From: Petr Baudis Subject: Re: ROOT_DEV usage in hostfs Date: Tue, 25 Jan 2005 11:11:26 +0100 Message-ID: <20050125101126.GC14056@pasky.ji.cz> References: <20050125091039.GA32409@lst.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: jdike@karaya.com, linux-fsdevel@vger.kernel.org Return-path: Received: from w241.dkm.cz ([62.24.88.241]:57219 "HELO machine.sinus.cz") by vger.kernel.org with SMTP id S261878AbVAYKL2 (ORCPT ); Tue, 25 Jan 2005 05:11:28 -0500 To: Christoph Hellwig Content-Disposition: inline In-Reply-To: <20050125091039.GA32409@lst.de> Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org Dear diary, on Tue, Jan 25, 2005 at 10:10:39AM CET, I got a letter, where Christoph Hellwig told me, that... > Could someone explain why hosts does thnings like: > > if((ino->i_sb->s_dev == ROOT_DEV) && (ino->i_uid == getuid())) > ino->i_uid = 0; > > (in fs/hostfs/hostfs_kern.c:read_name()) > > and > > if(attr->ia_valid & ATTR_UID){ > if((dentry->d_inode->i_sb->s_dev == ROOT_DEV) && > (attr->ia_uid == 0)) > attr->ia_uid = getuid(); > attrs.ia_valid |= HOSTFS_ATTR_UID; > attrs.ia_uid = attr->ia_uid; > } > if(attr->ia_valid & ATTR_GID){ > if((dentry->d_inode->i_sb->s_dev == ROOT_DEV) && > (attr->ia_gid == 0)) > attr->ia_gid = getuid(); > attrs.ia_valid |= HOSTFS_ATTR_GID; > attrs.ia_gid = attr->ia_gid; > } > > (hostfs_getattr) > > A filesystems shouldn't have special casing for the rootfs, and > chowning files from root to the current user sounds like an extremly > bad idea to me aswell. > > It's also the last thing preventing us from exporting ROOT_DEV Hmm, I pretty much forgot it all, but it seems to do a trick that if you run UML with a non-root UID (if that's actually possible), the files owned by you on the host system are owned by root inside of the UML (which does not seem to be so unreasonable, but should be probably configurable if it's the case) - depends on what context getuid() runs in, which I'm not sure of anymore neither. The gid handling is clearly buggy. It should test that in read_name() too and use getgid(). I don't know why it happens only on ROOT_DEV, perhaps some heuristic. Anyway, this comes (almost) verbatim from the 2.4.19 version by Jeff Dike, so he will hopefully come with a better explanation :-). -- Petr "Pasky" Baudis Stuff: http://pasky.or.cz/ 98% of the time I am right. Why worry about the other 3%.