linux-kernel - Re: [PATCH 5/8] media: cedrus: Detect first slice of a frame

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 30 Aug 2019 13:19:42 -0400
From:   Nicolas Dufresne <nicolas@...fresne.ca>
To:     Boris Brezillon <boris.brezillon@...labora.com>,
        Jernej Škrabec <jernej.skrabec@...l.net>
Cc:     mchehab@...nel.org, hverkuil-cisco@...all.nl,
        paul.kocialkowski@...tlin.com, mripard@...nel.org,
        pawel@...iak.com, m.szyprowski@...sung.com,
        kyungmin.park@...sung.com, tfiga@...omium.org, wens@...e.org,
        acourbot@...omium.org, gregkh@...uxfoundation.org,
        linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
        devel@...verdev.osuosl.org, linux-arm-kernel@...ts.infradead.org,
        ezequiel@...labora.com, jonas@...boo.se
Subject: Re: [PATCH 5/8] media: cedrus: Detect first slice of a frame

Le vendredi 30 août 2019 à 07:48 +0200, Boris Brezillon a écrit :
> On Thu, 29 Aug 2019 21:04:28 +0200
> Jernej Škrabec <jernej.skrabec@...l.net> wrote:
> 
> > Dne ponedeljek, 26. avgust 2019 ob 20:28:31 CEST je Boris Brezillon 
> > napisal(a):
> > > Hi Jernej,
> > > 
> > > On Thu, 22 Aug 2019 21:44:57 +0200
> > > 
> > > Jernej Skrabec <jernej.skrabec@...l.net> wrote:  
> > > > When codec supports multiple slices in one frame, VPU has to know when
> > > > first slice of each frame is being processed, presumably to correctly
> > > > clear/set data in auxiliary buffers.
> > > > 
> > > > Add first_slice field to cedrus_run structure and set it according to
> > > > timestamps of capture and output buffers. If timestamps are different,
> > > > it's first slice and viceversa.
> > > > 
> > > > Signed-off-by: Jernej Skrabec <jernej.skrabec@...l.net>
> > > > ---
> > > > 
> > > >  drivers/staging/media/sunxi/cedrus/cedrus.h     | 1 +
> > > >  drivers/staging/media/sunxi/cedrus/cedrus_dec.c | 2 ++
> > > >  2 files changed, 3 insertions(+)
> > > > 
> > > > diff --git a/drivers/staging/media/sunxi/cedrus/cedrus.h
> > > > b/drivers/staging/media/sunxi/cedrus/cedrus.h index
> > > > 2f017a651848..32cb38e541c6 100644
> > > > --- a/drivers/staging/media/sunxi/cedrus/cedrus.h
> > > > +++ b/drivers/staging/media/sunxi/cedrus/cedrus.h
> > > > @@ -70,6 +70,7 @@ struct cedrus_mpeg2_run {
> > > > 
> > > >  struct cedrus_run {
> > > >  
> > > >  	struct vb2_v4l2_buffer	*src;
> > > >  	struct vb2_v4l2_buffer	*dst;
> > > > 
> > > > +	bool first_slice;
> > > > 
> > > >  	union {
> > > >  	
> > > >  		struct cedrus_h264_run	h264;
> > > > 
> > > > diff --git a/drivers/staging/media/sunxi/cedrus/cedrus_dec.c
> > > > b/drivers/staging/media/sunxi/cedrus/cedrus_dec.c index
> > > > 56ca4c9ad01c..d7b54accfe83 100644
> > > > --- a/drivers/staging/media/sunxi/cedrus/cedrus_dec.c
> > > > +++ b/drivers/staging/media/sunxi/cedrus/cedrus_dec.c
> > > > @@ -31,6 +31,8 @@ void cedrus_device_run(void *priv)
> > > > 
> > > >  	run.src = v4l2_m2m_next_src_buf(ctx->fh.m2m_ctx);
> > > >  	run.dst = v4l2_m2m_next_dst_buf(ctx->fh.m2m_ctx);
> > > > 
> > > > +	run.first_slice =
> > > > +		run.src->vb2_buf.timestamp != run.dst-  
> > > vb2_buf.timestamp;
> > > 
> > > Can't we use slice->first_mb_in_slice to determine if a slice is the
> > > first? I'd expect ->first_mb_in_slice to be 0 (unless we decide to
> > > support ASO).  
> > 
> > I looked in all VPU documentation available to me (which isn't much) and there 
> > is no indication if ASO is supported or not. Do you have any sample video with 
> > out-of-order slices? It's my understanding that this is uncommon.
> 
> I'm not entirely sure, but my understanding was that it might be used
> when streaming over network where some packets might be lost and
> re-emitted later on.
> 
> > If it's 
> > supported, I would leave code as-is.
> 
> I remember seeing the ASO acronym mentioned in the hantro G1 spec, but
> at the same time we're doing frame-based decoding, so I guess the HW
> block expects slices to be ordered in that case. Honestly I don't know,
> so let's keep the code as-is.

We had an ASO interrupt when we tried to do slice decoding rather then
frame. I believe on Hantro, the way to do ASO is to actually re-order
in software.

ASO is a feature of baseline profile use to reduce latency. As an
example, VA-API does not support baseline profile (only constrained-
baseline, which excludes ASO).

Nicolas


Download attachment "signature.asc" of type "application/pgp-signature" (196 bytes)