Mysterious USB device ID change on Hauppauge HVR-900 (em28xx)

Mysterious USB device ID change on Hauppauge HVR-900 (em28xx)

[Frank Schäfer]

Hi,

I've got a Hauppauge HVR-900 (65008/A1C0) today. First,  the device
showed up as USB device 7640:edc1 (even after several unplug - replug
cycles), so I decided to add this VID:PID to the em28xx driver to see
what happens.
That worked fine, em2882/em2883, tuner xc2028/3028 etc. were detected
properly.
Later I noticed, that the device now shows up as 2040:6500, which is the
expected ID for this device.
Since then, the device maintains this ID. I also checked if Windows is
involved, but it shows up with the same ID there.

Does anyone have an idea what could have happened ???
I wonder if we should add this ID to the em28xx driver...

Regards,
Frank

Re: Mysterious USB device ID change on Hauppauge HVR-900 (em28xx)

[Antti Palosaari]

On 11/23/2012 07:01 PM, Frank Schäfer wrote:
> Hi,
>
> I've got a Hauppauge HVR-900 (65008/A1C0) today. First,  the device
> showed up as USB device 7640:edc1 (even after several unplug - replug
> cycles), so I decided to add this VID:PID to the em28xx driver to see
> what happens.
> That worked fine, em2882/em2883, tuner xc2028/3028 etc. were detected
> properly.
> Later I noticed, that the device now shows up as 2040:6500, which is the
> expected ID for this device.
> Since then, the device maintains this ID. I also checked if Windows is
> involved, but it shows up with the same ID there.
>
> Does anyone have an idea what could have happened ???
> I wonder if we should add this ID to the em28xx driver...

em28xx chip reads USB ID from the external eeprom using I2C just after 
it was powered. After USB ID is got it connects to the USB bus using 
that ID. If there is no external eeprom it uses chipset default USB ID, 
which is 0xeb1a as vendor ID and some other number defined for chip 
model as device ID. In that case those wrong IDs seems to be total 
garbage, which indicates there is some hardware problems when 
communicating towards eeprom.

That method is not only Empia USB interface chips but almost all chipset 
uses just similar method.


regard
Antti

-- 
http://palosaari.fi/

Re: Mysterious USB device ID change on Hauppauge HVR-900 (em28xx)

[Frank Schäfer]

Am 23.11.2012 18:20, schrieb Antti Palosaari:
> On 11/23/2012 07:01 PM, Frank Schäfer wrote:
>> Hi,
>>
>> I've got a Hauppauge HVR-900 (65008/A1C0) today. First,  the device
>> showed up as USB device 7640:edc1 (even after several unplug - replug
>> cycles), so I decided to add this VID:PID to the em28xx driver to see
>> what happens.
>> That worked fine, em2882/em2883, tuner xc2028/3028 etc. were detected
>> properly.
>> Later I noticed, that the device now shows up as 2040:6500, which is the
>> expected ID for this device.
>> Since then, the device maintains this ID. I also checked if Windows is
>> involved, but it shows up with the same ID there.
>>
>> Does anyone have an idea what could have happened ???
>> I wonder if we should add this ID to the em28xx driver...
>
> em28xx chip reads USB ID from the external eeprom using I2C just after
> it was powered. After USB ID is got it connects to the USB bus using
> that ID. If there is no external eeprom it uses chipset default USB
> ID, which is 0xeb1a as vendor ID and some other number defined for
> chip model as device ID. In that case those wrong IDs seems to be
> total garbage, which indicates there is some hardware problems when
> communicating towards eeprom.
>
Yeah, hardware problems, sure... but I wonder why the USB ID was/is
persistent ? Communication problems caused by circuit defect should lead
to a rather random behavior...
On power loss, the everything except (ee)prom(s) contents should be
reset to a default state.
And if the eeprom was corrupted, how could it magically recover ?
Also: the USB id is read from the eeprom even without a OS driver beeing
involved, but the first usage of the driver seems to have "fixed" the ID...

Regards,
Frank

> That method is not only Empia USB interface chips but almost all
> chipset uses just similar method.
>
>
> regard
> Antti
>

Re: Mysterious USB device ID change on Hauppauge HVR-900 (em28xx)

[Mauro Carvalho Chehab]

Em Fri, 23 Nov 2012 18:01:14 +0100
Frank Schäfer <fschaefer.oss@googlemail.com> escreveu:

> Hi,
> 
> I've got a Hauppauge HVR-900 (65008/A1C0) today. First,  the device
> showed up as USB device 7640:edc1 (even after several unplug - replug
> cycles), so I decided to add this VID:PID to the em28xx driver to see
> what happens.
> That worked fine, em2882/em2883, tuner xc2028/3028 etc. were detected
> properly.
> Later I noticed, that the device now shows up as 2040:6500, which is the
> expected ID for this device.
> Since then, the device maintains this ID. I also checked if Windows is
> involved, but it shows up with the same ID there.

This is a known hardware bug on some HVR devices. I have this problem with
one based on tm6000: sometimes, it is not able to read from the EEPROM.
When this happens, it gets the manufacturer's default USB ID.

You may force it to use the right em28xx entry using "card=" modprobe
parameter.

-- 
Regards,
Mauro