Re: USB should work now

Hi Christian,

> The tricky part here is that this is a chicken-n-egg situation.
>
>Only the people with a dual USB-Serial cable can troubleshoot
>a connection.
>
>If we could get either a LCD message "netusb loaded and is ok",
>or one of the 2 LEDs to switch itself on, this could be good
>for people who only have the USB cradle.
>
>
Good idea! We definitely need this!

>Michael, what do you think of this ? Could you describe more the
>tricky steps you had to go to ? I hope you did not have to
>plug-unplug the cable a few times...
>
No, I didn't have to do that... However, I just succeeded twice so far,
so it's a bit early for a recipe that always works. Here's what I
noticed so far:

    * Only worked when the USB connection was open in WinCE (I'm using
      synce through the USB cable to copy files to my CF card on my
      h2200). This corresponds to Koen's theory: the device needs to be
      initialized on WinCE in a certain way before starting Linux. Once
      the recipe gets stable, a few HaRET dumps may help in understand
      what's really needed.
    * The 2 times it worked, I had the "spurious IRQ for DMA channel 2"
      warning. I'll try the patch currently available on the
      linux-arm-kernel mailing list (I'll have to merge it with the
      current hh kernel branch).

Still trying to find a predictable sequence...

>I suggest we implement some
>hotplug as well on the ipaq side, so that the ipaq and the host
>pc do interact without sharing the same IP, for example...
>
>
Right! But how would you do that? In Familiar 0.7.2 with Linux 2.4, the
IP address for usbf on the iPAQ is fixed. Would you need a dhcp server
to implement dynamic addresses?

:-)

    Michael.

-- 
Michael Opdenacker
http://opdenacker.org/