From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nick Piggin Subject: Re: [PATCH 04/19] fs: Implement lazy LRU updates for inodes. Date: Sun, 17 Oct 2010 13:09:51 +1100 Message-ID: <20101017020951.GC3162@amd> References: <1287216853-17634-1-git-send-email-david@fromorbit.com> <1287216853-17634-5-git-send-email-david@fromorbit.com> <20101016092916.GA32197@amd> <20101016165930.GA20626@infradead.org> <20101016172924.GA3519@amd> <20101017004710.GC1614@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Nick Piggin , Dave Chinner , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org To: Christoph Hellwig Return-path: Content-Disposition: inline In-Reply-To: <20101017004710.GC1614@infradead.org> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Sat, Oct 16, 2010 at 08:47:10PM -0400, Christoph Hellwig wrote: > On Sun, Oct 17, 2010 at 04:29:24AM +1100, Nick Piggin wrote: > > > I don't think the pointer check will work either. By the time we retake > > > the lru lock the inode might already have been reaped through a call > > > to invalidate_inodes. There's no way we can do anything with it after > > > > I don't think you're right. If we re take inode_lock, ensure it is on > > the LRU, and call the can_unuse checks, there is no more problem than > > the regular loop taking items from the LRU, AFAIKS. > > As long as we have the global inode lock it should indeed be safe. > But once we have a separate lru lock (global or per-zone, with or > without i_lock during the addition) there is nothing preventing the > inode from getting reused and re-added to the lru in the meantime. > Sure this is an extremly unlikely case, but there is no locking to > prevent it once inode_lock is gone. No. There is nothing preventing that exact reuse from happening in mainline _today_ either, because the inode_lock is dropped there too. The point is that it is a heuristic that works (apparently) most of the time but if it gets it wrong then it's not a big deal, it's only the LRU position anyway. It would work exactly the same with separate global or per-zone lru locks.