Marcel Holtmann wrote:

>I don't know what the problem with 2.6.2 and SCO over HCI and at the
>moment I have no time to investigate. If your Bluetooth chip is CSR
>based you can check the SCO mapping with "hciconfig hci0 revision".
>  
>
If the SCO MTU for the device is too low prior to the URB submission, 
would it generate the -90?

It seems 90 is defined as EMSGSIZE in:

errno.h:63:#define      EMSGSIZE        90      /* Message too long */

However, not remapped as other errors are in usb.h.

A quick grep of  hci_usb.c reveals the error is generated when:

Line 275:  err = usb_submit_urb(urb, GFP_ATOMIC);

Where a bit of the logic determines 'urb->transfer_buffer_length', then 
submits with the resulting error.

Am I way off here?  I have attached the results of 'lsusb -v' to give 
more info.  If you think I may be on the right track, I can _try_ to 
play with it a bit, but it may take me some time since I am a bit rusty...

-- 

        Regards,
        
        Todd E. Johnson