From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [ANNOUNCE] iscsi-initiator-core 1.6.2.0-rc1 for 2.6.12-rc1 Date: Sat, 23 Apr 2005 21:49:15 -0500 Message-ID: <1114310956.4805.9.camel@mulgrave> References: <1112385701.3008.18.camel@haakon> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from stat16.steeleye.com ([209.192.50.48]:37855 "EHLO hancock.sc.steeleye.com") by vger.kernel.org with ESMTP id S262158AbVDXEZO (ORCPT ); Sun, 24 Apr 2005 00:25:14 -0400 In-Reply-To: <1112385701.3008.18.camel@haakon> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: "Nicholas A. Bellinger" Cc: iscsi-initiator-core-announce@vger.kernel.org, linux-scsi , iscsi-initiator-core-devel , Christoph Hellwig , Mike Christie On Fri, 2005-04-01 at 12:01 -0800, Nicholas A. Bellinger wrote: > A few minor fixes in this release, with the main feature addition being > support for persistent connection/session parameters as defined within > /etc/sysconfig/initiator. For example, it is now possible to define > specific connection/session parameters and have those values remain > throughout connection/session failures. Also, the first run of tests > with dm-multipath have been compelted successful. I am working on a > section within the documentation to describe how these two pieces work > together within linux 2.6. > > Also an important milestone has been reached with v1.0 of the > iscsi-initiator-core.org howto located at: > > http://iscsi-initiator-core.org/mediawiki/index.php/Howto > > This is a work in progress as data is imported from the ASCII version > from iscsi-initiator-core-tools v2.3. > > This release is now considered stable and no new features will be added > until 1.6.2.0-final is released. OK, I assessed this for inclusion. The feedback I had based on my last review was 1) Use the iscsi transport class 2) get rid of MC/S in favour of dm-multipath 3) Don't try to subvert the SCSI error handler Point 3) is much better, but still present in this: + /* + * This is completion of a given struct scsi_cmnd after an + * iSCSI exception occured. Based upon iSCSI exceptions and/or + * passed action parameter, struct scsi_cmnd->eh_timeout may have + * been stopped, and needs to be rstarted before completion to + * the SCSI stack. + */ + if (!(timer_pending(&sc->eh_timeout))) { + sc->eh_timeout.data = (unsigned long) sc; + sc->eh_timeout.expires = (jiffies + sc->timeout_per_command); + sc->eh_timeout.function = scsi_timeout_function; + add_timer(&sc->eh_timeout); However the other two still have not been addressed, so this patch is not suitable for inclusion. James