lockd: protect nlm_blocked access in nlmsvc_retry_blocked

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".

Cc: Bryan Schumaker <bjschuma@netapp.com>
Cc: stable@vger.kernel.org
Signed-off-by: David Jeffery <djeffery@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>

diff --git a/fs/lockd/svclock.c b/fs/lockd/svclock.c
index e703318..8ebd3f5 100644 (file)
--- a/fs/lockd/svclock.c
+++ b/fs/lockd/svclock.c
@@ -939,6 +939,7 @@ nlmsvc_retry_blocked(void)
        unsigned long   timeout = MAX_SCHEDULE_TIMEOUT;
        struct nlm_block *block;
 
+       spin_lock(&nlm_blocked_lock);
        while (!list_empty(&nlm_blocked) && !kthread_should_stop()) {
                block = list_entry(nlm_blocked.next, struct nlm_block, b_list);
 
@@ -948,6 +949,7 @@ nlmsvc_retry_blocked(void)
                        timeout = block->b_when - jiffies;
                        break;
                }
+               spin_unlock(&nlm_blocked_lock);
 
                dprintk("nlmsvc_retry_blocked(%p, when=%ld)\n",
                        block, block->b_when);
@@ -957,7 +959,9 @@ nlmsvc_retry_blocked(void)
                        retry_deferred_block(block);
                } else
                        nlmsvc_grant_blocked(block);
+               spin_lock(&nlm_blocked_lock);
        }
+       spin_unlock(&nlm_blocked_lock);
 
        return timeout;
 }