RDMA/rxe: insert mcg into mcg_tree only after rxe_mcast_add() succeeds

In the Linux kernel, the following vulnerability has been resolved:

RDMA/rxe: insert mcg into mcg_tree only after rxe_mcast_add() succeeds

rxe_get_mcg() publishes a newly allocated multicast group in
rxe->mcg_tree before programming the backing Ethernet multicast address
with rxe_mcast_add(), which runs outside mcg_lock. A local userspace
RDMA client reaches this path with ATTACH_MCAST on a UD QP; if
rxe_mcast_add() then returns an error (for example -ENODEV when the
backing netdev has been removed, or a propagated dev_mc_add() error),
the unwind frees the published group without removing it from the tree.
A later lookup of the same MGID dereferences the freed struct rxe_mcg
from __rxe_lookup_mcg().

Fix this by keeping the new mcg private until rxe_mcast_add() succeeds.
Split the tree publication into __rxe_publish_mcg(), call rxe_mcast_add()
before taking the tree reference, and free the still-private mcg on
failure. Because the group is never visible in mcg_tree until the
multicast address is programmed, no concurrent caller can look it up or
attach a QP to a group that is about to be torn down, so the error path
needs no conditional unwind. If another caller publishes the same MGID
while the address is being programmed, the post-add re-check under
mcg_lock finds the winner; this caller then drops its private object and
balances its own rxe_mcast_add() with rxe_mcast_del() before returning
the winner.

Reproduced by forcing the rxe_mcast_add() error return under KASAN:
without the change the next attach to the same MGID reports a
slab-use-after-free in __rxe_lookup_mcg(); with it the forced failure
returns cleanly. A no-injection attach/detach regression, including a
two-QP shared join/leave and re-attach, stays KASAN- and leak-clean.