From: Frederick W. Umminger (umminger@purina.berkeley.edu)
Date: 04/13/93


From: umminger@purina.berkeley.edu (Frederick W. Umminger)
Subject: Kernel crash: Need help fixing my Ext2fs
Date: 13 Apr 1993 22:55:16 GMT


        During "make install" for taylor-uucp-gamma1.04
I had my first kernel crash. The message was

Unable to handle kernel paging request at address C0840000
Oops: 0000
EIP: 0008 : D0034090D
EFLAGS: 0001010202
fs: 0017
base: 00000000, limit: C0000000
Pid: 3122, process nr:11
8b449f0489d1d3e00644

if this is of interest to anyone. I'd would appreciate
being enlightened as to what it means. Sorry if the
explanation of these messages is documented somewhere that
I should know about.

        I'm using kernel 0.99pre8 on a 486-33 with a 16M swap
partition and ext2fs as the root filesystem. I was running X
at the time, so the system load was fairly high.

        As a result of the crash, my ext2fs is damaged. I
tried to fix it with "e2fsck -r", but this didn't work and
seems to have mangled my lost+found directory. "ls" inside
this directory gives

ext2_readdir: bad directory entry (dev 0303, dir 11): rec_len
is smaller than minimal
offset=36, inode=0, rec_len=0, name_len=0

        I was using efs2ck 0.2c. I believe this is the version
of my entire e2fs. I seem to remember this problem being
discussed recently; do I just need to get a new efs2ck to
repair my filesystem? Is it safe to use in the meantime? The
problems listed by e2fsck were two inodes not used, not counted
in the bitmap, and a large number of blocks in use, no file
uses it. And now a bad directory entry.

        If this crash is due to a kernel bug rather than some
form of stupidity on my part, I will be happy to try to
reproduce it for debugging purposes after I get my filesystem
fixed.

                                Thanks,
                                        Frederick Umminger
                                        umminger@math.berkeley.edu