From mboxrd@z Thu Jan  1 00:00:00 1970
From: Al Viro <viro@ZenIV.linux.org.uk>
Subject: Re: [patch 00/13] vfs: add helpers to check r/o bind mounts
Date: Thu, 24 Apr 2008 16:12:35 +0100
Message-ID: <20080424151235.GJ15214@ZenIV.linux.org.uk>
References: <20080424113950.818688067@szeredi.hu> <20080424124245.GC15214@ZenIV.linux.org.uk> <E1Jp18b-0003iB-FK@pomaz-ex.szeredi.hu> <20080424134826.GD15214@ZenIV.linux.org.uk> <E1Jp28U-0003r3-8r@pomaz-ex.szeredi.hu> <20080424142857.GF15214@ZenIV.linux.org.uk> <E1Jp2Yw-0003xF-Fk@pomaz-ex.szeredi.hu> <20080424144448.GH15214@ZenIV.linux.org.uk> <E1Jp2pD-00043W-IF@pomaz-ex.szeredi.hu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: akpm@linux-foundation.org, torvalds@linux-foundation.org,
	dave@linux.vnet.ibm.com, ezk@cs.sunysb.edu, mhalcrow@us.ibm.com,
	linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org
To: Miklos Szeredi <miklos@szeredi.hu>
Return-path: <linux-fsdevel-owner@vger.kernel.org>
Received: from zeniv.linux.org.uk ([195.92.253.2]:49401 "EHLO
	ZenIV.linux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751623AbYDXPM6 (ORCPT
	<rfc822;linux-fsdevel@vger.kernel.org>);
	Thu, 24 Apr 2008 11:12:58 -0400
Content-Disposition: inline
In-Reply-To: <E1Jp2pD-00043W-IF@pomaz-ex.szeredi.hu>
Sender: linux-fsdevel-owner@vger.kernel.org
List-ID: <linux-fsdevel.vger.kernel.org>

On Thu, Apr 24, 2008 at 04:53:27PM +0200, Miklos Szeredi wrote:
> > > > RTFS.  permission() doesn't do "is that vfsmount read-only" checks, exactly
> > > > because it's 100% bogus - either you cover it with entire area where we
> > > > are guaranteed to stay r/w, or it's by definition racy.
> > > 
> > > I know that.
> > > 
> > > That does not mean, that fh_verify() needs to do vfsmount r/o checks.
> > > AFAICS it's perfectly OK to do that later, around the vfs_ call.
> > 
> > ... and around everything else that happens to be done after fh_verify
> > for write access, surely?
>
> What in particular?  You have managed to avoid answering this question
> for the last...I don't know how many emails.

Oh, for fuck sake...  grep and ye shall see.  Right next to setattr we
have nfsd4_set_nfs4_acl(), with pair of set_nfsv4_acl_one().  I'd rather
have those two covered by a single will/wont range, TYVM.

nfsd_create() will happily do vfs_mkdir() and nfsd_create_setattr().  Ditto.

And while we are at it, losing the check for r/o in fh_verify() will sure
as hell require at least handling it separately on the normal write path.

> >  Note that e.g. nfsd_setattr() does _not_ call
> > vfs_<anything>()...
> 
> Yes it does: notify_change().  It's vfs_setattr() under a pseudonym.

Are you going to move the will/wont in there?  Because there's a bunch of
stuff in fs/open.c that will disagree...