From: Theodore Ts'o Subject: Re: [PATCHv3 0/3] new APIs to allocate buffer-cache with user specific flag Date: Thu, 28 Aug 2014 09:48:04 -0400 Message-ID: <20140828134804.GC21925@thunk.org> References: <53FE9357.6000505@lge.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Alexander Viro , Andrew Morton , "Paul E. McKenney" , Peter Zijlstra , Jan Kara , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, Andreas Dilger , linux-ext4@vger.kernel.org, Minchan Kim , Joonsoo Kim , =?utf-8?B?7J206rG07Zi4?= To: Gioh Kim Return-path: Received: from imap.thunk.org ([74.207.234.97]:53896 "EHLO imap.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750801AbaH1NsZ (ORCPT ); Thu, 28 Aug 2014 09:48:25 -0400 Content-Disposition: inline In-Reply-To: <53FE9357.6000505@lge.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Thu, Aug 28, 2014 at 11:26:31AM +0900, Gioh Kim wrote: > > I have 3 patchs: > > 1. Patch 1/3: introduce a new API that create page cache with allocation flag > 2. Patch 2/3: have ext4 use the new API to read superblock > 3. Patch 3/3: have jbd/jbd2 use the new API to make journaling of superblock > > This patchset is based on linux-next-20140814. Looks good. Unless there are any objections from the mm folks, since the nearly all of the changes are in fs/buffer.c and in ext4/jbd2 code, I plan to carry this in the ext4 tree. I do plan to clean up the patch titles a little; from: fs/buffer.c: allocate buffer cache with user specific flag ext4: allocate buffer-cache for superblock in, non-movable area jbd/jbd2: allocate buffer-cache for superblock inode in non-movable area to: fs.c: support buffer cache allocations with gfp modifiers ext4: use non-movable memory for the ext4 superblock jbd/jbd2: use non-movable memory for the jbd superblock And do some minor english grammar/spelling cleanups in the commit description when I apply the patch. Thanks for this work; I'm going to need to use the interfaces you introduced in fs/buffer.c to guarantee that certain directory reads can be done with GFP_NOFAIL (since under heavy memory pressure, allocation failures there can currently lead to the file system getting declared corrupt. Interestingly, this bug has been around for a long time, and hasn't been noticed in over two cycles of enterprise distro qualifications by either RHEL or SLES, which leads me to wonder if there are other places where the error paths for GFP_NOFS allocations haven't been well tested....) - Ted