From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from perceval.ideasonboard.com (perceval.ideasonboard.com [213.167.242.64]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 4224629A9 for ; Wed, 5 Apr 2023 14:30:13 +0000 (UTC) Received: from pendragon.ideasonboard.com (softbank219059148246.bbtec.net [219.59.148.246]) by perceval.ideasonboard.com (Postfix) with ESMTPSA id 80F391718; Wed, 5 Apr 2023 16:30:09 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=ideasonboard.com; s=mail; t=1680705010; bh=Qs3POKuzAMl2FAwrRUyvpLBK6qJSzNKEzukle8jld14=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=MURgjpwOIUrJIP2D7Rtkc9A8Kxbv1mzNmPQvR40pLpJ2dTdxJTFhUDRn1ILC3QF/X DTL/lO/prlXUnU/52QubBohKPtzvTssDjp0yYnhPQHFsnYpBlxGwPzSsynDUY/aaiM RtPdJEXy7tJq0vMbsOjDR/FYTYpotdw9i5EwU06c= Date: Wed, 5 Apr 2023 17:30:18 +0300 From: Laurent Pinchart To: Hans Verkuil Cc: Luca Ceresoli , Thierry Reding , Jonathan Hunter , Sowjanya Komatineni , David Airlie , Daniel Vetter , Rob Herring , Krzysztof Kozlowski , Mauro Carvalho Chehab , Greg Kroah-Hartman , Dmitry Osipenko , linux-media@vger.kernel.org, linux-tegra@vger.kernel.org, dri-devel@lists.freedesktop.org, devicetree@vger.kernel.org, linux-kernel@vger.kernel.org, linux-staging@lists.linux.dev, Thomas Petazzoni , Paul Kocialkowski , Richard Leitner Subject: Re: [RESEND PATCH v4 03/21] staging: media: tegra-video: fix .vidioc_enum_fmt_vid_cap to return all formats Message-ID: <20230405143018.GC9915@pendragon.ideasonboard.com> References: <20230309144320.2937553-1-luca.ceresoli@bootlin.com> <20230309144320.2937553-4-luca.ceresoli@bootlin.com> <85268d69-3d3b-2c0f-ba26-073f09052362@xs4all.nl> <20230404161251.272cc78b@booty> <20230405023048.GD9915@pendragon.ideasonboard.com> <20230405103134.2ae10766@booty> Precedence: bulk X-Mailing-List: linux-staging@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: Hi Hans, On Wed, Apr 05, 2023 at 10:50:37AM +0200, Hans Verkuil wrote: > On 05/04/2023 10:31, Luca Ceresoli wrote: > > On Wed, 5 Apr 2023 05:30:48 +0300 Laurent Pinchart wrote: > >> On Tue, Apr 04, 2023 at 04:12:51PM +0200, Luca Ceresoli wrote: > >>> On Wed, 29 Mar 2023 13:16:22 +0200 Hans Verkuil wrote: > >>> > >>>> Hi Luca, > >>>> > >>>> I finally found the time to test this series. It looks OK, except for this patch. > >>> > >>> Thank you very much for taking the time! > >>> > >>>> The list of supported formats really has to be the intersection of what the tegra > >>>> supports and what the sensor supports. > >>>> > >>>> Otherwise you would advertise pixelformats that cannot be used, and the application > >>>> would have no way of knowing that. > >>> > >>> As far as I understand, I think we should rather make this driver fully > >>> behave as an MC-centric device. It is already using MC quite > >>> successfully after all. > >>> > >>> Do you think this is correct? > >> > >> Given the use cases for this driver, I agree. > > I disagree. > > This driver doesn't use the media controller for anything at the moment. The > /dev/mediaX device just shows the internal topology (i.e. connected sensors), > but otherwise it does nothing. > > While it would be great if we could unlock the ISP on the Tegra, the reality > is that it is entirely closed source and can't be used in a linux driver, and > that's not going to change, sadly. Never say never :-) > That leaves us with just a basic CSI capture driver. Rather than trying to > change this driver to a full MC device with no benefits, just drop this change > and get your code in. Can't the hardware support capturing different virtual channels or data types from the same CSI-2 source ? That would require MC support, the stream API requires subdev device nodes. > Note that this driver will stay in staging since it still fails when I try to > capture from two sensors at the same time: syncpoint errors start appearing > in that case. I think there are locking issues. I think I have someone to take > a look at that, but first I want your series to get merged. > > In the very unlikely event that the ISP can be implemented in a linux driver, > it will probably become a new driver. > > Regards, > > > Ok, thanks for the feedback. I will send a v5 with this change. -- Regards, Laurent Pinchart