Received-SPF: pass (google.com: domain of linux-kernel+bounces-87367-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Thu, 29 Feb 2024 11:42:37 -0800
From: Eric Biggers <ebiggers@kernel.org>
To: Fan Wu <wufan@linux.microsoft.com>
Cc: corbet@lwn.net, zohar@linux.ibm.com, jmorris@namei.org,
	serge@hallyn.com, tytso@mit.edu, axboe@kernel.dk, agk@redhat.com,
	snitzer@kernel.org, eparis@redhat.com, paul@paul-moore.com,
	linux-doc@vger.kernel.org, linux-integrity@vger.kernel.org,
	linux-security-module@vger.kernel.org,
	linux-fscrypt@vger.kernel.org, linux-block@vger.kernel.org,
	dm-devel@lists.linux.dev, audit@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Deven Bowers <deven.desai@linux.microsoft.com>
Subject: Re: [RFC PATCH v13 17/20] ipe: enable support for fs-verity as a
 trust provider
Message-ID: <20240229194237.GA1454@sol.localdomain>
References: <1709168102-7677-1-git-send-email-wufan@linux.microsoft.com>
 <1709168102-7677-18-git-send-email-wufan@linux.microsoft.com>
 <20240229044625.GA1946@sol.localdomain>
 <b73e3387-558f-4f40-8741-c6ed7965b25f@linux.microsoft.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <b73e3387-558f-4f40-8741-c6ed7965b25f@linux.microsoft.com>

On Thu, Feb 29, 2024 at 10:59:21AM -0800, Fan Wu wrote:
> > 
> > So IPE is interested in whether a file has an fsverity builtin signature, but it
> > doesn't care what the signature is or whether it has been checked.  What is the
> > point?
> > 
> > - Eric
> 
> It does make sure the signature is checked. This hook call can only be
> triggered after fsverity_verify_signature() succeed. Therefore, for files
> that are marked with the security blob inode_sec->fs_verity_sign as true,
> they must successfully pass the fsverity_verify_signature() check.
> 
> Regarding the other question, the current version does not support defining
> policies to trust files based on the inner content of their signatures
> because the current patch set is already too large.
> 
> We plan to introduce new policy grammars to enable the policy to define
> which certificate of the signature can be trusted after this version is
> accepted.

Ah, I see, you're relying on the fact that fsverity_verify_signature() verifies
the signature (if present) even if fs.verity.require_signatures hasn't been set.
That does happen to be its behavior, but this isn't clearly documented since
there previously wasn't really a use case for the builtin signatures without
setting fs.verity.require_signatures.  Can you please make sure this behavior is
documented properly in Documentation/filesystems/fsverity.rst and in function
comments?  Otherwise I worry that it could get changed and break your code.

- Eric