Sebastian Dröge
a42f27547d
omx: Implement new approach for locking that should solve all deadlocks on RPi
...
No mutex is locked while calling any OpenMAX functions anymore
and everything from the OpenMAX callbacks is inserted into a message
queue and handled from outside the callbacks.
Also there's only a single mutex and condition variable per component
now for handling anything from OpenMAX callbacks and a single mutex
for keeping our component/port state sane.
2013-01-14 10:36:32 +01:00
Sebastian Dröge
1d01c0ebf0
omx: Fix ununsed variable compiler warning
2012-12-20 12:30:05 +01:00
Sebastian Dröge
83b4c1d892
omx: No need to start the srcpad task in ::start() already
...
It will be started properly after the caps are set.
2012-12-20 12:27:47 +01:00
Sebastian Dröge
f90a6ed9e9
omx: Improve debug output
2012-12-20 12:20:31 +01:00
Sebastian Dröge
2481ceb8c4
omx: Update to new GLib thread API
2012-11-12 11:29:48 +01:00
Sebastian Dröge
e84fee6488
omx: Fix compilation after gst_pad_start_task() API changes
2012-06-20 11:09:13 +01:00
George Kiagiadakis
158775f497
Implement a new custom recursive mutex type and fix locking in callbacks so that in-context calls are allowed.
...
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.
2012-05-07 17:01:16 +03:00
Sebastian Dröge
3523724970
omxaudioenc: Use gst_buffer_extract() to copy the input buffer data to the OpenMAX buffer
2012-04-16 08:35:33 +02:00
Olivier Crête
ff2603a727
Port to Gst 0.11
2012-04-13 12:16:34 -04:00
Jonas Larsson
887d43c290
omxaudioenc: Add hack for encoder components that don't allow empty EOS buffers
2011-12-14 10:09:34 +01:00
Sebastian Dröge
766f5bd161
omxaudioenc: Don't forward EOS events immediately but let all other events be handled by the base class
...
Previously this logic was inversed, which did not make any sense at all.
2011-12-06 13:28:41 +01:00
Sebastian Dröge
064e4c628c
omxaudioenc: Use correct timestamp, duration and filled length for the EOS buffers
2011-12-06 12:47:25 +01:00
Sebastian Dröge
217ac7b3be
omxaudioenc: Use audio base classes from gst-plugins-base instead of having our own copies
2011-11-25 11:31:58 +01:00
Sebastian Dröge
9917fbe4c5
omxaudioenc: First set ports to flushing before waiting for the srcpad streaming thread to finish
2011-11-18 10:00:31 -08:00
Sebastian Dröge
8a5fbb876c
omxaudioenc: Signal the drain GCond even if downstream returned an error
2011-11-17 11:29:39 -08:00
Sebastian Dröge
7df465bfb3
omxaudioenc: Shutdown component in PAUSED->READY and deallocate buffers
2011-11-17 10:19:47 -08:00
Sebastian Dröge
1ab4e4b945
omxaudioenc: Make srcpad caps setting threadsafe
2011-11-10 15:18:08 +01:00
Sebastian Dröge
f0775fac6f
omxaudioenc: Make the srcpad task and the sinkpad streaming thread handling threadsafe
2011-11-10 15:10:14 +01:00
Sebastian Dröge
b62ffe1ada
omxaudioenc: Always flush the ports and make sure no processing is happening in ::flush
...
This fixes a race condition that happened when seeking
very often in a short period of time.
2011-11-10 15:03:05 +01:00
Sebastian Dröge
05e4b48855
omxaudioenc: Improve EOS handling
...
If downstream return UNEXPECTED we should still signal the
drain cond because nothing will trigger this again later.
2011-11-10 14:56:19 +01:00
Sebastian Dröge
1b7742ca98
omxaudioenc: Improve debugging of EOS and draining
2011-11-10 14:54:33 +01:00
Sebastian Dröge
8c1aee4171
omxaudioenc: No need to signal the drain cond when going from READY to PAUSED
...
Also the component is not started in READY
2011-11-10 14:41:50 +01:00
Sebastian Dröge
b45197c0d7
omxaudioenc: Release the audio encoder stream lock before acquiring an input buffer
...
Otherwise the srcpad task might block on this lock and
no buffers ever become available again.
2011-11-08 12:47:00 +01:00
Sebastian Dröge
17d0751887
omxaudioenc: Don't try to drain the component after EOS
...
And don't send EOS twice in any case. This most likely
will cause the component to not output it again and
is not necessary anyway.
2011-11-08 11:10:57 +01:00
Sebastian Dröge
093a27fea3
omxaudioenc: Minor code refactoring
2011-11-08 08:31:58 +01:00
Sebastian Dröge
0c0aa78373
omxaudioenc: Only drain the component a single time and only after processing started
2011-11-04 09:44:04 +01:00
Sebastian Dröge
1ef38751c7
omxaudioenc: Implement draining of the component and use it
...
This makes sure that all buffers are encoded and pushed downstream
before flushing the ports and losing some buffers.
2011-11-01 16:46:09 +01:00
Sebastian Dröge
f93a45fb66
omxaudioenc: Forward downstream flow returns to upstream
2011-11-01 15:10:12 +01:00
Sebastian Dröge
febb83ec54
omxaudioenc: Switch to Idle first and wait before switching to Loaded and deallocating buffers
...
Allocating buffers before the Idle state is reached can lead to crashes.
2011-09-28 10:35:38 +02:00
Sebastian Dröge
7dbf68a232
omxaudioenc: Add a check to prevent a zero-sized OMX buffer
2011-09-23 17:04:15 +02:00
Sebastian Dröge
f0fe1148b7
omx: Wait until the Executing state is reached before calling OMX_FillThisBuffer()
...
This correctly works around the QCOM race condition that happens when calling
FTB after setting the new state and before reaching it.
2011-09-14 10:16:41 +02:00
Sebastian Dröge
c5360dfae8
omxaudioenc: Always require at least OMX_MIN_PCMPAYLOAD_MSEC per input buffer
2011-08-17 14:33:31 +02:00
Sebastian Dröge
1daa53e554
omxaudioenc: Handle inbuf==NULL properly in ::handle_frame()
2011-08-17 14:17:18 +02:00
Sebastian Dröge
d1e1980e01
omxaudioenc: Add vfunc to get the number of samples inside a buffer
2011-08-17 13:03:50 +02:00
Sebastian Dröge
92545e554c
omxaudioenc: Release baseaudioencoder stream lock while waiting for a buffer in ::handle_frame()
...
This prevents deadlocks if no empty input buffers are available and
releasing input buffers requires the loop function to handle some
output buffers first.
2011-08-17 11:34:31 +02:00
Sebastian Dröge
c24533c8be
omxaudioenc: Remove hack that only applies to the video encoder class
2011-08-16 11:03:24 +02:00
Sebastian Dröge
3f33a577a9
omxaudioenc: Add initial version of audio encoder base class
2011-08-16 10:48:36 +02:00
