Following an upgrade to stable kernel 2.6.11-gentoo-r4 my system became (nearly) unbootable. My system consists of an LVM setup spanning two PATA disks (non-RAID) with the root and /boot filesystems separate (non-LVM). The LVM-managed filesystems were unable to be mounted, failing during startup. When I tried to mount them manually on the command line: # mount /dev/vg0/usr /usr mount: /dev/vg0/usr is not a valid block device The dm-mod kernel module is loaded normally. My kernel and system configs are identical. 2.6.10-gentoo-r6 boots normally and 2.6.11-gentoo-r4 fails badly for unknown reasons. emerge info: Portage 2.0.51.19 (default-linux/x86/2005.0, gcc-3.4.3, glibc-2.3.4.20040808-r1, 2.6.10-gentoo-r6 i686) ================================================================= System uname: 2.6.10-gentoo-r6 i686 AMD Athlon(tm) XP 2600+ Gentoo Base System version 1.4.16 Python: dev-lang/python-2.3.4-r1 [2.3.4 (#1, Feb 12 2005, 15:17:09)] ccache version 2.3 [enabled] dev-lang/python: 2.3.4-r1 sys-devel/autoconf: 2.59-r6, 2.13 sys-devel/automake: 1.7.9-r1, 1.8.5-r3, 1.5, 1.4_p6, 1.6.3, 1.9.4 sys-devel/binutils: 2.15.92.0.2-r1 sys-devel/libtool: 1.5.10-r4 virtual/os-headers: 2.6.8.1-r1, 2.6.8.1-r2 ACCEPT_KEYWORDS="x86" AUTOCLEAN="yes" CFLAGS="-O3 -march=athlon-xp -mfpmath=sse" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3.3/env /usr/kde/3.3/share/config /usr/kde/3.3/shutdown /usr/kde/3/share/config /usr/lib/X11/xkb /usr/lib/mozilla/defaults/pref /usr/share/config /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/terminfo /etc/env.d" CXXFLAGS="-O3 -march=athlon-xp -mfpmath=sse" DISTDIR="/usr/portage/distfiles" FEATURES="autoaddcvs autoconfig ccache distlocks sandbox sfperms userpriv usersandbox" GENTOO_MIRRORS="http://mirrors.acm.cs.rpi.edu/gentoo http://mirrors.tds.net/gentoo http://mirror.clarkson.edu/pub/distributions/gentoo" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="x86 3dnow X aac acpi alsa avi berkdb bitmap-fonts cdr crypt cups curl eds emboss encode faad flac foomaticdb gdbm gif gnome gphoto2 gtk gtk2 hal imagemagick imlib ipv6 jpeg junit libg++ libwww mad mikmod mmap motif mozilla mp3 mpeg ncurses nptl nptlonly offensive oggvorbis opengl pam pdflib perl png ppds python quicktime readline real sdl slang spell sse ssl svga tcpd tiff truetype truetype-fonts type1-fonts xml xml2 xmms xv zlib" Unset: ASFLAGS, CBUILD, CTARGET, LANG, LC_ALL, LDFLAGS
The kernel also breaks nvidia-kernel: http://bugs.gentoo.org/show_bug.cgi?id=85802
...and makes my laptop completely unusable. I'm going back to -r3, for the time being.
B. Keroack : Can you confirm that gentoo-sources-2.6.11-r3 is unaffected? (I don't see any changes which would affect this). Brendan Rankin: Is your problem identical? (can't mount the partition, etc..?) Could either of you please attach dmesg output from a working kernel, and the same for a broken one?
OK, this seems to be a strange conflict between fbspash and device-mapper. When I disable fbspash both 2.6.11-r4 and -r3 boot normally with no LVM errors. When fbsplash is compiled in (regardless of whether it's actually being used) the LVM errors occur. In addition, I noticed this error printed to the console during startup when booting the buggy kernels. It doesn't show up in dmesg/syslog so I missed it earlier: device-mapper ioctl cmd 0 failed: Invalid argument
Created attachment 55126 [details] dmesg output, 2.6.11-gentoo-r4 w/ fbsplash (buggy)
Created attachment 55127 [details] dmesg output, 2.6.11-gentoo-r3 w/ fbsplash (buggy)
Created attachment 55128 [details] dmesg output, 2.6.10-gentoo-r6 (working)
Created attachment 55129 [details] dmesg output, 2.6.11-gentoo-r4 (no fbsplash, working)
Ugh, think I know what this might be (ioctl numbers...)
I think this is a problem with your initrd: http://dev.gentoo.org/~spock/projects/gensplash/troubleshooting.php
Hmmm, so it probably seems from reading that document. The fix looks like far more trouble than fbsplash is worth, since I was only just experimenting with it anyway. Is there any way to put an ewarn in about the combination of LVM and fbsplash, at least until splashutils starts emitting workable initrds?
Michal, if you can do anything about this, please do so :) If not, it can be closed I guess.
Was this kernel built with genkernel? Are you using the initramfs image generated by splash_geninitramfs as an initrd or have you compiled it into the kernel? (this question is relevant only if you're not using genkernel). If you aren't using genkernel, then what kind of initrd do you use? (if any?)
These kernels were built manually. I was using a standalone initrd as generated by splash_geninitramfs (splash_geninitramfs -v -g /boot/fbsplash-emergence-1024x768 -r 1024x768 emergence). It should be noted that the LVM errors occur regardless of whether fbsplash and the initrd are actually being used. Simply having fbsplash compiled into the kernel triggers the bug.
Where is you device-mapper control device located? /dev/mapper/control? Is it created as a static device? If yes, this has to be changed. What device filesystem management solution are you using? devfs? udev? This /dev/mapper/control device should be created on-the-fly by the device filesystem manager, otherwise thing will break, for the reasons explained both on my devsite and on http://g-lite.kochen.nl/proj/initrd/
I'm assuming my explanation was good enough. Closing.
