linux-media.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Tomi Valkeinen <tomi.valkeinen@ideasonboard.com>
To: Hans Verkuil <hverkuil@xs4all.nl>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
	Sakari Ailus <sakari.ailus@linux.intel.com>,
	linux-media@vger.kernel.org
Subject: vb2_queue type question
Date: Mon, 22 Mar 2021 11:18:18 +0200	[thread overview]
Message-ID: <b86d16cc-e3b1-0db3-f544-9f630572126c@ideasonboard.com> (raw)

Hi Hans,

We were discussing this with Laurent and Sakari, I thought I'd ask if 
you have any feedback on this.

struct vb2_queue has 'type' field, so you can only use a queue for 
buffers of a single type. struct video_device has 'queue' field, so you 
can only use a single queue for a video_device instance.

TI's SoCs have a CSI-2 receiver, with a bunch of DMA engines. The HW 
doesn't care if we are currently capturing pixel buffers or metadata 
buffers (I don't have experience with other HW, but I imagine this 
shouldn't be a rare case). However, due to vb2_queue, the driver needs 
to decide which one to support, which limits the possible use cases.

I was browsing the code, and afaics the type field doesn't do much. It 
is, of course, used to reject queuing buffers of wrong type, and also 
(mostly in mem-2-mem code) to find out if functions are called in input 
or output context.

The latter one could be easily removed by just comparing the given queue 
pointer to a stored pointer (e.g. queue == priv->input_queue).

Do you see any problems if we were to change the type field to 
type_mask, allowing multiple buffer types per queue? Or even remove the 
vb2_queue->type. This raises some questions, like should a queue contain 
only buffers of a single type or can it contain a mix of buffers (I 
think it shouldn't contain a mix of buffers), or can a queue's type_mask 
contain both input and output types (I don't see why not).

An alternate which I tried was creating two vb2_queues, and switching 
the video_device->queue at runtime based on set_format. It kind of 
works, but I think the behavior is a bit unclear, and it might be 
difficult to catch all the corner cases.

  Tomi

             reply	other threads:[~2021-03-22  9:18 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-22  9:18 Tomi Valkeinen [this message]
2021-03-22  9:49 ` vb2_queue type question Hans Verkuil
2021-03-22  9:59   ` Laurent Pinchart
2021-03-22 10:08     ` Hans Verkuil
2021-03-22 10:30   ` Tomi Valkeinen
2021-03-26  9:57   ` Tomi Valkeinen
2021-03-26 10:18     ` Hans Verkuil
2021-03-30  8:18       ` Tomi Valkeinen
2021-03-30  8:56         ` Hans Verkuil
2021-03-31  5:22           ` Tomi Valkeinen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=b86d16cc-e3b1-0db3-f544-9f630572126c@ideasonboard.com \
    --to=tomi.valkeinen@ideasonboard.com \
    --cc=hverkuil@xs4all.nl \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-media@vger.kernel.org \
    --cc=sakari.ailus@linux.intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).