Re: [PATCH 5/7] seccomp_filter: Document what seccomp_filter is and how it works.

[Eric Paris <eparis@redhat.com>]

On Wed, 2011-05-04 at 19:03 +0200, Frederic Weisbecker wrote:
> On Wed, May 04, 2011 at 12:22:40PM -0400, Eric Paris wrote:
> So the following set of operations:
> > 
> > SECCOMP_FILTER_SET, __NR_foo, "a=0"
> > SECCOMP_FILTER_SET, __NR_read, "1"
> > SECCOPM_FILTER_APPLY
> >  
> > SECCOMP_FILTER_SET, __NR_foo, "b=0"
> > SECCOMP_FILTER_APPLY
> > 
> > SECCOMP_FILTER_SET, __NR_write, "1"
> > SECCOMP_FILTER_APPLY
> > 
> > Would return EPERM for the __NR_write entry since it was a new syscall
> > after a set.  I think we agree on all this.
> 
> No, why?
> 
> The default filter for a syscall, if none have been given for it, is "0".
> 
> Thus, if you write "1" later, the entire filter is going to be:
> 
> 	"0 && 1"
> 
> Which is fine, we are not overriding already applied permissions there.
> 
> So where is the need to return -EPERM in such a specific case? Is it
> worth the corner case to check in the kernel, and to handle in userspace?
> And for what reason?

I assumed without looking at the code (always a bad idea) that he wasn't
going to explicitly create a rule with "0" and was going to implicitly
deny anything without a rule.  If there is an explicit "0" rule then you
are right, i don't see a need to deny the set operation in the kernel.
But if it is implicit in the non-existence of a filter then it should be
easy to tell userspace it isn't allowed any more.

-Eric

-Eric