To do this we register a UidFrozenStateChangedCallback with
ActivityManager. The callback then queues up work on the Handler thread
to consume the original callback args and kill appropriate sockets.
Test: atest ConnectivityServiceTests
Bug: 277100090
Change-Id: I93f174811aef726145a18ea90443ddfc1456478e
Currently, native networks for non-VPN networks are created only
when the NetworkAgent goes into the CONNECTED state. This is not
ideal, because until the network is connected:
- Nothing routes to the network, so no IP packets can be sent on
it (except via packet sockets).
- Many network configuration changes (e.g., routes and DNS
servers) fail to apply because the network does not exist yet.
These cause ENONET errors in the logs.
Fix this starting in U by creating the native network as soon as
the agent registers. This should not impact apps because
NetworkCallbacks are not sent until networks are connected.
Bug: 143158421
Test: new coverage in NetworkAgentTest
Change-Id: I26bfa8630c085422175558645c47a6c64be96ae6
Lorenzo Colitticommited 2023-05-10 20:01:07 +09:00
From service-connectivity, NAMESPACE_TETHERING must be used, as
NAMESPACE_CONNECTIVITY is used for the NetworkStack module. Also the
isFeatureEnabled with module name must be used, otherwise the "android"
package version is queried instead of the tethering module version.
BpfNetMaps is using flag value 1 to workaround this issue.
flags in NsdService, AutomaticOnOffKeepaliveTracker, and ConnectivityFlags
were not pushed currently.
So, this change has no effect to devices.
Bug: 279108992
Test: TH
Change-Id: I2b4b6a13c048c20beef52b1f43b9e21aca0c637a
The ConnectivityResources class is only usable with QUERY_ALL_PACKAGES
permission, so it is not generally usable in framework-connectivity. It
is also backed by the ServiceConnectivityResources APK, which is
intended as resources for service-connectivity.
Move the class to service-connectivity and update callers. CTS needs a
way to determine the supported keepalive count as it used the
resources from KeepaliveUtils as @hide API, so provide a
ConnectivityManager @hide API for testing.
Bug: 279108992
Test: atest
Change-Id: I3c9a77c580b5ab87c922c32778bce15dc33b4d1d
Remi NGUYEN VANauthored and Motomu Utsumicommited 2023-04-28 19:18:39 +09:00
Cellular networks are now supported for marking wakeup packets. As a
result onWakeupEvent now also reports data relevant to the cell
transport.
These unit tests should add coverage for these new behaviors.
Test: atest FrameworksNetTests
Bug: 276498460
Change-Id: I0b8a03fff6255e067b736f3dbfb53de56420f90b
Starting with Android U, the config_networkWakeupPacketMark and
config_networkWakeupPacketMask values will be used to mark packets that
woke up the AP on the cellular transport also.
Test: Manual, by inspecting the output of `iptables -L -t -n -v mangle`
Bug: 276498460
Change-Id: I4c7d2801532604d436bfda96c4c52a791995a9ed
Netd is not updatable since it's not mainlined.
To make socket destroy code updatable, the code was re-implemented in
java and moved to Connectivity.
Bug: 270298713
Test: atest FrameworksNetTests
Change-Id: I5439c0c76c42a9f738a1b25a1f62e701755cbd05
This is a clean up commit to correct the javadoc of
NetworkAgent.CMD_START_SOCKET_KEEPALIVE. This event was used
between ConnectivityService/NetworkAgent, and also
ConnectivityService/KeepaliveTracker with different semantics.
This may confuse the developers to update the doc incorrectly,
so also separate the usage to different events.
Also move the event definitions together so that a new event
definition will not mis-use the existing values.
Test: atest FrameworksNetTests
Change-Id: I73ebd8abcc610531da3057da2d643a41befb2099
This is also needed in other classes in subsequent CLs, move to
a common place to prevent from code duplication.
Test: TH
Bug: 275532605
Change-Id: If7c757cecba8027b2f3b5209e6ea67733da50bf3
Using packageManager from ConnectivityService context to get the caller
application info in binder calls can cause the permission issue. Calling
the Binder.clearCallingIdentity() before interact with packageManager
can fix the permission issue.
Test: atest CtsHostsideNetworkTests
Bug: 272165411
Change-Id: I0a289afc6fb0e42a92aa3ffa0c77ed396c4485c3
This includes two restriction changes of simulateDataStall:
- Allow simulation on a network that has an underlying test
network
- Allow simulation on networks created by the system server
This commit allows VCN and VPN to CTS test their abilities of
handling suspected data stall.
This commit is safe because #simulateDataStall is already protected
by two signature permissions MANAGE_TEST_NETWORKS and NETWORK_STACK,
and those permissions are granted to only preinstalled apps.
Bug: 263415068
Test: atest VcnManagerTest (new tests)
Change-Id: Ic0cdb66abb291fbf4a7545efe4114179e965d2b6
API council has requested that the recently added BroadcastOptions
APIs be modified to better match API guidelines.
Bug: 267646347
Test: treehugger
Change-Id: Ia701ae31079476d0cc05079d62cd90ba81279cdd
When the network requester process has been killed, lots of
binderDied logs would be spewed out. These logs haven't been
useful for debugging but to be very spammy. Thus, remove this log
entirely.
Bug: 223128960
Test: Manual test
Change-Id: Ic3096bfee41167b1a54a9d04b474263d15217217
CM#setTestLowTcpPollingTimerForKeepalive() is hidden method for
CTS. The debuggable restriction will prevent it from running on
actual CTS. The method is protected by permission with parameter
range constraint. It should be enough for incorrect usage from
normal apps. Remove it to allow test case to run on actual CTS.
Bug: 259000745
Test: CTS is the follow up commit
Change-Id: I3d4627c75bae7411c6a72799643d3b29bbae4f2f
chiachangwangauthored and Chiachang Wangcommited 2023-03-02 07:54:21 +00:00
This is useful in the system server for better attribution of wakeup
events.
The logic for encoding the data in the prefix is still maintained ad-hoc
and should be standardized in the future.
Test: Builds, boots. Manually triggering a wifi wakeup doesn't result
in anything unexpected.
Test: atest FrameworksNetTests
Bug: 265742148
Change-Id: I7ef8a606459bb9f0630f629c01d3b34643ae5319
When the application wants to request network with
NET_CAPABILITY_PRIORITIZE_BANDWIDTH or
NET_CAPABILITY_PRIORITIZE_LATENCY, it has to declare
PackageManager.PROPERTY_NETWORK_SLICE_DECLARATIONS property and also
adds the declaration in a separate XML files. Otherwise, the request
will fail with a SecurityException being thrown.
Test: atest FrameworksNetTests CtsNetTestCases
Bug: 266524688
Change-Id: I6affc857b803211517368da288e1b2fdc06a955b
Test: 1. Patch local built module on T device
2. atest ConnectivityCoverageTests:android.net.connectivity.com.android.server.ConnectivityServiceTest
Fix: 267968887
Change-Id: I86e5be9a0cee0621ff9ef0996d56a51fc5f408c2
Take a Network parameter to have an one-to-one mapping between
keepalive and its underpinned network on the automatic keepalive.
Existing design could not really tell which network should the
automatic keepalive check for the TCP socket status if there are
multiple automatic keepalives enabled, e.g. Bothe IWLAN and VPN
on WiFi enable the automatic keepalive. The keepalive for IWLAN
should check if there are any TCP sockets on the IWLAN network
instead of VPN network.
Bug: 259000745
Test: atest FrameworksNetTests
Test: Cts in the follow up commit
Change-Id: I7353f4ef43e8fdad02c4d4a0bb5f6efa7d94c1b4
Print state and keepalive information in the dump.
Also update ConnectivityService and the corresponding test to
create AutomaticOnOffKeepaliveTracker from a Dependencies class.
ConnectivityService#dump() will require READ_DEVICE_CONFIG
permission because of the new dump() in AutomaticOnOffKeepaliveTreacker.
ConnectivityServiceTest should focus on verifying design in
ConnectivityService, so skipping the DeviceConfig check via a fake
Dependencies class in the unit test.
Sample dump:
AutomaticOnOff enabled: true
AutomaticOnOffKeepalive [KeepaliveInfo [ type=1 network=100 startedS ... ], state=STATE_ENABLED ]
Bug: 259000745
Test: m ; atest FrameworksNetTests CtsNetTestCases
Change-Id: I9ac725bcef7daec1d1a12bad592b7fa6856a1707
registerSystemDefaultNetworkCallback is required by WifiTrackerLib to
detect the system default network for UI purposes. Currently this API is
allowed for holders of the NETWORK_SETTINGS permission, but not
NETWORK_SETUP_WIZARD. Since SUW also uses WifiTrackerLib, we should
allow this API for NETWORK_SETUP_WIZARD as well.
Bug: 230643853
Test: atest ConnectivityServiceTest
Change-Id: Ic01a186f91ed67b4603b6b8964921be7473a483a
The default TCP polling alarm timer is very large(2 mins).
It's expensive in the CTS to wait for a couple alarms.
The polling alarm should be deprecated soon and replace
with callback design, so add the hidden method for testing
purpose to support the short term usage until design is
replaced with callbacks.
With the hidden method, the alarm timer will decrease to
1 second for a specified time period. The TCP sockets status
could be verified every 1 second.
Bug: 259000745
Test: m ; atest HostsideVpnTests with the follow up test
Change-Id: I99d494d3b50b2fbee73b926e92e97b1e194d43d4
This is much simpler and less error-prone, as well as less
subject to race conditions.
It also allows for cleaning up some TODOs.
Test: FrameworksNetTests
CtsNetTestCases
Bug: 267116236
Change-Id: I470c709446946ef35a0324427defe2f58b434339
Chalard Jeanauthored and chiachangwangcommited 2023-02-06 13:22:53 +00:00
This patch uses the binder token represented by the callback
to communicate the identity of the AutoOnOffKeepalive through
the alarm mechanism. This is a lot more robust and easy to
understand.
Addressing this, on top of the robustness and readability
advantages, corrects three bugs.
• The |obj| in the message to CMD_MONITOR_AUTOMATIC_KEEPALIVE
is now the binder token both on the sender side and the
receiver side, while the previous code had a Network in the
sender while the receiver expected AutoOnOffKeepalive,
crashing the system server with a wrong cast.
• The intent sent in the alarm would have a yet-uninitialized
value for the slot, so it would not be possible to find the
auto keepalive when the alarm fires.
• When the slot is reassigned, the alarm continued to fire
with the same slot value, so the auto keepalive would no
longer be found after a pause/resume cycle that would
change the slot.
Test: FrameworksNetTests
Change-Id: Ibdbfcd884e0d3559206cbaae7b6b7a524972c3ca
Chalard Jeanauthored and chiachangwangcommited 2023-02-04 08:05:00 +00:00
Upcoming CLs use this method to add chain status to failure log of
ConnectivityManagerTest#testFirewallBlocking which is a flaky test
Test: m
Bug: 262141231
Change-Id: Ieb6b4402a3435571decbd8a26fdf666b2db732fe
Even if there are no preferences, the allow list needs
to be updated upon adding a new user, because it's set
unconditionally when the network connects.
Adding all existing UIDs would address this issue too
and we probably want to do it, but immediately this
seems simpler.
Bug: 266136779
Test: new test for this
Change-Id: I77091a6c3d3ccf2b80ab0aaa01d09ef0df922501
Add SocketKeepalive.start with parameter to enable dynamic
keepalive mode based on the existence of TCP connections.
This supports IPSec mode to notify KeepaliveTracker to disable
keepalive when keepalive is unnecessary to improve battery life.
Keepalive is controlled by periodically TCP socket status check
for both enable and disable. This is a transition commit and
is expected to be updated based on the socket creation or
destroy.
Bug: 259000745
Test: m ; atest FrameworksNetTests
Change-Id: Ie4d598d69a73c4931c7d0b6dfde0e459e5dca6b4
chiachangwangauthored and Chalard Jeancommited 2023-01-20 13:51:20 +09:00
Add a new AutomaticOnOffKeepaliveTracker class between
ConnectivityService and KeepaliveTracker to handle the automatic
on/off keepalive. This commit only creates this new class and
move the TCP polling code to the new class as a preparation for
the following commit.
The original test file was created for testing the TCP polling
mechanism, so rename it to match the new class.
Bug: 259000745
Test: m ; atest FrameworksNetTests
Change-Id: I1b229f906283c0f5ef7a3efdb0572fcbfc5df72b
chiachangwangauthored and Chalard Jeancommited 2023-01-19 13:17:54 +09:00
This allows us to skip any older CONNECTIVITY_ACTION broadcasts
waiting to be delivered when a new broadcast is dispatched.
Bug: 255545931
Test: TH
Change-Id: Iac5da2f727e295d1775bfe91358900f316258be4
Motomu Utsumi
authored and
Gerrit Code Review
commited 
Lorenzo Colitti
authored and
Mark Fasheh
commited 
Remi NGUYEN VAN
authored and
Suprabh Shukla
commited 
Treehugger Robot
authored and
Chiachang Wang
authored and
Junyu Lai
commited 
Yan Yan
authored and
Yuyang Huang
commited 
Jeff Sharkey
commited 
Paul Hu
commited 
Treehugger Robot
authored and
Quang Luong
authored and
Jean Chalard
authored and
Michael Groover
commited 
Sudheer Shanka
commited