On 11/9/2022 3:33 PM, Paul Moore wrote:

On Tue, Oct 25, 2022 at 2:45 PM Casey Schaufler [off-list ref] wrote:

quoted

Add an integer member "id" to the struct lsm_id. This value is
a unique identifier associated with each security module. The
values are defined in a new UAPI header file. Each existing LSM
has been updated to include it's LSMID in the lsm_id.

The LSM ID values are sequential, with the oldest module
LSM_ID_CAPABILITY being the lowest value and the existing
modules numbered in the order they were included in the
main line kernel. The first 32 values (0 - 31) are reserved
for some as yet unknown but important use.

Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
---
 include/linux/lsm_hooks.h    |  1 +
 include/uapi/linux/lsm.h     | 32 ++++++++++++++++++++++++++++++++
 security/apparmor/lsm.c      |  2 ++
 security/bpf/hooks.c         |  2 ++
 security/commoncap.c         |  2 ++
 security/landlock/setup.c    |  2 ++
 security/loadpin/loadpin.c   |  2 ++
 security/lockdown/lockdown.c |  2 ++
 security/safesetid/lsm.c     |  2 ++
 security/selinux/hooks.c     |  2 ++
 security/smack/smack_lsm.c   |  2 ++
 security/tomoyo/tomoyo.c     |  2 ++
 security/yama/yama_lsm.c     |  2 ++
 13 files changed, 55 insertions(+)
 create mode 100644 include/uapi/linux/lsm.h

Unless you're getting paid by the patch, I'd rather you combine
patches 1/8 and 2/8 into a single patch.  They are both pretty small,
very related, and I don't want to see 1/8 merged anywhere without 2/8.

OK by me. One less compile+test cycle. It's nice to know there's some
limit on patch granularity.

quoted

diff --git a/include/linux/lsm_hooks.h b/include/linux/lsm_hooks.h
index e383e468f742..dd4b4d95a172 100644
--- a/include/linux/lsm_hooks.h
+++ b/include/linux/lsm_hooks.h

@@ -1607,6 +1607,7 @@ struct security_hook_heads {
  */
 struct lsm_id {
        const char      *lsm;           /* Name of the LSM */
+       int             id;             /* LSM ID */
 };

At the very least let's define lsm_id::id as an 'unsigned int' type,
but since we are going to see the lsm_id::id token used as part of the
kernel ABI (likely not in this struct) I agree with Greg's comments
about making the size more explicit.  I would suggest __u32/u32 as
32-bits should be plenty for this token.

Given the other upstream discussions we may want to do something
similar with lsm_id::lsm and __u8/u8.  I'm pretty sure I saw a similar
comment (by Greg?) elsewhere in this patchset when I was quickly
skimming these on my phone while away ...

Yes. The API friendly typing will be in the next revision.

--
paul-moore.com