[LinuxBIOS] EHCI stack to port to FILO?

[Corey Osgood]

Anyone know of a decent/simple EHCI stack to port to FILO? I'm working
from the 2.6 kernel driver at the moment, but I'm wondering if anyone
knows of anything else, there are way more features in here then we need.

And can anyone, perhaps with an EPIA or EPIA-M, confirm that UHCI is
working in FILO? I've tried disabling the EHCI device, but it still
fails on vt8237.

Thanks,
Corey

[Peter Stuge]

On Sun, Sep 30, 2007 at 04:08:56PM -0400, Corey Osgood wrote:
> Anyone know of a decent/simple EHCI stack to port to FILO?

Sorry, no.
> I'm working from the 2.6 kernel driver at the moment,

What about the early serial stuff? (Not debug port, but the other.)

Maybe there's some EHCI stuff there?
> And can anyone, perhaps with an EPIA or EPIA-M, confirm that UHCI
> is working in FILO? I've tried disabling the EHCI device, but it
> still fails on vt8237.

All EHCI controllers also provide one UHCI or OHCI (I forget if
there's a choice or if not, which it should be) per physical port in
order to have backwards compatibility for 1.x-only devices.

So you should find UHCI controller devices when EHCI is enabled. And
if you disable EHCI, the UHCI/OHCI controllers should be disabled
too.

Try plugging a low- or full-speed device into the controller.

//Peter

[Corey Osgood]

Peter Stuge wrote:
> On Sun, Sep 30, 2007 at 04:08:56PM -0400, Corey Osgood wrote:
> > I'm working from the 2.6 kernel driver at the moment,
>
> What about the early serial stuff? (Not debug port, but the other.)
>
> Maybe there's some EHCI stuff there?

You mean in the kernel or LB?
> > And can anyone, perhaps with an EPIA or EPIA-M, confirm that UHCI
> > is working in FILO? I've tried disabling the EHCI device, but it
> > still fails on vt8237.
>
> All EHCI controllers also provide one UHCI or OHCI (I forget if
> there's a choice or if not, which it should be) per physical port in
> order to have backwards compatibility for 1.x-only devices.
>
> So you should find UHCI controller devices when EHCI is enabled. And
> if you disable EHCI, the UHCI/OHCI controllers should be disabled
> too.

Hmm, vt8237r has a separate device for EHCI that can be disabled without
disabling the UHCI controllers. I'm assuming that this essentially makes
them act as regular UHCI controllers.
> Try plugging a low- or full-speed device into the controller.

Just tried it:

boot: uda1:/vmlinuz root=/dev/hda1 console=ttyS0,115200n8
earlyprintk=ttyS0,115200n8 ro            
LinuxLabs USB bootloader
New USB device, setting address 2
New USB device, setting address 2
[...]
New USB device, setting address 2
There is a USB device, but it won't init! This is a bad thing.

I'll try to disable EHCI and then use the 1.1 hard drive adapter, and
see what happens.

-Corey

[Stefan Reinauer]

* Corey Osgood  [071001 05:03]:
> > So you should find UHCI controller devices when EHCI is enabled. And
> > if you disable EHCI, the UHCI/OHCI controllers should be disabled
> > too.
>
> Hmm, vt8237r has a separate device for EHCI that can be disabled without
> disabling the UHCI controllers. I'm assuming that this essentially makes
> them act as regular UHCI controllers.

You don't have to disable the EHCI controller device to use the other.
If you don't use the EHCI device, the USB communications will just be
1.1 instead of 2.0
> boot: uda1:/vmlinuz root=/dev/hda1 console=ttyS0,115200n8
> earlyprintk=ttyS0,115200n8 ro            
> LinuxLabs USB bootloader
> New USB device, setting address 2
> New USB device, setting address 2
> [...]
> New USB device, setting address 2
> There is a USB device, but it won't init! This is a bad thing.

I think the driver is broken and/or not well tested, unfortunately.