Several people have mentioned that /mnt/.init.d fails to mount after a fresh and clean install. Search the forums, in particular these posts: - http://forums.gentoo.org/viewtopic.php?t=5487 (most informative) - http://forums.gentoo.org/viewtopic.php?t=5360 - http://forums.gentoo.org/viewtopic.php?t=197
From http://forums.gentoo.org/viewtopic.php?t=5487: I went through the whole deal setting the system up from stage 1. Still it wouln't mount tmpfs on init.d (or anywhere else either). If I comment the checking in /sbin/rc then the system starts fine (except tmpfs isn't mounted ofcourse). This lead me to think there might be due to some kernel option missing - but I know I have the virtual file system and the dev file system compiled in. However no /dev/shm is created - which should happen according to the documentation with the kernel. That might be an issue. I only have a vague idea of what these systems do and how they behaved so me troubleshooting them is more guessing than anything. /proc mounts fine and proc/filesystems show 'nodev tmpfs' but no 'shm'... I'm not sure if that should show up there either. Then... The solution !!! But there is something weird still going on! I had UNIX98 PTY support enabled, recompiling the kernel without it made the system boot!! Even after having uncommented the checking in /sbin/rc. And after having rebooted the /dev/shm was present (seems it's not when pty support is enabled?!) THE WEIRD PART: It still complains when /etc/init.d/checkroot tries to mount tmpfs at init.d because it has already been done in /sbin/rc !? Seems the startup scripts tries to mount it twice!? Both these scrips are checking bootoptions for notmpfs and proceed to mount tmpfs is the flag is not set. Strange? I think so. I hope somebody has an explanation for me - because I"m curious what's really going on. _________________ Ole Johansen
What gcc was used to compile kernel ? I have never seen this before, but it dont seem like a rc-script problem .. rather kernel/related problem. As for checkfs .. check that you did not go totally bos with hacking the scripts ... it reads: "mount -f ..." Meaning, it does a fake mount just to update /etc/mtab.
I "private message"-d Ole Johansen, though I bet the answer is gcc 2.95.3.
Just thinking about it ... you dont maybe use crypto support in the kernel ?
I don't have this problem (I'm just 'proxy reporting' it), but I would guess that most of the people don't use crypto-sources.
Actually referring to Ole Johansen :) BTW ... ramfs can be used .... just add "gentoo=notmpfs" to the kernel parameters. Maybe this can help until the tmpfs issue is solved. Also, did they try a vanilla 2.4.18 kernel ? Thanks.
Seems this was caused by a stale .config file from another distro. Look at: http://forums.gentoo.org/viewtopic.php?p=29962#29962 for more info.