Hi,


----- Original Message ----
> From: Stephen Smalley <[email protected]>
> To: Serge E. Hallyn <[email protected]>
> Cc: CAI Qian <[email protected]>; [email protected]; 
> [email protected]
> Sent: Friday, January 30, 2009 2:58:48 AM
> Subject: Re: [LTP] [PATCH] proc01: SELinux with attr/* Interface - version 2
> 
> > > 
> > > No, they do not belong there, because we are focus on testing of proc
> > > filesystem here. The test case here should not care about the content it
> > > read, but rather what the kernel behaves when to read those files.
> > 
> > Then the dummy lsm I suggested is the only way to really test what you
> > want :)
> 
> That's not really a serious suggestion, right?  If the proc01 test
> cannot be dependent on the presence/absence of any given security module
> (like SELinux), then it cannot be dependent on a fake test security
> module either as that would preclude running the test while SELinux was
> in fact enabled.
> 

Exactly. The test should be able to run in different testing environment if
possible, and expect correct results according to that particular environment
if they differ. The test code just need a little bit careful to make it not
overkill to maintain.

> > 
> > > > In fact, the logical course given your concern would be to write a
> > > > kernel module defining an LSM allowing random long write to and reads
> > > > from /proc/$$/attr/ so you can test the procfs bits of that API (if
> > > > you could still write an LSM as a module :).  It should be doable to
> > > > push a 'debug' LSM into the upstream kernel which just serves to
> > > > facilitate such testing.
> > > > 
> > > > Anyway I've made my own position clear, and I think Kamalesh's patch
> > > > implements precisely what Stephen suggested.
> > > > 
> > > 
> > > Stephen's suggestion is something we should take. I agree Kamalesh's
> > > patch has included his suggestion. Unfortunately, there are other issues
> > > with Kamalesh's patch that have been pointed out in the last email.
> > 
> > Sorry, I've looked back through the thread, but don't see the other
> > issues you're talking about.
> 
> I think his concern is that it will obscure an actual bug in SELinux in
> the future.
> 

My main concerns of Kamalesh's patch are,

* reduce testing coverage of the test with SELinux enabled testing
   environment, because all reads of attr/* files are not flagged
   EINVAL as errors any more.

* false positive and misleading messages may happen on
  SELinux-enabled machine.

If attr/* files returns EINVAL while SELinux is enabled happens
in the future, it can be a bug in kernel or SELinux or both.

> I don't care strongly about it either way - the selinux testsuite is
> what you should be using to exercise SELinux, and just booting and
> logging into a SELinux-enabled system will exercise many if not all of
> those interfaces.
> 

I agree. I suppose that proc01 is mainly to test how kernel behaves
when to read those entries -- may use various length read buffers
etc. Of course, people might use it differently. Someone may use
it as a latency kernel for realtime kernel testing. It would be nice
to keep that flexibiltiy other than saying that this test must be run
with a particular kernel or LSM configuration.

CAI Qian

> -- 
> Stephen Smalley
> National Security Agency


------------------------------------------------------------------------------
This SF.net email is sponsored by:
SourcForge Community
SourceForge wants to tell your story.
http://p.sf.net/sfu/sf-spreadtheword
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

Reply via email to