Handle CMD_IPV6_TETHER_UPDATE in UnavailableState
Technically, UnavailableState should not handle any messages at all (and
shouldn't exist). As the IpServer has already been removed from
mTetherStates. However, sendInterfaceState() cleans up IPv6 tethering
which eventually triggers a CMD_IPV6_TETHER_UPDATE. Instead of
preventing the CMD_IPV6_TETHER_UPDATE, ignore it for now and remove this
state entirely once we have rolled out SyncStateMachine.
Bug: 368713124
Test: TH
Change-Id: I32ae9583aa01411fa1fc299a0fc91138557b7ae2
diff --git a/Tethering/src/android/net/ip/IpServer.java b/Tethering/src/android/net/ip/IpServer.java
index b4c1b8a..506fa56 100644
--- a/Tethering/src/android/net/ip/IpServer.java
+++ b/Tethering/src/android/net/ip/IpServer.java
@@ -1394,8 +1394,28 @@
@Override
public void enter() {
mLastError = TETHER_ERROR_NO_ERROR;
+ // TODO: clean this up after the synchronous state machine is fully rolled out. Clean up
+ // can be directly triggered after calling IpServer.stop() inside Tethering.java.
sendInterfaceState(STATE_UNAVAILABLE);
}
+
+ @Override
+ public boolean processMessage(Message message) {
+ switch (message.what) {
+ case CMD_IPV6_TETHER_UPDATE:
+ // sendInterfaceState(STATE_UNAVAILABLE) triggers
+ // handleInterfaceServingStateInactive which in turn cleans up IPv6 tethering
+ // (and calls into IpServer one more time). At this point, this is the only
+ // message we potentially see in this state because this IpServer has already
+ // been removed from mTetherStates before transitioning to this State; however,
+ // handleInterfaceServiceStateInactive passes a reference.
+ // TODO: This can be removed once SyncStateMachine is rolled out and the
+ // teardown path is cleaned up.
+ return true;
+ default:
+ return false;
+ }
+ }
}
class WaitingForRestartState extends State {