[PATCH 1/2] fsl/qe: NULL dereference on error in ucc_of_parse_tdm()

[Dan Carpenter]

On Mon, Jul 24, 2017 at 09:39:32AM +0000, Qiang Zhao wrote:
> On Mon 7/24/2017 3:04 PM, Dan Carpenter <dan.carpenter at oracle.com>
> 
> > -----Original Message-----
> > From: Dan Carpenter [mailto:dan.carpenter at oracle.com]
> > Sent: Monday, July 24, 2017 3:04 PM
> > To: Qiang Zhao <qiang.zhao at nxp.com>
> > Cc: Leo Li <leoyang.li at nxp.com>; linuxppc-dev at lists.ozlabs.org; kernel-
> > janitors at vger.kernel.org
> > Subject: Re: [PATCH 1/2] fsl/qe: NULL dereference on error in
> > ucc_of_parse_tdm()
> > 
> > On Mon, Jul 24, 2017 at 02:24:14AM +0000, Qiang Zhao wrote:
> > > On Sat 7/22/2017 3:34 PM, Dan Carpenter <dan.carpenter at oracle.com>
> > wrote:
> > >
> > > > -----Original Message-----
> > > > From: Dan Carpenter [mailto:dan.carpenter at oracle.com]
> > > > Sent: Saturday, July 22, 2017 3:34 PM
> > > > To: Qiang Zhao <qiang.zhao at nxp.com>
> > > > Cc: Leo Li <leoyang.li at nxp.com>; linuxppc-dev at lists.ozlabs.org;
> > > > kernel- janitors at vger.kernel.org
> > > > Subject: [PATCH 1/2] fsl/qe: NULL dereference on error in
> > > > ucc_of_parse_tdm()
> > > >
> > > > If "pdev = of_find_device_by_node(np2);" fails then it would lead to
> > > > a NULL dereference.  This function is called from probe() and we're
> > > > using managed resources so we can just return without doing a manual
> > cleanup.
> > >
> > > You mean it will be cleaned up automatically?
> > 
> > Yes.  At module unload.
> 
> Do you mean when insmod it as a module, and when this module is removed, it will be cleaned up automatically?
> Do I understand correctly?
> Well, how about build-in?
> 

Sorry, I mispoke.  It's not really at module unload time.

In this case it's removed automatically when probe() fails, but normally
devm_ resources get released after the ->remove().  The
devres_release_all() function is what releases these, so do a search
for that function to see the details.

regards,
dan carpenter