WifiStateMachinePrime: centralize native failure handling
Instead of having each mode listen and react to underlying native daemon
failures, centralize the control in WifiStateMachinePrime. Now all
modes have the benefit of detection and SelfRecovery (when applicable).
Note: Multiple notifications for underlying failures is interfering with
recovery, tested with wifi hal death that only sends one callback.
Supplicant and wificond failures still trigger multiple callbacks and
recovery needs to be verified once those are fixed.
Bug: 75989180
Bug: 31346104
Test: frameworks/opt/net/wifi/tests/wifitests/runtests.sh
Test: triggered wifi hal failures with 'kill' while in all modes, wifi
re-entered scan and client modes as appropriate.
Change-Id: I2b606b25e463b1f5e152cf6a6ff2a188e5c48d6d
12 files changed