[PATCH v2 2/5] seccomp: Add wait_killable semantic to seccomp user notifier

Sargun Dhillon sargun at sargun.me
Sat May 1 00:09:26 UTC 2021

On Fri, Apr 30, 2021 at 4:23 PM Andy Lutomirski <luto at kernel.org> wrote:
> On Fri, Apr 30, 2021 at 1:49 PM Sargun Dhillon <sargun at sargun.me> wrote:
> >
> > The user notifier feature allows for filtering of seccomp notifications in
> > userspace. While the user notifier is handling the syscall, the notifying
> > process can be preempted, thus ending the notification. This has become a
> > growing problem, as Golang has adopted signal based async preemption[1]. In
> > this, it will preempt every 10ms, thus leaving the supervisor less than
> > 10ms to respond to a given notification. If the syscall require I/O (mount,
> > connect) on behalf of the process, it can easily take 10ms.
> >
> > This allows the supervisor to set a flag that moves the process into a
> > state where it is only killable by terminating signals as opposed to all
> > signals. The process can still be terminated before the supervisor receives
> > the notification.
> This is still racy, right?  If a signal arrives after the syscall
> enters the seccomp code but before the supervisor gets around to
> issuing the new ioctl, the syscall will erroneously return -EINTR,
> right?
> Can we please just fully fix this instead of piling a racy partial fix
> on top of an incorrect design?
> --Andy

I thought that you were fine with this approach. Sorry.

Maybe this is a dumb question, what's wrong with returning an EINTR if the
syscall was never observed by the supervisor?

I think that the only other reasonable design is that we add data to the
existing action which makes it sleep in wait_killable state.

More information about the Containers mailing list