Re: [9fans] usb questions

Re: [9fans] usb questions

[Francisco J Ballesteros]

We are working on a new usb set of drivers, with hot plugging.
usbd is now able to fire up drivers if necessary.

Ehci will come later.
But don´t hold your breath.

On Mon, Jan 19, 2009 at 4:33 AM,  <cinap_lenrek@gmx.de> wrote:
> 1)
> Is someone currently working on or has some stuff prepared for EHCI
> support for plan9?
>
> 2)
> I found that usbd has sometimes problems with replugging a usb
> device. After killing usbd and restrating it anything works fine. But
> this has the disadvantage that I loose other previously started usb
> drivers like for example a usb/usbaudio running in another window.
>
> It would be nice if can be made possible that you plug a device in and
> usbd activates it and then send a plumbing message. Then some plumber
> rules can start up the apropriate fileserver. In the same way 'Local mount ...'
> works. Here should be a generic rule that these usb driver filesystems just
> exit if they encouter problems during device pulling so the whole thing
> can start over when the device is plugged in again. Does that make any
> sense?
>
> --
> cinap
>
>
>
>
>

[9fans] usb questions

[cinap_lenrek]

1)
Is someone currently working on or has some stuff prepared for EHCI
support for plan9?

2)
I found that usbd has sometimes problems with replugging a usb
device. After killing usbd and restrating it anything works fine. But
this has the disadvantage that I loose other previously started usb
drivers like for example a usb/usbaudio running in another window.

It would be nice if can be made possible that you plug a device in and
usbd activates it and then send a plumbing message. Then some plumber
rules can start up the apropriate fileserver. In the same way 'Local mount ...'
works. Here should be a generic rule that these usb driver filesystems just
exit if they encouter problems during device pulling so the whole thing
can start over when the device is plugged in again. Does that make any
sense?

--
cinap

[9fans] usb questions

[Ronald G Minnich]

I'm looking at usb and trying to understand something. I have two
systems that only have BARs for memory mapped IO for the USB controllers
(i.e. they are memory BARs, not I/O BARs). As I read the usbuhci code,
it appears that the controller only gets added if BAR[4] has a valid I/O
address; on these controllers it is always zero. I am about to kill a
tree and print the spec, but am wondering what's up here -- why does
plan 9 kernel only add a controller if it has an I/O address?

ron