Hi all,
I'm having an odd problem with one of my disks. This disk has been through a lot
in the last week and a half, including being re-written with dd repeatedly (from
an image originally taken by dd), hexedited, etc. It started in an Ultrasparc,
and now lives in a PC, has a sun disklabel, and has stuff on it that I want from
when it was in the ultra. The ultra was running 2.4.19-rc1, the PC originally had
the same, has since been upgraded to 2.4.22-ac1.
The problem is this: the partition table is recognized, but the individual
partitions (the ones I care about) are zero, that is to say, they contain the
right size of zeros. The disk device itself, at the partition table boundaries,
is not zero, and I can't explain this discrepency. On the disk, there seem to be
correct and valid superblocks at the right places, they just don't exist in the
partition devices.
i.e. read from /dev/discs/disc2/part4 (or 6 or 7) is all zeros
/dev/discs/disc2/part1 (/) is a valid ext3 filesystem, as is part3 and disc.
part2 was swap, there is stuff in there, but I don't care about it.
Of course, the ones I want the information from is 4, 6, and 7.
A related question: If I have the dd image of the disk, shouldn't I be able to
cut it at the right places, put that in a new file, and mount that? i.e. 'mount
-o loop -r filename /mnt', but when I do that, e2fsck says bad magic number in
superblock, however, e2dump -s can read it fine. (none of the other flags to
e2dump works, however.) If I can slice up the disk into files and read those,
that works too, so solving either problem is adequate.
Any suggestions? I appreciate any help.
Thanks!
Aaron
On Wed, Sep 03, 2003 at 06:07:52PM -0600, Aaron Dewell wrote:
> The problem is this: the partition table is recognized, but the individual
> partitions (the ones I care about) are zero, that is to say, they contain the
> right size of zeros. The disk device itself, at the partition table boundaries,
> is not zero, and I can't explain this discrepency. On the disk, there seem to be
> correct and valid superblocks at the right places, they just don't exist in the
> partition devices.
Does the kernel have support for sun partition tables built in?
What are the kernel boot messages about the disk?
What does /proc/partitions say?
What does fdisk -l say?
On Thu, 4 Sep 2003, Andries Brouwer wrote:
> Does the kernel have support for sun partition tables built in?
Yes. Without it, the partition table is not recognized at all.
> What are the kernel boot messages about the disk?
Attached scsi disk sdc at scsi1, channel 0, id 15, lun 0
[...]
SCSI device sdc: 8498506 512-byte hdwr sectors (4351 MB)
/dev/scsi/host1/bus0/target15/lun0: p1 p2 p3 p4 p6 p7
> What does /proc/partitions say?
8 32 4249253 scsi/host1/bus0/target15/lun0/disc 498 30954 62904 7200 0 0 0 0 0 3640 7200
8 33 99696 scsi/host1/bus0/target15/lun0/part1 0 0 0 0 0 0 0 0 0 0 0
8 34 1046808 scsi/host1/bus0/target15/lun0/part2 0 0 0 0 0 0 0 0 0 0 0
8 35 4237080 scsi/host1/bus0/target15/lun0/part3 0 0 0 0 0 0 0 0 0 0 0
8 36 510942 scsi/host1/bus0/target15/lun0/part4 205 12795 26000 2880 0 0 0 0 0 1450 2880
8 38 1021884 scsi/host1/bus0/target15/lun0/part6 149 9267 18832 2100 0 0 0 0 0 1060 2100
8 39 1557750 scsi/host1/bus0/target15/lun0/part7 143 8889 18064 2210 0 0 0 0 0 1120 2210
> What does fdisk -l say?
Disk /dev/discs/disc2/disc (Sun disk label): 134 heads, 62 sectors, 1020 cylinders
Units = cylinders of 8308 * 512 bytes
Device Flag Start End Blocks Id System
/dev/discs/disc2/part1 0 24 99696 83 Linux native
/dev/discs/disc2/part2 u 24 276 1046808 82 Linux swap
/dev/discs/disc2/part3 0 1020 4237080 5 Whole disk
/dev/discs/disc2/part4 276 399 510942 83 Linux native
/dev/discs/disc2/part6 399 645 1021884 83 Linux native
/dev/discs/disc2/part7 645 1020 1557750 83 Linux native
Looks a little fishy between fdisk and /proc/partitions, but I don't
know enough to say what, or if it is weird for sure.
Aaron