[Pkg-bluetooth-maintainers] [wheezy] blueteeth adaptor cause kernel crash

[Jonathan Nieder]

fermat wrote:

> Yes, it can be easily reproduced, and I just prduced it two times.
>
> It is ok to plug in the adaptor, but when I send files to my phone
> through bluetooth, the screen frozes. capslock, magic sysrq and
> ctrl-alt-delete all don't works. What's more, several seconds later, the
> power of the keyboard get down, and then several seconds later, the
> power of the mouse get down again.

Thanks for these details.  Do I understand correctly that unlike RobinJ
in https://launchpad.net/bugs/986810 your keyboard LEDs didn't blink
(i.e., it's not a kernel panic)?

> On Wed, Oct 17, 2012 at 12:36:54PM -0700, Jonathan Nieder wrote:

>> Is this a regression?  Have you tried any other kernel versions, and
>> if so, how did they behave?
>
> I tried kernel 2.6.32-5-amd64, that kernel didn't break down, and I can
> send file from computer to my phone,

That's very good news.  It means one way to figure this out could be
to search through the precompiled kernels (testing the kernel halfway
between the newest known-good and oldest known-bad and repeating until
bored) from http://snapshot.debian.org/package/linux-2.6/ to narrow
down which introduced the bug.

> I tried to get the log. When I plug off the bluetooth adaptor and then
> plug in it, three messages appear on dmesg(1) output. these are the
> corresponding messages in /var/log/messages
[...]
> usb 2-1.5: new full-speed USB device number 4 using ehci_hcd
> usb 2-1.5: New USB device found, idVendor=0a12, idProduct=0001
> usb 2-1.5: New USB device strings: Mfr=0, Product=0, SerialNumber=0

So this bluetooth adapter is handled by the btusb driver:

	/* CSR BlueCore devices */
	{ USB_DEVICE(0x0a12, 0x0001), .driver_info = BTUSB_CSR },

[...]
> 08:54:28 bluetoothd[2614]: HCI dev 0 up
> 08:54:28 bluetoothd[2614]: Proximity GATT Reporter Driver: Operation not permitted (1)
> 08:54:28 bluetoothd[2614]: input-headset driver probe failed for device 54:F3:DE:69:66:01
> 08:54:28 bluetoothd[2614]: input-headset driver probe failed for device D8:71:57:64:5D:FD
> 08:54:28 bluetoothd[2614]: Unable to load keys to adapter_ops: Function not implemented (38)
> 08:54:28 bluetoothd[2614]: Adapter /org/bluez/2614/hci0 has been enabled
> 08:55:01 /USR/SBIN/CRON[5217]: (root) CMD (/etc/init.d/apf-firewall stop >> /dev/null 2>&1)
> 08:55:01 /USR/SBIN/CRON[5218]: (fermat) CMD (/usr/bin/notify-send foo bar)
> 08:55:17 obex-data-server: sdp_extract_seqtype: Unexpected end of packet
> 09:01:02 kernel: imklog 5.8.11, log source = /proc/kmsg started.

Cc-ing bluez maintainers in case they can make sense of this.  To
recap, what happens is that Fermat initiates file transfer and then
the machine hangs hard.

Thanks again,
Jonathan