On Mon, 2018-09-10 at 13:16 -0400, Nicolas Dufresne wrote: > Le lundi 10 septembre 2018 à 12:37 -0300, Ezequiel Garcia a écrit : > > On Mon, 2018-09-10 at 17:00 +0200, Hans Verkuil wrote: > > > From: Hans Verkuil <hans.verk...@cisco.com> > > > > > > state->info was NULL since I completely forgot to set state->info. > > > Oops. > > > > > > Reported-by: Ezequiel Garcia <ezequ...@collabora.com> > > > Signed-off-by: Hans Verkuil <hans.verk...@cisco.com> > > > > For both patches: > > > > Tested-by: Ezequiel Garcia <ezequ...@collabora.com> > > > > With these changes, now this gstreamer pipeline no longer > > crashes: > > > > gst-launch-1.0 -v videotestsrc num-buffers=30 ! video/x- > > raw,width=1280,height=720 ! v4l2fwhtenc capture-io-mode=mmap output- > > io-mode=mmap ! v4l2fwhtdec > > capture-io-mode=mmap output-io-mode=mmap ! fakesink > > > > A few things: > > > > * You now need to mark "[PATCH] vicodec: fix sparse warning" as > > invalid. > > * v4l2fwhtenc/v4l2fwhtdec elements are not upstream yet. > > * Gstreamer doesn't end properly; and it seems to negotiate > > Is the driver missing CMD_STOP implementation ? (draining flow) >
I think that's the case. Gstreamer debug log, right before it stalls: 0:00:16.929785442 180 0x5593bcbd18a0 DEBUG v4l2videodec gstv4l2videodec.c:375:gst_v4l2_video_dec_finish:<v4l2fwhtdec0> Finishing decoding 0:00:16.931866009 180 0x5593bcbd18a0 DEBUG v4l2videodec gstv4l2videodec.c:340:gst_v4l2_decoder_cmd:<v4l2fwhtdec0> sending v4l2 decoder command 1 with flags 0 0:00:16.934260349 180 0x5593bcbd18a0 DEBUG v4l2videodec gstv4l2videodec.c:384:gst_v4l2_video_dec_finish:<v4l2fwhtdec0> Waiting for decoder stop [stalls here] Regards, Ezequiel
