After compiling kernel 2.6.17-gentoo-r4 with sys-kernel/genkernel-3.4.0-r1, I tried to boot in and I get a message like: >> Activating mdev /init: 216: mdev: not found And it won't boot because it says "/dev/hda3 is not a valid root device". Any idea?
Please attach the genkernel.log file to this bug.
Just wanted to update: I've compiled my kernel w/o genkernel, and I just need to find the time to reproduce this (my latest genkernel.log isn't the one that completely died on me). So just give me a few days... ;-)
*** Bug 145551 has been marked as a duplicate of this bug. ***
Created attachment 95490 [details] genkernel.log, goes with bug 145551 report
(In reply to comment #0) > After compiling kernel 2.6.17-gentoo-r4 with sys-kernel/genkernel-3.4.0-r1, I > tried to boot in and I get a message like: > >> Activating mdev > /init: 216: mdev: not found > > And it won't boot because it says "/dev/hda3 is not a valid root device". > > Any idea? > I tried changing the initrd to have udev instead of mdev in the two commands in /etc/initrd.scripts that matter. Still nearly the same result, no /dev/hda*. But then I tried using the "shell" command and typing "udevstart" and CTRL-d. And after a couple of complaints about not being about to mount devfs, the system seems to be running fine. So maybe that'll provide a clue as to how to officially fix it. I guess we just stumbled across a transition that isn't quite finished yet. I don't know if those devfs complaints came from the initrd (there's also the genkernel->grub.conf problem of whether it's initrd or initramfs) or from my /etc/init.d scripts or some third place, but they need to be addressed, too. Maybe I'm just behind on etc-update, I have 94 of those ._cfg00* files to catch up on, too.
I also had this problem. Running etc-update and updating the genkernel config file (naturally together with kernel recompilation) solve the problem
So can anyone verify that there's a problem on a system that actually has their genkernel.conf updated or is this all caused by not updating the config file?
Could it be the case that this bug is really the same as http://bugs.gentoo.org/show_bug.cgi?id=142775
Sorry, I haven't had time to try genkernel and a reboot again with an updated genkernel.conf. I'll probably go through another sync, emerge world, and do all my etc-update files again (currently I think there are 105 of them) before I re-generate a kernel, and it may be weeks before I get to that. I hope that anybody else falling into the pit I fell into finds the advice here to update their genkernel.conf and get on with life, and it works for them. However, the mention of bug 142775 is interesting. Somehow I was assuming that udev is newer than mdev, but it appears from the discussion on 142775 that the opposite is true. I'd appreciate a pointer to some documentation, maybe a CHANGES files somewhere, on what the drift and thought about how gentoo approaches setting up /dev is. I think I still had devfs a couple of weeks ago on 2.6.15-gentoo-r1. Apparently there is an idea that udev is needlessly complex and fragile? I don't know much about any of it, but I'm anxious to read what pros and cons the experts have written. It seems worthy to me to replace mknod commands with a /dev/MAKEDEV script with something generated from hardware detection and dynamically updated with USB, firewire, hot-plug, etc. changes. But theory sometimes gets murky when it comes time to apply it. Also, my current boot is still fragile. The initrd I have runs out of space, so I have to delete a file or two to get the boot to finish. That may be unrelated to the other genkernel and/or udev vs. mdev issues, but that's my current situation. I'm living with it, although now I've forgotten which files I delete, so it'll be an adventure next time I have to boot. I think the error messages tell me which directory I have to create to finish the boot successfully, though, and once I delete a couple of unnecessary files, there is space for the required new directory and whatever it needs to contain.
Created attachment 97328 [details] genkernel.log Compiled after etc-update, still fails. Here's the log.
FYI, sys-kernel/genkernel-3.4.1 did the trick (or possibly it was just running over genkernel.conf). It works now :-) Feel free to close the bug...
Sicne you're the original bug reporter, I'm closing. =]
