Bluetooth rfcomm tty layer initialization

This was in 2.5 for a while. Happens when loading the rfcomm module.
Anything more needed for this? All bluetooth stuff is modular, and I'm using devfs.
I cannot capture the start of the rfcomm loading, the messages are scrolling far too fast and I don't have a serial console handy right now, sorry.

Thanks,

Jan

from dmesg:

Bluetooth: Core ver 2.2
Bluetooth: HCI device and connection manager initialized
Bluetooth: HCI socket layer initialized
Bluetooth: L2CAP ver 2.1
Bluetooth: L2CAP socket layer initialized
Bluetooth: BNEP (Ethernet Emulation) ver 1.0
Bluetooth: BNEP filters: protocol multicast

[ cutted some unrelated info ]

devfs_mk_cdev: could not append to parent for bluetooth/rfcomm/1 [ cutted 253 lines]
devfs_mk_cdev: could not append to parent for bluetooth/rfcomm/255
Bluetooth: RFCOMM TTY layer initialized

------------[ cut here ]------------
kernel BUG at include/linux/module.h:297!
invalid operand: 0000 [#1]
PREEMPT SMP
CPU:    0
EIP:    0060:[<d0cf3e2d>]    Not tainted VLI
EFLAGS: 00010246

EIP is at rfcomm_sock_alloc+0x59/0x10c [rfcomm]
eax: 00000000 ebx: cc9d9b00 ecx: 00000002 edx: d0cfb300
esi: 000000d0 edi: 00000003 ebp: 00000001 esp: cc987f14
ds: 007b es: 007b ss: 0068
Process rfcomm (pid: 1518, threadinfo=cc986000 task=ccd8f940)

Stack: 0000000c d0b0f384 d0cf3f17 ccc45180 00000003 000000d0 d0b02112 ccc45180
       00000003 0000007c ccc45180 00000001 c03ea720 ffffffa3 c026e056 
ccc45180
       00000003 0000001f 00000001 cc987fa8 00000000 ffffff9f c026e195 
0000001f
Call Trace:
 [<d0cf3f17>] rfcomm_sock_create+0x37/0x58 [rfcomm]
 [<d0b02112>] bt_sock_create+0xaa/0x104 [bluetooth]
 [<c026e056>] sock_create+0x14e/0x270
 [<c026e195>] sys_socket+0x1d/0x50
 [<c026ef3f>] sys_socketcall+0x9f/0x240
 [<c0109afd>] error_code+0x2d/0x38
 [<c0109073>] syscall_call+0x7/0xb

Code: b2 cf d0 83 bb 28 01 00 00 00 74 08 0f 0b cb 01 0c 71 cf d0 89 83 28 01 00 00 85 c0 74 40 50 e8 0e 11 44 ef 83 c4 04 85 c0 75 08 <0f> 0b 29 01 23 70 cf d0 b8 00 e0 ff ff 21 e0 ff 40 14 8b 50 10

Yep. I'm aware of that problem. It's on my TOFIX list :)

Max