Re: [PATCH 1/3] tpm: Disable TCG_TPM2_HMAC by default

[Jarkko Sakkinen]

On Tue May 28, 2024 at 2:44 AM EEST, James Bottomley wrote:
> On Tue, 2024-05-28 at 02:17 +0300, Jarkko Sakkinen wrote:
> > On Tue May 28, 2024 at 12:36 AM EEST, James Bottomley wrote:
> > > On Mon, 2024-05-27 at 22:53 +0300, Jarkko Sakkinen wrote:
> > > > On Mon May 27, 2024 at 8:57 PM EEST, James Bottomley wrote:
> > > > > On Mon, 2024-05-27 at 18:34 +0300, Jarkko Sakkinen wrote:
> > > [...]
> > > > > > While looking at code I started to wanted what was the
> > > > > > reasoning for adding *undocumented* "TPM2_OA_TMPL" in
> > > > > > include/linux/tpm.h.It should really be in tpm2-sessions.c
> > > > > > and named something like TPM2_NULL_KEY_OA or similar.
> > > > > 
> > > > > Well, because you asked for it. I originally had all the flags
> > > > > spelled out and I'm not a fan of this obscurity, but you have
> > > > > to do stuff like this to get patches accepted:
> > > > > 
> > > > > https://lore.kernel.org/linux-integrity/CZCKTWU6ZCC9.2UTEQPEVICYHL@suppilovahvero/
> > > > 
> > > > I still think the constant does make sense.
> > > 
> > > I'm not so sure.  The TCG simply defines it as a collection of
> > > flags and every TPM tool set I've seen simply uses a list of flags
> > > as well.  The original design was that the template would be in
> > > this one place and everything else would call into it.  I think the
> > > reason all template construction looks similar is for ease of
> > > auditing (it's easy to get things, particularly the flags, wrong).
> > > 
> > > If it only has one use case, it should be spelled out but if
> > > someone else would use it then it should be in the tpm.h shared
> > > header.
> > 
> > It is used only in tpm2-sessions.c and for the null key so there it
> > should be. And it is also lacking the associated documentation. Now
> > both name and context it is used is lost.
>
> The comment above the whole thing says what it is and where it comes
> from:
>
> 	/*
> 	 * create the template.  Note: in order for userspace to
> 	 * verify the security of the system, it will have to create
> 	 * and certify this NULL primary, meaning all the template
> 	 * parameters will have to be identical, so conform exactly to
> 	 * the TCG TPM v2.0 Provisioning Guidance for the SRK ECC
> 	 * key H template (H has zero size unique points)
> 	 */
>
> If we put the broken out flags back it's all fully documented.

Not the most productive conclusion when refusing to follow properly a
trivial request in the review feedback tbh.

BR, Jarkko