Hi.
Some days ago i had the silly idea to make a backup. I have a small server
with a IBM-DTLA-307045 harddisk and i wanted to backup some dirs in /home
(which is a reiserfs partition and exported via NFS -- i use the
kernel-server). The backup should be saved on another reiserfs partition on
the same harddisk:
Filesystem Size Used Avail Use% Mounted on
/dev/scsi/host0/bus0/target0/lun0/part2
167M 14M 143M 9% /
/dev/discs/disc1/part7
29G 2.7G 26G 10% /home
/dev/discs/disc1/part5
8.8G 132M 8.6G 2% /backup
I tried to backup not complete /home but about 900MB of it.
The server is an AMD K6-200 on a Gigabyte GA586 board with 128MB of memory, a
Adaptec AHA2940 SCSI-Controller and a Promise PDC20267 IDE-Controller to which
only the IBM harddisk is attachted (DMA activated:
/sbin/hdparm -c1 /dev/discs/disc1/disc
/sbin/hdparm -W 0 /dev/discs/disc1/disc ).
# hdparm -v /dev/discs/disc1/disc
/dev/discs/disc1/disc:
multcount = 0 (off)
I/O support = 1 (32-bit)
unmaskirq = 0 (off)
using_dma = 1 (on)
keepsettings = 0 (off)
nowerr = 0 (off)
readonly = 0 (off)
readahead = 8 (on)
geometry = 23819/16/63, sectors = 90069840, start = 0
When i start the backup, tar reaches about 100MB archive size (sometimes only
50MB, sometime about 150MB) and then the kernel oopses:
Oct 3 19:02:41 mauzi kernel: Unable to handle kernel paging request at virtual address 00002107
Oct 3 19:02:41 mauzi kernel: printing eip:
Oct 3 19:02:41 mauzi kernel: 00002107
Oct 3 19:02:41 mauzi kernel: *pde = 00000000
Oct 3 19:02:41 mauzi kernel: Oops: 0000
Oct 3 19:02:41 mauzi kernel: CPU: 0
Oct 3 19:02:41 mauzi kernel: EIP: 0010:[<00002107>]
Oct 3 19:02:41 mauzi kernel: EFLAGS: 00010287
Oct 3 19:02:41 mauzi kernel: eax: c4c6da20 ebx: c4c6da20 ecx: c4c6da20 edx: c5837000
Oct 3 19:02:41 mauzi kernel: esi: c1160dc0 edi: 00000000 ebp: c1160dc0 esp: c3a8dea8
Oct 3 19:02:41 mauzi kernel: ds: 0018 es: 0018 ss: 0018
Oct 3 19:02:41 mauzi kernel: Process tar (pid: 15124, stackpage=c3a8d000)
Oct 3 19:02:41 mauzi kernel: Stack: c0d81e2c 00001000 c0130152 c1160dc0 00001000 00000001 c1259fe0 c1160dc0
Oct 3 19:02:41 mauzi kernel: c0130647 c1160dc0 00002107 00001000 c1259fe0 c1160dc0 c0d81e2c 00000000
Oct 3 19:02:41 mauzi kernel: c01fe09c 00000000 00000001 c01fdf28 c0d81d80 c1160d80 00000000 c01fdf28
Oct 3 19:02:42 mauzi kernel: Call Trace: [create_empty_buffers+22/76] [block_read_full_page+75/520] [__alloc_pages+61/452] [reiserfs:__insmod_reiserfs_S.text_L116592+34834/116672] [reiserfs:__insmod_reiserfs_S.text_L116592+24324/116672]
Oct 3 19:02:42 mauzi kernel: [do_generic_file_read+810/1220] [generic_file_read+126/404] [file_read_actor+0/96] [sys_read+149/204] [system_call+51/64]
Oct 3 19:02:42 mauzi kernel:
Oct 3 19:02:42 mauzi kernel: Code: Bad EIP value.
I tried Kernel 2.4.10, 2.4.9 and 2.4.6 -- with all 3 version (all three
vanilla from kernel.org) the same error. Ah, /home was exported while making
the backup, but no client was working. Any ideas? (If you need more data,
please ask! I've no experience debugging the kernel.)
P.S. The oops is reproduceable.
-- Until the next mail..., Stefan. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sun Oct 07 2001 - 21:00:29 EST