Re: [PATCH v10 0/8] Enroll kernel keys thru MOK

[Mimi Zohar]

On Tue, 2022-02-22 at 13:26 +0100, Jarkko Sakkinen wrote:
> On Tue, Feb 22, 2022 at 06:59:25AM -0500, Mimi Zohar wrote:
> > On Sun, 2022-02-20 at 20:00 +0100, Jarkko Sakkinen wrote:
> > > On Wed, Jan 26, 2022 at 05:06:09PM -0500, Mimi Zohar wrote:
> > 
> > > > > 
> > > > > Mimi brought up that we need a MAINTAINERS update for this and also
> > > > > .platform.
> > > > > 
> > > > > We have these:
> > > > > 
> > > > > - KEYS/KEYRINGS
> > > > > - CERTIFICATE HANDLING
> > > > > 
> > > > > I would put them under KEYRINGS for now and would not consider further
> > > > > subdivision for the moment.
> > > > 
> > > > IMA has dependencies on the platform_certs/ and now on the new .machine
> > > > keyring.  Just adding "F: security/integrity/platform_certs/" to the
> > > > KEYS/KEYRINGS record, ignores that dependency.  The discussion wouldn't
> > > > even be on the linux-integrity mailing list.
> > > > 
> > > > Existing requirement:
> > > > - The keys on the .platform keyring are limited to verifying the kexec
> > > > image.
> > > > 
> > > > New requirements based on Eric Snowbergs' patch set:
> > > > - When IMA_KEYRINGS_PERMIT_SIGNED_BY_BUILTIN_OR_SECONDARY is enabled,
> > > > the MOK keys will not be loaded directly onto the .machine keyring or
> > > > indirectly onto the .secondary_trusted_keys keyring.
> > > > 
> > > > - Only when a new IMA Kconfig explicitly allows the keys on the
> > > > .machine keyrings, will the CA keys stored in MOK be loaded onto the
> > > > .machine keyring.
> > > > 
> > > > Unfortunately I don't think there is any choice, but to define a new
> > > > MAINTAINERS entry.  Perhaps something along the lines of:
> > > > 
> > > > KEYS/KEYRINGS_INTEGRITY
> > > > M:     Jarkko Sakkinen <jarkko@xxxxxxxxxx>
> > > > M:     Mimi Zohar <zohar@xxxxxxxxxxxxx>
> > > > L:      keyrings@xxxxxxxxxxxxxxx
> > > > L:      linux-integrity@xxxxxxxxxxxxxxx
> > > > F:      security/integrity/platform_certs
> > > > 
> > > 
> > > This would work for me.
> > 
> > Thanks, Jarkko.  Are you planning on upstreaming this change, as you
> > previously said, or would you prefer I do it?
> > 
> This is the problem I'm encountering:
> 
> https://lore.kernel.org/keyrings/YhLNYxBTbKW62vtC@xxxxxx/

That's the answer to a different question. :)  I was asking about the
MAINTAINERS record.