils.getAndProcessNetlinkDumpMessages() - sends dump request
   and expecting multiple reply messages.

Bug: 320420453

Test: atest NetworkStaticLibTests
atest FrameworksNetTests

Change-Id: I3dcb76963cf2646eb311602b50ab4e44b8ad3e8f

The current query scan modes, active and passive, have
limitations. Unicast queries only originate from the first query
session, multicast queries should be sent simultaneously with
unicast, and both modes suffer from infrequent scans. To
address these shortcomings, a aggressive query mode is proposed.

Bug: 271067818
Test: atest FrameworksNetTests NsdManagerTest
Change-Id: I87100e03a5b6306dcae79f05c034f000e6dc2889

Test: TH
Fix: 299876562
Change-Id: I347a269889ea6cf39218b6322dcba92a14535e9c