[9fans] usb: keyboards vs. hubs

[9fans] usb: keyboards vs. hubs

[Lyndon Nerenberg]

What is it with keyboards plugged in to USB hubs that the kernel doesn't 
like?  I thought it was just a broken hub when I first noticed this, but 
now I've seen it on two completely distinct systems.

If the keyboard has anything between it and the CPU (e.g. a hub or KVM)
it doesn't work.  Yet on the exact same path, a USB mouse works just 
fine.  In at least 50% of the cases, the same keyboard directly attached 
works fine :-P  Am I just in a vortex field?


--lyndon

   Specifications are for the weak and timid!

Re: [9fans] usb: keyboards vs. hubs

[Bruce Ellis]

yes you are in a USB vortex.  no human can implement the
standard.  no standards dudes can implement anything.

brucee

On 5/4/07, Lyndon Nerenberg <lyndon@orthanc.ca> wrote:
> What is it with keyboards plugged in to USB hubs that the kernel doesn't
> like?  I thought it was just a broken hub when I first noticed this, but
> now I've seen it on two completely distinct systems.
>
> If the keyboard has anything between it and the CPU (e.g. a hub or KVM)
> it doesn't work.  Yet on the exact same path, a USB mouse works just
> fine.  In at least 50% of the cases, the same keyboard directly attached
> works fine :-P  Am I just in a vortex field?
>
>
> --lyndon
>
>   Specifications are for the weak and timid!
>

Re: [9fans] usb: keyboards vs. hubs

[Charles Forsyth]

plan 9 supports the mouse on usb using usb/mouse, running the full protocol,
so it can find it on the hub.

it does nothing to support the keyboard, so when it works, i'd suspect it's
because you've enabled legacy mode in the bios, which leaves the keyboard
in special `boot device mode', and (perhaps depending on which bios)
that possibly only works when it's directly visible on one of the system's notional root hubs
(ie, directly wired ports without a hub at all).

Re: [9fans] usb: keyboards vs. hubs

[Douglas A. Gwyn]

"Lyndon Nerenberg" <lyndon@orthanc.ca> wrote in message 
news:20070503185314.B22416@orthanc.ca...
> If the keyboard has anything between it and the CPU (e.g. a hub or KVM)
> it doesn't work.

Assuming there is a USB driver for the keyboard device, then
check whether the problem occurs if the keyboard is already
connected when the computer is powered on.  At least for
PS/2 keyboards, there is an initialization sequence that has to
be performed (KVM switches generally do that).

Re: [9fans] usb: keyboards vs. hubs

[Wes Kussmaul]

Bruce Ellis wrote:
>   no standards dudes can implement anything.
...in a message that was answered expertly by Charles Forsyth, who once 
again shows that he, a consummate implementer with a real passion for 
simplicity and elegance, should be the sole source of standards for 
everything in the world.

If the world cannot, with suitable inducements not necessarily excluding 
some arm twisting, induce Charles to take the job then there are another 
few dozen suitable candidates. But one really smart talented implementer 
surrounded by some smart guards to keep the supplicants at bay should be 
responsible for all standards, including how to turn on the freaking 
wipers on that rental car.

Wes

PS No, I really mean it.

Re: [9fans] usb: keyboards vs. hubs

[Steve Simon]

> including how to turn on the freaking 
> wipers on that rental car.

had a Renault Megan from work for a few days and couldn't work out
where the "key" (credit card like thing) went - had to look and idiot
and ask somone.

OSs I expect but when cars start getting such crazy feature creep
its time to break out the long knives.

</rant>

-Steve

Re: [9fans] usb: keyboards vs. hubs

[Wes Kussmaul]

Steve Simon wrote:
> OSs I expect but when cars start getting such crazy feaure creep
> its time to break out the long knives.
>   
I hope you join me in eagerly awaiting the czar's edict on

1. required placement/symbology etc. in car starting devices (actually 
they should be zigbee or other wireless protocol)
2. putting real duly constituted public authority into the expression 
"certification authority"
3. proper enrollment procedures for binding key pairs to human beings
4. binding the resulting certificate to one's driver's license & 
embedding it in the key fob (or wristwatch or whatever)
5. standards for software that defines when and how police may ping your 
car to see who's driving;
6. building codes for the facility from which you control the use of 
information that is associated with the key pair

Wes Kussmaul

Re: [9fans] usb: keyboards vs. hubs

[ron minnich]

On 5/4/07, Charles Forsyth <forsyth@terzarima.net> wrote:
> plan 9 supports the mouse on usb using usb/mouse, running the full protocol,
> so it can find it on the hub.
>
> it does nothing to support the keyboard, so when it works, i'd suspect it's
> because you've enabled legacy mode in the bios, which leaves the keyboard
> in special `boot device mode', and (perhaps depending on which bios)
> that possibly only works when it's directly visible on one of the system's notional root hubs
> (ie, directly wired ports without a hub at all).

and, of course, there's always the issue that many BIOS USB stacks
will support *some*, but not *all*, keyboards. I had this recently
with a brand new Sun Ultra40. No PS/2 keyboards for them, no sir! USB!
It's the 21st century!

So, the hunt for a working USB keyboard turned up several NFG units
and one working unit.

Summary: Just because the BIOS won't support *a* USB keyboard, does
not mean it can not support some other USB keyboard.

Just to keep it interesting.

ron