RDMA/core: Fix race between destroy and release FD object

The call to ->lookup_put() was too early and it caused an unlock of the
read/write protection of the uobject after the FD was put. This allows a
race:

     CPU1                                 CPU2
 rdma_lookup_put_uobject()
   lookup_put_fd_uobject()
     fput()
				   fput()
				     uverbs_uobject_fd_release()
				       WARN_ON(uverbs_try_lock_object(uobj,
					       UVERBS_LOOKUP_WRITE));
   atomic_dec(usecnt)

Fix the code by changing the order, first unlock and call to
->lookup_put() after that.

Fixes: 3832125624b7 ("IB/core: Add support for idr types")
Link: https://lore.kernel.org/r/20200423060122.6182-1-leon@kernel.org
Suggested-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c
index 2947f4f83561d7c776bbeaa47955d0149b4c5ddc..177333d8bcdaeeb3a8e455248715776e892adb40 100644 (file)
--- a/drivers/infiniband/core/rdma_core.c
+++ b/drivers/infiniband/core/rdma_core.c
@@ -678,7 +678,6 @@ void rdma_lookup_put_uobject(struct ib_uobject *uobj,
                             enum rdma_lookup_mode mode)
 {
        assert_uverbs_usecnt(uobj, mode);
-       uobj->uapi_object->type_class->lookup_put(uobj, mode);
        /*
         * In order to unlock an object, either decrease its usecnt for
         * read access or zero it in case of exclusive access. See
@@ -695,6 +694,7 @@ void rdma_lookup_put_uobject(struct ib_uobject *uobj,
                break;
        }
 
+       uobj->uapi_object->type_class->lookup_put(uobj, mode);
        /* Pairs with the kref obtained by type->lookup_get */
        uverbs_uobject_put(uobj);
 }