linux-kernel - Re: [PATCH v15 18/19] media: uvcvideo: implement UVC v1.5 ROI

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: <CANiDSCvGeCHaw2xfZg__UU=fVCWADSQiMw8od5ofBesWjF-7tA@mail.gmail.com>
Date: Fri, 15 Nov 2024 09:22:12 +0100
From: Ricardo Ribalda <ribalda@...omium.org>
To: Gergo Koteles <soyer@....hu>
Cc: Laurent Pinchart <laurent.pinchart@...asonboard.com>, 
	Mauro Carvalho Chehab <mchehab@...nel.org>, Hans de Goede <hdegoede@...hat.com>, 
	Ricardo Ribalda <ribalda@...nel.org>, Sakari Ailus <sakari.ailus@...ux.intel.com>, 
	Hans Verkuil <hverkuil@...all.nl>, Yunke Cao <yunkec@...omium.org>, linux-media@...r.kernel.org, 
	linux-kernel@...r.kernel.org, Yunke Cao <yunkec@...gle.com>
Subject: Re: [PATCH v15 18/19] media: uvcvideo: implement UVC v1.5 ROI

On Fri, 15 Nov 2024 at 01:04, Gergo Koteles <soyer@....hu> wrote:
>
> Hi Ricardo,
>
> On Thu, 2024-11-14 at 21:28 +0100, Ricardo Ribalda wrote:
> > Hi
> >
> > On Thu, 14 Nov 2024 at 21:16, Gergo Koteles <soyer@....hu> wrote:
> > >
> > > Hi Ricardo,
> > >
> > > On Thu, 2024-11-14 at 21:03 +0100, Ricardo Ribalda wrote:
> > > > Hi Gergo
> > > >
> > > > Sorry, I forgot to reply to your comment in v14.
> > > >
> > > > On Thu, 14 Nov 2024 at 20:53, Gergo Koteles <soyer@....hu> wrote:
> > > > >
> > > > > Hi Ricardo,
> > > > >
> > > > > On Thu, 2024-11-14 at 19:10 +0000, Ricardo Ribalda wrote:
> > > > > >
> > > > > > +     },
> > > > > > +     {
> > > > > > +             .id             = V4L2_CID_UVC_REGION_OF_INTEREST_AUTO,
> > > > > > +             .entity         = UVC_GUID_UVC_CAMERA,
> > > > > > +             .selector       = UVC_CT_REGION_OF_INTEREST_CONTROL,
> > > > > > +             .size           = 16,
> > > > > > +             .offset         = 64,
> > > > > > +             .v4l2_type      = V4L2_CTRL_TYPE_BITMASK,
> > > > > > +             .data_type      = UVC_CTRL_DATA_TYPE_BITMASK,
> > > > > > +             .name           = "Region Of Interest Auto Controls",
> > > > > > +     },
> > > > > >  };
> > > > > >
> > > > >
> > > > > Wouldn't be better to use 8 V4L2_CTRL_TYPE_BOOLEAN controls for this?
> > > >
> > > > If I create 8 Booleans, they will always be shown in the device. And
> > > > the user will not have a way to know which values are available and
> > > > which are not.
> > > >
> > > > We will also fail the v4l2-compliance test, because there will be up
> > > > to 7 boolean controls that will not be able to be set to 1, eventhough
> > > > they are writable.
> > > >
> > >
> > > And can't it be that only those returned by GET_MAX be added?
> > >
> > > ```
> > > The bmAutoControls bitmask determines which, if any, on board features
> > > should track to the region of interest. To detect if a device supports
> > > a particular Auto Control, use GET_MAX which returns a mask indicating
> > > all supported Auto Controls.
> > > ```
> > >
> > > Sorry for the misunderstanding, I just don't quite understand.
> >
> > I guess we could, but we would have to make a big change in the way
> > the controls are probed today. uvc does not use the control framework.
> >
> > What will be the benefit of using 8 controls?
> > - Applications still have to know what those controls do, they should
> > not rely on the control name.
>
> Applications like v4l2-ctl are not aware of every controls, work by
> control type, and let the user decide what to do, based on the name.
>
> To avoid having to know each bitmask type control, they need to be able
> to query which bit means what and what to display to the user.
>
> Could VIDIOC_QUERYMENU be supplemented with this?

I believe that violates compliance. VIDIOC_QUERYMENU should only be
used on menus.
https://git.linuxtv.org/v4l-utils.git/tree/utils/v4l2-compliance/v4l2-test-controls.cpp?h=v4l-utils-1.28.1#n143

>
>
> > - Changing from lets say AUTO_EXPOSURE to AUTO_FOCUS, will require to
> > send at least 2 controls via v4l2_s_ext_control... I think it is more
> > practical and less prone to errrors to send just one control
> >
>
> Yes, that could be a good reason.
>
> Thanks,
> Gergo
>
>


-- 
Ricardo Ribalda