Commit graph

134 commits

Author SHA1 Message Date
Sebastian Dröge
9c464924a5 omxvideodec: Extract data from the input buffer, not the codec data 2012-12-19 12:44:31 +01:00
Sebastian Dröge
7fa4616598 omx: Fix refcount problem with frames being dropped because of decoder bugs 2012-11-12 15:14:09 +01:00
Sebastian Dröge
2481ceb8c4 omx: Update to new GLib thread API 2012-11-12 11:29:48 +01:00
Sebastian Dröge
bced70525a omx: Use open/close vfuncs 2012-10-22 14:34:53 +02:00
Sebastian Dröge
9d0e582484 omx: Handle video meta correctly 2012-10-22 14:28:04 +02:00
Sebastian Dröge
5fdb490f1a omx: Port to video base classes from -base 2012-10-22 14:20:41 +02:00
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
George Kiagiadakis
7ca067be28 omxvideodec: Implement no-empty-eos-buffer hack, as in omxvideoenc.
Conflicts:

	omx/gstomxvideodec.c
2012-05-03 12:24:43 +02:00
George Kiagiadakis
4aa3fa8a0d omxvideodec: Fix coding style in the drain-may-not-return hack code 2012-05-03 12:20:05 +02:00
Arun Raghavan
9e4cddee7e omxvideodec: Add hack for Ducati components not returning from drain
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.
2012-04-26 09:42:52 +05:30
Olivier Crête
ff2603a727 Port to Gst 0.11 2012-04-13 12:16:34 -04:00
Sebastian Dröge
4c225f9cdd omxvideodec: Remove workaround for basevideocodec bug 2011-12-12 14:26:34 +01:00
Sebastian Dröge
05ad3343bf omxvideodec: Use correct timestamp, duration and filled length for the EOS buffers 2011-12-06 12:47:12 +01:00
Sebastian Dröge
b4e632cefe omxvideodec: Properly drop frames due to QoS
Instead of finishing them they should be passed to drop(), which
will then send QoS messages.
2011-12-01 16:20:36 +01:00
Sebastian Dröge
1c0396de0e omxvideodec: First set ports to flushing before waiting for the srcpad streaming thread to finish 2011-11-18 09:58:58 -08:00
Sebastian Dröge
1371dafae4 omxvideodec: Signal the drain GCond even if downstream returned an error 2011-11-17 11:29:18 -08:00
Sebastian Dröge
30610a746e omxvideodec: Shutdown component in PAUSED->READY and deallocate buffers 2011-11-17 10:19:47 -08:00
Sebastian Dröge
c99af4a32e omxvideodec: Always push buffers downstream, even if we didn't find a corresponding GstVideoFrame 2011-11-16 11:21:25 -08:00
Sebastian Dröge
0f07403ff2 omxvideodec: Don't release buffers twice if dropping because of QoS 2011-11-14 12:50:26 -08:00
Sebastian Dröge
30d2f970f9 omxvideodec: Make srcpad caps setting threadsafe 2011-11-10 15:17:41 +01:00
Sebastian Dröge
b36db47cba omxvideodec: Log if acquiring buffer for EOS failed 2011-11-10 14:51:06 +01:00
Sebastian Dröge
3412ecf538 omxvideodec: The component is not started in READY 2011-11-10 14:42:13 +01:00
Sebastian Dröge
6e41e31e6c omxvideodec: Always flush the ports and make sure no processing is happening in ::reset
This fixes a race condition that happened when seeking
very often in a short period of time.
2011-11-09 15:46:39 +01:00
Sebastian Dröge
60a6c1b49f omxvideodec: Fix minor race condition when draining after upstream signalled EOS 2011-11-09 15:45:20 +01:00
Sebastian Dröge
92c26f684a omxvideodec: Improve EOS handling
If downstream return UNEXPECTED we should still signal the
drain cond because nothing will trigger this again later.
2011-11-09 15:44:11 +01:00
Sebastian Dröge
c026bee8c6 omxvideodec: Improve debugging of EOS and draining 2011-11-09 15:43:32 +01:00
Sebastian Dröge
59faafc30b omxvideodec: The component is not started already when going from READY to PAUSED 2011-11-09 15:42:46 +01:00
Sebastian Dröge
6c2cca3e94 omxvideodec: No need to signal the drain cond when going from READY to PAUSED 2011-11-09 15:42:13 +01:00
Sebastian Dröge
cf22a12156 omxvideodec: Improve debugging in case of QoS-related frame drops 2011-11-09 09:00:57 +01:00
Sebastian Dröge
3da6d8bfa3 omxvideodec: Release the video codec 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
9251e0004c omxvideodec: 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 11:08:06 +01:00
Sebastian Dröge
b10c2cfc72 omxvideodec: Implement dropping of too late frames via QoS 2011-11-08 09:09:28 +01:00
Sebastian Dröge
d4eb77fa2b omxvideodec: Make sure to always release buffers back to OMX 2011-11-08 08:31:32 +01:00
Sebastian Dröge
9151617d39 omxvideodec: Free pending frames after draining component 2011-11-07 14:00:47 +01:00
Sebastian Dröge
95b06504f9 omxvideoenc: Make handling and usage of the base video codec frames threadsafe 2011-11-07 11:04:27 +01:00
Sebastian Dröge
be1f75d2f4 omxvideodec: Always free all pending frames when caps changes require reconfiguration 2011-11-07 10:58:24 +01:00
Sebastian Dröge
d5371b9208 omxvideodec: Only drain the component a single time and only after processing started 2011-11-04 09:44:04 +01:00
Sebastian Dröge
ec9dde48e5 omxvideodec: If no stride was set for the OMX output port assume GStreamer stride
This is not really correct but there's nothing else we could do.
2011-11-02 10:39:50 +01:00
Sebastian Dröge
e4b7ffcc1b omxvideodec: Implement draining of the component and use it
This makes sure that all buffers are decoded and pushed downstream
before flushing the ports and losing some buffers.
2011-11-01 16:23:22 +01:00
Sebastian Dröge
2f2c17a1b5 omxvideodec: Fix deadlock if ::reset is called before finding the corresponding frame of a decoded buffer 2011-11-01 16:23:22 +01:00
Sebastian Dröge
3b4260e54f omxvideodec: Forward downstream flow returns to upstream 2011-11-01 13:58:38 +01:00
Sebastian Dröge
10a5680692 omxvideodec: Fix deadlock caused by calling reset while the loop function does something with the base video codec stream lock 2011-10-20 15:21:07 +02:00
Sebastian Dröge
cf6f14d03c omxvideodec: Move locking at the correct place 2011-10-20 14:30:38 +02:00
Sebastian Dröge
bb1dd8aaba omxvideodec: Add API for subclasses to prepare/convert frames 2011-09-29 10:44:20 +02:00
Sebastian Dröge
ab9ce7db7d omxvideodec: 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 10:35:37 +02:00
Sebastian Dröge
792cdb7d1f omxvideodec: New hack for QCOM to recreate the component instead of reconfiguring it on caps changes 2011-09-28 10:35:31 +02:00
Sebastian Dröge
5659bd3b4f omxvideodec: Add some more checks for OMX buffer sizes 2011-09-23 16:40:30 +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
3ba45aa0af omxvideodec: Negotiate video format with downstream and what the component claims to support 2011-09-02 14:43:43 +02:00
Sebastian Dröge
e583c48222 omxvideodec: Release basevideocodec stream lock while waiting for a buffer
This prevents deadlocks if no empty input buffers are available and
releasing input buffers requires the loop function to handle some
output buffers first.
2011-08-19 09:23:54 +02:00
Sebastian Dröge
7a09447406 omxvideodec: Use "video/x-raw-yuv" as src template caps instead of strict I420 caps 2011-08-10 10:21:15 +02:00
Sebastian Dröge
2ea5bdf553 omxvideodec: Set the state back to StateLoaded even if an error happened 2011-08-10 08:52:25 +02:00
Sebastian Dröge
febc2d99cb omxvideodec: Don't abort if the color format is not supported but give a useful error message 2011-08-03 13:10:33 +02:00
Sebastian Dröge
8ac445abd8 omxvideo{enc,dec}: Only set/unset flushing state on ports if they were created already 2011-07-25 13:19:06 +02:00
Sebastian Dröge
87587dd6d2 omxvideodec: Try harder to deallocate the buffers after errors happened 2011-07-25 10:48:58 +02:00
Sebastian Dröge
71d176f71a omx: Improve debug output a bit 2011-07-19 12:29:51 +02:00
Sebastian Dröge
b6558570bf omx: Rework port reconfiguration again and only use the Bellagio specific hacks with Bellagio
We only reconfigure ports that need to be reconfigured now instead of
always all ports.
2011-07-19 10:33:54 +02:00
Sebastian Dröge
4616d804ac omx: Add infrastructure to enable special hacks for broken OpenMAX implementations 2011-07-19 10:33:15 +02:00
Sebastian Dröge
9377aefb87 omxvideodec: Add support for converting between omx and gst rowstrides 2011-07-18 08:41:20 +02:00
Sebastian Dröge
41feed55b7 omxvideodec: Add support for NV12 / OMX_COLOR_FormatYUV420SemiPlanar 2011-07-14 08:29:03 +02:00
Sebastian Dröge
961445aa1b omxvideodec: Only flush the component ports after we passed input to them 2011-07-14 07:58:41 +02:00
Sebastian Dröge
252624c7cc omxvideodec: Only change states downwards if an upper state was reached 2011-07-13 21:19:34 +02:00
Sebastian Dröge
f0cbbad0f1 omx: Add support for setting the component-role 2011-07-13 20:37:02 +02:00
Sebastian Dröge
0a9fe2f146 omx: Improve error reporting by formatting the error codes better and also providing their string representation 2011-07-13 20:22:51 +02:00
Sebastian Dröge
8a78da68a6 omxvideodec: Fix typo 2011-07-13 14:02:50 +02:00
Sebastian Dröge
fb0ca24654 omxvideodec: Set SYNCFRAME flag on the OMX buffers for non-delta units 2011-07-13 12:46:50 +02:00
Sebastian Dröge
f5690ff412 omxvideodec: Free all pending frames when resetting the decoder
Workaround for https://bugzilla.gnome.org/show_bug.cgi?id=654529
2011-07-13 12:37:44 +02:00
Sebastian Dröge
af159705c9 omxvideodec: Handle output buffers without a corresponding GstVideoFrame more gracefully
This can happen on EOS in some cases and when the input is not
properly framed.
2011-07-13 10:02:20 +02:00
Sebastian Dröge
1b08dfa2a6 omxvideodec: Fix deadlock when finishing old frames that are left over by the decoder 2011-07-13 09:31:22 +02:00
Sebastian Dröge
a24cdd41bc omxvideodec: Make sink/src pad template caps configurable 2011-07-12 11:36:42 +02:00
Sebastian Dröge
113d18caea omxvideodec: Make core/component-name and in/outport index configurable 2011-07-12 11:04:16 +02:00
Sebastian Dröge
2042e08b33 omxvideodec: Use the destroy notify to free the coder_hook 2011-07-09 11:06:06 +02:00
Sebastian Dröge
9c9315059e omxvideodec: Fix some minor memory leaks 2011-07-09 11:06:06 +02:00
Sebastian Dröge
db08890edd omx: Rework port reconfiguration
We always reconfigure all ports now if the settings of one
port changes to prevent lots of race conditions, dropped
frames and similar issues.
2011-07-09 11:06:06 +02:00
Sebastian Dröge
0fbff1000f omxvideodec: Use the frames storage of the base class instead of implementing our own
They could get out of sync and we could store already destroyed frames.
2011-07-09 11:06:06 +02:00
Sebastian Dröge
c8c1c7f10f omx: Add more checks to acquire_buffer() and return the current state additional to the buffer
Also refactor the code flow in the video decoder for this. This makes
the usage of acquire_buffer() easier and more atomic.
2011-07-09 11:06:06 +02:00
Sebastian Dröge
934fac7946 omxvideodec: Also flush/unflush the input port when changing the state PAUSED<->READY 2011-07-09 11:06:06 +02:00
Sebastian Dröge
3ac2bfc976 omxvideodec: Don't leak the codec_data after sending it 2011-07-09 11:06:06 +02:00
Sebastian Dröge
939d30ed17 omxvideodec: Remove reconfiguration test hack 2011-07-09 11:06:05 +02:00
Sebastian Dröge
9a3753bd61 omxvideodec: Error out if the GStreamer allocated buffer is smaller than the OpenMAX output buffer
Usually this must never happen but currently it happens during reconfigurations
because of a race condition. Still it's better than crashing.
2011-07-09 11:06:05 +02:00
Sebastian Dröge
11d2e806c2 omxvideodec: Remove obsolete FIXME comment 2011-07-09 11:06:05 +02:00
Sebastian Dröge
08181d86dc omxvideodec: Make the inport and outport index configurable by the subclass 2011-07-09 11:06:05 +02:00
Sebastian Dröge
bc1e73e8c8 omx: Add initial version of OpenMAX framework, video decoder base class and MPEG4 video decoder
This currently hardcodes a lot of stuff but works at least.

Also adds a generic framework for handling OpenMAX cores, components
and ports.
2011-07-09 11:06:05 +02:00