follow up to http://aosp/2965923

 to hopefully eliminate
that class of bug in the future...

Test: TreeHugger
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: Ie09db001ff15a6b2b15bb1bb0c1ce8228e162337

Test: atest CtsHostsideNetworkTests

Bug: 304347838
(cherry picked from https://googleplex-android-review.googlesource.com/q/commit:4cef70d592e91c2ccd3419f52bf779e29fcb4a79)
Merged-In: I88ffa390a890a09d95109518329ba6abcb162bc1
Change-Id: I88ffa390a890a09d95109518329ba6abcb162bc1

The enum is already using 12 bits, so using fewer bits loses information.
This is in sync with the UidOwnerValue struct used to contain this data.

Test: atest resolv_integration_test
Test: atest resolv_unit_test

Bug: 325150155
Change-Id: Ide0ccea9614cda803b3700a0cc22977e9608209e

Rename API methods and update API docs as per API
council feedbacks

Bug: 316603231
Test: make update-api && make
Change-Id: Ibb1f58792d2e40ef1084c2b532e999a516a87b37

Bug: 324326824
Test: TH
Change-Id: I6d5f9c27b2fc2f2290e27f63b933b9ccacfffa85

Make TetheringRequest parcelable so we can pass it via Wifi APIs and map
Soft AP state changes to the correct request.

Bug: 216524590
Test: build
Change-Id: I63a45b14e9abc288b353159dfdcbe96fde485cb7

CtsNetTestCasesLatestSdk was originally used for MTS to prevent
installation issues with CUR_DEVELOPMENT target SDK modules (as is the
case for CtsNetTestCases) not installing on released devices.

However setting the min_sdk to a given value should be enough for that
purpose. While there may have been other blockers requiring separation
of CtsNetTestCases and CtsNetTestCasesLatestSdk at the time (in the Q/R
timeframe), more recent modules like CtsAdServicesDeviceTestCases seem
to be using CUR_DEVELOPMENT target SDK in MTS just fine.

Not specifying a target SDK means that the branch default will be used,
depending on the release configuration. So for example trunk_staging
would typically be CUR_DEVELOPMENT=10000, and "next" may use the SDK
version of that release.
This makes sense as a test suite targeting CUR_DEVELOPMENT is necessary
to verify a development build, while a production "next" configuration
should generally use the latest released SDK. But in any case, for
CtsNetTestCases purposes, this would cause the exercised connectivity
APIs to use their latest non-compatibility behavior, which is what
tests expect.

Considering this, add CtsNetTestCases to MTS, so that
CtsNetTestCasesLatestSdk can be removed. This will allow having one test
suite for both CTS and MTS, which in turns allows marking
CtsNetTestCases as MCTS, so that the version of the test to run is
autodetected based on installed modules.

Note this change also removes the unnecessary "platform_api" statement
and comment as this is already handled in
framework-connectivity-test-defaults.

Bug: 317913702
Test: m, aapt dump badging, verify (min)sdkVersion=30, targetSdk=10000
Change-Id: I2267e5c9219f059872928cc681547c3bd5fbc394

Apps will now lose network access once they are in top-sleeping or
higher process state. Adding some cts tests to ensure correct behavior.

The tests are using temporarily using junit Assume to skip if the
feature flag is enabled to enable a following code migration.

Test: atest CtsHostsideNetworkTests

BYPASS_INCLUSIVE_LANGUAGE_REASON=Existing method

Bug: 304347838
Change-Id: Iba7c7bd01309fa7a301fdc1524033f0fad5ae512

* changes:
  Introduce Powered Off Finding API
  Add Powered Off Finding operations to NearbyService

Bug: 307898240
Test: atest CtsNearbyFastPairTestCases
Test: atest NearbyIntegrationPrivilegedTests
Test: atest NearbyIntegrationUntrustedTests
Change-Id: I57b37ea796c0791d72fec931f731a0143e816dac

On V, this feature is enabled on all devices.
On U, this feature is enabled if the flag is enabled.

Test: TH
Change-Id: I025b9a8098a0896be51cf06c5b8262243ba439b7

Raise the max listener per client limit from 10 to 200.

A frequent use-case for NsdManager is to file a DiscoveryListener,
and try to resolve/listen for every service using a ServiceInfoCallback.

The current limit of 10 callbacks is easy to hit in that case, since it
would happen if there are at least 10 devices on the network.

Raise the limit to 200 to avoid such problems. In practice it is very
unlikely that there would be more than 200 services on a network; if
advertised by different devices, they would barely fit in a /24 IPv4
prefix, so a limit of 200 should generally be high enough.

Note that for discovery 200 listeners does not mean that batches of 200
queries will be sent: listeners on the same service type are grouped
together so only one query is sent for the type, and results are
distributed to matching listeners.

Note also that the limit is currently only checked when registering
discovery or advertising requests. This should be also checked when
registering other kinds of listeners in a future change.

Bug: 324377460
Test: atest
Change-Id: I86fc00777e9dea3e5192b575581ba562e7401e42

(this will land in M-2024-05 ie. May mainline tethering module)

Test: TreeHugger
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: I250b691eb03751da05ce10d41d6721f4964dd1d5

The flag com.android.net.flags.ipsec_transform_state gates APIs
exposed by the Tethering module, and thus should also be included
in Connectivity/common/flags.aconfig

Bug: 324278950
Test: make
Change-Id: Ia1fe733a4971ac56cae65870a5339362594322e6

These can be used by manual or automated tests.

Test: Manual by using the `cmd connectivity` shell commands
Test: atest CtsNetTestCases:ConnectivityManagerTest
Test: atest netd_integration_test

Bug: 322562125
Change-Id: I1de55e430141f566e510b37554a9416c4444213b

(super useful for debugging)

Sample output:
  I NetBpfLoad: NetBpfLoad '/apex/com.android.tethering/bin/netbpfload' starting...
  I NetBpfLoad: NetBpfLoad api:10000/34 kver:6010019 platform:0 mainline:1
  D NetBpfLoad: Found Tethering Apex mounted from blockdev /dev/block/dm-55
  I NetBpfLoad: Tethering APEX version 340090000

Test: TreeHugger
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: Iaf1bc22b02913ea21ff9368abd9933e6ef4017f1

Bug: 319532485
Test: Verified the return value on husky (arm)
      and cuttlefish (x86_64)
Change-Id: I8fd53b0f137ad701e051644eafaa12c2a5cf3e7f

This fulfills a "TODO: remove this when there is a riscv64 libcronet"
because with https://android-review.googlesource.com/2952606 there is
now a RISC-V64 libcronet. 🙂

Bug: 276962315
Change-Id: Ia69657b13a35e47852bf2bc46b2aac95fcf46e0a

This class is in the Tethering module, so any time we update it,
the tests are likely to fail unless Tethering is installed.

Fix: 324367069
Test: TH (test-only change)
Change-Id: I3d906e261939ff70225deeddff91cc6a83efe264

Allow all printable ASCII characters in subtypes, instead of only the
set of characters allowed in service types.

This is still more restrictive than what the spec allows, but allowing
random bytes would be harder to manage (especially when displaying the
subtypes in logs for example) and there is no known use-case for it.
Having the NsdManager API impose more restrictions is probably also best
for apps to minimize interoperability problems.

Bug: 324377460
Test: atest
Change-Id: Id75879a1c3ea6af8d7995b0e8e0dcc06e7160a0a