Yes, I tried both initrd and initramfs. The IDE stuff is included as modules by default by mkinitrd since there is an active IDE controller. However, looking at dmesg on a mixed scsi / ide system (linux on scsi) and on a failed upgrade ide only system indicates that the modules are ignoring the command line parm. And if you put the parm into modprobe.conf.local, it is picked up but then the modules puke since they don't know what to do with it. So, the decision to put a install a different kernel than the one used for the cdrom boot means that people will get a non-functional, unbootable system if their prime disk has an overlay. I could tolerate it if the kernel on the cdrom boot image also had this problem and it just failed to detect the partitions before the upgrade started. I'll drop the issue for now and go back to 9.1. It seems to be the last version that works with my configurations. Thanks for all who read my messages. Back to lurking. Dave -----Original Message----- From: meister@netz00.com [mailto:meister@netz00.com] Sent: Friday, November 11, 2005 6:40 PM on -maybe stupid- question: Did you try to load the IDE-module in initrd and pass hda=remap on kernel commandline? The initrd get loaded before the kernel so the commandline parameter should hit the loaded IDE-support from initrd. And did you try to use initrd not initramfs? On ppc I was only able to boot a xfs-root-partition if I load the module xfs in an initrd...initramfs did not work...