segfault every other boot since baselayout 1.12

Message

UberLord · Post by **UberLord** » Wed Sep 06, 2006 2:59 pm

If that's the line that causes it then there's a good chance that our svn trunk repo (link in prior post) may work as we've refactored that bit of code to the extent of removing it.

Seelenfeuer · Post by **Seelenfeuer** » Thu Sep 07, 2006 8:57 am

okay, ill just replace the function-call, i dont want to run my system-base with software without testing... if it works it may be a good idea to put this snippet in the current baselayout-version and create a new release (12.5 or 12.4-r8?). anyway, ill report back in a hurry...

Regards

edit: the rc-script is totally rewritten, so no way there... maybe i find another solution for the moment, but unless one updates baselayout to svn there is no way atm (its unsure whether svn fixes it!)

Seelenfeuer · Post by **Seelenfeuer** » Thu Sep 07, 2006 9:17 am

One more thing i noticed: it still fails with the retval-thingy, but if rc segfaults each module which fails to be loaded by the modules-initscript (which is loaded before the crash) causes a segfault in /sbin/functions.sh line 181, so it may also be related to this script..... *sigh*

NoControl · Post by **NoControl** » Mon Sep 18, 2006 3:24 am

First of all, I don't post this demanding an answer, I know reiser4 is not supported I'm just posting what I found out (which is mostly more weird stuff) in hopes it helps someone find the exact problem and solution to it.
Second, it seems not everyone using reiser4 + baselayout-1.12 is having this problem, see this thread and my bug report.
Third, the same person who started that thread has also posted on the reiserfs mailinglist: the gory details.
Fourth, only a 'proper' reboot seems to cause this. When rebooting using"/bin/mount -a -o remount,ro &>/dev/null && /sbin/reboot -f", the segfault never occurs.

I don't know how far Seelenfeuer got in the mean time, but I did some testing of my own. (BTW: thanks for the hint about brackets starting a new environment, UberLord

)

With baselayout-1.12.5 installed, in /etc/init.d/modules I changed the last lines of start() (that's around line 118) to become:

Code: Select all

	einfo "in /etc/init.d/modules; only command after this is return 0"
	return 0
}

In /etc/init.d/clock (find out why clock further on) I changed the last lines of start() (that's around line 100) to become:

Code: Select all

	einfo "in /etc/init.d/clock; only command after this is eend ${ret} \"${errstr}\" \"You will need to set the clock yourself\""
	eend ${ret} "${errstr}" "You will need to set the clock yourself"

	einfo "in /etc/init.d/clock; only command after this is return 0"
	return 0
}

Then, in /sbin/rc I changed the lines around line 400 (this is at the end of start_critical_service()) to become:

Code: Select all

		start
		original_return=$?
		einfo "after start, original return is ${original_return}"
		exit $original_return
		)

		return_after_close=$?		
		if [[ $return_after_close == "0" ]] ; then
			splash "svc_started" "${service}" "0"
			return 0
		fi

		eerror "Failed to start /etc/init.d/${service}, return code was ${return_after_close}"

The output I get with these modified scripts is (also available as JPEG for your viewing pleasure):

Code: Select all

[snip]
 * in /etc/init.d/modules; only command after this is return 0
 * after start, original return is 0
/sbin/rc: line 377: 2392 Segmentatie fout [SNIP... here it outputs the entire part between round brackets]
 * Failed to start /etc/init.d/modules, return code was 139
[snip]

If I change the critical service listing in /sbin/rc (line 169) "checkroot modules checkfs localmount clock bootmisc" to exclude "modules" completely, checkfs and localmount do start correctly, without segfault, and it's only after /etc/init.d/clock that the same scenario plays as after /etc/init.d/modules. (Artist's impression of output.)

So /etc/init.d/modules' start() returns 0, it's only after the closing ')' (so when the sub environment shell ends) that the segfault occurs. And even weirder: why don't checkfs and localmount crash, but modules and clock do? When the segfault occurs, it's always with return value 139. Not that I'm not way over my head here, but is there a way to run that subshell in gdb or something similar? Is a debugger even possible at this moment in booting? If so please let me know how...

NoControl · Post by **NoControl** » Mon Sep 18, 2006 4:37 am

UberLord wrote:Interesting. Well, we've recently totally re-written that part of baselayout. It may fix it, it may not. And it's not going to be available for a while yet unfortunately. However, you can get the source via http://sources.gentoo.org/viewcvs.py/baselayout/trunk/
You'll probably have todo a manual install - the 1.12 ebuilds will not work with it.
Simply copy all rc-* and init-* in sbin to /lib/rcscripts/sh, (copy init.linux.sh as init.sh) and the rest to /sbin
mkdir's init.d and tmp in /lib/rcscripts
Copy all the init.d scripts across - net-scripts are in their own folder.

No guarantees it will work - and no support either. Better have a copy of baselayout's distfile around

I tried this and I screwed it up, but no worries

I have a copy of baselayout local and on the network, and a Knoppix CD for emergencies

So far I've realized I forgot to mark the new /sbin/rc as executable (as far as stupid things go...

).

If you ever feel like it, I'd be willing to retry this with some extra instructions. I appreciate the time you've given this considering its low priority.

tbender · Post by **tbender** » Mon Nov 20, 2006 3:59 pm

just to let everybody know:

I have reiser4 and baselayout-1.12.6 and I am experiencing the same problems.

Code: Select all

Portage 2.1.2_rc2 (default-linux/x86/2006.0, gcc-4.1.1, glibc-2.4-r4, 2.6.16-beyond4 i686)
=================================================================
System uname: 2.6.16-beyond4 i686 AMD Duron(tm) Processor
Gentoo Base System version 1.12.6
Last Sync: Sat, 18 Nov 2006 09:30:01 +0000
distcc 2.18.3 i686-pc-linux-gnu (protocols 1 and 2) (default port 3632) [disabled]
ccache version 2.3 [enabled]
dev-java/java-config: 1.3.7, 2.0.30
dev-lang/python:     2.4.3-r4
dev-python/pycrypto: 2.0.1-r5
dev-util/ccache:     2.3
sys-apps/sandbox:    1.2.17
sys-devel/autoconf:  2.13, 2.60
sys-devel/automake:  1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2
sys-devel/binutils:  2.16.1-r3
sys-devel/gcc-config: 1.3.13-r4
sys-devel/libtool:   1.5.22
virtual/os-headers:  2.6.17-r1
ACCEPT_KEYWORDS="x86"
AUTOCLEAN="yes"
CBUILD="i686-pc-linux-gnu"
CFLAGS="-O3 -pipe -march=athlon-xp -fomit-frame-pointer"
CHOST="i686-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config"
CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/revdep-rebuild /etc/terminfo /usr/lib/x11/xkb/"
CXXFLAGS="-O2 -mcpu=i686 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="autoconfig candy ccache distlocks metadata-transfer sandbox sfperms strict"
GENTOO_MIRRORS="http://ftp-stud.fht-esslingen.de/pub/Mirrors/gentoo/ http://mirrors.sec.informatik.tu-darmstadt.de/gentoo/ http://pandemonium.tiscali.de/pub/gentoo/"
LANG="de_DE@euro"
LC_ALL="de_DE@euro"
LINGUAS="de en"
PKGDIR="/usr/portage/packages"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/home/tbender/portage"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="x86 3dnow X aac acpi aim alsa apache2 apm arts asf avi bash-completion bitmap-fonts bootsplash cdr cli cracklib crypt css cups curl divx dllloader dlloader dri dvb dvd dvdr dvdread eds elibc_glibc emboss encode esd firefox foomaticdb fortran gif gpm gstreamer gtk gtk2 iconv icq imap imlib input_devices_evdev input_devices_keyboard input_devices_mouse isdnlog java jpeg kde kdeenablefinal kernel_linux libg++ libwww linguas_de linguas_en lirc lirc_devices_atiusb mad maildir mikmod mmx motif mp3 mpeg musepack musicbrainz mythtv ncurses nls nptl nptlonly nsplugin nvidia ogg opengl oss pam pcre perl php pic png ppds pppd python qt qt3 qt4 quicktime readline reflection samba sasl sdl session slp spell spl sse ssl symllink tcltk tcpd thread threads truetype truetype-fonts type1-fonts udev unicode usb userland_GNU v4l vcd video_cards_nv video_cards_nvidia video_cards_v4l vidix vorbis win32codecs wxwindows xfce xml xorg xv xvid zlib zvbi"
Unset:  CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LDFLAGS, MAKEOPTS, PORTAGE_RSYNC_EXTRA_OPTS

Is it true, that reiser4 will will make it way into 2.6.20-vanilla?

Regards,
Tobias

Seelenfeuer · Post by **Seelenfeuer** » Fri Dec 22, 2006 11:48 pm

Reiser4 will go into the kernel when the devs decide that the code fullfills the qualitystandards, not a single day earlier.

Back to thread: ive recently set up my whole box from scratch and as i had some problems with grub on sata i switched to lilo. Since then this error was gone, BUT.... a few days ago the error popped up again...but i think this time i have another idea, though i dont know how to make any use of it: i think the error was there again after i plugged in a new bluetooth-stick O.o weird i know, but since i did this my usb-hub doesnt work correct anymore, too, even if the stick isnt plugged in, and i just dont want to believe that one event isnt connected to the other. So my next guess would be to have a closer look at udev.... as i updated udev on december the 15th to v103 ill downgrade tomorrow to make sure it wasnt because of this. Oh and btw, the stick gives two errors in dmesg, maybe this is usefull too (i doubt it is, but hey, this shall be solved someday

)

Code: Select all

usb 5-1: new full speed USB device using uhci_hcd and address 2
usb 5-1: device descriptor read/64, error -71
usb 5-1: device descriptor read/64, error -71
usb 5-1: new full speed USB device using uhci_hcd and address 3
usb 5-1: configuration #1 chosen from 1 choice

Regards
Seelenfeuer

PS: another idea, maybe some users who experience this problem could post which modules are loaded, maybe it is linked to one foul module... to keep posts small use something like

Code: Select all

cat /etc/modules.autoload.d/kernel-2.6 | grep -v '#'

(or sed if u know how to do it (i dont

) )
my modules are:

Code: Select all

via-rhine
uinput
ohci-hcd
uhci-hcd
hci_uart
hci_usb
hci_vhci
usblp
usb-storage
dv1394
bluetooth
hidp
l2cap
rfcomm
sco
agpgart
via-agp
nvidia
bttv
saa7134 card=6
video-buf
videodev
tuner
tvaudio
tveeprom
v4l1-compat
v4l2-common
saa7146_vv
dpc7146
joydev
grip

stuorguk · Post by **stuorguk** » Wed Aug 22, 2007 8:59 am

Anyone got a fix for this? My raid1 is not unmounting when I reboot.

Seelenfeuer · Post by **Seelenfeuer** » Wed Aug 22, 2007 10:19 am

WELL......
no ^^ my hd went to nirvana and i switched because of this error back to reiserfs (without any problems

) so you can't expect any further investigation from me in the near future

Regards

CrimsonKing · Post by **CrimsonKing** » Wed Aug 22, 2007 7:16 pm

Ive been looking into the problem a bit, since I've been experiencing it on recent builds of the baselayout 2 alpha.

For me, if I stop services after clock, and manually run checkroot, I will see the segfault.
The script runs through in its entirety, but after it runs it segfaults.
If I simply remount ro and run the script again, it won't segfault and I can resume booting normally.

I hibernate my laptop though, so I don't run into it often, and haven't really sat down to figure it out.