Re: Venus v4l2-compliance failures

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Thu, Feb 17, 2022 at 9:35 AM Hans Verkuil <hverkuil@xxxxxxxxx> wrote:
>
> On 17/02/2022 15:12, Stanimir Varbanov wrote:
> > Hi Hans,
> >
> > Presently we have two failures while running v4l2-compliance on venus
> > stateful decoder:
> >
> > 1. fail: v4l2-compliance.cpp(753): !ok
> >         test for unlimited opens: FAIL
> >
> > 2. fail: v4l2-test-formats.cpp(1668): IS_DECODER(node)
> >         test Cropping: FAIL
> >    fail: v4l2-test-codecs.cpp(104): node->function !=
> > MEDIA_ENT_F_PROC_VIDEO_DECODER
> >         test VIDIOC_(TRY_)DECODER_CMD: FAIL
> >
> > Failure #1 is related to the limitation we made in decoder open(). The
> > maximum parallel decoding sessions is limited to 16 and the check
> > for this maximum is made in decoder open() because the clients wanted to
> > know that earlier. For example, Chromium browser can open 16 hw
> > accelerated decoder sessions (in separate Tabs) and from 17 and upward
> > it will fallback to sw decoder. I wonder how that failure can be fixed.
>
> I'm wondering if this isn't better done via a read-only control that
> reports the max number of parallel sessions.
>
Do you see this as a constant value?  It would be burdensome if the
client had to keep track of how many contexts are in use.  Or do you
see this as a number of currently available contexts?

> I really hate artificial open() limitations, it's very much against the
> v4l2 philosophy.
>

[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux