Until now a leak was possible under the following scenario:
Create a GLSurfaceView
Register a renderer (this automatically starts a GLThread).
Discard the GLSurfaceView without installing it in the view system.
This scenario can occur when a device is rotated rapidly from
orientation A to orientation B to orientation C. In that scenario,
orientation B's GLSurfaceView might be discarded without ever being
attached to a window.
If this issue had been identified before GLSurfaceView had clients, one
possible fix would have been to delay the construction of the GLThread
until the GLSurfaceView was attached to a window. Unfortunately, it's
too late, and so making that change would lead to observable changes in
behavior, possibly breaking some clients.
Instead, fixed by making GLThread and EGLHelper static classes that hold
onto a weak reference to the GLSurfaceView. This allows the GLSurfaceView
to be garbage collected when it is no longer used, even if the GLThread
is active. GLSurfaceView's finalize method will manually stop the GLThread
if it is still running when the GLSurfaceView exits.
Part of this change was to remove the Renderer reference from GLThread,
because Renderer is a user-supplied class that could contain a reference
chain that points back to the GLSurfaceView.
Fixes b/5606613 "GLSurfaceView that's never added to a window will
leak threads and views, can leak activities"
Change-Id: Iafdc329eb6e9e40062358e7c119f5547ffe23d5e
Addtion of synchronized to the methods triggers an API change.
Per council advice, pushing synch block into the function body.
Change-Id: Iaa395d9720bce499259ab750fe97149715e9f271
The msg is now constructed to try to bin these reports in
interesting ways. We'll see. Also change the tag name from
watchdog to lowmem, since sharkey is kindly taking care of
the back-end to handle this.
Improve how we put processes into low memory states to better
poke things like home and the previous app.
Also clean up some debug output, and add a few new am comment
options for controlling the current debug app.
Change-Id: I562a931a95244a2727bb7a6e1fd80dec259cdae2
Applies to docks, lid switch, HDMI and rotation lock.
We always choose a mode that is compatible with the application's
request, so if in a landscape dock, we might override a request
for seascape but we leave requests for portrait alone.
Bug: 5620454
Change-Id: Ib0c4f60f7f9e3aeafaba9c717233a950fccb8af2
* commit '716747f7cb29b45e41c2870f12d39095f4ed80c8':
Fix for issue 5309336 -add videoeditor maximum prefetch YUV frames in media_profiles.xml to limit the total memory usage.
This change is useful to ensure that injected POWER key events
always wake the device even if the right policy flags were
not set.
Change-Id: I05b8b35f984f5c9f58b0e9d44bcb6f41a9562fc8
When a dialog is activated, we won't show the blackscreen
until after the lockscreen comes back in focus, which occurs
when the dialog closes.
The user may toggle the power button, causing onScreenTurnedOff to
reset mHasOverlay, even though the dialog is still on top of the window.
Thus, the dialog will have a black box behind it without this change.
Change-Id: I6bec7984b88454e9fff22147cbbfc27b9c0a74f2
Surface Texture dequeue logic is modified to return the oldest of the
free buffers to Client on dequeue call.
Currently dequeue method is returning the first buffer index which is free.
The parsing is done in ascending order of the buffer slot indices.
This leads to returning the buffer which has been just queued to composer,
and hence display, and this defeats the purpose of having minimum dequeue count
as 2 in asynchrnouse mode.
This is fixed by checking all the free slots and returning the oldest buffer.
Change-Id: Ibbac10593c3994c278c601af0480b171635ecdd4
Signed-off-by: Sunita Nadampalli <sunitan@ti.com>
