According to the OMX specification, implementations are allowed to call
callbacks in the context of their function calls. However, our callbacks
take locks and this causes deadlocks if the unerlying OMX implementation
uses this kind of in-context calls.
A solution to the problem would be a recursive mutex. However, a normal
recursive mutex does not fix the problem because it is not guaranteed
that the callbacks are called from the same thread. What we see in Broadcom's
implementation for example is:
- OMX_Foo is called
- OMX_Foo waits on a condition
- A callback is executed in a different thread
- When the callback returns, its calling function
signals the condition that OMX_Foo waits on
- OMX_Foo wakes up and returns
The solution I came up with here is to take a second lock inside the callback,
but only if recursion is expected to happen. Therefore, all calls to OMX
functions are guarded by calls to gst_omx_rec_mutex_begin_recursion() / _end_recursion(),
which effectively tells the mutex that at this point we want to allow calls
to _recursive_lock() to succeed, although we are still holding the master lock.
This happens on the Galaxy Nexus, and causes the pipeline to hang waiting
endlessly for a drain. The hack replaces the wait with a wait + 500ms timeout.
gst_omx_port_set_flushing() calls OMX_FillThisBuffer at the end of a flush
without releasing the port lock, and this can cause a deadlock with the
EventHandler. This patches fixes this by dropping the lock for the duration of
the fill buffer call.
This change adds prelimary buildsystem hooks to
build gst-omx with the Android buildsystem. Like
the rest of GStreamer's Android hooks, the process
relies on the availability of androgenizer. A tool
developed by Collabora to automatically generate
Android.mk files from within the auto* setup.
Androgenizer is currently available at:
http://cgit.collabora.com/git/user/derek/androgenizer.git/
The event will be forwarded downstream from the srcpad
loop function after the last buffer was generated by the
component. Forwarding it after ::finish() will use the
sinkpad streaming thread and does not guarantee that
the encoder is completely drained.
Get rid of weird code that copies a list manually, taking
ownership of the elements and then frees the old list. Instead,
just take over the old list entirely. (If the intent was to
reverse the list, one could use g_list_reverse() instead).
Then, push events in the list out from last to first (since they
were prepended as they came in) instead of just pushing out the
last in the list and leaking the others.
