raid5 2.1.90 Oops

Andrea Arcangeli (arcangeli@mbox.queen.it)
Fri, 20 Mar 1998 17:30:49 +0100 (CET)


I tried to run a raid5 over two partition on a ZIP disk and I got an
oops running 'mdrun -p0 /dev/md0':

Unable to handle kernel NULL pointer dereference at virtual address 00000004
current->tss.cr3 = 004be000, %cr3 = 004be000
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<c016b6c8>]
EFLAGS: 00010282
eax: 00000000 ebx: c01cb7f4 ecx: 00000001 edx: 00000801
esi: c01cb7f4 edi: 00000002 ebp: 00000000 esp: c0401ea0
ds: 0018 es: 0018 ss: 0018
Process mdrun (pid: 3724, process nr: 37, stackpage=c0401000)
Stack: 00000000 c01cb7f4 00000002 00000000 00001c90 ffffffff 00001de5 00000000
c016b838 c01cb7f4 00000000 00000018 00000002 00000000 00000010 00000000
c01cb7f4 00000000 00000002 c01cb7f4 c016bdda 00000000 00000004 00040000
Call Trace: [<c016b838>] [<c016bdda>] [<c01199cf>] [<c0119a4e>] [<c011b820>] [<c0119cc9>] [<c016c343>]
[<c016c1c8>] [<c010e613>] [<c012bf4c>] [<c0109916>]
Code: 8b 04 88 24 c0 83 c0 c0 89 43 14 68 00 10 00 00 81 e2 ff ff

root@dragon:~# dmesg |ksymoops /System.map
Using `/System.map' to map addresses to symbols.

>>EIP: c016b6c8 <analyze_one_sb+24/114>
Trace: c016b838 <analyze_sbs+80/2ac>
Trace: c016bdda <do_md_run+1a6/288>
Trace: c01199cf <do_no_page+57/26c>
Trace: c0119a4e <do_no_page+d6/26c>
Trace: c011b820 <filemap_nopage>
Trace: c0119cc9 <handle_mm_fault+e5/198>
Trace: c016c343 <md_ioctl+17b/398>
Trace: c016c343 <md_ioctl+17b/398>
Trace: c010e613 <do_page_fault+143/328>
Trace: c012bf4c <sys_ioctl+13c/18c>
Trace: c0109916 <system_call+3a/40>
Code: c016b6c8 <analyze_one_sb+24/114>
Code: c016b6c8 <analyze_one_sb+24/114> 8b 04 88 movl (%eax,%ecx,4),%eax
Code: c016b6cb <analyze_one_sb+27/114> 24 c0 andb $0xc0,%al
Code: c016b6cd <analyze_one_sb+29/114> 83 c0 c0 addl $0xffffffc0,%eax
Code: c016b6d0 <analyze_one_sb+2c/114> 89 43 14 movl %eax,0x14(%ebx)
Code: c016b6d9 <analyze_one_sb+35/114> 68 00 10 00 00 pushl $0x1000
Code: c016b6de <analyze_one_sb+3a/114> 81 e2 ff ff 00 andl $0x9000ffff,%edx
Code: c016b6ea <analyze_one_sb+46/114> 90 nop
Code: c016b6eb <analyze_one_sb+47/114> 90 nop

This is caused by the ppa driver of the ZIP drive that was not loaded
during the mdrun command (it was rmmodded by kmod since it was not used at
that time).

The md driver don' t increase the ppa module count also after I have mounted
/dev/md0 so I am able to rmmod ppa also when /dev/md0 is mounted and at
the first sync ...

Andrea[s] Arcangeli

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu