Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20210208051749.1785246-1-sergey.senozhatsky@gmail.com>
 <20210208051749.1785246-4-sergey.senozhatsky@gmail.com> <CAPybu_10Uz0Eb2U5ZohNV1t0gf98ZBZOa3KFCzdi1RJ0k3c1yQ@mail.gmail.com>
 <YFFiizDjNBMG3uI+@google.com> <CAPybu_0ruoc-w3402j-vVNs2-xq8=-_XzVKSxiG+iuyB=eNimA@mail.gmail.com>
 <YFLbYjm0VyzaEMkr@google.com>
In-Reply-To: <YFLbYjm0VyzaEMkr@google.com>
From:   Ricardo Ribalda <ribalda@chromium.org>
Date:   Thu, 18 Mar 2021 22:19:15 +0100
Message-ID: <CANiDSCtCaVUxrYeceWXEJ3o61ze8uNyce69xW_KcQTirhWfZaQ@mail.gmail.com>
Subject: Re: [PATCHv2 3/3] media: uvcvideo: add UVC 1.5 ROI control
To:     Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc:     Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>,
        Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
        Tomasz Figa <tfiga@chromium.org>,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        linux-media <linux-media@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Sergey Senozhatsky <senozhatsky@chromium.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

Hi Sergey

On Thu, Mar 18, 2021 at 5:47 AM Sergey Senozhatsky
<sergey.senozhatsky.work@gmail.com> wrote:
>
> On (21/03/17 08:58), Ricardo Ribalda Delgado wrote:
> [..]
> > >
> > > GET_CUR?
> > yep
> >
> > >
> > > > https://www.kernel.org/doc/html/v4.13/media/uapi/v4l/vidioc-g-selection.html?highlight=vidioc_s_selection
> > > > On success the struct v4l2_rect r field contains the adjusted
> > > > rectangle.
> > >
> > > What is the adjusted rectangle here? Does this mean that firmware can
> > > successfully apply SET_CUR and return 0, but in reality it was not happy
> > > with the rectangle dimensions so it modified it behind the scenes?
> >
> > I can imagine that some hw might have spooky requirements for the roi
> > rectangle (multiple of 4, not crossing the bayer filter, odd/even
> > line...) so they might be able to go the closest valid config.
>
> Hmm. Honestly, I'm very unsure about it. ROI::SET_CUR can be a very
> hot path, depending on what user-space considers to be of interest
> and how frequently that object of interest changes its position/shape/etc.
> Doing GET_CUR after every SET_CUR doubles the number of firmware calls
> we issue, that's for sure; is it worth it - that's something that I'm
> not sure of.
>
> May I please ask for more opinions on this?

Could you try setting the roi in a loop in your device and verify that
it accepts all the values with no modification. If so we can implement
the set/get as a quirk for other devices.

>
>         -ss


-- 
Ricardo Ribalda