From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Luis R. Rodriguez" Subject: Re: [RFC 0/2] vfs / btrfs: add support for ustat() Date: Wed, 16 Jul 2014 23:34:12 +0200 Message-ID: <20140716213412.GP10393@wotan.suse.de> References: <1405465625-1122-1-git-send-email-mcgrof@do-not-panic.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: viro@zeniv.linux.org.uk, clm@fb.com, jbacik@fb.com, linux-fsdevel@vger.kernel.org, linux-btrfs@vger.kernel.org, linux-kernel@vger.kernel.org, jeffm@suse.com, fdmanana@suse.com To: "Luis R. Rodriguez" Return-path: Content-Disposition: inline In-Reply-To: <1405465625-1122-1-git-send-email-mcgrof@do-not-panic.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Tue, Jul 15, 2014 at 04:07:03PM -0700, Luis R. Rodriguez wrote: > From: "Luis R. Rodriguez" > > ustat() does not work for btrfs, the reason is the VFS layer > cannot get to the super block from the anonymous bdevs. The > way these are used within btrfs though are to actually use a > series of them per superblock. This modifies the way the > VFS exposes these bdevs by pegging them onto the super > block on a linked list. This should also allow us to grow > the subdevices on the super block with any other information > that might be better suited underneath so that the VFS > layer is aware. If this approach is reasonable then perhaps > we should next remove the exported symbol for get_anon_bdev() > as btrfs was the only user. > Based on some private review I got this can be simplified leaving the embedded the struct on the fs data structure and just using it to expose it on the VFS. Will send a v2. Luis