man/io_uring_setup.2: improve SQPOLL wakeup example

This makes it consistent with how liburing uses it, instead of using
the sledge hammer that is a fully SMP memory barrier on archs that
don't need it.

Reported-by: Joran Dirk Greef <joran@coil.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

man/io_uring_setup.2

@@ -97,11 +97,14 @@ call with the following code sequence:
 .in +4n
 .EX
 /*
- * Ensure that the wakeup flag is read after the tail pointer has been
- * written.
+ * Ensure that the wakeup flag is read after the tail pointer
+ * has been written. It's important to use memory load acquire
+ * semantics for the flags read, as otherwise the application
+ * and the kernel might not agree on the consistency of the
+ * wakeup flag.
  */
-smp_mb();
-if (*sq_ring->flags & IORING_SQ_NEED_WAKEUP)
+unsigned flags = atomic_load_relaxed(sq_ring->flags);
+if (flags & IORING_SQ_NEED_WAKEUP)
  io_uring_enter(fd, 0, 0, IORING_ENTER_SQ_WAKEUP);
 .EE
 .in