From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: with ECARTIS (v1.0.0; list xfs); Thu, 14 Feb 2008 15:45:46 -0800 (PST) Received: from larry.melbourne.sgi.com (larry.melbourne.sgi.com [134.14.52.130]) by oss.sgi.com (8.12.11.20060308/8.12.11/SuSE Linux 0.7) with SMTP id m1ENjaA4002817 for ; Thu, 14 Feb 2008 15:45:40 -0800 Date: Fri, 15 Feb 2008 10:45:59 +1100 From: David Chinner Subject: Re: [patch] Prevent AIL lock contention during transaction completion Message-ID: <20080214234559.GO155259@sgi.com> References: <20080121052330.GG155259@sgi.com> <4796E8C8.3030702@sgi.com> <20080123073446.GU155259@sgi.com> <479986F5.7070800@sgi.com> <20080125074235.GI155407@sgi.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20080125074235.GI155407@sgi.com> Sender: xfs-bounce@oss.sgi.com Errors-to: xfs-bounce@oss.sgi.com List-Id: xfs To: David Chinner Cc: Timothy Shimmin , xfs-dev , xfs-oss On Fri, Jan 25, 2008 at 06:42:35PM +1100, David Chinner wrote: > On Fri, Jan 25, 2008 at 05:51:33PM +1100, Timothy Shimmin wrote: > > So do we really need to call xlog_assign_tail_lsn() then? > > Or are we just being conservative in case we missed something? > > Conservative - the last thing I want is to introduce a subtle > difference to the tail lsn in the log record because we didn't > update it immediately before writing it to disk. I think we are > probably safe removing it, but lets leave that until we got some > wider test coverage on this change first.... Tim - did you finish the review of this? Testing on the 2048p machine appears to have been successful, so I'm just waiting on review ACKs now.... Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group