linux-kernel - Re: [PATCH] media: i2c: max9286: Pass default bus type when parsing fwnode endpoint

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] [day] [month] [year] [list]

Message-ID: <20191002085622.GN896@valkosipuli.retiisi.org.uk>
Date:   Wed, 2 Oct 2019 11:56:22 +0300
From:   Sakari Ailus <sakari.ailus@....fi>
To:     Jacopo Mondi <jacopo@...ndi.org>
Cc:     Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>,
        jacopo+renesas@...ndi.org, kieran.bingham+renesas@...asonboard.com,
        laurent.pinchart+renesas@...asonboard.com,
        niklas.soderlund+renesas@...natech.se, mchehab@...nel.org,
        linux-media@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] media: i2c: max9286: Pass default bus type when parsing
 fwnode endpoint

Hi Jacopo, Manivannan,

On Wed, Oct 02, 2019 at 09:54:32AM +0200, Jacopo Mondi wrote:
> Hello Manivannan,
> 
> On Tue, Oct 01, 2019 at 08:25:03PM +0530, Manivannan Sadhasivam wrote:
> > The caller of v4l2_fwnode_endpoint_alloc_parse() is expected to pass a
> > valid bus_type parameter for proper working of this API. Hence, pass
> > V4L2_MBUS_CSI2_DPHY as the bus_type parameter as this driver only supports
> > MIPI CSI2 for now. Without this commit, the API fails on 96Boards
> > Dragonboard410c connected to MAX9286 deserializer.
> >
> > Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
> > ---
> >
> > This patch depends on the latest "MAX9286 GMSL Support" series posted
> > by Kieran Bingham.
> >
> >  drivers/media/i2c/max9286.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/media/i2c/max9286.c b/drivers/media/i2c/max9286.c
> > index 9390edf5ad9c..6e1299f15493 100644
> > --- a/drivers/media/i2c/max9286.c
> > +++ b/drivers/media/i2c/max9286.c
> > @@ -976,7 +976,9 @@ static int max9286_parse_dt(struct max9286_device *max9286)
> >
> >  		/* For the source endpoint just parse the bus configuration. */
> >  		if (ep.port == MAX9286_SRC_PAD) {
> > -			struct v4l2_fwnode_endpoint vep;
> > +			struct v4l2_fwnode_endpoint vep = {
> > +				.bus_type = V4L2_MBUS_CSI2_DPHY
> > +			};
> 
> Isn't the bus_type just an hint ? __v4l2_fwnode_endpoint_parse()
> should try to parse CSI2 first, and should be able to to deduct the bus
> type from the properties specified in DT...

It's basically telling the V4L2 fwnode parser to parse that bus only, and
not guess what the caller wanted.

This is what drivers should actually be doing. It also allows setting the
defaults that many drivers declare in their DT bindings but very few
actually implement.

> 
> Anyway, I'm not against this change, the contrary actually, I'm just
> afraid it might hide some more subtile bug, as we don't need this on
> our platforms...

I'd expect it to be safer this way, as there's no guessing involved.

-- 
Regards,

Sakari Ailus