Daniel Mack | 10 Jun 22:26 2012
Picon

Re: snd-usb-audio probe problem (Teac UD-H01)

On 08.04.2012 16:23, Daniel Mack wrote:
> On 02.04.2012 10:12, Clemens Ladisch wrote:
>> bEN wrote:
>>> the new USB DAC from Teac seems detected as 'Human Interface Device'
>>> instead of 'Audio'.
>>
>> It has both HID and audio interfaces.
>>
>>> Despite the fact it is based on a standard USB Audio
>>> class v2 controller (Tenor 8802).
>>
>> Which doesn't prevent Teac from writing buggy firmware.
>>
>>> $ dmesg
>>> usb 2-4: new high speed USB device using ehci_hcd and address 3
>>> usb 2-4: config 1 has an invalid interface number: 3 but max is 2
>>> usb 2-4: config 1 has no interface number 1
>>
>> This violates the USB spec.  But I guess it shouldn't prevent the driver
>> from attaching to the device.
> 
> Right, that's just a warning.
> 
>>> Audio class v2 interfaces need an interface association
>>
>> This looks like a bug in the driver; there certainly is such
>> an descriptor.  Daniel?
> 
> The USB audio driver actually just uses functions from the USB driver
> core to access the interface association. However, the USB core didn't
> assign it during its probe, which is most probably a result of the
> broken descriptor set.

And it turned out it is.

Ben, can you try the patch below, please? It's a pretty obvious flaw in
the USB core system which is triggered by buggy descriptors.

Daniel

Attachment (intf-assoc-fix.patch): text/x-patch, 707 bytes
On 08.04.2012 16:23, Daniel Mack wrote:
> On 02.04.2012 10:12, Clemens Ladisch wrote:
>> bEN wrote:
>>> the new USB DAC from Teac seems detected as 'Human Interface Device'
>>> instead of 'Audio'.
>>
>> It has both HID and audio interfaces.
>>
>>> Despite the fact it is based on a standard USB Audio
>>> class v2 controller (Tenor 8802).
>>
>> Which doesn't prevent Teac from writing buggy firmware.
>>
>>> $ dmesg
>>> usb 2-4: new high speed USB device using ehci_hcd and address 3
>>> usb 2-4: config 1 has an invalid interface number: 3 but max is 2
>>> usb 2-4: config 1 has no interface number 1
>>
>> This violates the USB spec.  But I guess it shouldn't prevent the driver
>> from attaching to the device.
> 
> Right, that's just a warning.
> 
>>> Audio class v2 interfaces need an interface association
>>
>> This looks like a bug in the driver; there certainly is such
>> an descriptor.  Daniel?
> 
> The USB audio driver actually just uses functions from the USB driver
> core to access the interface association. However, the USB core didn't
> assign it during its probe, which is most probably a result of the
> broken descriptor set.

And it turned out it is.

Ben, can you try the patch below, please? It's a pretty obvious flaw in
the USB core system which is triggered by buggy descriptors.

Daniel


Gmane