capabilities
cover capabilities
This reintroduces useful capabilities.
In this model, the inheritable mask is a limit on what capabilities the task or any of its children may have. Permitted and effective masks have their old meanings.
Init gets all capabilities (even CAP_SETPCAP) since it seems absurd to do otherwise.
Part 1 shouldn't change user-visible behavior; it just moves the vfs capability logic into fs/exec.c (where it IMHO belongs) and the setuid logic into apply_creds (where it IMHO belongs). I made no effort to make apply_creds pretty since it all gets deleted in the next patch anyway.
Part 2 redoes the capability logic.
This code is paranoid about setuid. A later patch will allow that to be overridden by a sysctl so that securelevel can be done usefully.
I left cap_bset in, even though I can't see a use for it anymore.
I put some user tools at http://www.stanford.edu/~luto/cap/; I've used themn to exercise this code somewhat.
It applies to 2.6.6-mm1. It does _not_ apply to 2.6.6 vanilla, but the fix should be trivial (it conflicts with something in fs/exec.c). It will not apply to earlier versions, as it depends on the compute_creds race fix.
Please let me know what you think and if you see any holes in this code or possibilities of exposing / introducing bugs in other programs (think Sendmail bug).
This should also eliminate the Oracle magic-group mess :)
Andrew- is this sufficiently non-scary for -mm?
Where this is going:
There is probably some dead code now in sys_capset. I'll check on that. Also, I have an ext3 xattr caps patch lying around somewhere. I can try to get it working again.
--Andy
Quoted 1 line · * Andy Lutomirski ([email protected]) wrote:
This reintroduces useful capabilities.
Thanks for doing this Andy. I'd like to give it a close look, but won't have any time until tomorrow.
thanks,
-chris
Signature · 1 line
Linux Security Modules http://lsm.immunix.org http://lsm.bkbits.net
Quoted 3 lines · Andy Lutomirski
This reintroduces useful capabilities.
What if there are existing applications which are deliberately or inadvertently relying upon the current behaviour? That seems unlikely, but the consequences are gruesome.
If I'm right in this concern, the fixed behaviour should be opt-in. That could be via a new prctl() thingy but I think it would be better to do it via a kernel boot parameter. Because long-term we should have the fixed semantics and we should not be making people change userspace for some transient 2.6-only kernel behaviour.
Quoted 1 line
Andrew- is this sufficiently non-scary for -mm?
Scares the shit out of me, frankly ;)
Quoted 9 lines · Andrew Morton
Quoted 2 lines · Andy Lutomirski
This reintroduces useful capabilities.
What if there are existing applications which are deliberately or inadvertently relying upon the current behaviour? That seems unlikely, but the consequences are gruesome.
Like something that turns KEEPCAPS on then setuid()s then executes an untrusted program? It's obviously wrong, but it's secure currently since the exec wipes capabilities. And no one would notice. Ugh!
Quoted 6 lines
If I'm right in this concern, the fixed behaviour should be opt-in. That could be via a new
prctl()thingy but I think it would be better to do it via a kernel boot parameter. Because long-term we should have the fixed semantics and we should not be making people change userspace for some transient 2.6-only kernel behaviour.
The prctl would defeat the purpose (imagine if bash forgot the prctl -- then the whole thing is pointless). I'll cook up the boot parameter in the next couple days (probably with a config option and some kind of warning that the old behavior is deprecated). Is it a problem if I make the changes to init's state unconditional? (I still don't see why CAP_SETPCAP is dangerous for root to have...)
The only concern is that some new code relies on the new inheritable semantics. That shouldn't be so bad, though, since that's just an extra precaution (if there are insecure setuid binaries around, you already have problems).
--Andy
Quoted 8 lines · Andy Lutomirski
Quoted 5 lines
What if there are existing applications which are deliberately or inadvertently relying upon the current behaviour? That seems unlikely, but the consequences are gruesome.
Like something that turns KEEPCAPS on then setuid()s then executes an untrusted program?
Or if it simply has caps and then does exec.
Quoted 8 lines
Quoted 5 lines
If I'm right in this concern, the fixed behaviour should be opt-in. That could be via a new
prctl()thingy but I think it would be better to do it via a kernel boot parameter. Because long-term we should have the fixed semantics and we should not be making people change userspace for some transient 2.6-only kernel behaviour.The prctl would defeat the purpose (imagine if bash forgot the prctl -- then the whole thing is pointless).
yup, lots of apps would need to be changed to interface with something which won't be present in 2.8 kernels.
Quoted 3 lines
I'll cook up the boot parameter in
the next couple days (probably with a config option and some kind of warning that the old behavior is deprecated).
I wouldn't bother with a config option.
Quoted 3 lines
Is it a problem if I make
the changes to init's state unconditional? (I still don't see whyCAP_SETPCAPis dangerous for root to have...)
I'd be more comfortable (ie: comfort level non-zero) if there was zero behaviour change if the boot option isn't enabled.
Quoted 4 lines
The only concern is that some new code relies on the new inheritable semantics. That shouldn't be so bad, though, since that's just an extra precaution (if there are insecure setuid binaries around, you already have problems).
Yes, we can live with that. The price of prior sins.
Quoted 4 lines · Andy Lutomirski
Like something that turns KEEPCAPS on then setuid()s then executes an untrusted program? It's obviously wrong, but it's secure currently since the exec wipes capabilities. And no one would notice. Ugh!
Definitely wrong.
Quoted 2 lines
The prctl would defeat the purpose (imagine if bash forgot the prctl -- then the whole thing is pointless).
Capabilities aware programmes are most likely already setting PR_SET_KEEPCAPS anyway if they're doing anything half-fancy. Another prctl() wont hurt too much if it is the only way to guarantee backward compatible security (?).
regards,
Signature · 5 lines
Paul Jakma [email protected] [email protected] Key ID: 64A2FF6A warning: do not ever send email to [email protected] Fortune: "I go on working for the same reason a hen goes on laying eggs." - H. L. Mencken