From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: Requested changes for the SCSI error handler Date: 08 Jun 2004 11:26:30 -0500 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <1086711994.2105.18.camel@mulgrave> References: <1086122775.2061.87.camel@mulgrave> <20040604232344.GA3214@us.ibm.com> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from stat1.steeleye.com ([65.114.3.130]:6045 "EHLO hancock.sc.steeleye.com") by vger.kernel.org with ESMTP id S265245AbUFHQ0x (ORCPT ); Tue, 8 Jun 2004 12:26:53 -0400 In-Reply-To: <20040604232344.GA3214@us.ibm.com> List-Id: linux-scsi@vger.kernel.org To: Mike Anderson Cc: Alan Stern , SCSI development list On Fri, 2004-06-04 at 18:23, Mike Anderson wrote: > Looks ok. I know your patch is in line with James requests, but I had a > counter idea that allows for setting the settle delays up, down, or to > zero post scsi_host_alloc. If the LLDD does nothing behavior should not > change. This would allow increasing settle if some LLDD wanted it and > also set it through sysfs if we wanted that in the future. You can > throw this patch in the bit bucket if you want to go the previously > stated direction. Well, how about a compromise? I like the ability to alter the host settle time; however, I think turning these over to the driver to sort out how it pleases gives the LLD more control in this situation, certainly for the HOST reset. I'm less sure that a bus reset timeout belongs in the LLD, but I suppose that could be pulled into the SPI transport class eventually. So, what about Alan's approach using a single flag to give the LLD control, but give it control for both the host and bus resets? James