From mboxrd@z Thu Jan  1 00:00:00 1970
From: Christoph Hellwig <hch@infradead.org>
Subject: Re: iget4/read_inode2 race in get_new_inode
Date: Mon, 29 Apr 2002 21:13:37 +0100
Sender: linux-fsdevel-owner@vger.kernel.org
Message-ID: <20020429211337.A32523@infradead.org>
References: <5.1.0.14.2.20020429201020.04a2eec0@pop.cus.cam.ac.uk> <20020429164844.GA4846@ravel.coda.cs.cmu.edu> <Pine.GSO.4.21.0204291344070.2918-100000@weyl.math.psu.edu> <5.1.0.14.2.20020429201020.04a2eec0@pop.cus.cam.ac.uk> <20020429205113.A31052@infradead.org> <5.1.0.14.2.20020429205850.04929c00@pop.cus.cam.ac.uk>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Alexander Viro <viro@math.psu.edu>,
	Jan Harkes <jaharkes@cs.cmu.edu>, linux-fsdevel@vger.kernel.org
Return-path: <linux-fsdevel-owner@vger.kernel.org>
To: Anton Altaparmakov <aia21@cantab.net>
Content-Disposition: inline
In-Reply-To: <5.1.0.14.2.20020429205850.04929c00@pop.cus.cam.ac.uk>; from aia21@cantab.net on Mon, Apr 29, 2002 at 09:07:30PM +0100
List-Id: linux-fsdevel.vger.kernel.org

On Mon, Apr 29, 2002 at 09:07:30PM +0100, Anton Altaparmakov wrote:
> >Don't do this - rather use the XFS-style icreate() plus per-filesystem
> >exclusion.  The VFS shouldn't have to worry about this kind of problems,
> >especially with such an ugly API.
> 
> Well, I find the icreate ugly... The name is extremely misleading IMO.

Find a better name.. (and no, please not 'iget_without_read_inode').

> Why do you want to bring a massive per-sb-global lock onto each fs?!? That 
> is ridiculous, IMHO. And will certainly cause scalability problems on SMP.

How do you think it can scale worse than the global inode lock?