RDS: fix race condition when sending a message on unbound socket
authorQuentin Casasnovas <quentin.casasnovas@oracle.com>
Tue, 24 Nov 2015 22:13:21 +0000 (17:13 -0500)
committerBen Hutchings <ben@decadent.org.uk>
Wed, 30 Dec 2015 02:25:58 +0000 (02:25 +0000)
commit6240b18899c2ff1bfe29a1475a527f4e10662cb0
tree0dd014ce30786b7ab1ee9282ef28a5949d9c5d96
parent062d75332dfb0a31d97c702b2cf31cc0705cdb3b
RDS: fix race condition when sending a message on unbound socket

commit 8c7188b23474cca017b3ef354c4a58456f68303a upstream.

Sasha's found a NULL pointer dereference in the RDS connection code when
sending a message to an apparently unbound socket.  The problem is caused
by the code checking if the socket is bound in rds_sendmsg(), which checks
the rs_bound_addr field without taking a lock on the socket.  This opens a
race where rs_bound_addr is temporarily set but where the transport is not
in rds_bind(), leading to a NULL pointer dereference when trying to
dereference 'trans' in __rds_conn_create().

Vegard wrote a reproducer for this issue, so kindly ask him to share if
you're interested.

I cannot reproduce the NULL pointer dereference using Vegard's reproducer
with this patch, whereas I could without.

Complete earlier incomplete fix to CVE-2015-6937:

  74e98eb08588 ("RDS: verify the underlying transport exists before creating a connection")

Cc: David S. Miller <davem@davemloft.net>
Reviewed-by: Vegard Nossum <vegard.nossum@oracle.com>
Reviewed-by: Sasha Levin <sasha.levin@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
net/rds/connection.c
net/rds/send.c