Martin-Éric Racine wrote: > > So usb-discover is not working then? It should load hid (it's not called > > usbhid in 2.4) when it runs, unconditionally, assuming there is a > > /proc/bus/pci/devices. Don't you have that file? > > Hard to tell without a way to logon to the system while it boots. You could mount the initrd and modify usr/sbin/usb-discover to always load the module and set -x it at the same time. > > > 2.6 kernels don't try finding any AT keyboard and forget usbhid too. > > > > Until today when Kamion put it back in, we didn't have usb-discover on > > 2.6 images, relying on just udev to load usb stuff for 2.6. It sounds > > like udev is not working on your hardware; maybe you need to file a bug > > on udev? > > Based on today's kernels, I'm no longer sure of exactly what causes it. > Are daily d-i builds based on experimental udev builds? Daily d-i builds are based on unstable. > Btw, one problem I notice with the USB stick recipe (as applied here to > copying everything to the CF from which d-i gets launched) is that one > cannot wipe off the installation media if that's where the ISO resides, > because parted complains about the device being already mounted. Yes, that's a limitation of hd-media installs. > I tried > having the ISO on the USB stick as an alternative, but that currently > fails too, presumably because that's where the bootstrap kernel resides. You should be able to partition the CF if the ISO is on USB. -- see shy jo
Attachment:
signature.asc
Description: Digital signature