Georg Acher wrote:
>
> On Tue, Apr 11, 2000 at 06:26:25PM +0000, David Brownell wrote:
> > Seeking insight here ... I append a kernel debug log,
> > against 2.3.99 pre4-5, showing something I observed
> > last night. This is with a PIIX4 UHCI.
> >
> > Regardless of whether this keyboard/hub is buggy or not,
> > there's at least one kernel problem that it's tweaking.
> >
> > Any insights into what's really happening?
>
> The first thing that came into my mind:
>
> Is the keyboard-hub based on an Atmel chip?
Don't think so -- the biggest chip I can see through the plastic
says "Alcor" like the vendor ID string. That's about 10cm long.
I don't want to tear it apart while I can still return it.
Vendor ID 058f, product codes 9213 for hub, 9410 for keybd.
"STRONG MAN KEYBD HUB" sez the product string, "ALCOR" says the
vendor string. It's got a Mac-style "power on" button.
To look at its descriptors, nothing strange. The hub says it's
a hub (USB 1.1), it's got some strings (US English only), and
the trouble seems to arise only when I try to use that hub.
The keyboard seems OK, but I get lowlevel glitches like repeated
characters.
> I have a standalone hub with
> Atmel here, it doesn't like to be asked for string descriptors
> after the polling interrupt is running. I have looked at it with
> the CAT, it simply doesn't respond to the interrupt packets anymore.
It usually responds just fine to such requests, but not in
this case. Not responding to string descriptor requests
appears to be an indication that other stuff is going wrong
too ... I think that it follows other failures, but can't
yet be sure of the ordering.
Seeing an undending stream of interrupts reported is odd,
and given that I saw it with OHCI when the unit wasn't
even plugged in any more ... it feels like there's a bug
somewhere outside the {U,O}HCI code that's somehow been
tweaked. Or else the *HCI code is starting to get more
compatible (bug-for-bug :-) than we anticipated!
Is that what's happening -- "not responding to interrupt packets"?
If so, it's not responding to _anything_ suddenly. I've no USB
analyser, but it'd be interesting to see what it says about the
actual sequence of USB events leading up to trouble. (Anyone
have a lendable analyser in the SF Bay area?)
- Dave
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
