From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike Christie Subject: Re: Creating a subdirectory under fc_host statistics using fc_function_template Date: Sun, 04 Jan 2009 11:17:41 -0600 Message-ID: <4960EF35.2090502@cs.wisc.edu> References: <494A86DE.7020908@emulex.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from sabe.cs.wisc.edu ([128.105.6.20]:50451 "EHLO sabe.cs.wisc.edu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754429AbZADRRx (ORCPT ); Sun, 4 Jan 2009 12:17:53 -0500 In-Reply-To: <494A86DE.7020908@emulex.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: James Smart Cc: "James.Bottomley@HansenPartnership.com" , Krishna Gudipati , "linux-scsi@vger.kernel.org" James Smart wrote: > James, > > What Krishna describes, is a need to add driver-specific attributes (and > optionally a directory of attributes) to a class-level object. I > attempted patches of a similar form when I was originally doing the > fc_rports, which were rejected. > > I believe the position is that we do not allow driver-specific things on > the class objects, and drivers are limited in scope to only adding > individual attributes to the scsi_host via the attribute list in the > scsi_host_template. > > I'd like to re-propose the patches to add driver-specific data to the FC > class objects. Do you believe there's a change in heart to accepting > this functionality before I start the work to create them ? > > For example, I know that Emulex would have liked to add driver-specific > state data to the rports. And Krishna's proposal is another request. > We are having the same problem for iscsi. For qla4xxx and cxgb3i and bnx2i, we would like to add driver specific attrs to the iscsi session/connection which are like FC's rport struct. I was hoping if at least the classes provide an API to add driver specific attrs similar to how scsi-ml does, then it would be ok?