Re: [PATCH v3 5/6] usb: chipidea: let chipidea core device of_node equal's glue layer device of_node - Russell King

From: Russell King - ARM Linux <linux@armlinux.org.uk>
To: Peter Chen <hzpeterchen@gmail.com>
Cc: Peter Chen <peter.chen@nxp.com>,
	gregkh@linuxfoundation.org, stern@rowland.harvard.edu,
	ulf.hansson@linaro.org, broonie@kernel.org, sre@kernel.org,
	robh+dt@kernel.org, shawnguo@kernel.org, dbaryshkov@gmail.com,
	dwmw2@infradead.org, mark.rutland@arm.com,
	devicetree@vger.kernel.org, k.kozlowski@samsung.com,
	stephen.boyd@linaro.org, oscar@naiandei.net, arnd@arndb.de,
	pawel.moll@arm.com, linux-pm@vger.kernel.org,
	linux-kernel@vger.kernel.org, s.hauer@pengutronix.de,
	linux-usb@vger.kernel.org, mail@maciej.szmigiero.name,
	troy.kisky@boundarydevices.com, stillcompiling@gmail.com,
	Peter Chen <peter.chen@freescale.com>,
	p.zabel@pengutronix.de, festevam@gmail.com,
	linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v3 5/6] usb: chipidea: let chipidea core device of_node equal's glue layer device of_node
Date: Thu, 21 Jul 2016 10:41:28 +0100	[thread overview]
Message-ID: <20160721094128.GN1041@n2100.armlinux.org.uk> (raw)
In-Reply-To: <20160721092012.GA22749@shlinux2>

On Thu, Jul 21, 2016 at 05:20:12PM +0800, Peter Chen wrote:
> On Thu, Jul 21, 2016 at 10:14:38AM +0100, Russell King - ARM Linux wrote:
> > On Wed, Jul 20, 2016 at 05:40:28PM +0800, Peter Chen wrote:
> > > diff --git a/drivers/usb/chipidea/core.c b/drivers/usb/chipidea/core.c
> > > index 69426e6..0d05812 100644
> > > --- a/drivers/usb/chipidea/core.c
> > > +++ b/drivers/usb/chipidea/core.c
> > > @@ -914,6 +914,16 @@ static int ci_hdrc_probe(struct platform_device *pdev)
> > >  	if (!ci)
> > >  		return -ENOMEM;
> > >  
> > > +	/*
> > > +	 * At device tree, we have no device node for chipidea core,
> > > +	 * the glue layer's node is the parent node for host and udc
> > > +	 * device. But in related driver, the parent device is chipidea
> > > +	 * core. So, in order to let the common driver get parent's node,
> > > +	 * we let the core's device node equals glue layer's node.
> > > +	 */
> > > +	if (dev->parent && dev->parent->of_node)
> > > +		dev->of_node = dev->parent->of_node;
> > 
> > This is a dangerous thing to do.  You're changing the dev->of_node of
> > _this_ device, which means that _this_ driver will no longer match
> > the device if you remove and reinsert the driver module, or unbind
> > and try to re-bind the device to this driver.
> > 
> 
> Thanks for commenting it.
> 
> I have tested load/unload, it does not show any problems.
> 
> The chipidea core device is created by code at runtime, not by device node.
> And we have NO device node for this chipidea core device at dts.

Okay, so we still probably have the bind/unbind problem, where "dev"
can be matched by the driver which claimed "dev->parent".  Remember,
in an OF environment, driver matching is done by the compatible
property, which is accessed via dev->of_node.

Therefore, I would suggest that you NULL dev->of_node in the error
cleanup paths and in the remove function, so you don't have an
unbound device with a duplicated (but inappropriate) dev->of_node
pointer.

-- 
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up
according to speedtest.net.