Commit graph

63 commits

Author SHA1 Message Date
Nicolas Dufresne
1bef43f9d4 v4l2decoder: Track pending request
With the asynchronous slice decoding, we only queue up to 2 slices
per frames. That side effect is that now we are dequeuing bitstream
buffers in both decoding and presentation order. This would lead to
a bitstream buffer from a previous frame being dequeued instead of
the expected last slice buffer and lead to us trying to queue an
already queued bitstream buffer.

We now fix this by tracking pending requests. As request are executed
in decoding order, we marking a request done, we can effectively
dequeue bitstream buffer from all previous request, as they have been
executed already.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
a88e63dd2f v4l2decoder: Improve debug tracing
Add some missing traces and move per-slice operation to TRACE level to
reduce the noise level.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
d5a205cff4 v4l2decoder: Convert request pool to GstQueueArray
The decoder is not being access from multiple threads, instead it is
always protected by the streaming lock. For this reason, a
GstAtomicQueue for the request pool is overkill and may even introduce
unneeded overhead. Use a GstQueueArray in replacement, the
GstQueueArray is a good fit since the number of item is predictable and
unlikely to vary at run-time.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
a2eb1b57ff v4l2slh264dec: Wait on previous pending request in slice mode
In slice mode, we'll do one request per slice. In order to recycle
bitstream buffer, and not run-out, wait for the last pending
request to complete and mark it done.

We only wait after having queued the current slice in order to reduce
that potential driver starvation and maintain performance (using dual
buffering).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
b20c6fe6c4 v4l2slh264dec: Renew bitstream buffer after submitting slice
Submitting a slice actually clears the bitstream buffer. Ensure we
have a newly allocated bitstream buffer for the next slice.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
bc1a0323a9 v4l2slh264dec: Factor out bitstream allocation
No functional changes.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
779f331bd4 v4l2slh264dec: Add a helper to ensure output buffer
In preparation of multi-slice decoding, we will decode multiple
slices into the same buffer. This will ensure we have a buffer to
decode to, queued into the driver.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
d65f7de650 v4l2slh264dec: Factor out request wait
This will be reused to wait for previous slices to be complete
when dealing with following slices (in slice decoding mode).

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
176a860169 v4l2slh264dec: Remove double return in submit_bitstream()
This is code cleanup, no functional changes.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
1f48e60bde v4l2slh264dec: Fix typo in debug trace
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1395>
2020-07-02 12:21:51 -04:00
Nicolas Dufresne
483539f45c v4l2slh264dec: Fix reading mode and start code type
These two controls are not pointer based, so we don't need to pass any size or
pointer and need to copy the values afterward. This fixes H264 decoding
regression with Hantro and RKVDEC drivers.

Fixes 037730a787

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1330>
2020-06-09 10:52:44 -04:00
Nicolas Dufresne
d16367f046 v4l2codecs: h264: Add missing break
There was a missing break for the 4:4:4 case which would break the sizeimage
calculation. We don't currently have hardware that supports 4:4:4, so this
code wasn't tested. This was detected by Coverity.

CID 1463592 1463591

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1283>
2020-05-21 20:12:07 +00:00
Nicolas Dufresne
2160239548 v4l2slh264dec: Request large enough bitstream buffer
The Cedrus driver would otherwise choose 1KB buffer, which is too small.
This follows what some drivers do, which is simply to use the size a
packed raw image would have. Specifications do not really guaranty any minimum
compression ratio.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1268>
2020-05-19 17:51:11 +00:00
Nicolas Dufresne
037730a787 v4l2slh264dec: Add slice based decoder support
This adds support for slice based decoder like the Allwinner/Cedrus driver. In
order to keep things efficient, we hold the sink buffer until we reach the end
of the picture. Note that as we don't know which one is last, we lazy queue the
slices. This effectively introduces one slice latency.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1268>
2020-05-19 17:51:11 +00:00
Nicolas Dufresne
661f0c161e v4l2codecdec: Fix error handling
If none of the format the HW produce is supported, the fiter will be NULL,
which would lead to assertion when trying to release it.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1268>
2020-05-19 17:51:11 +00:00
Nicolas Dufresne
26264163de v4l2decoder: Add legacy non-multiplanar support
The Cedrus driver uses the lagacy buffer type (non-mplane). This automatically
detect and use the right v4l2_buf_type. This also affect code using
v4l2_buffer and v4l2_format structures.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1268>
2020-05-19 17:51:11 +00:00
Nicolas Dufresne
14b0b68354 v4l2codecs: Update kernel headers
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1268>
2020-05-19 17:51:11 +00:00
Víctor Manuel Jáquez Leal
d2aae6bb96 codecs: h264decoder: ref pic lists as decode_slice parameters
Pass reference picture lists to decode_slice() vmethods

Change gstv4l2codech264dec and gstnvh264dec accordingly.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1238>
2020-05-19 16:57:09 +00:00
Ting-Wei Lan
d10e2381c7 v4l2codecs: Fix compilation error on FreeBSD
This commit does the following things to fix compilation on FreeBSD:

1. Add required typedefs to linux/types-compat.h.
2. Remove unnecessary include linux/ioctl.h and replace linux/types.h
   with linux/types-compat.h. Both files do not exist on FreeBSD.
3. Check the header including makedev macro. FreeBSD does not have
   sys/sysmacros.h, and including it unconditionally causes error.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1259>
2020-05-11 15:17:31 +00:00
Nicolas Dufresne
5ece17ecbd v4l2slvp8dec: Flip the meaning of segment_feature_mode
In section 9.3.4 a), segment_feature_mode have 0 for absolute and 1 for delta,
while in 19.2, it says the opposite. But the reference code, which usually
rules over the text state that 1 means absolute:

 if (hdr->update_data)
 {
     hdr->abs = bool_get_bit(bool);

And uses it with that meaning to decide weither to override the existing value
or just add the detla. This fixes multiple decoding issues.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1245>
2020-05-05 17:57:08 +00:00
Nicolas Dufresne
11c62eb015 v4l2slvp8dec: Copy header version
This field was not copied.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1245>
2020-05-05 17:57:08 +00:00
Nicolas Dufresne
8c56822936 v4l2slvp8dec: Add debugging for reference frames
This simply trace the frame number of the references used for decoding.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1245>
2020-05-05 17:57:08 +00:00
Nicolas Dufresne
a449815f95 v4l2slvp8dec: Ensure width/height is always set
Our parser strictly read the bitstream. As it's known from DXVA that always
having a valid width/height might be needed, use the cached width/height
instead of the value from the parser. This didn't impact Hantro driver.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1245>
2020-05-05 17:57:08 +00:00
Nicolas Dufresne
ffd74fb5ec v4l2slvp8dec: Fix debug category name
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1245>
2020-05-05 17:57:08 +00:00
Nicolas Dufresne
bf9f82c8c3 v4l2codecs: Factor out plugin registration
This introduce a common place for generic functions and factor out the plugin
registration code. This code is nearly identical between implementation.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1216>
2020-04-28 22:53:11 +00:00
Nicolas Dufresne
de2cccaa47 v4l2decoder: Relax width/height return value check
The driver adjust the width/height to coded size. This was not an issue for
H264, as the coded size is in the bitstream, but is an issue with VP8.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1216>
2020-04-28 22:53:11 +00:00
Nicolas Dufresne
72c0b3e5ae v4l2codecs: Add VP8 decoder
This is derived from the H264 decoder, some boiler plate will be factored out
in the following commits.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1216>
2020-04-28 22:53:11 +00:00
Seungha Yang
ac850fd25d v4l2slh264dec: Support avc and avc3 formats
h264decoder baseclass provides parsed sps/pps (from codec data) for
subclass. Also, since current implementation is putting start code prefix
manually, the format of stream should not be matter for subclass.

Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1214>
2020-04-23 23:38:03 +00:00
Víctor Manuel Jáquez Leal
e96435dfea v4l2codecs: allocator: use gst_clear_object
So tracer can do its job.
2020-04-20 12:23:32 +00:00
Víctor Manuel Jáquez Leal
691fdb85c0 v4l2codecs: fix v4l2codecdevice get type
Currently the GType of v4l2codecdevice is hardcoded to zero, but it
rather should be delivered by the GType system.
2020-04-10 21:05:19 +02:00
Nicolas Dufresne
8f0ceba251 v4l2codecs: Fix FD leak during device enumeration
This was revealed by Coverity.

CID 1461248
2020-04-08 20:25:44 +00:00
Nicolas Dufresne
f5da12bcf5 v4l2decoder: Fix file descriptor leak
A copy paste error was leading to file descriptor leak. This was detected by
Coverity.

CID 1461285
2020-04-07 17:10:08 -04:00
Nicolas Dufresne
4cb871eb53 v4l2codecs: Only build this plugin on Linux
This is not useful on any other OSs, it will also avoid potential build
failure as this code uses Linux specific calls.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
7254a18f0f v4l2codecs: Add plugin dependency
This ensure that the registry cache get updated when a meaningful change is
made in /dev for files named media*.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
e70993bf43 v4l2codecs: Wait for buffers to come back
This code add required mechanism to try and allocate (not implemented yet)
otherwise wait for more buffers. This also comes with mechanism to terminate
the wait on flush or PAUSED_TO_READY transitions.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
09a9ffcda6 v4l2codecs: Implement flushing sequence
This simply consit of cycling through STREAMOFF/STREAMON with stateless
decoders.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
f69283819a v4l2codecs: allocator: Add method to wait for more buffers
This add function to wait for buffers to get back into the pool along with a
set_flushing() method to allow unblocking this wait.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
bc0a8ff40d v4l2codecs: pool: Create new buffer when pool is empty
This simply create an empty GstBuffer when the pool is empty. This way it's up
to the allocator to grow or wait if we ran out of memory.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
ee26c9c9a1 v4l2slh264dec: Improve end_picture() robustness
Use a goto to ensure that for all cases we cleanup the current picture state.
And move the src buffer allocation higher, so we don't queue a bitstream
buffer if we don't have a picture buffer to decode into.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
390cbe1f00 v4l2slh264dec: Add output format negotiation
This allow negotiating the output format through caps. Some drivers can
pipeline the decoder buffer through an image processor. This only support
colorspace conversion for now.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
6494d7b056 v4l2format: Convert between V4L2 and GST video format
This will be needed in the output format negotiation.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
2a1836f183 v4l2codecs: Read driver provided stride
This implements driver stride support but only for single allocation buffers.
This code is imported from the original v4l2 plugin and adapted to the new
helper context.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
21652a8e52 v4l2h264dec: Copy frames when GstVideoMeta is not supported
In some case, when downstream does not support GstVideoMeta, we need to
normalize the stride and offset of the buffer so that downstream can render
properly with a GstVideoMeta. This code is not called when GstVideoMeta is
supported downstream.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
3c2d25eebf v4l2codec: Add initial GstVideoMeta support
In this patch we strictly set the GstVideoMeta width/height to the coded width
and height. Further patches will add stride support and frame copying when
downstream does not support GstVideoMeta.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
f6690420ce v4l2slh264dec: Prevent spurious renegotiation
Don't let downstream cause a renegotiation at random point in time. This would
lead to spurious renegotiation and the decoder state may not be recoverable.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
b47929a74d v4l2slh264dec: Hold on picture buffers
These buffers should not be reused for decoding until they are no longer
referenced.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
b107c20198 v4l2codecs: Fix bytesused value
Pass the actual amount of bytes we have copied into the bitstream buffer. Also
unmap the memory before queuing.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
0d740a184e v4l2codech264dec: Implement finish frame
This enables the request to be processed by the HW (STREAMON) and to be pushed
downstream for further processing.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
3cc6643050 v4l2codecdecoder: Add support for dequeuing buffers
This includings polling of the request and streamon/streamoff.
2020-03-31 09:34:05 -04:00
Nicolas Dufresne
4305771178 v4l2codech264dec: Implement queuing
We now pass the controls, associated to a request, queue the bitstream, qeueue
a picture buffer to decode into and finally queue the request. This now runs
until the buffer pool is exhausted. The next step will be to dequeue.
2020-03-31 09:34:05 -04:00