From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: RE: [patch 08/10] scsi/dpt_i2o: replace schedule_timeout() with msleep_interruptible() Date: 21 Oct 2004 08:50:32 -0400 Sender: linux-scsi-owner@vger.kernel.org Message-ID: <1098363038.1776.4.camel@mulgrave> References: <60807403EABEB443939A5A7AA8A7458B40B5B1@otce2k01.adaptec.com> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from stat16.steeleye.com ([209.192.50.48]:34733 "EHLO hancock.sc.steeleye.com") by vger.kernel.org with ESMTP id S269088AbUJUMuu (ORCPT ); Thu, 21 Oct 2004 08:50:50 -0400 In-Reply-To: <60807403EABEB443939A5A7AA8A7458B40B5B1@otce2k01.adaptec.com> List-Id: linux-scsi@vger.kernel.org To: "Salyzyn, Mark" Cc: janitor@sternwelten.at, SCSI Mailing List , nacc@us.ibm.com On Thu, 2004-10-21 at 07:44, Salyzyn, Mark wrote: > The timeout granularity of 1 second should be the first clue that this > is not important. Commands complete and wake up the task under normal > operation; the timeout is to deal with an errant controller. Erm, yes, but what you're doing is clearly incorrect, and the reason we went to msleep variants anyway. Your code forgets to set the current state, so you stand a good chance of returning immediately from the schedule_timeout() without even yielding the processor. James