Fix deadlock conditions in Camera3Device.
Potential deadlock conditions this addresses, include:
- Not waking up waiting threads for several situations where
the status had been updated.
- Not waking up all waiting thread when status had been updated
(only one thread was awoken due to use of signal).
- Threads clear status transitions before other waiting threads
have a chance to examine them.
Bug: 22448586
Change-Id: I53ba669d333a83d2bfa1ca3170d34acc6d8fe6e3
diff --git a/services/camera/libcameraservice/device3/Camera3Device.h b/services/camera/libcameraservice/device3/Camera3Device.h
index eea34af..0ea9b8b 100644
--- a/services/camera/libcameraservice/device3/Camera3Device.h
+++ b/services/camera/libcameraservice/device3/Camera3Device.h
@@ -205,7 +205,11 @@
STATUS_CONFIGURED,
STATUS_ACTIVE
} mStatus;
+
+ // Only clear mRecentStatusUpdates, mStatusWaiters from waitUntilStateThenRelock
Vector<Status> mRecentStatusUpdates;
+ int mStatusWaiters;
+
Condition mStatusChanged;
// Tracking cause of fatal errors when in STATUS_ERROR
@@ -276,6 +280,13 @@
virtual CameraMetadata getLatestRequestLocked();
/**
+ * Update the current device status and wake all waiting threads.
+ *
+ * Must be called with mLock held.
+ */
+ void internalUpdateStatusLocked(Status status);
+
+ /**
* Pause processing and flush everything, but don't tell the clients.
* This is for reconfiguring outputs transparently when according to the
* CameraDeviceBase interface we shouldn't need to.