Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20191012191602.45649-1-dancol@google.com> <20191012191602.45649-5-dancol@google.com>
 <CALCETrVmYQ9xikif--RSAWhboY1yj=piEAEuPzisf+b+qEX4uA@mail.gmail.com> <CAKOZuevQD-xsy_PrvT7F3Pqaoo5apZFukj2ZKLLQKup1cwgZ-A@mail.gmail.com>
In-Reply-To: <CAKOZuevQD-xsy_PrvT7F3Pqaoo5apZFukj2ZKLLQKup1cwgZ-A@mail.gmail.com>
From:   Andy Lutomirski <luto@kernel.org>
Date:   Sat, 12 Oct 2019 17:46:16 -0700
Message-ID: <CALCETrWFFbgdR3qtsQhxdMbaocoeHHkwpTQ-mQKJ4mgWBHz-Ng@mail.gmail.com>
Subject: Re: [PATCH 4/7] Teach SELinux about a new userfaultfd class
To:     Daniel Colascione <dancol@google.com>
Cc:     Andy Lutomirski <luto@kernel.org>,
        Linux API <linux-api@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Lokesh Gidra <lokeshgidra@google.com>,
        Nick Kralevich <nnk@google.com>,
        Nosh Minwalla <nosh@google.com>,
        Tim Murray <timmurray@google.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sat, Oct 12, 2019 at 5:12 PM Daniel Colascione <dancol@google.com> wrote:
>
> On Sat, Oct 12, 2019 at 4:09 PM Andy Lutomirski <luto@kernel.org> wrote:
> >
> > On Sat, Oct 12, 2019 at 12:16 PM Daniel Colascione <dancol@google.com> wrote:
> > >
> > > Use the secure anonymous inode LSM hook we just added to let SELinux
> > > policy place restrictions on userfaultfd use. The create operation
> > > applies to processes creating new instances of these file objects;
> > > transfer between processes is covered by restrictions on read, write,
> > > and ioctl access already checked inside selinux_file_receive.
> >
> > This is great, and I suspect we'll want it for things like SGX, too.
> > But the current design seems like it will make it essentially
> > impossible for SELinux to reference an anon_inode class whose
> > file_operations are in a module, and moving file_operations out of a
> > module would be nasty.
> >
> > Could this instead be keyed off a new struct anon_inode_class, an
> > enum, or even just a string?
>
> The new LSM hook already receives the string that callers pass to the
> anon_inode APIs; modules can look at that instead of the fops if they
> want. The reason to pass both the name and the fops through the hook
> is to allow LSMs to match using fops comparison (which seems less
> prone to breakage) when possible and rely on string matching when it
> isn't.

I suppose that whoever makes the first module that wants to use this
mechanism can have the fun task of reworking it.  There's nothing
user-visible here that would make it hard to change in the future.