Revert "[VCN07] Bypass VCN for non-internet app accessible cellu..."
Revert submission 1579872-vcn07
Reason for revert: Possible culprit for b/179768816, broken ConnectivityControllerTest
Reverted Changes:
Id09b19c13:Address comments on aosp/1550755
I245bd69e0:[VCN07.1] Add test for bypassing VCN for non-inter...
I9936894b9:[VCN07] Bypass VCN for non-internet app accessible...
Change-Id: I7902a221c8597427e48d3e32fa5668b9c1a47420
Revert "[VCN07] Bypass VCN for non-internet app accessible cellu..."
Revert submission 1579872-vcn07
Reason for revert: Possible culprit for b/179768816, broken ConnectivityControllerTest
Reverted Changes:
Id09b19c13:Address comments on aosp/1550755
I245bd69e0:[VCN07.1] Add test for bypassing VCN for non-inter...
I9936894b9:[VCN07] Bypass VCN for non-internet app accessible...
Change-Id: I3fded28691e05a3dbf25bb7361c72acaa61b29c7
SubscriptionManager.setSubscriptionOverrideUnmetered() has been updated
to add NET_CAPABILITY_TEMPORARILY_NOT_METERED instead of
NET_CAPABILITY_NOT_METERED. So, we can't rely on that for changing the
meteredness.
Bug: 179664982
Test: atest tests/cts/hostside/src/com/android/cts/net/HostsideRestrictBackgroundNetworkTests.java
Test: atest tests/cts/hostside/src/com/android/cts/net/HostsideNetworkCallbackTests.java
Change-Id: I4163b2f9c6f80b4e639d9977ef7e2b444ab12e81
Merged-In: I4163b2f9c6f80b4e639d9977ef7e2b444ab12e81
This method does not behave correctly when wifi is connected but
the last CONNECTIVITY_ACTION broadcast was not for wifi. This
could happen due to another network connecting or disconnecting,
such as VPN.
Bug: 179774433
Test: test-only change
Change-Id: Ib2010b3871133c38b6d508bf508134dd9b814ce2
On R and below, when an app that is not the active VPN app calls
getConnectionOwnerUid, a SecurityException is thrown. On S and
above, the method just returns INVALID_UID, which is what would
happen if the app was the active VPN app but was not authorized
to find that particular socket.
Accept both behaviours.
Bug: 173331190
Test: atest HostsideVpnTests#testGetConnectionOwnerUidSecurity passes on R and S
Change-Id: Ie6de2ef505b0d7bc253c55215a698aeaa210785c
This CL bypass adding NOT_VCN_MANAGED on R device. And add TODOs
for refactoring and adding test coverage.
Test: atest CtsNetTestCasesLatestSdk:NetworkRequestTest on R device
Bug: 175662146
Change-Id: Id09b19c13f9a74e25d45689d2211418172dfaec8
This would allow us to run both metered and unmetered networkpolicy
related tests when the device is on a cellular network.
Bug: 165343126
Test: atest CtsHostsideNetworkTests:HostsideRestrictBackgroundNetworkTests
Test: atest CtsHostsideNetworkTests:HostsideNetworkCallbackTests
Change-Id: Id70856dffc920e74bda28583dba4dd851a832397
Merged-In: Id70856dffc920e74bda28583dba4dd851a832397
NetworkAgent is not updatable on R-, so there is no need to run tests
compiled against S on older platforms. They are also incompatible
because NetworkAgent stopped using AsyncChannel in S.
This only affects MTS, and CTS used to test S+ devices. CTS for R-
devices is already cut in a branch that this change will not merge to.
Bug: 177290955
Change-Id: I85c137a1b5fd7ef701fba4623566df5a8fb707a3
Test: atest CtsNetTestCasesLatestSdk
NsdManagerTest is not owned by Wifi, so it doesn't
belong in Wifi CTS.
Bug: 176902925
Test: atest CtsNetTestCases
Change-Id: I8eb3cba7a3b2ff24a012cca8d3fdff173e88c69a
Use two oneway binder interfaces instead.
The interfaces post messages to handlers as was implemented before, but
provide a more strictly defined interface, with less hops between
NetworkAgent, AsyncChannel, and ConnectivityService.
Test: atest FrameworksNetTests CtsNetTestCasesLatestSdk
Merged-In: Ica51d0179bcb3b4e314d2c3e85709aead6ca5657
Change-Id: Ica51d0179bcb3b4e314d2c3e85709aead6ca5657
All callbacks in the class use the DEFAULT_TIMEOUT_MS, except for this recently
added test.
The test is currently flaky in various branches.
Fixes: 174716714
Fixes: 174691729
Change-Id: I9f4446bb83eb4402c912292ac31d46e1d2cb6a52
Test: TH
The tests should just be run on any change in packages/modules/Connectivity,
not just changes in the CTS package.
Change-Id: I1c86c402f39a8a9b84e7422445df289c3ea7842f
Test: TEST_MAPPING needs to be verified by TH
Currently, when an agent calls setUnderlyingNetworks, the
transports of the network are hardcoded to TRANSPORT_VPN only.
This is being fixed in another CL in this topic. The new code
reports the same transports sent by the agent.
This test registers an agent with TRANSPORT_TEST|TRANSPORT_VPN.
Update the test to expect both transports in the callbacks it
receives.
Bug: 173331190
Test: passes when run with other CL in topic
Change-Id: Iede61d74fa02a4fec1f345f6d18835c1acab7337
- Use a more realistic routing config (instead of a LinkAddress
with prefix length 0, an address with a mask of 32 and a
default route).
- Use InetAddresses to parse IP addresses since it's a public API.
- Import the NET_CAPABILITY_xxx and TRANSPORT_xxx constants for
readability.
Bug: 173331190
Test: test-only change
Change-Id: I9703d7dd5781f60fb3d531267d3f5370cbccd64f
The CL that introduced testSetUnderlyingNetworks allowed tests to
pass in NetworkCapabilities when registering an agent, but
incorrectly always set the agent's capabilities anyway. This
resulted in testSetUnderlyingNetworks registering an agent with
TRANSPORT_VPN and NET_CAPABILITY_NOT_VPN.
Fix this by ensuring that createNetworkAgent either uses the
passed-in capabilities unmodified, or creates its own. Assert
that the test VPN network does not have NET_CAPABILITY_NOT_VPN.
While I'm at it, remove the manual unregistration of the callback
by using the registerNetworkCallback helper method.
Also add a little bit of extra test coverage.
Bug: 173331190
Test: test-only change
Change-Id: I114d876a1b2bf5344dd7c6fa23862df1c0a412c3
Also reorganize the files to have one toplevel CTS owners file, and only
additional owners in subdirectories.
Change-Id: I28ce193896e1ade2034589327805870774a4d608
Test: m
We were running cp to create a file and then running a command to read
the file. However were weren't waiting for the cp to finish (or,
indeed, start), so occasionally the read would fail saying the file
didn't exist.
Also added some logging, because diagnosing failure without it is
painful. Also simplify & improve file closing logic.
Some of this was previously submitted on internal master as change
I1c875102f0cce32cbbe2e3b36de913741c9abb92, but I've reverted that in
favor of this.
Bug: 168216494
Test: atest CtsNetTestCases:android.net.cts.NetworkWatchlistTest
Change-Id: I46a9db8b7a1885a9829f00bbd3233b863cfd1f5d
The new configuration runs tests on physical devices with SIM cards.
This is only done in postsubmit, as such tests can only run a few (~2) times a day due to
availability of the devices.
This configuration allows running all tests, so there is no annotation filter.
Change-Id: If95e4deeed8098327023f99e6e610331b8e86a0f
Test: TH validates TEST_MAPPING configurations
Ikev2VpnTest#testStartStopVpnProfileV4 and testStartStopVpnProfileV6
contains the IKE sequences and requires the phone to send and receive
the required packet in 1000ms.
On some low end devices, the response time may exceeds this TIMEOUT
sometimes.
So we need to use larger timeouts.
Bug: 171176723
Change-Id: I005b83ffecd8ade5063000690bfd4136d82f1a07
