> > is it possible to fix partition 26 from its saved partition table,
> > in principle?
> > For certain reasons I need to have many partitions eg. on a
> > 80 GB IDE.
> > ...
> > Would this work:
> > * since I have a record of the sizes of partitions 24, 25, 26, 27..,
> > if I recreate partition 26 at the same size as previously, without
> > reformatting, the contents may be saved ?
> > * but then apparently the 'tail' of partition 26 [probably via
> > its table, will (wrongly be marked as the END partition) ] ?
> > Creating partitons by linux's fdisk or cfdisk obviously rounds-off
> > the sector counts for the given MB size. I guess I used the
> > same utility to create the consecutive partitions, so if
> > partn 24, 25 are say 16 MB, then I can check exactly how many
> > sectors they have, and assume that damaged partn 26 has the
> > same number of sectors ?
> > This is what the IDE looks like now:-
> > IDE0#00 76319MB --- (Whole disk)
> > IDE0#01 1004MB 11 * Win 95/98, FAT32
> > ...
> > IDE0#23 125MB 131 |Linux fs
> > IDE0#24 62MB 76 |Native Oberon, Aos
> > IDE0#25 62MB 76 |Native Oberon, Aos
> > IDE0#26 29455MB --- (Free)
> > IDE0#27 1270MB --- |(Free)
> > and I can find the listing of the sizes up to the previous tail of
> > partition-chain, before the problem occured.
> > I'm guessing that if I can just replace the last 'link' [now a tail-link]
> > with a link to the next partition. it would be OK ?
> I would recommend that before you do any mucking around, you first copy
> the contents of the missing partition to a file. You should be able to do
> it easily using something like
> dd if=/dev/hda of=backupfile skip=nnnn count=nnnn
That's what it's all about ? The rest is easy ?
I'm guessing that if I keep a safety copy of the start-sector-number of
all my logical-partitions, then I can always restore if the 'chain breaks' ?
Well no, it doesn't 'scan' as per searching each block and makingQuote:> then if something goes awry you can always recreate the partition in any
> old manner and copy the data back. You may want to copy all the partitions
> out this way, perhaps to an external USB drive, then recreate the whole
> mess. I'd also check for bad blocks on the device - things don't disappear
> for no reason.
> You can also use sfdisk to scan the disk for partitions.
heuristic guesses [like some utilities try]; it just 'links through' like
fdisk & cfdisk does ?
OK once the partition boundry's are determined ?Quote:> If I were doing this, I would do the following:
> 1) sfdisk to scan for partitions
> 2) dd to copy partitions to files - at least for missing partitions
My 'man badblocks' want to refer to partitions.Quote:> 3) badblocks to check the drive
I need to refer to a range of blocks.
It seems that only dd can do this ?
I've never understood "loop", I'll have to seek some docu. explanation.Quote:> 4) attempt to recreate missing partitions - if the process doesn't work
> quite right, then simply repartition to approximate values and one
> partition at a time, mount via loop the backup file and copy the data back.
== Chris Glur.