From mboxrd@z Thu Jan 1 00:00:00 1970 From: "J. Bruce Fields" Subject: Re: [RFC] Remove BKL from fs/locks.c Date: Fri, 4 Jan 2008 18:41:56 -0500 Message-ID: <20080104234156.GH14827@fieldses.org> References: <20071230061615.GS11638@parisc-linux.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-fsdevel@vger.kernel.org To: Matthew Wilcox Return-path: Received: from mail.fieldses.org ([66.93.2.214]:36874 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752251AbYADXl6 (ORCPT ); Fri, 4 Jan 2008 18:41:58 -0500 Content-Disposition: inline In-Reply-To: <20071230061615.GS11638@parisc-linux.org> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Sat, Dec 29, 2007 at 11:16:15PM -0700, Matthew Wilcox wrote: > > I've been promising to do this for about seven years now. > > It seems to work well enough, but I haven't run any serious stress > tests on it. This implementation uses one spinlock to protect both lock > lists and all the i_flock chains. It doesn't seem worth splitting up > the locking any further. I think you missed the code in lockd and nfsd4 code that walks the i_flock lists, and you might want to grep for i_flock to make sure that's all. In fact, lockd runs entirely under the bkl, so it may take a careful review to make sure there aren't some other odd places where it depends on that for mutual exclusion with code in locks.c. Yipes. Thanks for working on this. --b.