Re: [PATCH RESEND] KEYS: fix parsing invalid pkey info string

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Mon, 2018-12-17 at 12:02 -0800, Linus Torvalds wrote:
> On Mon, Dec 17, 2018 at 11:51 AM James Bottomley
> <James.Bottomley@hansenpartnership.com> wrote:
> > 
> > If this is to replace Eric's patch, didn't you want to set
> > token_mask to (1<<Opt_err)?
> 
> No, let's not add any extra code that is trying to be subtle. Subtle
> interactions was where the bug came from.

Sure; I suppose liking the TPM gives me a taste for subtlety.

> The code already checks the actual Opt_xyz for errors in a switch
> statement. The token_mask should be _purely_ about duplicate options
> (or conflicting ones).
> 
> Talking about the conflicting ones: Opt_hash checks that
> Opt_policydigest isn't set. But Opt_policydigest doesn't check that
> Opt_hash isn't set, so you can mix the two if you just do it in the
> right order.
>
> But that's a separate bug, and doesn't seem to be a huge deal.

Actually, I'm afraid it's not a bug, it's a command line ordering
thing.  To check the policy digest length, we need to know which hash
algorithm you're using, so if you're specifying a hash algorithm, it
has to appear *before* a policydigest as a command input.

I take it this is another subtlety you'd like documenting ...?

> But it *is* an example of how bogus all of this stuff is. Clearly
> people weren't really paying attention when writing any of this code.

Hey, not going to argue here.  The whole policy session passed in is
questionable because some of the actions the kernel takes on the key
have to depend on what was in the policy (which you don't know and
can't deduce from the hash).  The only way to get it to work
universally is to pass the actual policy in and have the kernel
construct the policy session itself.  Fortunately fixing it can be low
priority because we don't seem to have any users.

James