On 2021-08-27 15:49, Paul Moore wrote:

On Fri, Aug 27, 2021 at 9:36 AM Richard Guy Briggs [off-list ref] wrote:

quoted

On 2021-08-26 15:14, Paul Moore wrote:

quoted

On Thu, Aug 26, 2021 at 12:32 PM Richard Guy Briggs [off-list ref] wrote:

quoted

I'm getting:
        # ./iouring.2
        Kernel thread io_uring-sq is not running.
        Unable to setup io_uring: Permission denied

        # ./iouring.3s
        >>> server started, pid = 2082
        >>> memfd created, fd = 3
        io_uring_queue_init: Permission denied

I have CONFIG_IO_URING=y set, what else is needed?

I'm not sure how you tried to run those tests, but try running as root
and with SELinux in permissive mode.

Ok, they ran, including iouring.4.  iouring.2 claimed twice: "Kernel
thread io_uring-sq is not running." and I didn't get any URING records
with ausearch.  I don't know if any of this is expected.

Now that I've written iouring.4, I would skip the others; while
helpful at the time, they are pretty crap.

Ok.

I have no idea what kernel you are running, but I'm going to assume
you've applied the v2 patches (if not, you obviously need to do that
<g>).  Beyond that you may need to set a filter for the
io_uring_enter() syscall to force the issue; theoretically your audit
userspace patches should allow a uring op specifically to be filtered
but I haven't had a chance to try that yet so either the kernel or
userspace portion could be broken.

I'm running audit/next (on 5.14-rc1) with your v2 patches.

I did set a syscall filter for
	-a exit,always -F arch=b64 -S io_uring_enter,io_uring_setup,io_uring_register -F key=iouringsyscall
and that yielded some records with a couple of orphans that surprised me
a bit.  I've attached that log.  I was a bit surprised there were no
records for ./iouring.3*.

I'm now testing the new "-a uring,always -U ..." to get that userspace
code working as expected...

At this point if you are running into problems you'll probably need to
spend some time debugging them, as I think you're the only person who
has tested your audit userspace patches at this point (and the only
one who has access to your latest bits).

Yes, I'll do some basic debugging and then publish to avoid wasting
people's time on silly bugs, but to get help on the more serious ones.

paul moore

- RGB

--
Richard Guy Briggs [off-list ref]
Sr. S/W Engineer, Kernel Security, Base Operating Systems
Remote, Ottawa, Red Hat Canada
IRC: rgb, SunRaycer
Voice: +1.647.777.2635, Internal: (81) 32635