From mboxrd@z Thu Jan 1 00:00:00 1970 From: Boaz Harrosh Subject: Re: [PATCH] tcm: Convert backstore ->set_configfs_dev_params() to use parser.h Date: Tue, 09 Nov 2010 19:58:01 +0200 Message-ID: <4CD98BA9.7010102@panasas.com> References: <1289296122-13152-1-git-send-email-nab@linux-iscsi.org> <20101109100359.GA26324@infradead.org> <1289297001.19946.92.camel@haakon2.linux-iscsi.org> <20101109134522.GA4548@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: Received: from exprod5og102.obsmtp.com ([64.18.0.143]:46964 "HELO exprod5og102.obsmtp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with SMTP id S1754123Ab0KIR6E (ORCPT ); Tue, 9 Nov 2010 12:58:04 -0500 In-Reply-To: <20101109134522.GA4548@infradead.org> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Christoph Hellwig Cc: "Nicholas A. Bellinger" , linux-scsi , Christoph Hellwig On 11/09/2010 03:45 PM, Christoph Hellwig wrote: > On Tue, Nov 09, 2010 at 02:03:21AM -0800, Nicholas A. Bellinger wrote: >> thanks for the pointer here.. >> >> Btw, the kstrdup() is required because the configfs attribute store >> function will be passing a 'const char *page' that is propigated from >> target_core_configfs.c: target_core_store_dev_control() out to >> ->set_configfs_dev_params() backend code. > > strsep doesn't modify the data it points to, just the pointer itself. > We could get around this by casting the pointer, which at least avoids > the memory allocation. But it's probably not worth bothering. > If it is so, then the proper fix is to "const" the parameter to strsep() most certainly don't hide a const_cast like that. This is a landmine waiting to explode. Boaz