Following race may occur because of the call_srcu and the placement of
the synchronize_srcu vs the xa_erase.

CPU0				   CPU1

mlx5_ib_free_implicit_mr:	   destroy_unused_implicit_child_mr:
 xa_erase(odp_mkeys)
 synchronize_srcu()
				    xa_lock(implicit_children)
				    if (still in xarray)
				       atomic_inc()
				       call_srcu()
				    xa_unlock(implicit_children)
 xa_erase(implicit_children):
   xa_lock(implicit_children)
   __xa_erase()
   xa_unlock(implicit_children)

 flush_workqueue()
				   [..]
				    free_implicit_child_mr_rcu:
				     (via call_srcu)
				      queue_work()

 WARN_ON(atomic_read())
				   [..]
				    free_implicit_child_mr_work:
				     (via wq)
				      free_implicit_child_mr()
 mlx5_mr_cache_invalidate()
				     mlx5_ib_update_xlt() <-- UMR QP fail
				     atomic_dec()

The wait_event() solves the race because it blocks until
free_implicit_child_mr_work() completes.

Fixes: 5256edcb ("RDMA/mlx5: Rework implicit ODP destroy")
Link: https://lore.kernel.org/r/20200227113918.94432-1-leon@kernel.orgSigned-off-by: Artemy Kovalyov <artemyko@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>