Re: strace of io_uring events?

[Andy Lutomirski <luto@kernel.org>]

> On Jul 15, 2020, at 4:07 PM, Kees Cook <keescook@chromium.org> wrote:
>
> Earlier Andy Lutomirski wrote:
>> Let’s add some seccomp folks. We probably also want to be able to run
>> seccomp-like filters on io_uring requests. So maybe io_uring should call into
>> seccomp-and-tracing code for each action.
>
> Okay, I'm finally able to spend time looking at this. And thank you to
> the many people that CCed me into this and earlier discussions (at least
> Jann, Christian, and Andy).
>
> It *seems* like there is a really clean mapping of SQE OPs to syscalls.
> To that end, yes, it should be trivial to add ptrace and seccomp support
> (sort of). The trouble comes for doing _interception_, which is how both
> ptrace and seccomp are designed.
>
> In the basic case of seccomp, various syscalls are just being checked
> for accept/reject. It seems like that would be easy to wire up. For the
> more ptrace-y things (SECCOMP_RET_TRAP, SECCOMP_RET_USER_NOTIF, etc),
> I think any such results would need to be "upgraded" to "reject". Things
> are a bit complex in that seccomp's form of "reject" can be "return
> errno" (easy) or it can be "kill thread (or thread_group)" which ...
> becomes less clear. (More on this later.)

My intuition is not to do this kind of creative reinterpretation of
return values. Instead let’s have a new type of seccomp filter
specifically for io_uring. So we can have SECCOMP_IO_URING_ACCEPT,
ERRNO, and eventually other things. We probably will want a user
notifier feature for io_uring, but I'd be a bit surprised if it ends
up ABI-compatible with current users of user notifiers.

> - There appear to be three classes of desired restrictions:
>  - opcodes for io_uring_register() (which can be enforced entirely with
>    seccomp right now).

Agreed.

>  - opcodes from SQEs (this _could_ be intercepted by seccomp, but is
>    not currently written)

As above, I think this should be intercepted by seccomp, but in a new
mode.  I think that existing seccomp filters should not intercept it.

>  - opcodes of the types of restrictions to restrict... for making sure
>    things can't be changed after being set? seccomp already enforces
>    that kind of "can only be made stricter"

Agreed.

>
> - How does no_new_privs play a role in the existing io_uring credential
>  management? Using _any_ kind of syscall-effective filtering, whether
>  it's seccomp or Stefano's existing proposal, needs to address the
>  potential inheritable restrictions across privilege boundaries (which is
>  what no_new_privs tries to eliminate). In regular syscall land, this is
>  an issue when a filter follows a process through setuid via execve()
>  and it gains privileges that now the filter-creator can trick into
>  doing weird stuff -- io_uring has a concept of alternative credentials
>  so I have to ask about it. (I don't *think* there would be a path to
>  install a filter before gaining privilege, but I likely just
>  need to do my homework on the io_uring internals. Regardless,
>  use of seccomp by io_uring would need to have this issue "solved"
>  in the sense that it must be "safe" to filter io_uring OPs, from a
>  privilege-boundary-crossing perspective.
>
> - From which task perspective should filters be applied? It seems like it
>  needs to follow the io_uring personalities, as that contains the
>  credentials. (This email is a brain-dump so far -- I haven't gone to
>  look to see if that means io_uring is literally getting a reference to
>  struct cred; I assume so.) Seccomp filters are attached to task_struct.
>  However, for v5.9, seccomp will gain a more generalized get/put system
>  for having filters attached to the SECCOMP_RET_USER_NOTIF fd. Adding
>  more get/put-ers for some part of the io_uring context shouldn't
>  be hard.

Let's ignore personalities for a moment (and see below).  Thinking
through the possibilities:

A: io_uring seccomp filters are attached to tasks.  When an io_uring
is created, it inherits an immutable copy of its creating task's
filter, and that's the filter set that applies to that io_uring
instance.  This could have somewhat bizarre consequences if the fd
gets passed around, but io_uring already has odd security effects if
fds are passed around.  It has the annoying property that, if a
library creates an io_uring and then a seccomp filter is loaded, the
io_uring bypasses the library.

B: The same, but the io_uring references the creating task so new
filters on the task apply to the io_uring, too.  This allows loading
and then sandboxing.  Is this too bizarre overall?

C: io_uring filters are attached directly to io_urings.  This has the
problem where an io_uring created before a task sandboxes itself isn't
sandboxed.  It also would require that a filter be able to hook
io_uring creation to sandbox it.

Does anyone actually pass io_urings around with SCM_RIGHTS?  It would
be really nice if we could make the default be that io_urings are
bound to their creating mm and can't be used outside it.  Then
creating an mm-crossing io_uring could, itself, be restricted.

In any case, my inclination is to go for choice B.  Choice C could
also be supported if there's a use case.