Sorry to jump on you Radek but it makes me angry when people do not read the manual or notes supplied, e.g. (taken from the top of DaniS506.DOC
Daniela's S506 ADD - Release 1.8.5
----------------------------------
NAME
DaniS506.ADD - replacement for IBM1S506.ADD
SYNOPSIS
EIDE driver with support for Intel, VIA, ALi, SiS, CMD, Promise, Cyrix,
HighPoint, AMD, Artop, SMSC, ServerWorks, Opti, Nvidia, National Semiconduc-
tor, SiliconImage, ITE, Adaptec, ATI, Broadcom, JMicron, Marvell chipsets,
and generic busmaster support for all SFF-8038i compliant chips.
This covers virtually *any* EIDE controller built/sold in the past few
years and even some just about to be released in the near future. Don't be
confused by the generic chip names below! If you are uncertain about your
actual controller to be supported, run the DumpIDE utility - it will tell
you.
If we get wrong information here it will live on and cause more problems down the years.
So far it appears the problem revolves about being able to clear the 'dirty' bit on the drives and so make them visible to OS/2. No amount of copying , cloning, or anything that does a sector-by-sector copy is going to clear that 'dirty' flag and make them readable. The only thing it will do is provide a backup should damage happen in some way.
Mounting the drives on a working OS/2 system will first of all allow chkdsk /f to be run which may, or may not, be all that is needed. It also allows DFSee to run in native mode which will raise the possibility of a successful recovery of the contents of the disks.