> I'm also running running Gentoo Linux, it applies to me ;-) Whee! Much rejoice :-). > > I modified linuxrc to copy the device-mapper nodes from the initrd > > filesystem (where dmraid "created" them) to the "static" /dev > I thought about that, but I don't like the "newroot" to be mounted read/write > during initrd-time :( Hm? Please enlighten me. Why is that bad? Just diffed my linuxrc against Gerte's, he does the same thing. Which reminds me, I get a message on every boot up saying that "reiserfs assumes mounted read-only, skipping journal playback". I wonder... (I also get "attempt to access beyond end of device" and "ichxrom_init_one(): Unable to register resource 0xffb00000-0xffffffff - kernel bug?", but that's a whole other discussion. Humn =)). > > Also, I modified Gentoo's /sbin/rc script to save the device-mapper > > nodes before it mounts udev and restore them afterwards (line 184): > I don't like hacking /sbin/rc ;-) ... And I think this stuff is udev's > resposibility, isn't it possible to learn udev to do the trick? Good question. No idea how udev works :-(. > > Works like a charm. > I believe that! :) ... I'm really suprised to see someone got through the same > initrd hell as me ;-) We should form a support group for other unfortunates! (We could use it as an excuse for drinking if nothing else.) > I've wrote a new linuxrc inspired by Gerte's linuxrc, simplified and less > external binaries needed: [script snipped] Neat! Except for... > /bin/mount -t ext3 -o ro $REAL_ROOT /root This one. Won't work for eg. reiserfs users (me). (Think I'll be reinstalling with ext2 anyway, to get noflushd working, but anyhow.)
