Currently 5G usage includes both metered and unmetered usage.
To support 5G unmetered usage metrics, unmetered usage should
be separated from the 5G usage. Thus, ConnectivityService needs
to inform NetworkStatsService if TEMPORARILY_NOT_METERED has
changed. Then NSS will count temporarily not metered usage as
unmetered.
Bug: 222339311
Bug: 183776809
Test: FrameworksNetTests:ConnectivityServiceTest
Change-Id: I0c7e1eb98925184f36e322e09b4d3d8c6f056fa6
This commit is part of a large scale change to fix errorprone
errors that have been downgraded to warnings in the android
source tree, so that they can be promoted to errors again.
The full list of changes include the following, but not all
will be present in any one individual commit:
BadAnnotationImplementation
BadShiftAmount
BanJNDI
BoxedPrimitiveEquality
ComparableType
ComplexBooleanConstant
CollectionToArraySafeParameter
ConditionalExpressionNumericPromotion
DangerousLiteralNull
DoubleBraceInitialization
DurationFrom
DurationTemporalUnit
EmptyTopLevelDeclaration
EqualsNull
EqualsReference
FormatString
FromTemporalAccessor
GetClassOnAnnotation
GetClassOnClass
HashtableContains
IdentityBinaryExpression
IdentityHashMapBoxing
InstantTemporalUnit
InvalidTimeZoneID
InvalidZoneId
IsInstanceIncompatibleType
JUnitParameterMethodNotFound
LockOnBoxedPrimitive
MathRoundIntLong
MislabeledAndroidString
MisusedDayOfYear
MissingSuperCall
MisusedWeekYear
ModifyingCollectionWithItself
NoCanIgnoreReturnValueOnClasses
NonRuntimeAnnotation
NullableOnContainingClass
NullTernary
OverridesJavaxInjectableMethod
ParcelableCreator
PeriodFrom
PreconditionsInvalidPlaceholder
ProtoBuilderReturnValueIgnored
ProtoFieldNullComparison
RandomModInteger
RectIntersectReturnValueIgnored
ReturnValueIgnored
SelfAssignment
SelfComparison
SelfEquals
SizeGreaterThanOrEqualsZero
StringBuilderInitWithChar
TreeToString
TryFailThrowable
UnnecessaryCheckNotNull
UnusedCollectionModifiedInPlace
XorPower
See https://errorprone.info/bugpatterns for more
information on the checks.
Bug: 253827323
Test: m RUN_ERROR_PRONE=true javac-check
Change-Id: I7625fa386afe93823b97cb2ecb8fd09a5856c05b
Some callers (e.g. iwlan) need to learn about system default
network but they cannot have NETWORK_SETTINGS permission.
To allow them to use this API but prevent from misuse by
unprivileged apps, enforce USE_RESTRICTED_NETWORKS for this API.
Test: atest com.android.server.ConnectivityServiceTest#testRegisterPrivilegedDefaultCallbacksRequireNetworkSettings android.net.cts.ConnectivityManagerTest#testRegisterNetworkCallback
Fix: 242456635
Change-Id: I80ab27445af874328c9c0f4814a8fbf035ae5df4
This use was checked in after the cleanup that changed the latter into the former
Bug: 157405399
Test: FrameworksNetTests
Change-Id: I2b27ad712ee218322209bbc6af6e8147a0a41430
Chalard Jeanauthored and Jean Chalardcommited 2022-11-02 05:40:03 +00:00
This change adds testing for VPN MTU calculations, and in particular
cases where IPv6 is not supported due to MTUs being too small.
Bug: 245612827
Test: atest FrameworksNetTests:VpnTest
Change-Id: Ib064e6670244cf3b53b5d03bd62c86b2139ff32a
lucaslinauthored and Lucas Lincommited 2022-11-01 08:09:37 +00:00
Fix the policy leak by deleting forwarding policies when deleting
the IPsec tunnel interface.
Bug: 254566085
Test: atest IpSecServiceParameterizedTest (new tests added)
Change-Id: I29dd4810abb978fe886776b2fbacdfc74325436d
The interface index is only propagated if there is an A/AAAA
record in the response. Since the responses here are not entirely
controlled in the test, we just want to verify that we get a
response with the target we expect.
Verified by running the test 1400 times via --runs_per_test flag.
Bug: 254155029
Fix: 255922416
Test: atest MdnsSocketClientTests#startDiscovery_andPropagate\
InterfaceIndex_includesInterfaceIndex --iteration 100
Change-Id: Icb0fe7a58a1e626b7f781ff24c2389a3095d4dc6
aosp/2267683 adds the new param "mark" in generateIpv6Address and
TestDependencies#generateIpv6Address needs to override for this change
correctly.
Change-Id: Ia526bcdf80a428948d55a28bd5adec405530a634
Test: TH
Add tests to verify ike session being triggered when data stall.
Bug: 238692379
Test: atest FrameworksNetTests
Change-Id: I656ea0973072b0bbdd54863eccbe105adb02d6ad
In ConnectivityService, updateLinkProperties calls NetworkStats
even though the ip address or interface name has not changed and
only the tcp buffer size has changed. This is noisy and could
be problematic when RAT change occurs frequently, since when
RAT changes tcp buffer size configuration also changes.
This CL also fixes a wrong nullability annotation where the oldLp
of updateLinkProperties could be null when updateNetworkInfo
is called.
Test: atest ConnectivityServiceTest#testStatsIfacesChanged
Fix: 232048480
Change-Id: Ic226eb4a8aa1f38cba293510813f1cb55f0ea658
This allows clat to initialize properly when VPNs are configured
with "Block connections without VPN", rather than to error out with
"no IPv6 addresses were available for clat".
This issue primarily affects particular mobile networks configured
with NAT64 (without direct IPv4 connectivity).
Bug: 255040839
Change-Id: I4a8ee0295e0f5d1e330f7529856347b8bd10360c
t-m-wauthored and Maciej Żenczykowskicommited 2022-10-25 01:54:51 +00:00
shouldIgnoreValidationFailureAfterRoam will incorrectly return
true in the first few seconds after boot even if the network
never roamed. This is extremely unlikely to happen, but add a
check for that just in case.
Fix: 230450214
Test: new unit test
Change-Id: I0789d9bdaa0bd9e78673e8f4248a2ca610052f1e
Lorenzo Colitticommited 2022-10-24 23:28:56 +09:00
- Confirm that the default network switches to cell for the right
reason.
- Test the behaviour on T as well (always disabled).
- Minor code cleanups.
Bug: 216567577
Test: test-only change
Change-Id: Idf4cf74aec3d40246f67519d211dbd45e739a6dc
Lorenzo Colitticommited 2022-10-24 23:28:54 +09:00
For Matter, a device may have a single host with multiple A/AAAA
records, but multiple PTR/SRV pairs that represent different
operational identities for the same device.
Currently, MdnsResponseDecoder looks for only a single matching
MdnsResponse by hostname when looping through the A/AAAA records
(which will leave the second response incomplete). Updated with a
flag that will allow returning all MdnsResponses which match the
given hostname and update the Inet records for all responses.
Bug: 254155029
Test: atest FrameworksNetTests
Verified with a Matter device with two identities that the
second response is incomplete without this change.
Additionally, added unit tests to verify the before and
after behavior with a real packet capture from the device.
Change-Id: Ic4a101b305c039f6cd7fd557246a2bfdaf0c781a
For Matter, in addition to the InetAddress and port, we also need
to know the associated NetworkInterface index. With this change,
the interface index is propagated from the MdnsSocketTypeClient
through to MdnsServiceInfo based on a flag (defaulting to -1 for
unspecified, if the flag is disabled).
Validated end-to-end against a commissioned Matter device,
enabling the flags, verifying the returned network index against
an adb shell 'ip link show' output.
Bug: 254155029
Test: atest FrameworksNetTests
Change-Id: I96e804b32cdeeff3ed22da35a030df4d5dbb179a
RFC 6763 defines that TXT value can accept both utf-8 string and
binary data. Current implementation will always cast the TXT
value to a utf-8 string and will cause data lose when there are
non-utf-8 chars in the TXT value. This commit fixes this by
having the browser passing the TXT values back as byte[].
Also fixed the TXT key&value parsing issues per RFC 6763
section 6.5: accept cases of no '=' and reject empty key.
Bug: 254155029
Test: atest FrameworksNetTests CtsNetTestCases
Change-Id: I4b755e60ad6e59db19faa41556dd214993d73896
Before this patch, ConnectivityServiceTest#TestNetworkCallback
relies on TestableNetworkCallback calling this overridden methods
for all expectCallback calls. This is very confusing :
- The code for TestableNetworkCallback might be refactored so it
no longer calls this, we'd lose the checks and nobody would
notice.
- Anyone using TestableNetworkCallback instead of
TestNetworkCallback would get a different behavior but would
not notice as the interface for these two classes is exactly
the same
This is also bad for performance because all callback checks will
always look whether it's a check for LOSING, which is rare.
This patch also only generates the error message when the error
actually happens.
Test: ConnectivityServiceTest
Bug: 157405399
Change-Id: Ic9566b815dc4f9b001986ed1376d31a1b97ac8c5
Currently, the stats returned by getNetworkStatsUidDetail()
does not filter iface for mUidOperations, it will cause this
API returns the stats that might contain data with unexpected
iface. Thus filter iface for mUidOperations before combining
it into the return stats.
Bug: 235444141
Test: FrameworksNetTests
Change-Id: I81aaacb5682d69022e521a793516373037510742
This will be useful for diagnosing default network changes.
Logs may look like, for example :
Update capabilities for net 100 : -NET_CAPABILITY_OEM_PAID+NET_CAPABILITY_TRUSTED
Or :
Update score for net 100 : -POLICY_EVER_EVALUATED+POLICY_IS_VPN+POLICY_IS_DESTROYED
Test: FrameworksNetTests
manual
Change-Id: Ic4788c599573aae9daeca995b8853800aaaba0b8
Add code to verify which networks are passed in, and check that
the default interface appears in the LinkProperties of one of the
snapshots.
Test: test-only change
Change-Id: I3c1a483b89564b1c994b8e644ece5b903f549475
Lorenzo Colitticommited 2022-10-13 16:06:54 +09:00
Treehugger Robot
authored and
Gerrit Code Review
commited 
Junyu Lai
authored and
Paul Hu
authored and
Aaron Huang
authored and
Remi NGUYEN VAN
commited 
Cole Faust
commited 
Chalard Jean
authored and
Yan Yan
authored and
lucaslin
authored and
Nucca Chen
commited 
Chiachang Wang
authored and
Maciej Żenczykowski
authored and
t-m-w
authored and
Lorenzo Colitti
authored and
Motomu Utsumi
authored and