Re: [BUG] usb problems in .38-rc3+

On Sunday 20 February 2011 12:09:20 Ed Tomlinson wrote:

On Wednesday 16 February 2011 13:47:27 Gustavo F. Padovan wrote:

quoted

Hi Ed,

* Ed Tomlinson [off-list ref] [2011-02-16 13:38:00 -0500]:

quoted

On Wednesday 16 February 2011 09:00:53 Gustavo F. Padovan wrote:

quoted

Hi Ed,

* Ed Tomlinson [off-list ref] [2011-02-05 14:17:57 -0500]:

quoted

Hi,

My bluetooth usb dongle has stopped working in .38-rc3+

Bus 002 Device 008: ID 0a12:0001 Cambridge Silicon Radio, Ltd Bluetooth Dongle (HCI mode)

and in dmesg

[108495.369056] usb 5-1: new full speed USB device using ohci_hcd and address 3
[108495.573077] usb 5-1: New USB device found, idVendor=0a12, idProduct=0001
[108495.580099] usb 5-1: New USB device strings: Mfr=0, Product=0, SerialNumber=0
[108652.971977] usb 5-1: USB disconnect, address 3
[108652.972718] btusb_bulk_complete: hci0 urb ffff8800684e99c0 failed to resubmit (19)
[108652.973725] btusb_bulk_complete: hci0 urb ffff8800684e9c00 failed to resubmit (19)
[108652.974707] btusb_intr_complete: hci0 urb ffff8800684e9840 failed to resubmit (19)
[108653.001690] btusb_send_frame: hci0 urb ffff8800bb42a240 submission failed
[108657.007541] usb 2-4.7: new full speed USB device using ehci_hcd and address 8
[108657.147736] usb 2-4.7: New USB device found, idVendor=0a12, idProduct=0001
[108657.154954] usb 2-4.7: New USB device strings: Mfr=0, Product=0, SerialNumber=0

Its been working for years here.

Any ideas on what to try or additional info to gather?

What's the symptom? what do you mean by stopped to work?

With .37 kernel my bluetooth apple magicmouse is detected at boot and works as expected & when 
kde 4.6 is active bluedevil sees my bluetooth adaptor.  With .38-rc my bluetooth mouse is not detected
nor does bluedevil see the bluetooth adaptor.

quoted

Can you provide me output of "hciconfig -a"?

This gives nothing when the doogle is not plugged (it runs but prints out nothing).

quoted

And if possible enable dynamic debug for your kernel, mount the debugfs 
and then

echo -n "module btusb +p" > /sys/kernel/debug/dynamic_debug/control
echo -n "module bluetooth +p" > /sys/kernel/debug/dynamic_debug/control

Hey a new debug option to enable!  With it enabled and the system booted with the bluetooth doogle
unpluged here is what happens when it gets plugged (with 38-rc5):

grover ed # hciconfig -a
hci0:   Type: BR/EDR  Bus: USB
        BD Address: 00:0A:3A:55:07:5A  ACL MTU: 192:8  SCO MTU: 64:8
        UP RUNNING 
        RX bytes:356 acl:0 sco:0 events:13 errors:0
        TX bytes:53 acl:0 sco:0 commands:13 errors:0
        Features: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
        Packet type: DM1 DH1 HV1 
        Link policy: 
        Link mode: SLAVE ACCEPT 
        Name: 'grover-0'
        Class: 0x000000
        Service Classes: Unspecified
        Device Class: Miscellaneous, 
        HCI Version: 1.1 (0x1)  Revision: 0x20d
        LMP Version: 1.1 (0x1)  Subversion: 0x20d
        Manufacturer: Cambridge Silicon Radio (10)

and the device works.  So either enabling dynamic debug, something between rc4 and rc5 or booting with the
adaptor unplugged makes things work.  Will do some more checking to figure out which.

Great that it is working now. It might be a usb bug that you had, because we
had no changes in the Bluetooth subsystem between rc4 and rc5.

But does booting with the adapter work as well?

I have not been able to get it to work a second time no mater what combo I try - some sort of race I would guess:

Plugged at boot			fails
Plugged after boot			fails
Plugged after, btusb traced	fails
Plugged after, bluetooth traced	fails
Plugged after, both traced	fails

Gustavo,

I've figured out how to make things work though I am not sure why it fixes things.  

If I power on my bluetooth devices after running an 'hciconifg -a' they work.  

I can leave the adapter plugged at boot or plug it after.  The important thing seems to
be running the hciconfig before attempting to connect devices.  From the manpage
hciconfig -a should just be a query.  Instead it seems to initialize something and
makes things work here.  

Any idea why?
Ed