scaled-back caps, take 4
Quoted 12 lines · Chris Wright
Quoted 10 lines · * Olaf Dietsche ([email protected]) wrote:
Quoted 6 lines · Andy Lutomirski
cap_2.6.6-mm2_4.patch: New stripped-back capabilities.
fs/exec.c | 15 ++++- include/linux/binfmts.h | 9 ++- security/commoncap.c | 130 ++++++++++++++++++++++++++++++++++++++++++------ 3 files changed, 136 insertions(+), 18 deletions(-)
Quoted 8 lines
Quoted 4 lines
[patch]
Why don't you provide this as a configurable
andycap.cmodule? I think, this is the whole point of LSM.I agree, if we can't find a clean way to do it. However, note this includes changes to core. And it's nice to fix this for the base case.
On the other hand, this version has minimal changes to core (it adds a new field to linux_binprm and makes fs/exec.c fill in some extra information).
These changes shouldn't break any existing code, as the current behavior
is for bprm->cap_* to be undefined when bprm_set_security is called. None of this is strictly necessary for my patch, but it makes it a lot cleaner.
So, if the core changes were merged, my caps semantics could be maintained as a (fairly simple) separate LSM. That prevents it working with SELinux or other (non-stacking) LSMs loaded.
--Andy
Quoted 3 lines · Andy Lutomirski
So, if the core changes were merged, my caps semantics could be maintained as a (fairly simple) separate LSM. That prevents it working with SELinux or other (non-stacking) LSMs loaded.
SELinux supports stacking with the existing capability module (SELinux registers first, then the capability module registers as a secondary module under it).
Signature · 2 lines
Stephen Smalley <[email protected]> National Security Agency