Re: RFC PATCH: apply security_syslog() only to the syslog() syscall, not to /proc/kmsg

["Zack Weinberg" <zackw@panix.com>]

On 11/9/06, Stephen Smalley <sds@tycho.nsa.gov> wrote:
> Unless I missed something, your plan above would disable SELinux
> syslog-related permission checking upon reads of a previously opened
> file descriptor to /proc/kmsg.  So it would change SELinux behavior in a
> way that is directly contrary to the notion of mandatory access control.

Yes, it would do that; no, I don't see why that change is contrary to
the notion of mandatory access control.  An open fd on /proc/kmsg
(with my changes applied) offers strictly fewer privileges than
SYSTEM__SYSLOG_MOD (no access to opcodes 4 and 5), and with SELinux
active, you can't get that open fd without having had
SYSTEM__SYSLOG_MOD at some prior time.  SELinux does not (as far as I
can tell) do MAC checks for access to normal files at read() time,
only open().

I see this as bringing /proc/kmsg in line with standard Unix file
permission semantics, overall.

> Part 4 appears to further expose /proc/kmsg to access by any uid 0
> process even if it has no capabilities (think privilege shedding or
> containers).

Only in the default privilege model, not in SELinux.  (CAP_SYS_ADMIN
is a hell of a lot more powerful than SYSTEM__SYSLOG_MOD.)  And I
could be talked out of part 4.

> But having a mapping in the core to a much
> smaller set of permissions would be even better, and help with
> maintenance; the next time someone added a new code, they would more
> likely see the mapping table in the core and update it than go digging
> into the individual security modules.

But that mapping is itself a security policy decision, and could
plausibly need to be done differently in different security modules...

zw