linux-kernel - Re: camera subdevice support was Re: [PATCH 1/4] v4l2: device_register_subdev

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: <20170304190505.GA31766@amd>
Date:   Sat, 4 Mar 2017 20:05:05 +0100
From:   Pavel Machek <pavel@....cz>
To:     Sakari Ailus <sakari.ailus@....fi>
Cc:     sre@...nel.org, pali.rohar@...il.com, linux-media@...r.kernel.org,
        linux-kernel@...r.kernel.org, laurent.pinchart@...asonboard.com,
        mchehab@...nel.org, ivo.g.dimitrov.75@...il.com
Subject: Re: camera subdevice support was Re: [PATCH 1/4] v4l2:
 device_register_subdev_nodes: allow calling multiple times

On Sat 2017-03-04 14:30:11, Sakari Ailus wrote:
> On Sat, Mar 04, 2017 at 09:55:51AM +0100, Pavel Machek wrote:
> > Dobry den! :-)
> 
> Huomenta! :-)

Dobry vecer! :-).

> > > Good point. Still there may be other ways to move the lens than the voice
> > > coil (which sure is cheap), so how about "flash" and "lens-focus"?
> > 
> > Ok, so something like this? (Yes, needs binding documentation and you
> > wanted it in the core.. can fix.)
> > 
> > BTW, fwnode_handle_put() seems to be missing in the success path of
> > isp_fwnodes_parse() -- can you check that?
> 
> Where exactly? I noticed that if notifier->num_subdevs hits the limit the
> last node isn't put properly. I'll fix that. Is that what you meant?

I guess I'm confused. I see no put on the success path. Maybe it is
put somewhere out of the function where I did not look... is the
reference held while the driver is running

> > --- a/drivers/media/platform/omap3isp/isp.c
> > +++ b/drivers/media/platform/omap3isp/isp.c
> > @@ -2114,7 +2114,7 @@ static int isp_fwnode_parse(struct device *dev, struct fwnode_handle *fwn,
> >  			buscfg->bus.ccp2.lanecfg.data[0].pol =
> >  				vfwn.bus.mipi_csi1.lane_polarity[1];
> >  
> > -			dev_dbg(dev, "data lane %u polarity %u, pos %u\n", i,
> > +			dev_dbg(dev, "data lane %u polarity %u, pos %u\n", 0,
> 
> Why?

I was printing uninitialized / unused variable, which is a no-no (and
gcc complains). I guess I should do a separate patch.

> >  				buscfg->bus.ccp2.lanecfg.data[0].pol,
> >  				buscfg->bus.ccp2.lanecfg.data[0].pos);
> >  
> > @@ -2162,10 +2162,64 @@ static int isp_fwnode_parse(struct device *dev, struct fwnode_handle *fwn,
> >  	return 0;
> >  }
> >  
> > +static int camera_subdev_parse(struct device *dev, struct v4l2_async_notifier *notifier,
> > +			       const char *key)
> > +{
> > +	struct device_node *node;
> > +	struct isp_async_subdev *isd;
> > +
> > +	printk("Looking for %s\n", key);
> > +	
> > +	node = of_parse_phandle(dev->of_node, key, 0);
> 
> There may be more than one flash associated with a sensor. Speaking of which
> --- how is it associated to the sensors?

Ok, more then one flash I can understand (will fix). 

> One way to do this could be to simply move the flash property to the sensor
> OF node. We could have it here, too, if the flash was not associated with
> any sensor, but I doubt that will ever be needed.
> 
> This really calls fork moving this part to the framework away from
> drivers.

The rest I don't get :-(. The flash is likely connected over i2c, so
it should not become child node of omap3isp.

And yes, I agree we want to move it into the framework. Lets agree on
how it should work and where to put it, I'll debug it here then move it... 

> > +	if (!node)
> > +		return 0;
> > +
> > +	printk("Having subdevice: %p\n", node);
> > +		
> > +	isd = devm_kzalloc(dev, sizeof(*isd), GFP_KERNEL);
> > +	if (!isd)
> > +		return -ENOMEM;
> > +
> > +	notifier->subdevs[notifier->num_subdevs] = &isd->asd;
> > +
> > +	isd->asd.match.of.node = node;
> > +	if (!isd->asd.match.of.node) {
> 
> You should check node here first.

Umm. I did, above. This can't happen, AFAICT.

> > +		dev_warn(dev, "bad remote port parent\n");
> > +		return -EIO;
> > +	}
> > +
> 
> And then assign it here.
> 
> isd->asd.match.fwnode.fwn = of_fwnode_handle(node);
> 
> > +	isd->asd.match_type = V4L2_ASYNC_MATCH_OF;
> 
> V4L2_ASYNC_MATCH_FWNODE, please.

Ok. Lets see if it still works after the changes :-)... it does, good.

> > +static int camera_subdevs_parse(struct device *dev, struct v4l2_async_notifier *notifier,
> > +				int max)
> > +{
> > +	int res = 0;
> 
> No need to assign res here.

Ok.

Thanks and best regards,
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Download attachment "signature.asc" of type "application/pgp-signature" (182 bytes)