OK, this is still a work in progress, but I'm getting farther.
ATM, I do a connect_change in hub.c at the end of hub_configure for each
port on the hub, ensuring that it deals with each device on the hub as soon
as we're done configuring it.
It now recognizes the pegasus device and configures it as eth0, but due to
other issues it gives me some errors (
party FAIL ffff and eth0:set Rx mode) I'm still not finished.
How does the loop at the end of hub_configure strike everyone? Is this an
"OK" solution or is it wrong for reasons I don't understand (e.g., some
other initialization may not be done yet)
Will provide a dmesg if requested.
Thanks,
Ben
--
Ben Von Handorf
[EMAIL PROTECTED]
> Well, it'd need to work for both controller types,
> so it shouldn't be specific to that UHCI driver.
>
> And it really should allow for devices that are
> connected to hubs, too. You don't know that the
> NIC you're trying to enumerate is connected to the
> root hub; some root hubs have only one connector,
> so you need an external hub to use USB very much.
>
> The current mechanism is interrupt driven, so one
> workaround might be to pause for a while (maybe
> two seconds) in the thread that just found a new
> host controller and set it up. Very late in each
> of three HCD initialization sequences. Or just
> once, after PCI inits a new USB controller (ugh).
>
> Comments from folk who know booting, and USB, more?
>
> - Dave
>
> p.s. The same issue comes up with booting from a
> USB disk, I'd expect. Hmm.
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
