Re: [RFC PATCH] MAINTAINERS: add an explicit credentials entry

[Paul Moore]

On Sun, Mar 9, 2025 at 7:12 AM Günther Noack <gnoack3000@xxxxxxxxx> wrote:
>
> Hello Paul and Serge!
>
> On Tue, Mar 04, 2025 at 05:23:05PM -0500, Paul Moore wrote:
> > The lack of an explicit credential (kernel/cred.c) entry has caused
> > confusion in the past among new, and not-so-new developers, about where
> > to send credential patches for discussion and merging.  Those patches
> > that are sent can often rot on the mailing lists for months as there
> > is no clear maintainer tasked with reviewing and merging patches.
> >
> > I'm volunteering for the cred maintainer role to try and reduce the
> > confusion and help cred patches find their way up to Linus' tree.  As
> > there generally aren't a lot of cred patches I'll start with simply
> > folding them into the LSM tree, but if this changes I'll setup a
> > dedicated cred tree.
> >
> > Signed-off-by: Paul Moore <paul@xxxxxxxxxxxxxx>
> > ---
> >  MAINTAINERS | 8 ++++++++
> >  1 file changed, 8 insertions(+)
> >
> > diff --git a/MAINTAINERS b/MAINTAINERS
> > index 896a307fa065..68e4656c15ea 100644
> > --- a/MAINTAINERS
> > +++ b/MAINTAINERS
> > @@ -6139,6 +6139,14 @@ L:     linux-input@xxxxxxxxxxxxxxx
> >  S:   Maintained
> >  F:   drivers/hid/hid-creative-sb0540.c
> >
> > +CREDENTIALS
> > +M:   Paul Moore <paul@xxxxxxxxxxxxxx>
> > +L:   linux-security-module@xxxxxxxxxxxxxxx
> > +S:   Supported
> > +T:   git https://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/lsm.git
> > +F:   include/linux/cred.h
> > +F:   kernel/cred.c
>
> Maybe also add the documentation:
>
> +F:     Documentation/security/credentials.rst
>
> This documents the prepare_creds()/commit_creds()/abort_creds()
> "transactional" API that tasks should use to change credentials.

Thanks.  Serge already posted a patch to add the doc page, we should be all set.

> Acked-by: Günther Noack <gnoack3000@xxxxxxxxx>
>
> Thank you both for stepping up to establish a clearer ownership for
> credentials!  There is a need for authoritative decisions in that
> area, and it has been difficult to find the right contacts for
> credentials on earlier patches as well, such as:
>
> https://lore.kernel.org/all/20240805-remove-cred-transfer-v2-0-a2aa1d45e6b8@xxxxxxxxxx/
> (patch by Jann Horn: "get rid of cred_transfer")
>
> https://lore.kernel.org/all/20250221184417.27954-2-gnoack3000@xxxxxxxxx/
> (patch by me, multithreaded Landlock enablement)

Yeah, the cred_transfer/keyctl issue is particularly nasty and needs
to be revisited.  If memory serves there was still a compatibility
issue with Jann's patch, but we may want to consider merging that into
-next just to see if userspace still cares.  It's on my todo list to
take a closer look when I have the time.

-- 
paul-moore.com