[driver-discuss] bluetooth prototype panic on fr4k

[Pete Dennis]

Just reporting a panic with the bluetooth prototype code (code picked
up last 9-Jan-2007)

The stacks are something like:

 > $c
vpanic()
mutex_panic+0x73(fffffffffbc0c958, ffffffff8b86dc18)
mutex_vector_enter+0x4a4(ffffffff8b86dc18)
usba_rm_from_list+0x2e(ffffffff8b86dc08, ffffffff81766d20)
usba_req_wrapper_free+0x2d(ffffffff81766d20)
usb_free_intr_req+0x34(ffffffff81766db0)
usbhci_release_dev+0x29()
usbhci_close+0x20()
qdetach+0xbf(ffffffff846bbd50, 1, 2003, ffffffff87e92be0, 0)
strclose+0x35b(ffffffff8b7fd540, 2003, ffffffff87e92be0)
device_close+0xba(ffffffff8b7fd640, 2003, ffffffff87e92be0)
spec_close+0x1c8(ffffffff8b7fd640, 2003, 1, 0, ffffffff87e92be0)
fop_close+0x3f(ffffffff8b7fd640, 2003, 1, 0, ffffffff87e92be0)
closef+0x56(ffffffff846c12e0)
closeall+0x67(ffffffff8383bf68)
proc_exit+0x3d1(1, 0)
exit+0x15(1, 0)
rexit+0x18(0)
sys_syscall32+0x101()

And the corresponding panic string:

panic[cpu0]/thread=ffffffff845b2ee0:
mutex_enter: bad mutex, lp=ffffffff8b86dc18 owner=fffffe800012bc80 
thread=ffffff
ff845b2ee0

and for the other type of panic:

 > $c
turnstile_block+0x23d(0, 0, ffffffff81cd6498, fffffffffbc04bc8, 0, 0)
mutex_vector_enter+0x315(ffffffff81cd6498)
usba_rm_from_list+0x2e(ffffffff81cd6488, ffffffff838efb60)
usba_req_wrapper_free+0x2d(ffffffff838efb60)
usba_check_for_leaks+0x144(ffffffff84b41080)
hubd_post_detach+0x82(ffffffff816d9180, 2, fffffe8000a9e688)
usba_hubdi_bus_ctl+0x220(ffffffff8042cd68, ffffffff811e9be0, 17,
fffffe8000a9e688, fffffe8000a9e6ac)
ddi_ctlops+0x40(ffffffff811e9be0, ffffffff811e9be0, 17, fffffe8000a9e688,
fffffe8000a9e6ac)
i_detach_ctlop+0x5b(ffffffff811e9be0, 0, 1, 0)
devi_detach+0xa4(ffffffff811e9be0, 0)
detach_node+0x5f(ffffffff811e9be0, 40011)
i_ndi_unconfig_node+0xd2(ffffffff811e9be0, 4, 40011)
i_ddi_detachchild+0x2c(ffffffff811e9be0, 40011)
devi_detach_node+0x73(ffffffff811e9be0, 40011)
devi_unconfig_one+0xd4(ffffffff8042cd68, ffffffff82790681, 40011)
ndi_busop_bus_unconfig+0x48(ffffffff8042cd68, 40011, 5, ffffffff82790681)
hubd_bus_unconfig+0x65(ffffffff8042cd68, 40011, 5, ffffffff82790681)
ndi_devi_unconfig_one+0x14e(ffffffff8042cd68, ffffffff82790681, 0, 40011)
usba_destroy_child_devi+0xb7(ffffffff811e9be0, 1)
hubd_delete_child+0xb7(ffffffff816d9180, 2, 1, 1)
hubd_check_disconnected_ports+0xcb(ffffffff8042cd68, 0)
walk_devs+0x4d(ffffffff8042cd68, fffffffff3bb9970, 0, 1)
ddi_walk_devs+0x1e(ffffffff8042cd68, fffffffff3bb9970, 0)
hubd_root_hub_cleanup_thread+0xb2(ffffffff816d9180)
thread_start+8()

and the corresponding panic string:

panic[cpu0]/thread=fffffe8000a9ec80:
BAD TRAP: type=e (#pf Page fault) rp=fffffe8000a9e230 addr=ffffffff00000001
...


To reproduce - make sure that the bluetooth drivers are loaded.
Press the bluetooth button on the fr4k to turn it on and run the
supplied hciconfig.py and then press the bluetooth button to turn
it off.

The machine will panic either immediately or a short time later.

Looks like there is a structure being cleaned up while it is being
used.

Aside should there be a new alias for the bluetooth development ?

Cheers
pete