[PATCH 1/3] media: rockchip: Introduce the rkvdec2 driver

[Detlev Casanova]

On Monday, June 17, 2024 10:04:59 A.M. EDT Nicolas Dufresne wrote:
> Hi,
> 
> Le lundi 17 juin 2024 à 17:47 +0800, Jianfeng Liu a écrit :
> > Hi Detlev,
> > 
> > Thanks a lot for your work! I try to use rkvdec2 with chromium but it
> > can't play h264 video. Here is the log of chromium:
> > 
> > [5799:5873:0617/171224.850061:VERBOSE2:video_decoder_pipeline.cc(473)]
> > Initialize(): config: codec: h264, profile: h264 high, level: not
> > available, alpha_mode: is_opaque, coded size: [1920,1080], visible rect:
> > [0,0,1920,1080], natural size: [1920,1080], has extra data: true,
> > encryption scheme: Unencrypted, rotation: 0°, flipped: 0, color space:
> > {primaries:BT709, transfer:BT709, matrix:BT709, range:LIMITED}
> > [5799:5886:0617/171224.850915:VERBOSE2:v4l2_video_decoder.cc(182)]
> > V4L2VideoDecoder():
> > [5799:5886:0617/171224.851218:VERBOSE1:v4l2_device.cc(128)] Open(): No
> > devices supporting H264 for type: 0
> > [5799:5886:0617/171224.851346:VERBOSE4:v4l2_queue.cc(1069)] This queue
> > does  support requests.: No such file or directory (2)
> This one indicates that V4L2_BUF_CAP_SUPPORTS_REQUESTS might be missing in
> the REQBUFS implementation. I suspect GStreamer simply assumes this today
> for driver exposing stateless formats (which is fair, its not a compliance
> test, and we don't have a codec compliance yet).

There is a src_vq->supports_requests = true; Do I also need to set a flag on 
the device caps ?

> I'd suggest to check and fix this one, and retry, might only be noise, might
> be the main cause, we cannot tell.
> 
> > [5799:5886:0617/171224.851426:VERBOSE1:v4l2_video_decoder.cc(476)]
> > InitializeBackend(): Using a stateless API for profile: h264 high and
> > fourcc: S264
> > [5799:5886:0617/171224.851687:VERBOSE1:v4l2_video_decoder.cc(598)]
> > SetupInputFormat(): Input (OUTPUT queue) Fourcc: S264
> > [5799:5886:0617/171224.851797:VERBOSE1:v4l2_video_decoder.cc(636)]
> > AllocateInputBuffers(): Requesting: 17 OUTPUT buffers of type
> > V4L2_MEMORY_MMAP
> > [5799:5886:0617/171224.867687:VERBOSE1:v4l2_queue.cc(1511)] Streamon():
> > (OUTPUT_MPLANE) VIDIOC_STREAMON failed: Invalid argument (22)
> > [5799:5886:0617/171224.867902:VERBOSE1:v4l2_video_decoder.cc(937)]
> > StartStreamV4L2Queue(): Failed to streamon V4L2 queue.
> > [5799:5886:0617/171224.868009:VERBOSE1:v4l2_video_decoder.cc(1377)]
> > SetState(): Error occurred, stopping queues.
> > [5799:5886:0617/171224.868105:ERROR:v4l2_video_decoder.cc(120)]
> > StartStreamV4L2Queue failed at
> > Decode at media/gpu/v4l2/v4l2_video_decoder.cc:915
> > [5799:5886:0617/171224.871898:WARNING:v4l2_video_decoder_backend_stateles
> > s.cc(126)] There is/are 0 pending CAPTURE queue buffers pending dequeuing.
> > This might be fine or a problem depending on the destruction semantics
> > (of theclient code.
> > 
> > Here is the chromium code failed when calling VIDIOC_STREAMON:
> > https://github.com/chromium/chromium/blob/125.0.6422.60/media/gpu/v4l2/v4l
> > 2_queue.cc#L1508
> > 
> > I'm running chromium v125.0.6422.60 and I can decode 1080p h264 with
> > hantro g1 decoder on rk3588.
> 
> Are you using minigbm ? Because if you do, we don't have minigbm code for
> this driver (and have no plan to do so, since we don't aim for ChromeOS
> support).
> 
> Nicolas
> 
> > Best regards,
> > Jianfeng

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.infradead.org/pipermail/linux-rockchip/attachments/20240617/6187a9b2/attachment.sig>