Message-ID: <CAAFQd5DiAnP5pc4SoPV3aDM7k71wYmPtAHzN7thquf47EOioHw@mail.gmail.com>
Date:   Fri, 8 Nov 2019 20:52:02 +0900
From:   Tomasz Figa <tfiga@...omium.org>
To:     Boris Brezillon <boris.brezillon@...labora.com>
Cc:     Ezequiel Garcia <ezequiel@...labora.com>,
        Hans Verkuil <hverkuil@...all.nl>,
        Linux Media Mailing List <linux-media@...r.kernel.org>,
        kernel@...labora.com,
        Nicolas Dufresne <nicolas.dufresne@...labora.com>,
        "open list:ARM/Rockchip SoC..." <linux-rockchip@...ts.infradead.org>,
        Heiko Stuebner <heiko@...ech.de>,
        Jonas Karlman <jonas@...boo.se>,
        Philipp Zabel <p.zabel@...gutronix.de>,
        Alexandre Courbot <acourbot@...omium.org>,
        fbuergisser@...omium.org,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Douglas Anderson <dianders@...omium.org>
Subject: Re: [PATCH v2 for 5.4 1/4] media: hantro: Fix s_fmt for dynamic
 resolution changes

On Fri, Nov 8, 2019 at 7:20 PM Boris Brezillon
<boris.brezillon@...labora.com> wrote:
>
> On Mon,  7 Oct 2019 14:45:02 -0300
> Ezequiel Garcia <ezequiel@...labora.com> wrote:
>
> > Commit 953aaa1492c53 ("media: rockchip/vpu: Prepare things to support decoders")
> > changed the conditions under S_FMT was allowed for OUTPUT
> > CAPTURE buffers.
> >
> > However, and according to the mem-to-mem stateless decoder specification,
> > in order to support dynamic resolution changes, S_FMT should be allowed
> > even if OUTPUT buffers have been allocated.
> >
> > Relax decoder S_FMT restrictions on OUTPUT buffers, allowing a resolution
> > modification, provided the pixel format stays the same.
> >
> > Tested on RK3288 platforms using ChromiumOS Video Decode/Encode Accelerator Unittests.
> >
> > Fixes: 953aaa1492c53 ("media: rockchip/vpu: Prepare things to support decoders")
> > Signed-off-by: Ezequiel Garcia <ezequiel@...labora.com>
> > --
> > v2:
> > * Call try_fmt_out before using the format,
> >   pointed out by Philipp.
> >
> >  drivers/staging/media/hantro/hantro_v4l2.c | 28 +++++++++++++++-------
> >  1 file changed, 19 insertions(+), 9 deletions(-)
> >
> > diff --git a/drivers/staging/media/hantro/hantro_v4l2.c b/drivers/staging/media/hantro/hantro_v4l2.c
> > index 3dae52abb96c..586d243cc3cc 100644
> > --- a/drivers/staging/media/hantro/hantro_v4l2.c
> > +++ b/drivers/staging/media/hantro/hantro_v4l2.c
> > @@ -367,19 +367,26 @@ vidioc_s_fmt_out_mplane(struct file *file, void *priv, struct v4l2_format *f)
> >  {
> >       struct v4l2_pix_format_mplane *pix_mp = &f->fmt.pix_mp;
> >       struct hantro_ctx *ctx = fh_to_ctx(priv);
> > +     struct vb2_queue *vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx, f->type);
> >       const struct hantro_fmt *formats;
> >       unsigned int num_fmts;
> > -     struct vb2_queue *vq;
> >       int ret;
> >
> > -     /* Change not allowed if queue is busy. */
> > -     vq = v4l2_m2m_get_vq(ctx->fh.m2m_ctx, f->type);
> > -     if (vb2_is_busy(vq))
> > -             return -EBUSY;
> > +     ret = vidioc_try_fmt_out_mplane(file, priv, f);
> > +     if (ret)
> > +             return ret;
> >
> >       if (!hantro_is_encoder_ctx(ctx)) {
> >               struct vb2_queue *peer_vq;
> >
> > +             /*
> > +              * In other to support dynamic resolution change,
>
>                       ^ order
>
> > +              * the decoder admits a resolution change, as long
> > +              * as the pixelformat remains. Can't be done if streaming.
> > +              */
> > +             if (vb2_is_streaming(vq) || (vb2_is_busy(vq) &&
> > +                 pix_mp->pixelformat != ctx->src_fmt.pixelformat))
> > +                     return -EBUSY;
>
> Sorry to chime in only now, but I'm currently looking at the VP9 spec
> and it seems the resolution is allowed to change dynamically [1] (I
> guess the same applies to VP8). IIU the spec correctly, coded frames
> using the new resolution can reference decoded frames using the old
> one (can be higher or lower res BTW). If we force a streamoff to change
> the resolution (as seems to be the case here), we'll lose those ref
> frames (see the hantro_return_bufs() in stop streaming), right?
> Hans, Tomasz, any idea how this dynamic resolution change could/should
> be supported?

The same problem applies to stateful decoders as well. This is an
inherent limitation of the current V4L2 API. To handle this kind of
streams we would have to make the format a per-buffer parameter,
rather than per-queue as it is defined currently.

Best regards,
Tomasz

>
> >               /*
> >                * Since format change on the OUTPUT queue will reset
> >                * the CAPTURE queue, we can't allow doing so
> > @@ -389,12 +396,15 @@ vidioc_s_fmt_out_mplane(struct file *file, void *priv, struct v4l2_format *f)
> >                                         V4L2_BUF_TYPE_VIDEO_CAPTURE_MPLANE);
> >               if (vb2_is_busy(peer_vq))
> >                       return -EBUSY;
> > +     } else {
> > +             /*
> > +              * The encoder doesn't admit a format change if
> > +              * there are OUTPUT buffers allocated.
> > +              */
> > +             if (vb2_is_busy(vq))
> > +                     return -EBUSY;
> >       }
> >
> > -     ret = vidioc_try_fmt_out_mplane(file, priv, f);
> > -     if (ret)
> > -             return ret;
> > -
> >       formats = hantro_get_formats(ctx, &num_fmts);
> >       ctx->vpu_src_fmt = hantro_find_format(formats, num_fmts,
> >                                             pix_mp->pixelformat);
>
> [1] Section "5.16 Reference frame scaling" of
>     https://storage.googleapis.com/downloads.webmproject.org/docs/vp9/vp9-bitstream-specification-v0.6-20160331-draft.pdf

Powered by blists - more mailing lists