Baslinux mailing list

[sindi keesan <keesan AT sdf.lonestar.org>]

My scsi controller on an otherwise perfect curbside-find computer was 
failing and while I was trying to copy the files a few at a time to 
another scsi drive (hoping the drive was the problem) I seem to have 
overwritten part of BL2 with BL3 (issue and inittab, among others), so I 
moved the scsi drive to a Compaq (which is a rather strange beast) and 
copied it to an IDE drive, then copied all of /etc from another BL2 
installation and changed fstab to point to /dev/hdb2.  The copying /etc 
might be irrelevant, the scsi failure really made a mess of the file 
system and I kept running e2fsck which said it fixed things but kept 
putting files into Lost+Found (which I mostly deleted).  I was accessing 
both drives from loop BL3 on an IDE drive during the copy.

Now when I boot I get:

rc.d/rc.S:  Testing filesystem status:  Read-only file system
Parallelizing fsck version 1.18 (11 Nov 1991)
Warning... fsck.ext2 for device /dev/hdb2 exited with signal 11

[I can run BL3 e2fsck on this partition and it is 'clean']

fsck returned error code - REBOOT NOW configuration error - unkown item 
'USERGROUPS_ENAB' (notify administration)

[This string is in login.defs, which I don't even have in BL3 - do I need 
the file at all?]

none (login):

[I can now type root and it logs me in, or if I wait it tells me:]
Login timed out after 60 seconds.
Remounting root device with read-write enabled.
EXTS-fs warning:  mounting unchecked fs, running e2fsck is recommended

Then I get another login prompt:
Basiclinux login:  [root]
configuration error:  unkown item "USERGROUPS_ENAB" (notify administrator)

Then BL2 boots normally and I am online with it.  So it is usable but how 
can I avoid the above warnings and slow boot?


I got 89 hits with google/linux for 'exited with signal 11' and fsck.  One 
says something about a kernel bug triggering a GPF in kernel space.  Dmesg 
only has that one line about mounting unchecked fs.  Someone else had 
'power cycled' while installing Redhat and e2fsck again works when used 
from a rescue disk (I used BL3 loop).  Their solution was to boot with 
'fastboot' option.

Someone else said it has something to do with himem-enabled machines. 
(The problem only occurred after I upped my RAM from 96MB to 256MB by 
switching machines, then put more RAM in the original machine).  They 
suggest reducing RAM from 1GB to 896MB. (I have just upped mine to 256MB).

The kernel was 2.4.9, mine is 2.4.31 but the same kernel works in BL3.

Signal 11 is segfault.  The program accessed a memory location that was 
not assigned and is usually a bug in the program.  But probably not in 
this case.  The problem may only be encountered in extreme situations 
(crazy Compaq Deskpro?).

Someone says to check my RAM, but that is for a case where e2fsck gives a 
Kernel Panic.  And my problem follows the hard drive between computers, 
and BL3 boots correctly on both.

Someone else said to check whether Lost and Found exists.  I have three 
files and three directories in mine.  I had deleted bunches of files from 
here while trying to copy between scsi drives when things kept crashing. 
(I had no other scsi setup to copy on, needed to use this controller).

'if lots of the output of fsck was with relation to orphaned inodes the OP 
may be able to find his/her data under lost+found'.   I have three 
directories and two files in there but less cannot read them.

Would it help to run e2fsck using an alternate superblock?
Someone else trying to fix bad inodes said that did not help.

BL2 also works perfectly when I chroot to it, only thing not working is 
fsck.ext2 during boot.

Is the SCSI controller definitely bad if it had trouble spinning up the 
drive (which works every time in another setup)?  I had some RAM fail 
shortly after I gave up on the controller (it hit 85 degrees in here).

keesan AT sdf.lonestar.org
SDF Public Access UNIX System - http://sdf.lonestar.org