loop: queue_lock NULL pointer derefence in blk_throtl_exit
authorSergey Senozhatsky <sergey.senozhatsky@gmail.com>
Wed, 19 Jan 2011 15:25:02 +0000 (08:25 -0700)
committerJens Axboe <jaxboe@fusionio.com>
Wed, 19 Jan 2011 15:25:02 +0000 (08:25 -0700)
Performing
$ sudo mount -o loop -o umask=0 /dev/sdb1 /mnt/
mount: wrong fs type, bad option, bad superblock on /dev/loop0,
       missing codepage or helper program, or other error
       In some cases useful info is found in syslog - try
       dmesg | tail  or so

$ sudo modprobe -r loop

results in oops:

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
 IP: [<ffffffff812479d4>] do_raw_spin_lock+0x14/0x122
 Process modprobe (pid: 6189, threadinfo ffff88009a898000, task ffff880154a88000)
 Call Trace:
  [<ffffffff81486788>] _raw_spin_lock_irq+0x4a/0x51
  [<ffffffff8123404b>] ? blk_throtl_exit+0x3b/0xa0
  [<ffffffff8105b120>] ? cancel_delayed_work_sync+0xd/0xf
  [<ffffffff8123404b>] blk_throtl_exit+0x3b/0xa0
  [<ffffffff81229bc8>] blk_release_queue+0x21/0x65
  [<ffffffff8123bb06>] kobject_release+0x51/0x66
  [<ffffffff8123bab5>] ? kobject_release+0x0/0x66
  [<ffffffff8123ce1e>] kref_put+0x43/0x4d
  [<ffffffff8123ba27>] kobject_put+0x47/0x4b
  [<ffffffff8122717c>] blk_cleanup_queue+0x56/0x5b
  [<ffffffffa01c3824>] loop_exit+0x68/0x844 [loop]
  [<ffffffff8107cccc>] sys_delete_module+0x1e8/0x25b
  [<ffffffff814864c9>] ? trace_hardirqs_on_thunk+0x3a/0x3f
  [<ffffffff81002112>] system_call_fastpath+0x16/0x1b

because of an attempt to acquire NULL queue_lock.
I added the same lines as in blk_queue_make_request -
index 44e18c0..49e6a54 100644`fall back to embedded per-queue lock'.

Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
drivers/block/loop.c

index 44e18c073c44105d03c1f7e0b9cd9bda229cac0f..49e6a545eb63fde25bc1984e3b2c165432f984b9 100644 (file)
@@ -1641,6 +1641,9 @@ out:
 
 static void loop_free(struct loop_device *lo)
 {
+       if (!lo->lo_queue->queue_lock)
+               lo->lo_queue->queue_lock = &lo->lo_queue->__queue_lock;
+
        blk_cleanup_queue(lo->lo_queue);
        put_disk(lo->lo_disk);
        list_del(&lo->lo_list);