EXT2 hole problems

Andy Norman (ANorman@datamonitor.com)
Tue, 19 Nov 1996 12:12:26 +0000

I am having problems with one of my servers, it is regularly displaying
messages on the console like:

EXT2-fs error (device 08:06): ext2_readdir: directory #56597 contains a
hole at offset 9216

The only process accessing this partition is afpd, which also keeps dying:

Unable to handle kernel paging request at virtual address f26d0b4c
current->tss.cr3 = 006fc000,
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<00120acf>]
EFLAGS: 00010206
eax: 326d0b30 ebx: 00dbee40 ecx: 00d60064 edx: 00d67894
esi: 00000000 edi: 00ee9000 ebp: bffff030 esp: 0048ff88
ds: 0018 es: 0018 fs: 002b gs: 002b ss: 0018
Process afpd (pid: 15374, process nr: 25, stackpage=0048f000)
Stack: 00000000 00000000 00000000 00d67894 00120b96 00ee9000 00000000 00000000
005bd810 bffff0a8 080145d4 00ee9000 0010a412 080145d4 00000000 00000000
atp_sreq to (5) 256.26:249
bffff0a8 080145d4 bffff030 ffffffda 0000002b 0000002b 00000
02b 0000002b
Call Trace: [<00120b96>] [<0010a412>]
Code: 8b 50 1c 85 d2 74 11 53 8b 44 24 10 50 ff d2 89 c6 83 c4 08

I have unmounted and checked the partition and fscked it but no problems
are found, restarting the machine doesn't help either.

Any suggestions ? (I am going to install Redhat 4 on it later this week and
add some memory to it to see if that helps, I'll probably also go for a
later version of the kernel).

Linux 2.0.0, Slackware 3.0 (largely un-upgraded), netatalk 1.3.3, Adaptec
2940, Fujitsu 1Gb SCSI disk, 2xSMC Ultra, 2x3Com590, Pentium 90, No-name
Intel PCIset m/board, 16Mb RAM (soon to be upped to 64), 32Mb swap.

Andy Norman Datamonitor Group IT Co-ordinator
ANorman@datamonitor.com andy@focus.demon.co.uk
Tel: 0171 625 8548
Fax: 0171 625 5080