From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Vasquez Subject: Re: [PATCH] SCSI/qla2xxx: handle sysfs errors Date: Thu, 12 Oct 2006 10:31:38 -0700 Message-ID: <20061012173138.GC4296@andrew-vasquezs-computer.local> References: <20061012014538.GA12894@havoc.gtf.org> <20061012165734.GG3638@andrew-vasquezs-computer.local> <452E7966.7030206@garzik.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from pat.qlogic.com ([198.70.193.2]:53124 "EHLO avexch1.qlogic.com") by vger.kernel.org with ESMTP id S1750806AbWJLRbm (ORCPT ); Thu, 12 Oct 2006 13:31:42 -0400 Content-Disposition: inline In-Reply-To: <452E7966.7030206@garzik.org> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Jeff Garzik Cc: linux-scsi@vger.kernel.org, Andrew Morton , LKML On Thu, 12 Oct 2006, Jeff Garzik wrote: > Andrew Vasquez wrote: > >NACK, please don't do this. SYSFS entries, albiet important, aren't > >necessarilly critical to a functioning driver. I'd rather the driver > >not error out. > > As discussed before, the only errors thrown are either ENOMEM or EFAULT, > both of which are quite serious. Absolutely. But, given the relatively late-stage initialization of these attributes (it's the last thing that gets done before probe() completes)-- the driver has already allocated (successfully) memory, intialized hardware and (possibly) presented storage, the complete unwinding, seems over-kill. > >Here's what I had stewing to address the must_check directives and > >qla2xxx: > > If you're gonna change it that much, might as well use attribute groups. sysfs_create_group() and friends don't appear to support binary attributes.