From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sudip Mukherjee Subject: Re: [PATCH 1/4] parport: modify parport subsystem to use devicemodel Date: Wed, 15 Apr 2015 15:58:04 +0530 Message-ID: <20150415102804.GA3803@sudip-PC> References: <1429084124-2271-1-git-send-email-sudipm.mukherjee@gmail.com> <1429084124-2271-2-git-send-email-sudipm.mukherjee@gmail.com> <20150415082746.GI10964@mwanda> <20150415092055.GA3198@sudip-PC> <20150415094500.GZ16501@mwanda> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20150415094500.GZ16501@mwanda> Sender: linux-doc-owner@vger.kernel.org To: Dan Carpenter Cc: Jonathan Corbet , Jean Delvare , Wolfram Sang , Willy Tarreau , Greg Kroah-Hartman , One Thousand Gnomes , linux-i2c@vger.kernel.org, devel@driverdev.osuosl.org, linux-kernel@vger.kernel.org, linux-doc@vger.kernel.org List-Id: linux-i2c@vger.kernel.org On Wed, Apr 15, 2015 at 12:45:00PM +0300, Dan Carpenter wrote: > On Wed, Apr 15, 2015 at 02:50:55PM +0530, Sudip Mukherjee wrote: > > > > > > + tmp->name = name; > > > > > > I wonder who frees this name variable. My concern is that it gets > > > freed before we are done using it or something. (I have not looked at > > > the details). > > it will be done in free_port() the release callback of parport device. > > That doesn't work. Some of the callers pass a string literal. oops, when i replied to your first mail about this I looked at parport_register_port(), but the one that you have mentioned is the one I gave in parport_register_dev() and it seems that the name is not freed anywhere except the driver that has called parport_register_device() or parport_register_dev(). I think I should better copy it and then give it to tmp->name and free it also in the unregister function. And this should be fixed in the original parport_register_device() also. I will wait for some more reviews send the v2 series tomorrow. regards sudip > > regards, > dan carpenter >