On Sat, Jan 01, 2005 at 05:56:32PM +0100, Geert Stappers wrote: > > > > The installed initrd image misses files. > > > > - a sleep command is referenced in the /sbin/init script but is not on the > > image (see DELAY= parameter in the generated linuxrc.conf) > > May first thought was to clone this bug, > but "sleep" could be on the disk that can't be read. My thought went more towards this: busybox is used anyway on the initrd image. Adding sleep to the busybox commands costs only very few bytes and a link. > > So not yet cloned. > > > - the ohci1334 module needed to drive the external firewire disk is > > missing from the image. > > Is this simular as http://bugs.debian.org/283712, > "Reboot from USB Hard Disk failed"? Yes. The firewire driver needs some seconds as well until the disk is announced to the kernel. The 2.6.8 version also restarts three times in order to work around something not quite standard, which takes even more time. Anyway, selecting the disk as installation target works, only reboot from it fails. The difference to # 283712 is the the ppc bootloader (yaboot) is installed on the ide-hd (as it happens, the disk had some spare bytes at the right position for the bootloader). The openfirmare could load yaboot from the firewire (and most like usb drives as well) directly, but this would not change the initrd content problem. BTW, someone might be interested in http://www.dorchain.net/~joerg/code/cramfsswap* to ease initrd development across endian barriers. > > The modification of initrd was last year considered a can of worms. I can understand, but despite of tempting rfc1925 I think that the proposals for making installation for my particular case are a) easy and b) well within the boot concept. Normally I wanted to send patches, but I'm currently very busy, so something more precise to discuss might follow. Bye, Joerg
Attachment:
signature.asc
Description: Digital signature