From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bart Van Assche Subject: Re: [PATCH] make dev_loss_tmo checks uniform across drivers Date: Tue, 04 Jun 2013 08:54:02 +0200 Message-ID: <51AD8F0A.9050007@acm.org> References: <1370283919-11884-1-git-send-email-gurinder.shergill@hp.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from juliette.telenet-ops.be ([195.130.137.74]:33011 "EHLO juliette.telenet-ops.be" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751359Ab3FDGyH (ORCPT ); Tue, 4 Jun 2013 02:54:07 -0400 In-Reply-To: <1370283919-11884-1-git-send-email-gurinder.shergill@hp.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: gurinder.shergill@hp.com Cc: linux-scsi@vger.kernel.org On 06/03/13 20:25, gurinder.shergill@hp.com wrote: > The dev_loss_tmo parameter controls how long the FC transport layer > insulates a loss of a remote port. It is defined to be between 1 and > SCSI_DEVICE_BLOCK_MAX_TIMEOUT (600). Different drivers do slightly different > checks when passed this parameter. This patch makes the dev_loss_tmo setting > uniform across all the drivers by pulling the common logic into the > transport layer function which is responsible for invoking the driver > provided callouts. > > In addition, this patch also fixes a issue (see below), where dev_loss_tmo > could end up with a value more than SCSI_DEVICE_BLOCK_MAX_TIMEOUT when > fast_io_fail_tmo is enabled and then disabled. The change is to never > allow dev_loss_tmo to be out of its defined range. And, once that is a > given, then fast_io_fail_tmo can be capped by dev_loss_tmo. Hello Sunny, As far as I know some users set fast_io_fail_tmo such that dev_loss_tmo can be set to a very large value in order to prevent that the SCSI host number changes. Sorry but I'm not sure it's a good idea to remove that functionality. Bart.