There was bug in the logic that calculated the relative timeout, the start time was
reset each time an event was received, which caused the timeout to never occur if
an application was constantly redrawing.
Now we always check for a timeout when we come back from the waitEvent() and
process the "anti-freeze" if needed, regardless of whether an event was received.
The problem comes from a deadlock with AudioPolicyService mutex: When the second ringtone starts,
this mutex is locked by AudioPolicyService::startOutput() which in turn calls setParameters() to change the output device.
Audioflinger::ThreadBase::setParameters() signals the parameter change to the AudioFlinger mixer thread and waits for a condition
indicating that the parameter change has been processed.
At the same time, the mixer thread detects that the audio track corresponding to the first ring tone has been killed and calls its destructor.
This calls AudioPolicyService::releaseOutput() which tries to lock the AudioPolicyService mutex.
If this happens before the mixer thread can process the setParameters() command we are deadlocked.
The deadlock ends because setParameters() uses a timeout when waiting for the condition.
This regression was introduced by change 33736 fixing issue 2265163.
The fix consists in calling AudioPolicyService::releaseOutput() from Track::destroy() instead of from Track destructor: as detroy() is never called from the mixer thread loop (as opposed to the destructor) the deadlock described above cannot occur.
There is a delay between registering the two profiles,
and handsfree profile is a superset of the headset profile.
So some devices do an SDP and get the headset profile record
before we have registered the handsfree profile.
a) We can reject all incoming connections till all profiles are
registered, but then this would mean we connect later in some cases.
Registering profiles in this order seems fine to me.
Note: There is a also the need to fix forking sdptool to register
profiles, which would obliviate the need to wait 500 msecs between
profile registrations.
Bug: 2293792
Dr No: Eastham
This fixes another case where the screen would turn on when the keyguard is open but hidden by another activity.
Change-Id: I2b7c8a329036401709e96ded4f4c138041192a71
Signed-off-by: Mike Lockwood <lockwood@android.com>
This is needed since its no longer copied to /data/dontpanic by init.
Change-Id: I5217da73ec470653824b7fb9a31e093e263a8dc9
Signed-off-by: Dima Zavin <dima@android.com>
* changes:
Rename WebChromeClient.addMessageToConsole to WebChromeClient.onConsoleMessage. Do not merge.
Update JavaDoc for CacheManger.CacheResult, WebChromeClient.getDefaultVideoPoster and WebChromeClient.getVideoLoadingProgressView. Do not merge.
Improves documentation for GeolocationPermissions class. Do not merge.
In this case, as opposed to what was happening in 1977685, the focused view wasn't the one that was
directly removed, it was a child of the removed view.
...Binary XML file line #37: Error inflating class <unknown> after adding a secondary account
The problem was that we weren't dealing well with the situation where we start a transition
from activity A to B, then transition back to A before B is shown (it finishes before being
shown), then transition from A to C. At this point we had some state showing that we
were in the process of showing A from it being hidden (due to the middle transition from
B to A), which would cause the layout pass to ensure its window is hidden before the
transition starts.
The solution is to detect the case where we are showing a token and it is already actually
shown, and in this case not do all of the token setup for it to wait for its windows to
be displayed before it is shown. This isn't needed, the windows are already displayed
or the token is already set up to wait for them to be displayed.
Change-Id: I16925b91e1e2449dd65ade162a5758173c6e2695
