lockd: protect nlm_blocked access in nlmsvc_retry_blocked
authorDavid Jeffery <djeffery@redhat.com>
Wed, 10 Jul 2013 17:19:50 +0000 (13:19 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sun, 28 Jul 2013 23:26:17 +0000 (16:26 -0700)
commite2f918414ca99c5321ad352fa6dfa3147e3e31d8
tree6ecec69e41cf7b2135506a5f1911a3d3a6ce654f
parent5eae08bf9d6feff605b055d5f0964fc31a86b18e
lockd: protect nlm_blocked access in nlmsvc_retry_blocked

commit 1c327d962fc420aea046c16215a552710bde8231 upstream.

In nlmsvc_retry_blocked, the check that the list is non-empty and acquiring
the pointer of the first entry is unprotected by any lock.  This allows a rare
race condition when there is only one entry on the list.  A function such as
nlmsvc_grant_callback() can be called, which will temporarily remove the entry
from the list.  Between the list_empty() and list_entry(),the list may become
empty, causing an invalid pointer to be used as an nlm_block, leading to a
possible crash.

This patch adds the nlm_block_lock around these calls to prevent concurrent
use of the nlm_blocked list.

This was a regression introduced by
f904be9cc77f361d37d71468b13ff3d1a1823dea  "lockd: Mostly remove BKL from
the server".

Signed-off-by: David Jeffery <djeffery@redhat.com>
Cc: Bryan Schumaker <bjschuma@netapp.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/lockd/svclock.c