Device categories for testing purposes

Marc Tardif marc.tardif at
Fri Nov 25 15:00:04 UTC 2011

* Colin Ian King <colin.king at> [2011-11-24 23:30 +0000]:
> On 24/11/11 21:58, Marc Tardif wrote:
> > * Colin Ian King <colin.king at> [2011-11-24 09:51 +0000]:
> >> On 23/11/11 22:04, Marc Tardif wrote:
> >>> Your feedback would be much appreciated to come up with a list of
> >>> categories for testing the kernel over a minimal set of systems with
> >>> unique devices. For example, lets say I'm interested in testing audio
> >>> across many systems, I would like to be given the minimal set of systems
> >>> with unique devices having the PCI class/subclass:
> >>>
> >>>   Multimedia controller/Audio device, or
> >>>   Multimedia controller/Multimedia audio controller
> >>>
> >>> or having the USB class/subclass:
> >>>
> >>>   Audio/Control Device, or
> >>>   Audio/MIDI Streaming, or
> >>>   Audio/Streaming
> >>>
> >>> At this point, I'm only interested in coming up with a mapping of PCI
> >>> and USB device class/subclass to a list of testing categories that are
> >>> most likely to be relevant for testing purposes. The following table is a
> >>> first draft where we currently have the categories: modem, video, audio,
> >>> storage, capture, network, wireless, processor, video, bluetooth. The
> >>> category "x" is used to indicate it might not be relevant.
> >> Does this include testing on-board USB3 controllers that appear on
> >> modern machines?
> > If this is of particular interest, we could add the "usb" testing category
> > for the following PCI device class/subclass:
> I'm not sure if we class it "of interest" or not. However, I think we
> need to be sure we cover USB 3.0 for certification, I'd hate to buy a
> brand new certified box and find USB 3.0 does not work. And this is true
> for most new Intel kit I believe...

The objective is to generate a minimal set of systems with unique
devices for testing purposes. If you expect to see several unique
USB 3.0 devices which should all be tested, then a testing category
should be defined. Otherwise, if you just want to make sure that the
USB 3.0 protocol is working correctly on just one compatible device,
then there should be no need for such a category. Please let me know
which of the two would be desirable.

PS. I have subscribed you to a bug in order to make sure your request
    to test for USB 3.0 as part of certification is not forgotten.

Marc Tardif <marc.tardif at>
Freenode: cr3, Jabber: cr3 at
1024D/72679CAD 09A9 D871 F7C4 A18F AC08 674D 2B73 740C 7267 9CAD

More information about the kernel-team mailing list