From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Bluetooth lockdep warning Date: Fri, 29 Jun 2007 14:37:53 +0200 Message-ID: <4684FD21.6030609@trash.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: Linux Netdev List To: marcel@holtmann.org Return-path: Received: from stinky.trash.net ([213.144.137.162]:43793 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755429AbXF2MiF (ORCPT ); Fri, 29 Jun 2007 08:38:05 -0400 Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org I'm getting this on current -git after adding an obexfs mount to my fstab: ============================================= [ INFO: possible recursive locking detected ] 2.6.22-rc6 #2 --------------------------------------------- obexfs/3786 is trying to acquire lock: (sk_lock-AF_BLUETOOTH){--..}, at: [] l2cap_sock_bind+0x37/0xf3 [l2cap] but task is already holding lock: (sk_lock-AF_BLUETOOTH){--..}, at: [] rfcomm_sock_connect+0x47/0xe0 [rfcomm] other info that might help us debug this: 2 locks held by obexfs/3786: #0: (sk_lock-AF_BLUETOOTH){--..}, at: [] rfcomm_sock_connect+0x47/0xe0 [rfc omm] #1: (rfcomm_mutex){--..}, at: [] mutex_lock+0x25/0x2a stack backtrace: Call Trace: [] __lock_acquire+0x162/0xbbe [] :l2cap:l2cap_sock_bind+0x37/0xf3 [] lock_acquire+0x7c/0xa0 [] :l2cap:l2cap_sock_bind+0x37/0xf3 [] lock_sock_nested+0xd6/0xea [] module_put+0x4c/0x68 [] :l2cap:l2cap_sock_bind+0x37/0xf3 [] kernel_bind+0xb/0xd [] :rfcomm:rfcomm_dlc_open+0xc7/0x260 [] iomem_open+0x17/0x39 [] :rfcomm:rfcomm_sock_connect+0xa0/0xe0 [] sys_connect+0x68/0x8e [] fd_install+0x5e/0x67 [] trace_hardirqs_on_thunk+0x35/0x37 [] trace_hardirqs_on+0x11c/0x147 [] trace_hardirqs_on_thunk+0x35/0x37 [] system_call+0x7e/0x83 The warning appears to be bogus (things work fine), the reason seems to be that the rfcomm socket and the rfcomm l2 socket share the same lockdep class. Can someone more familiar with that code please have a look?