Mail Archives: opendos/2001/04/05/06:34:45
OK, here's a tricky one ...
I was "playing around" with some bits and pieces and had an
unexpected failure of the MB (yep, there was smoke! ;-) ...
This failure corrupted the hard disk, wiping the MBR/PT, boot
sector, FAT1, and half of FAT2. I have rebuilt the MBR/PT, the
boot sector and sort-of rebuilt the FAT, using what remained
of FAT2 and the root directory information (I built cluster chains
from the starting cluster information in the root directory, as if
the disk had no fragmentation and no spare clusters in the
first half of the disk).
The disk is/was DR-DOS 6.0, with just a few necessary files
in the normal disk space, the bulk of the disk was a SSTOR
compressed "drive". Now, I can easily replace any files in
the normal disk space if they aren't exactly right, as they are
(were) just a few O/S files. My main concern therefore is to
ensure that the SSTOR "drive" *is* exactly right ...
Well, that's the background ... now to the specific problem.
With the loss of half the original FAT information, I don't
know what clusters were previously marked as "BAD". The
starting cluster of the SSTOR file (SSTORsomething.SWP,
I think) is known, from the root directory information. The
second half of the cluster chain is known, from the remainder
of FAT2. However, the total number of clusters I have chained
together for the SSTOR file is greater, by 6 clusters, than the
file size information from the root directory indicates. I'm sure
this must be due to clusters, that were previously marked as
"BAD", that are now linked in the FAT chain.
So ... is there any way to determine, from the internal
SSTOR file structure, what clusters (or whatever) have
actually been used by SSTOR for data storage? In other
words, is there anything equivalent to a FAT within the
SSTOR file, that might be used to identify either "BAD"
clusters or expendable clusters?
OK, before you ask :
No, there are no defects listed on the drive sticker, and
no, none of the sectors or clusters presently give read
errors. So I can't identify the "BAD" clusters that way ...
Oh, and please, no smart-alec remarks about backing up!
<bg>
Joe.
- Raw text -