From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [GIT PATCH] SCSI bug fixes for 2.6.23-rc1 Date: Sun, 29 Jul 2007 21:01:49 -0500 Message-ID: <1185760909.3434.22.camel@localhost.localdomain> References: <1185720592.3434.7.camel@localhost.localdomain> <46AD19F7.7040905@garzik.org> <1185756759.3434.19.camel@localhost.localdomain> <46AD390B.2080400@garzik.org> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Return-path: Received: from hancock.steeleye.com ([71.30.118.248]:56113 "EHLO hancock.sc.steeleye.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1761570AbXG3CBv (ORCPT ); Sun, 29 Jul 2007 22:01:51 -0400 In-Reply-To: <46AD390B.2080400@garzik.org> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Jeff Garzik Cc: Andrew Morton , Linus Torvalds , linux-scsi , linux-kernel On Sun, 2007-07-29 at 21:04 -0400, Jeff Garzik wrote: > James Bottomley wrote: > > msleep_interruptible -> ssleep is a > > change with zero practical impact for this driver, > > eh, how do you figure? > > A signal can clearly cause the abort-related functions to delay far > shorter than the driver wishes. > > The msleep_interruptible() in arcmsr_wait_msgint_ready() probably isn't > fatal -- unless it's there to ensure the hardware isn't pounded -- but > again, a signal can cause a CPU-wasting busy loop that could last for > quite a while (in CPU terms). I didn't say the usage wasn't a bad example or couldn't cause problems in certain cases. I said "zero practical effect". This is because the two possible threads that enter these routines already have all user signals blocked. James