On T+ this is mainline code, and the bpf program (netd.c)
does not support collecting tcp specific stats.

(note: this *is* a behaviour change, as it currently
just always returns 0, but UNSUPPORTED/-1 seems to make
more sense)

Test: TreeHugger
Bug: 294604315
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: Ifda15e67d9167a2be98ab6f4e3817ff7c0bb3fea

(next step is to replace use of Stats struct with
 identical (except field order) StatsValue struct)

Test: TreeHugger
Bug: 294604315
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: I9be3c411f9592bf4edc75386b1c5b386ebeb5905

This patch is based on aosp/2535559 from maze@.

Add source prefix into the upstream key such that only packets which
source IPv6 address matches it will be forwarded to the upstream
interface.

In this patch, the source prefix is set to zero so there is no
behavior changes. Next CL in patch series will use the real source
prefixes retrieved from upstream interface.

Test: atest TetheringTests
Bug: 261923493
Change-Id: I43d068a29b937c7dfeb6fab632a8effb47ee2263

Fix flaky test by only catch relevant the AddOrUpdateEvent and
RemoveEvent.

Bug: 294685851
Test: atest CtsNetTestCases
Change-Id: Ia12e51a6527f1e23610fdb0bea1068ebb5b80a73

The NsdService will throw a NPE if a new client is registered
with a null INsdManagerCallback object. To avoid this, perform
a null check before registering a new client and throw an
IllegalArgumentException if the callback is null.

Bug: 293285797
Test: atest FrameworksNetTests
Change-Id: Id61e27873591031c3fe383879aee0d40eebc08b3

The nsd metrics implementation needs to be updated to support
the new field sent_query_count on NetworkNsdReported.

Bug: 287546772
Test: m
Merged-In: Ie1d8d5f2cce4f6040dcfb980db6ab11dfe01fb72
Change-Id: Ie1d8d5f2cce4f6040dcfb980db6ab11dfe01fb72

Test: TreeHugger
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: Ibcb4fd53b0bbb04dc52f9fcbb5ab6bdae9efe02c

Remove the dependency on services-connectivity-shared-srcs
filegroup gradually.

Bug: 290885904
Test: TH
Change-Id: I6474563a3a0b521945e6467a2832862764ad4770

Test main thread mocked settings object but use that on the handler
thread as well, which is not thread-safe.
Create a real object instead of mocking it instead.

Test: atest ConnectivityCoverageTests:android.net.connectivity.com.android.server.net.NetworkStatsServiceTest#testDataMigration_differentFromFallback \
      --rerun-until-failure 500
Fix: 289705810
Change-Id: Ibfb722818467fdf2b69a4b8fa87b4ac04a713573

These tests were disabled from presubmit because of flaky test.
It's no longer flaky now, so add these tests back to presubmit.

Bug: 293233390
Test: atest FrameworksNetTests
Change-Id: I7f1f36385d2ca6ed210e1d9b7399b6509fd26519

When the device is very slow to run this test, cell might end
its lingering period in the middle of the test while wifi is
the default. VPN will then ignore it as an underlying network,
causing the test to fail.

Test: testVpnSetUnderlyingNetworks
      Added sleep in a strategic place to make sure the problem
      was correctly identified, and observed the test failing.
      After the fix, the test passes with the sleep.
Fixes: 280627836
Fixes: 293844384
Change-Id: Iaf56de963eb17845c1d76a0310b55f7f9321f7e4

Bug: 294326360
Test: atest FrameworksNetTests
Change-Id: I9a3a84f907362260bdcf6be19e2c4d5345788d2e

Move scheduling logic into a standalone MdnsQueryScheduler class to
simplify the MdnsServiceTypeClient class.

Bug: 292470176
Test: atest FrameworksNetTests CtsNetTestCases
Change-Id: I31130c239bc54f6dc0efde2921ce51df35076a74

This test assumes the data saver mode is initially disabled, but WearOS enables the mode on boot. This change fixes the precondition in the setUp method.

Test: run cts -m CtsHostsideNetworkTests
Bug: 280542664
(cherry picked from https://android-review.googlesource.com/q/commit:2cdeffc7e97a915a847cfc40d89e091aeadf20b3)
Merged-In: Ibcceaf0f44821100c861ef7b3c61d0888ae30481
Change-Id: Ibcceaf0f44821100c861ef7b3c61d0888ae30481

This change adds the nativeGetIfIndexStat JNI method, which can be used
to read the traffic stats associated with a specific iface index.

Test: Included unit test.
Bug: 241098920
Change-Id: I7a4adcb9610562ff01fe5cd1234d6dfc57f9408d

This change adds InitStats for explictly zero-initializing rx/tx
bytes/packets, and -1-initializing rx/tx tcp packets.

Test: Build, existing unit tests.
Bug: 241098920
Change-Id: Ib9598dab924b48240461b31906a4b5698db255d8

API-Coverage-Bug: 282618639
Bug: 269240366
Test: atest CtsNetTestCases
Change-Id: I7ac72b675ff3715891f31bc03c453e14b1264e40

Components that can provide offload like IpClient (packet
filter offloading) can use the API to register a callback to be notified
when offload is necessary.

Bug: 269240366
Test: atest CtsNetTestCases
Change-Id: I8080702f5b530001b88e79e504f4722ac01bc576

The MTU from the socket is a much easier way to get the
correct MTU to any given destination. In particular this
is important for CLAT which used to be wrong :
on my device, with a perfectly working GoogleGuest :
50: v4-wlan0: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1472 qdisc pfifo_fast state UNKNOWN group default qlen 500

  F  ICMPv4 payloadLength{1472} dst{8.8.8.8} src{192.0.0.4:27}: FAILED: 0/15 (4811ms)

After this patch, the payload is correctly 1444.

Test: Manual on GoogleGuest
      Manual on a network with a custom set MTU
Change-Id: I6cd2398572b3fe08f88832ab6d0aa715c118cd4f

This change updates 'mBpfMap' to 'mIndexToIfaceBpfMap' to
better-describe what the bpf map contains. This is done as a
prerequisite change for adding a name->indices map.

Test: Build
Bug: 241098920
Change-Id: Idaab7072a14ea3c88e9a5df808e768f355d83782

Close the fd based on the automaticOnOffState increases the
complexity. Replace it with always dup the fd in the constructor.

Bug: 290094178
Test: atest FrameworksNetTests
Test: atest android.net.cts.ConnectivityManagerTest
Change-Id: I0133588c41dd6aca9f0ee0fcc8e4b4418e152284