The correct return code to keep on processing any further TC
attached programs is 'TC_ACT_PIPE' and not 'TC_ACT_OK' (which
is terminal).
Without this the ipv6 tether offload program causes termination
of processing and the ipv6 clatd offload program never actually
handles any packets (while tethering is active).
This results in lack of bpf xlat64 offloading for tethered ipv4
traffic on an ipv6-only (cellular) network.
This in turn means incoming TCP packets get GRO'ed, do not get
bpf offloaded, and get delivered to the clat daemon, which
due to them being bigger than the mtu (due to gro) cannot
handle them and discards them.
This results in poor performance, since tcp falls back to 1 mss/mtu
sized packet per rtt.
Tested via tethering a linux laptop on an ipv6-only cellular connection
and downloading the linux kernel from kernel.org via 'wget -6' and 'wget -4'.
Before:
IPv6: over 2MB/s, observed:
5805 packets, including 4 sackOK
IPv4: under 1MB/s, observed:
9300 packets, including 8 sackOK, 387 sack 1, 501 sack 2, 2310 sack 3
After:
IPv6: over 7MB/s, observed:
16702 packets, including 4 sackOK
IPv4: over 9MB/s, observed:
32755 packets, including 2 sackOK
Test: builds, TreeHugger, see above
Bug: 195624908
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: I623dacb5a37dc689cea34499c3906c11fcaf946c
This test fails on devices where physical Ethernet interfaces are
available but marked restricted, like cuttlefish.
Bug: 197462993
Test: test-only change
Change-Id: I15c991b2e43e2d5e823dcdcfbd74adfd9b2f6f08
Bypass the IPv4 TCP packets with port 21 (ftp) and 1723 (pptp) from
BPF offload because these packets need the netfilter conntrack helper.
Bug: 195914327
Test: manual test as the follows
1. Connect to ftp.slackware.com with port 21 in active mode.
2. Check the PORT command success.
Command: PORT 192,168,62,128,174,17
Response: 200 PORT command successful.
3. Download a file.
Change-Id: I8e3b8d9323eb0e572f20c74442b55d4ee95abc2f
When the avoidBadWifi configuration is false and not overridden,
a WiFi network that was validated in the past but becomes
unvalidated needs to outscore a cell network that is validated.
This is happening correctly when the stack compares two networks.
However, when the stack compares an existing network to an offer
for a cellular network, the offer was automatically considered
not to yield. This would mean the stack would be requesting cell
out of the telephony factory, only for that network to lose to
WiFi and be discarded immediately, then recreated again etc.
When there is some other reason cell should be up (such as the
"mobile always on" setting being active), this would not be
visible because the cell network would have another reason not
to be torn down.
Have offers correctly account for the current value of the
configuration and setting. This has the ranking of the offer
lose against WiFi like the actual network loses, meaning the
offer is not needed.
This also requires updating the offers whenever the value of
the setting changes.
Test: new test for this, also ConnectivityServiceTest
Bug: 195441367
Change-Id: I4fe5de98bc15bcf9bbbe25c6c7c8a7ba382f8db7
This test requires a cell network so the test will be failed if
the device does not support telephony. Add a condition to check
if the device supports telephony and skip cellular battery stats
test if telephony is not supported.
Bug: 196231205
Test: atest CtsNetTestCases:BatteryStatsManagerTest
Change-Id: I9ddc1da2a3f83f3fd2ab59059185f2f7a8d08701
Wifi link layer is an optional feature so this test will be
failed on wifi stats check if a device does not support it.
Add a check to know if the device supports wifi link layer
stats and skip it if it is not supported.
Bug: 195518957
Test: CtsNetTestCases:BatteryStatsManagerTest
Change-Id: I592dd5f1d6e13b020beadb11b9d913857a82e524
Merged-In: I592dd5f1d6e13b020beadb11b9d913857a82e524
Currently, when a VPN app calls registerDefaultNetworkCallback,
it will always get its own VPN, even if the VPN app called
VpnService.Builder#addDisallowedApplication to take itself out
of the VPN's UID ranges.
Add a test for the current incorrect behaviour.
Also fix an indentation error elsewhere.
Bug: 195265065
Test: test-only change
Change-Id: Id9648ea71fc7ae10855aa311beeb7975569d17f2
restrictCapabilitiesForTestNetwork was renamed after S release to be
included in a mainline release, but this causes the MTS test to fail on
S if the Connectivity module is not updated, and CTS to fail if the
connectivity module is updated.
Mark the test as @ConnectivityModuleTest so it can be skipped on
non-connectivity module MTS tests (such as NetworkStack tests), and add
back the previous method name to keep CTS passing.
Bug: 196755836
Test: atest NetworkCapabilitiesTest
Change-Id: Ibd6c2e62e5949ec6d93e9f6e4fc05129c29b94f8
There are IpServerTest test case fail because the code check the wrong
code name for T pre-release builds.
Bug: 194812819
Test: atest TetheringTests
Change-Id: I73bf58c60b6e790fe0060b60c023c10f5d8cbfa8
The directories were imported by mistake from old git log in the
frameworks/base repository, and are unused.
Bug: 192995539
Test: m
Merged-In: I0d4bd06e79adedbf582d2fed551d9f1bde521fd1
Merged-In: Ie3a058b904cb2e3704c3a0725c8f720f81bb235e
Change-Id: Ic3ed5d6d25fd87b136d9ae359e7444c4b1289744
When run as part of NetworkStack MTS, ConnectivityDiagnosticsManagerTest
may fail as it verifies behavior that is only present will the latest
tethering (connectivity) module installed.
There is no guarantee that connectivity will be up-to-date for a given
NetworkStack version, especially considering that on R->S device upgrade
NetworkStack is likely to have been updated, but connectivity will be
the factory version. Therefore NetworkStack tests need to cover devices
that do not have an up-to-date connectivity module.
Fix the test by observing that at least on T both module will have newer
behavior. On S, accept both factory and updated behavior.
Bug: 195727283
Test: atest ConnectivityDiagnosticsManagerTest
Change-Id: Ifc0b09c884419d28817cfe619940c979ee1b0b9e
The check intends to do the bit & operation. The net cap value
should be shifted against the original capabilities.
Also fix the typo in the method name.
Bug: 191918212
Test: atest FrameworksNetTests
Change-Id: I98396b2538f36fe8b29d27a544a2dfb3060bc9c5
testAvoidBadWifi could pass without issue. The refactor should
also be done. The test does not need to be ignored now. Remove
the annotation to bring the test coverage back.
Bug: 178071397
Test: atest ConnectivityServiceTest
Change-Id: I5652fa817f16b8c241f1e2066a0b04ad2156a3b7
The resolverOptions member of the ResolverParamsParcel has never
been set by AOSP code but was only used by OEMs modifying
DnsManager. Now that DnsManager is mainline code, this is no
longer possible. So the DNS resolver introduces a new
setResolverOptions IPC to allow OEMs to set the options and makes
the resolverOptions nullable.
Make DnsManager set resolverOptions to null, to ensure that when
DnsManager calls setResolverConfiguration, it does not overwrite
any options set by the OEM.
Bug: 194048056
Test: Device boots and has connectivity
Change-Id: I310a79521f5a365e50e2c65e9dd87d9b68f105d7
Ideally, unregisterCallbackQuietly() is only needed when the
callback is registered in the try block. The callbacks
registration located outside the try block should be
registered theoretically. Otherwise, it may be a bug if the
tests get the IllegalArgumentException.
Use unregisterCallbackQuietly() in all finally may cause bug
in unregisterNetworkCallback invisible.
However, it may cause different tests with different code to
do the unregistration. Thus, do a refactor to add some methods
to wrap the NetworkCallback registration. This allows the tests
to do the unregister automatically in the tearDown().
Bug: 195364778
Test: atest android.net.cts.ConnectivityManagerTest
Change-Id: If06993f643ac7aeb23638a283347c427934a2f99
The resources may have different values depending on devices, and
resources IDs may mismatch if the test was not built at the same time as
the installed module, so mock the resources to allow running the
integration tests on more builds.
Bug: 193847396
Test: atest FrameworksNetIntegrationTests
Merged-In: Ic33d897690a8ea84a78c01dc5f2b2e2c473d57df
Change-Id: Ic33d897690a8ea84a78c01dc5f2b2e2c473d57df
Catch exception of unregisterNetworkCallback to
avoid reset operations afterwards being skipped by it.
Change-Id: Idb1161abfd143fb1305982338007dac169d372b5
Bug: 195364778
Add an option to display the no internet dialog directly instead of
showing a notification when the notification would have been high
priority (typically when the network was explicitly selected). This is
disabled by default, but allows device manufacturers to use a slightly
more disruptive UX to ensure that the user is aware that the network has
no connectivity, and can take action.
Also add an option to show the same notification as "no internet"
instead of the "partial connectivity" notification. This is also
disabled by default, but allows device manufacturers to use the "no
internet" text if they feel that "partial connectivity" text is hard
to understand for the user.
Bug: 193847396
Test: atest NetworkNotificationManagerTest
Change-Id: Ib5bd74d8cf973bf70d373dd63648c178fae0ebae
Needed because an invalid file descriptor should notify the caller to
stop using the object.
Bug: 190523685
Test: atest BpfMaptest
Change-Id: I70fb08b54b1c0caa4be4e3f07b59bf4f2397f39d
Instead, use a dedicated MockConnection class.
This should improve performance of the test, and works around failures
where mockito crashes when creating mocks when openConnection is called.
The failures are believed to be a bug in mockito-extended, but are not
likely to be fixed soon.
Bug: 185083316
Test: atest FrameworksNetIntegrationTests
Change-Id: I32deaaaaa5ce9876611314c8e8b9d8bdd6325df0
Using jarjar on intermediate libraries causes issues when they are
combined with other libraries that use different jarjar rules, as the
resulting binary may contain multiple incompatible copies of the
classes. Instead, jarjar should be used on the final artifact, after
combining the libraries.
Ensure NetworkStackJarJarRules is not applied on intermediate libraries
like TetheringIntegrationTestsLib, but instead apply it on
TetheringIntegrationTests or CtsTetheringTest.
Bug: 192535368
Test: atest TetheringCoverageTests TetheringIntegrationTests \
CtsTetheringTest
Change-Id: I248a533651984de81f676336c0e74e3788610bbb
DisplayMetrics are used in some downstream branches and need to be
mocked. This avoids merge conflicts in NetworkNotificationManagerTest.
Partial cherry-pick of change IDs:
I35d28c8df341dbbac2774026c6ca749e296c0482
I1aeed1c79e4a829d8829eb08224f9b21fafc50fe
Test: atest NetworkNotificationManagerTest
Bug: 176239013
Merged-In: I3c2563d4ae4e3715d0c6270344ba8f7ef067872f
Change-Id: I118ec1af615aef36293bf0b480389f766122ccab
The crash occurs when some app has more than half its limit
in requests that will need to be moved to some other default
network upon changing the preferences.
This will send the requests for this app over the limit
temporarily when creating new requests for the reevaluated
ones.
While ConnectivityService has a provision for making a
transaction-like addition/removal of requests that is meant
to avoid exactly this kind of crash with the transact()
method on PerUidCounter, the code only transacts on
mSystemNetworkRequestCounter. But these requests are counted
in the mNetworkRequestCounters, which is not part of the
transaction, causing the crash anyway.
To avoid the problem, this patch allows the request counters
to go over the max if and only if the system server is
updating the request counts for a UID other than its own.
This should allow only the case where ConnectivityService is
moving the requests over to the new per-uid default, while
keeping the exception when registering from an app (then the
calling UID is not the system server), or when the system
server registers its own requests (then the UID inside the
request is that of the system server).
A much better solution than this patch would be to completely
eliminate the transact() method by somehow unregistering the
old ones before creating the new ones.
However this would be a much bigger and difficult patch than
this, and much more dangerous, because callers depend on the
list of requests to find out the old requests to remove, so
they have to be created first.
Another possible clean solution would be to count the
requests not in the NRI constructor, but later. This would be
more error-prone though because it would be very easy to
create an NRI without counting it.
Bug: 192470012
Test: ConnectivityServiceTest. Improve tests so they catch
this case.
Change-Id: Ia482e6fbf2bf300ce6cbaca72810d394ed201b98
Wifi is not guaranteed to connect by just disabling airplane mode
or turning on wifi. Use ensureWifiConnected in the test to force
the test to reconnect to its access point.
Bug: 194413134
Change-Id: I82540a3c6fa1d50ffa39e925bdb3d214876e2083
Test: atest ConnectivityManagerTest
