[Bug 209253] Loss of connectivity on guest after important host <-> guest traffic

[bugzilla-daemon@bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=209253

--- Comment #7 from Alex Williamson (alex.williamson@redhat.com) ---
Color me suspicious, but there are backtraces from two configurations in the
comments here that have no vfio devices, the original post and Justin's second
trace.  The identified commit can only affect vfio configurations.

All of the backtraces seem to be from triggering this warning:

__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
{
        unsigned long flags;

        /*
         * Deadlock or stack overflow issues can happen if we recurse here
         * through waitqueue wakeup handlers. If the caller users potentially
         * nested waitqueues with custom wakeup handlers, then it should
         * check eventfd_signal_count() before calling this function. If
         * it returns true, the eventfd_signal() call should be deferred to a
         * safe context.
         */
        if (WARN_ON_ONCE(this_cpu_read(eventfd_wake_count)))
                return 0;

This cpu-local counter is only incremented while holding a spinlock with IRQs
disabled while handling the wait queue.

It's not obvious to me how the backtraces shown can lead to recursive eventfd
signals.  I've setup a configuration for stress testing, but any detailed
description of a reliable reproducer would be appreciated.

-- 
You are receiving this mail because:
You are watching the assignee of the bug.