Fix test breakages I caused when adding cell
support for NATT keepalives.
-Make the minimum keepalive interval a constant in
ConnectivityManager and use it in tests.
-Re-Disallow IPv6 Keepalives
Bug: 73327535
Test: 'runtest -x ConnectivityServiceTest' now passes
Change-Id: I5ec4367d250ee371014e65c897c3897a25a05e2d
NOT_SUSPENDED and FOREGROUND are capabilities that need to
be public so as to reach feature parity with what information
can be gotten through the use of CONNECTIVITY_ACTION and
synchronous calls to ConnectivityManager. This change makes
them public, and wires up the NOT_SUSPENDED capability.
This deprecates in effect the old onSuspended and onResumed
callbacks, but these have never been public.
This also converts the onAvailable path from a multiple
binder call design to a simpler, single binder call. This
is only for internal convenience
Test: runtest frameworks-net
Test: cts
Test: also manual testing
Change-Id: I6ea524bb361ecef0569ea2f9006c1e516378bc25
Prior to this change ConnectivityManager used to patch in the UID
of the requesting app inside the NetworkCapabilities sent to it.
The rationale was that the app may not know what other apps may
use the network, so the view it should have of the network should
always say the network only applies to that app.
But this has an unfortunate side effect : apps can't match the
received network against a default NetworkCapabilities. Ostensibly
this only applies to the system because all involved calls are
@hide, but still : system code would get some NetworkCapabilities,
for example using networkCapabilitiesForType, and then try to
match the capabilities of an available network using
satisfiedByNetworkCapabilities. Because the passed network is
declared to only apply to one's own UID and the UIDs of the
NetworkCapabilities are set to null meaning "I need this network
to apply to all UIDs", the answer will be "false".
While this is WAI in a sense, it is very counter-intuitive that
code trying to match a network would be required to patch in its
own UIDs.
There are three ways of fixing this :
1. Require all apps to do the above. It's correct, but it's
cumbersome and counterintuitive. Multiple places in existing
code needs to be fixed, Tethering is an example.
2. Write the UIDs of the caller in any NetworkCapabilities object
that is created. This is not very practical, because it imposes
the converse requirement on all NetworkAgents, which would then
have to clear the UIDs before they send the capabilities to
ConnectivityService. All NetworkAgents need to be fixed.
3. Instead of sending an object with a list of one UID to apps,
send a null list. The drawback is that the networks nominally
look to apps like they apply to all apps. I argue this does
not matter ; what matters is that the UID lists do not leak.
Clients just see a null list of UIDs (and third party can't
even access them without using reflection). No other changes
are required besides this two-line patch.
This patch implements 3. I believe it is the saner approach, with
both the most intuitive behavior and the best backward compatibility
characteristics, as well as the easiest change.
This does not encroach on the future plans to make the actual
UID list available to apps with NETWORK_SETTINGS.
Test: runtest frameworks-net
Change-Id: I978d91197668119e051c24e1d04aafe1644a41cf
This will allow NetworkStatsService to treat traffic on these
networks differently from traffic where the app selects a network
that is not the default.
Bug: 35142602
Test: runtest frameworks-net
Change-Id: I5ea9d200d9fb153490c6108bb9390bf152f297da
This change more strictly accounts for onCapabilitiesChanged
callbaks and their values. It exposes several cases where we the
callbacks we send are spurious.
Test: ConnectivityServiceTest continues to pass
Change-Id: Ifb9b00b6f0cae48f8ed41a525100d1744b5f429b
In future, managing DNS-over-TLS hostname lookup and netd programming
can be encapsulated here.
Test: as follows
- built
- flashed
- booted
- runtest frameworks-net passes
Bug: 64133961
Change-Id: I47ccfa99c30c780524c45c4af605e720ccba34a0
The "roaming" state of a network really belongs on NetworkCapabilities
instead of being published through NetworkInfo.isRoaming(). One major
reason is to support developers creating NetworkRequests for a
non-roaming network.
Watch for any capability changes that network statistics are
interested in (either metered or roaming) and notify it to perform
an update pass; fixes bug where we previously only triggered on
roaming changes.
Fix bug in VPNs where metered/roaming capabilities of underlying
networks weren't being propagated; this was probably preventing
some jobs from running over unmetered networks, and causing other
jobs to run over roaming networks! Also passes along link bandwidth
information from underlying networks, and propegates any changes
to underlying networks.
Fix race condition by reading prevNc inside lock. Utility methods
correctly calculate min/max link bandwidth values.
Test: bit FrameworksNetTests:android.net.,com.android.server.net.,com.android.server.connectivity.,com.android.server.ConnectivityServiceTest
Bug: 68397798, 16207332
Change-Id: I3e1a6544c902bf3a79356b72d3616af1fd2b0f49
This patch extracts the logging of DefaultNetworkEvent from inside
ConnectivityService and move it to a new DefaultNetworkMetrics class.
The DefaultNetworkMetrics is a singleton owned by the
IpConnectivityMetrics singleton implementing the metrics service for
core networking. ConnectivityService has access to this singleton via
LocalServices.
This class layout will allow to remove the Parcelable interface of
DefaultNetworkEvent and will instead let the IpConnectivityMetrics
service grab metrics from the DefaultNetworkMetrics directly.
Bug: 34901696
Test: runtest frameworks-net
Change-Id: I55694d89124272732aba114198776462372de18b
Although commit 893a762c2f fixed some flakyness issues in
testNetworkCallbackMaximum so that it became stable when ran on its own,
it introduced a new source of random failures because instead of
registering callbacks after callbacks until a limit was reached, commit
893a762c2f changed the test logic to push the assertions right up to
the theoretical limit.
More precisely when registering and unregistering PendingIntents in a
loop, not introducing some delay for checking that previous
PendingIntents have been effectively unregistered can cause the test to
fail. This patch fixes this issue.
Bug: 32561414
Bug: 62918393
Test: runtest frameworks-net
testNetworkCallbackMaximum now succeeds 100 in a row on sailfish
Change-Id: I086817a738ab99fd53ba76ca8faada6151f46472
This patch is a batch of mechanical changes to test classes to migrate
away from AndroidTestCase and TestCase.
Bug: 62918393
Test: runtest frameworks-net
Change-Id: I74134609e511f22c4d9ecd65780e981f9ba7ae3f
Registered requests are not keyed by PendingIntents in
ConnectivityService, which means that unregistering a request with a
PendingIntent causes a linear search in all registered requests.
testNetworkRequestMaximum was registering too many PendingIntents
simultaneously, causing the unregistration loop to have n^2
complexity and to take a long time to take effect.
To make the unregistering loop less likely to trigger a timeout on
waitForIdle, this patch changes the test to not register MAX_REQUEST
number of PendingIntent, but instead mixes a small number of
PendingIntents with NetworkCallbacks to reach MAX_REQUEST number of
simultaneously registered requests.
When unregistering these requests, callbacks are unregistered first.
Bug: 32561414
Test: runtest frameworks-net
Change-Id: I48b882c884abe20b388190b7f28baee293446f37
On some devices, support for TYPE_ETHERNET is not specified in
the networkAttributes config resource, even though the device is
capable of supporting Ethernet (e.g., via USB host adapters).
This leads to Ethernet working but various connectivity APIs
behaving as if it was not - for example, no CONNECTIVITY_ACTION
broadcasts will be issues when it connects or disconnects.
Ensure that ConnectivityService always treats Ethernet as
available if the service is running. Currently the service is
started if the device supports FEATURE_ETHERNET or
FEATURE_USB_HOST.
Bug: 37359230
Test: bullhead builds, boots
Test: ConnectivityServiceTest passes
Test: Ethernet is available even if removed from networkAttributes resource
Test: ConnectivityManagerTest CTS test passes
Change-Id: I58801bf4f0bbdc3ff6345ec6bfdc911ce045c8ab
This will finally allow to write captive portal detection unit tests.
Bug: 32561414
Bug: 62918393
Test: runtest frameworks-net
Change-Id: I38db1bb79ae80a82b4199dc9cb1b56257e0cf222
For some networks such as mobile data connections, its LinkProperties
does not contain routes for the local subnet so no such route is added
to the interface's routing table. This can be problematic especially
if the device is in VPN lockdown mode where there exists high-priority
PROHIBIT routing rule which in turn blocks the network's default gateway
route from being added (next hop address hitting the prohibit rule).
We fix this by patching LinkProperties to always include direct connected routes
when they are received by ConnectivityService. This has the added advantage that
when apps get LinkProperties, they see the directly connected routes as well.
Bug: 63662962
Test: runtest frameworks-core -c android.net.LinkPropertiesTest
Test: runtest frameworks-services -c com.android.server.ConnectivityServiceTest
Test: Start with device with mobile data, set up ics-OpenVPN in always-on
lockdown mode. Turn off mobile data then turn it back on, observe
mobile data connectivity is restored and VPN successfully reconnects.
Change-Id: I35b614eebccfd22c4a5270f40256f9be1e25abfb
This patch fixes a couple of flakyness issues with
testNetworkInfoOfTypeNone. It also fixes some typos and naming issues.
Bug: 62918393, 62918393
Test: runtest frameworks-net
Change-Id: I1c56557ab113d3ef57dbc06a6e882634d03c5b09
This patch allows to use TYPE_NONE for the legacy network type variable
of NetworkInfo. This usage is "safe" with respect to legacy APIs using
network types as most of them already returns null or do nothing for
TYPE_NONE.
Of the existing APIs in ConnectivityManager that accept a network type
argument, those which were already returning null or doing nothing for
TYPE_NONE are:
getNetworkInfo(int)
getNetworkForType(int)
stopUsingNetworkFeature(int, String)
networkCapabilitiesForType(int)
requestRouteToHostAddress(int, InetAddress)
reportInetCondition(int, int)
isNetworkSupported(int)
getLinkProperties(int)
Only setProvisioningNotificationVisible needs an additional guard
against TYPE_NONE.
Bug: 30088447
Bug: 62844794
Test: runtest frameworks-net
Change-Id: I112596fcd03d3c2cd42a2a84d265adb38e3944bb
ConnectivityServiceTest was still using sleep() in a few places although
these were unnecessary:
- in testSatisfiedThenLostNetworkRequestDoesNotTriggerOnAvailable(),
expectNoCallback() and expectAvailableCallback() already include
waitForIdleHandler calls that drain the message queues and make
sleep no-ops.
- in testTimedoutAfterUnregisterdNetworkRequest, the sleeps were
introduced before unregisterNetworkCallback was changed to have a
synchronous effect for callback unregistration, therefore the sleep
becomes simply non-sensical. To reflect this the name of the method
is also changed.
Bug: 62918393
Test: runtest frameworks-net
Change-Id: I7b701ecf5846a5e1890e86107b8d2544b419ce44
ConnectivityServiceTest was still using sleep() in a few places although
these were unnecessary:
- in testSatisfiedThenLostNetworkRequestDoesNotTriggerOnAvailable(),
expectNoCallback() and expectAvailableCallback() already include
waitForIdleHandler calls that drain the message queues and make
sleep no-ops.
- in testTimedoutAfterUnregisterdNetworkRequest, the sleeps were
introduced before unregisterNetworkCallback was changed to have a
synchronous effect for callback unregistration, therefore the sleep
becomes simply non-sensical. To reflect this the name of the method
is also changed.
Bug: 62918393
Test: runtest frameworks-net
Change-Id: I78426665670f702304212753f417b3d5a8a2c107
This patch allows to use TYPE_NONE for the legacy network type variable
of NetworkInfo. This usage is "safe" with respect to legacy APIs using
network types as most of them already returns null or do nothing for
TYPE_NONE.
Of the existing APIs in ConnectivityManager that accept a network type
argument, those which were already returning null or doing nothing for
TYPE_NONE are:
getNetworkInfo(int)
getNetworkForType(int)
stopUsingNetworkFeature(int, String)
networkCapabilitiesForType(int)
requestRouteToHostAddress(int, InetAddress)
reportInetCondition(int, int)
isNetworkSupported(int)
getLinkProperties(int)
Only setProvisioningNotificationVisible needs an additional guard
against TYPE_NONE.
Bug: 30088447
Bug: 62844794
Test: runtest frameworks-net
Change-Id: I4455f2726d06406047086368628c1f253d854d8d
- less strict regex for SharedLogTest: the subsecond part of the
timestamp can have 0, 1, 2 or 3 digits.
- refactor NetworkStatsServiceTest and NetworkStatsObserversTest to use
waitForIdleHandler facility of ConnectivityServiceTest.
NetworkStatsServiceTest was using a flaky custom version of
waitForIdleHandler.
Bug: 62918393
Bug: 32561414
Test: runtest frameworks-net
Change-Id: I634acfb5f4fe1bd5267e3f14b9f645edc32d5d12
Recent continuous testing runs indicates that commit 849b81b7 did not
completely fixed the issue with testRequestBenchmark.
This patch changes the name of the test to not include "test" and
removes @SmallTest annotation, which should do the job of @Ignore while
ConnectivityServiceTest still extends AndroidTestCase.
In addition timeouts are adjusted to take into account recent failures
observed.
This is the last pending action before turning on FrameworksNetTests on
presubmits.
Bug: 32561414
Test: no functional change
Change-Id: I56ef334e19e99e5a3483418330e5f0ccd6eb31bb
