WHQL and unclassified category: What's the real story

Hi,

currently I have the task of taking a WDM driver written by the customer through
WHQL testing. It is a PCI Express based device which will be submitted under
‘unclassified’ and only for Win 7/x64 . It is essentially a specialised piece of
test equipment.

Looking at the manual (help) from the DTM studio, it says somewhere that the
driver reliability tests are enough for an ‘unclassified’ approval. Looking at the
webpage http://msdn.microsoft.com/de-de/windows/hardware/gg487407#EBB however, it
would seem to indicate that I also need to pass Device Fundamentals ->
Connectivity PCI.

Currently there is one issue here, the subvendor ID in the customer logic is zero
which along with 0xffff is not allowed.

So what is the real story, do I also need to pass PCI connectivity or is the
‘Driver Reliability’ enough.

Charles

> Currently there is one issue here, the subvendor ID in the customer logic is

zero which along with 0xffff is not allowed.

So what is the real story, do I also need to pass PCI connectivity or is the
‘Driver Reliability’ enough.

If it is anything like unclassified for USB devices, then yes it needs to pass.

Regards,
Tim Green
Senior Development Engineer
DisplayLink (UK) Limited
Registered in England No. 04811048

Charles wrote:

currently I have the task of taking a WDM driver written by the customer through
WHQL testing. It is a PCI Express based device which will be submitted under
‘unclassified’ and only for Win 7/x64 . It is essentially a specialised piece of
test equipment.

If this is a specialized piece of test equipment, then why go to the
trouble of a WHQL signature at all?

Looking at the manual (help) from the DTM studio, it says somewhere that the
driver reliability tests are enough for an ‘unclassified’ approval. Looking at the
webpage http://msdn.microsoft.com/de-de/windows/hardware/gg487407#EBB however, it
would seem to indicate that I also need to pass Device Fundamentals ->
Connectivity PCI.

Right. An unclassified USB device has to pass the USB compatibility
test suite, and an unclassified PCI device has to pass the PCI
connectivity test suite. The hardware tests are not that hard to pass;
if you fail those, then your device does not meet the hardware specs,
and you shouldn’t be selling it anyway.

Currently there is one issue here, the subvendor ID in the customer logic is zero
which along with 0xffff is not allowed.

Correct. Such a device violates the PCI specifications and cannot be
sold as a PCI device. When a company receives a PCI vendor ID, it signs
certain agreements that it is legally obligated to follow.

So what is the real story, do I also need to pass PCI connectivity or is the
‘Driver Reliability’ enough.

You need to pass the PCI connectivity tests. You cannot legitimately
sell a PCI device that does not meet the PCI specifications.


Tim Roberts, xxxxx@probo.com
Providenza & Boekelheide, Inc.

Am 14.12.2011 18:27, schrieb Tim Roberts:

If this is a specialized piece of test equipment, then why go to the
trouble of a WHQL signature at all?

The customer’s customer insists.

> Currently there is one issue here, the subvendor ID in the customer logic is zero
> which along with 0xffff is not allowed.

Correct. Such a device violates the PCI specifications and cannot be
sold as a PCI device. When a company receives a PCI vendor ID, it signs
certain agreements that it is legally obligated to follow.

As far as I remember, the legal part governs Vendor ID and Device ID (well, along
with not being allowed to sublicense subvendor IDs and a few other rules). These
are ok, as the customer is a PCISIG member. Subsystem ID, subvendor ID is optional
and purely a Microsoft requirement (here).

> So what is the real story, do I also need to pass PCI connectivity or is the
> ‘Driver Reliability’ enough.

You need to pass the PCI connectivity tests. You cannot legitimately
sell a PCI device that does not meet the PCI specifications.

That’s what I was afraid of. Just wanted to make sure after the contradiction in
the docs.

Thanks.