To:	opendos AT delorie DOT com
X-Comment-To:	Rob McGee <i812 AT iname DOT com>
Cc:	i812 AT iname DOT com
References:	<20010314002926 DOT B3024 AT sl7>
Message-Id:	<2.07b7.14Z6R.GA6TFH@belous.munic.msk.su>
From:	"Arkady V.Belousov" <ark AT belous DOT munic DOT msk DOT su>
Date:	Wed, 14 Mar 2001 15:30:53 +0300 (MSK)
Organization:	Locus
X-Mailer:	dMail [Demos Mail for DOS v2.07b7]
Subject:	Re: [ot] disk geometry mismatch?
Lines:	69
MIME-Version:	1.0
Reply-To:	opendos AT delorie DOT com

X-Comment-To: Rob McGee

Hi!

14-ํมา-2001 00:29 i812 AT iname DOT com (Rob McGee) wrote to opendos AT delorie DOT com:

RM> could no longer access the hard drive. It's a WDC AC35100L which is
RM> C/H/S 10672/15/63, LBA'ed to 627/255/63. Identify drive from a Linux
___________________^^____________________^^

RM> I had a CD of Norton Util. 95 v. 2 (FAT32-aware) and tried ndd.exe. It
RM> offered to try to recover a partition of 812MB. I declined the offer and
RM> tried diskedit.exe instead. The first 63 sectors were all 0's. (63 spt,
RM> of course. Is that normal?)

     Yes. FDISKs usually start first partition from next free head, not
right after MBR (from the next sector):

______________O\_/_________________________________\_/O______________
No  Drive  Start  End   MB   Status  Type
 1    C:      0   203  100,3    A    PRI DOS
 2    --    204  1022  403,1    N    EXT DOS
_____________________________________________________________________
              O/~\                                 /~\O

______________O\_/_________________________________\_/O______________
BIGDOS ไม   1       0      1   15     203     63          63    205569
_______________^_______________________________________________^^
EXTEND ๎ลิ  0     204      1   15    1022     63      205632    825552
_____________________________________________________________________
              O/~\                                 /~\O

This picture from diskedit better to slightly modify by swapping heads and
tracks:

  0 1 1    203 15 63          63    205569
____^
204 0 1   1022 15 63      205632    825552

This disk pointed in BIOS setup as 1023/16/63.

RM> The 64th sector showed what looked to my untrained eye like some
RM> filesystem information. There was a string "FAT32" and the volume label
RM> as it was before the crash.

     There most probaly boot sector of first partition.

RM> Finally this afternoon I came up with the hypothesis of a disk geometry
RM> mismatch. That some parameters had been manually entered in the dead
RM> motherboard's CMOS settings, and they weren't quite correct, but close
RM> enough that the drive went along with it.
RM> Does this hypothesis sound right?

     Most probably. I already meet such case - if point disk, partitioned in
CHS mode, as LBA disk in BIOS setup, then it look worngly.

RM> If so is there anywhere I might be
RM> able to find a clue about the disk geometry parameters used?

     I'm affraid this may be done only manually. But you may try to decide
disk geometry from partitions parameters (look above) or try special disk
recovery utility like Tiramisu. You may also try to talk with Brian
Refsnyder - author of FreeFDISK.

RM> I'm afraid
RM> that my "fdisk /mbr" might have rendered the thing unbootable though.

     Most probably no. /MBR option brings FDISK to refresh MBR _code_ and
should remain partition table itself untouched.

- Raw text -