linux-kernel - Re: [PATCH] [media] vb2: move dma-buf unmap from __vb2_dqbuf() to vb2_buffer

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]

Message-ID: <57927A64.6000307@osg.samsung.com>
Date:	Fri, 22 Jul 2016 20:56:20 +0100
From:	Luis de Bethencourt <luisbg@....samsung.com>
To:	Javier Martinez Canillas <javier@....samsung.com>,
	linux-kernel@...r.kernel.org
CC:	Sakari Ailus <sakari.ailus@....fi>,
	Marek Szyprowski <m.szyprowski@...sung.com>,
	Mauro Carvalho Chehab <mchehab@...pensource.com>,
	Kyungmin Park <kyungmin.park@...sung.com>,
	Pawel Osciak <pawel@...iak.com>, linux-media@...r.kernel.org,
	Hans Verkuil <hverkuil@...all.nl>,
	Shuah Khan <shuahkh@....samsung.com>
Subject: Re: [PATCH] [media] vb2: move dma-buf unmap from __vb2_dqbuf() to
 vb2_buffer_done()

On 20/07/16 19:22, Javier Martinez Canillas wrote:
> Currently the dma-buf is unmapped when the buffer is dequeued by userspace
> but it's not used anymore after the driver finished processing the buffer.
> 
> So instead of doing the dma-buf unmapping in __vb2_dqbuf(), it can be made
> in vb2_buffer_done() after the driver notified that buf processing is done.
> 
> Decoupling the buffer dequeue from the dma-buf unmapping has also the side
> effect of making possible to add dma-buf fence support in the future since
> the buffer could be dequeued even before the driver has finished using it.
> 
> Signed-off-by: Javier Martinez Canillas <javier@....samsung.com>
> 
> ---
> Hello,
> 
> I've tested this patch doing DMA buffer sharing between a
> vivid input and output device with both v4l2-ctl and gst:
> 
> $ v4l2-ctl -d0 -e1 --stream-dmabuf --stream-out-mmap
> $ v4l2-ctl -d0 -e1 --stream-mmap --stream-out-dmabuf
> $ gst-launch-1.0 v4l2src device=/dev/video0 io-mode=dmabuf ! v4l2sink device=/dev/video1 io-mode=dmabuf-import
> 
> And I didn't find any issues but more testing will be appreciated.
> 
> Best regards,
> Javier
> 

Hello all,

Tested this using the same GStreamer pipeline as Javier mentions above.
It works nicely.

Thanks,
Luis

Tested-by: Luis de Bethencourt <luisbg@....samsung.com>