KEYS: fix refcount_inc() on zero
authorMark Rutland <mark.rutland@arm.com>
Thu, 8 Jun 2017 13:47:41 +0000 (14:47 +0100)
committerJames Morris <james.l.morris@oracle.com>
Fri, 9 Jun 2017 03:29:50 +0000 (13:29 +1000)
If a key's refcount is dropped to zero between key_lookup() peeking at
the refcount and subsequently attempting to increment it, refcount_inc()
will see a zero refcount.  Here, refcount_inc() will WARN_ONCE(), and
will *not* increment the refcount, which will remain zero.

Once key_lookup() drops key_serial_lock, it is possible for the key to
be freed behind our back.

This patch uses refcount_inc_not_zero() to perform the peek and increment
atomically.

Fixes: fff292914d3a2f1e ("security, keys: convert key.usage from atomic_t to refcount_t")
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Cc: David Windsor <dwindsor@gmail.com>
Cc: Elena Reshetova <elena.reshetova@intel.com>
Cc: Hans Liljestrand <ishkamiel@gmail.com>
Cc: James Morris <james.l.morris@oracle.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: James Morris <james.l.morris@oracle.com>
security/keys/key.c

index cbae368..83da68d 100644 (file)
@@ -660,14 +660,11 @@ not_found:
        goto error;
 
 found:
-       /* pretend it doesn't exist if it is awaiting deletion */
-       if (refcount_read(&key->usage) == 0)
-               goto not_found;
-
-       /* this races with key_put(), but that doesn't matter since key_put()
-        * doesn't actually change the key
+       /* A key is allowed to be looked up only if someone still owns a
+        * reference to it - otherwise it's awaiting the gc.
         */
-       __key_get(key);
+       if (!refcount_inc_not_zero(&key->usage))
+               goto not_found;
 
 error:
        spin_unlock(&key_serial_lock);