From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH 2/2] ->write_super lock_super pushdown Date: Tue, 12 May 2009 15:25:04 +0200 Message-ID: <20090512132504.GA19872@lst.de> References: <20090511213503.GB19326@lst.de> <4A094B02.20606@panasas.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Christoph Hellwig , viro@zeniv.linux.org.uk, linux-fsdevel@vger.kernel.org To: Boaz Harrosh Return-path: Received: from verein.lst.de ([213.95.11.210]:49533 "EHLO verein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750842AbZELNZM (ORCPT ); Tue, 12 May 2009 09:25:12 -0400 Content-Disposition: inline In-Reply-To: <4A094B02.20606@panasas.com> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Tue, May 12, 2009 at 01:10:10PM +0300, Boaz Harrosh wrote: > Please I have a question about this? > > lock_super(): > I do not see any other lock_super() in exofs, so all this might "lock" is > race against itself, right? Yes. > Should I make sure that concurrent > exofs_write_super are protected some other way and remove this? Yes. > > lock_kernel(); > What is that used for? What should I check so this can be removed? lock kernel is the big kernel lock as in the very first lock that was added when Linux grew SMP support. If you filesystem does internal locking you most likely don't need this one at all. The only superblock method still called with the BKL in the vfs tree is remount, but it will most likely be gone before 2.6.31, too. After that you can do a quick audit if there was anything it protected against and remove it.