From: Christoph Hellwig Subject: Re: [PATCH v3 2/4] xfs: initialize inode security on tmpfile creation Date: Wed, 16 Apr 2014 07:14:09 -0700 Message-ID: <20140416141409.GA21743@infradead.org> References: <1397578706-5385-1-git-send-email-bfoster@redhat.com> <1397578706-5385-3-git-send-email-bfoster@redhat.com> <20140415175033.GB26404@infradead.org> <534D90D0.9090805@tycho.nsa.gov> <20140415202222.GA10928@infradead.org> <534D94E4.8070606@tycho.nsa.gov> <534E7CDA.2060805@tycho.nsa.gov> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Brian Foster , linux-fsdevel@vger.kernel.org, linux-security-module@vger.kernel.org, xfs@oss.sgi.com, Eric Paris , Paul Moore To: Stephen Smalley Return-path: Content-Disposition: inline In-Reply-To: <534E7CDA.2060805@tycho.nsa.gov> Sender: linux-security-module-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Wed, Apr 16, 2014 at 08:51:38AM -0400, Stephen Smalley wrote: > Maybe I spoke too soon. IIUC, I_LINKABLE doesn't necessarily > distinguish tmpfiles from other files, as some tmpfiles may be linkable > and others not. But what we want is a way to identify all tmpfiles when > security_inode_init_security() is called if we are going to label them > independently of the provided dir. Oh, right. If O_EXCL is specified (another annoying overload of the flag..) the tmpfile can't ever be linked back into the filesystem and thus doesn't have I_LINKABLE set. I guess the best way to fix this is using the magic qstr you suggested before. That means security_inode_init_security would need to be called after d_tmpfile, which most filesystems don't do right now.