home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Cream of the Crop 3
/
Cream of the Crop 3.iso
/
utility
/
sfs100.zip
/
README.1ST
< prev
next >
Wrap
Text File
|
1994-02-20
|
8KB
|
168 lines
Overview
--------
The SFS 1.0 distribution contains the following files:
readme.txt This file
sfs.sys SFS device driver (386 version)
sfs486.sys SFS device driver (486+ version)
sfs.sig \ PGP digital signature for
sfs486.sig / sfs.sys and sfs486.sys
mksfs.exe \
mountsfs.exe | SFS support
chsfs.exe | software
adminsfs.exe [1] /
mksfs.sig \
mountsfs.sig | PGP digital signatures
chsfs.sig | for support software
adminsfs.sig [1] /
winsfs.exe \
winsfs.pif | SFS for Windows
winsfs.sig /
sfs1.doc \
sfs2.doc | SFS documentation
sfs3.doc /
sfs.ps Large DOT "SFS Encrypted Data" warning sign
sfs_disk.ps [2] Many smaller versions of the above for floppy disks
[1] Currently available only to a few alpha-testers as
some details are still being worked out
[2] Some laser printers won't print this file, possibly
due to the amount of detail contained in it.
Getting Started
---------------
Before you start, you should read the docs (or at least the first few sections
which outline how to use the software). I know that all software says this,
but with SFS it's especially important. You should read the section on
incompatibilities in the docs before you start. Among other things this will
show you how to run the built-in self-tests to check disk access and make sure
SFS won't run into any problems with disk I/O. I recommend *very strongly*
that you run mksfs with the -c option before you create an encrypted SFS
partition on your hard drive.
In addition, before you rush into using SFS, if you plan to use it on a hard
drive I would recommend backing up anything which you regard as valuable.
People all over the world have been using it for some months without problems,
but I really can't 100% guarantee that it'll work perfectly on your system -
there are just too many strange pieces of hardware and software out there for
that.
OK, now you can go ahead and install it. I'll wait here.
Possible Problems
-----------------
Finished? OK.... here are some (rather obscure) possible problems which have
been reported to date:
There has been a report of problems with a 1.2 GB SCSI drive using an Adaptec
controller with the sector remapping for > 1 Gb drives enabled and running in
conjunction with LILO and Linux 0.99 patchlevel 14 or above. After running
mksfs, LILO reported some sort of error. The problem was corrected by
re-installing LILO. Despite various tests, no clue as to what's happening here
has appeared. A hex dump of the sectors after the SFS partition indicate that
the following partition was untouched by mksfs, so it may be some interaction
between LILO and the partition table (maybe LILO writes to the partition table
like the OS/2 boot manager does, and doesn't expect to find an SFS partition.
Who knows...).
There seems to be some bizarre interaction with version 6.30 of the Logitech
Mouse driver and SFS. In particular, mounting and unmounting SFS volumes can
disable the mouse driver and even cause it to not recognise the mouse any more.
With nothing but the mouse driver loaded, simply running mksfs or mountsfs is
enough to lose the mouse under DOS. These problems have occurred when using a
serial mouse and QEMM 7.03 with stealth enabled, but don't seem to occur with a
bus mouse and no QEMM. There are no interrupt conflicts, and neither the mouse
driver nor SFS are overwriting each others data or code areas. The mouse can
be reenabled simply by running Windows. In addition, running mountsfs in a
secondary shell spawned from Windows (not just a secondary shell under DOS, but
specifically one where Windows is acting as an intermediary) seems to isolate
the mouse from mountsfs, and the mouse isn't deactivated. If anyone has any
ideas on this, I'd like to hear from them.
There has been a report of some sort of odd problem with an older version of
QEMM when EMS suport is turned on. The driver reports an internal consistency
check failure indicating that something has corrupted it, and the problem goes
away without EMS support enabled. Possibly a newer version of QEMM will fix
this, or it may be that the problem doesn't involve QEMM but is due to some
other software behaving unusually while under the influence of QEMM with EMS.
If the FAT on an encrypted disk is edited to create artificial bad sectors, and
the volume subsequently fixed with Norton Disk Doctor, and NDD finds a problem
on the disk, the SFS volume will be unmounted after NDD has finished fixing the
problem. The only way to do this is with a write to the control channel on the
SFS driver using a special packet format, so how NDD manages to perform this
unmount is a puzzle.
SFS has been tested with (among other things) Smartdrv 4.0, Windoze 3.1, QEMM
7.0 (with Stealth), SHARE, Stacker 3.1, and assorted disk utilities such as
Norton DiskEdit, Norton Disk Doctor, Microsoft and Norton Antivirus (with
innoculation), Norton Speed Disk, and various others, and has so far not caused
any problems. If anyone tests it with any other system software, please let me
know.
If you need to contact me, you can mail me at pgut1@cs.aukuni.ac.nz. If you
need to contact me really badly, try calling +64 9 426-5097 from 9am-2am NZT
(that's about GMT+13). I'll leave you to fight with the modem for possession
of the phone line.
Finally, if anyone ever writes about SFS anywhere (apart from on the internet),
I'd appreciate hearing about it, since it's annoying sometimes getting
third-hand references to articles I'll never be able to find which were
published months beforehand.
Peter.
Revision levels
---------------
0.90 First public beta
0.97 Second public beta
Explict support for DRDOS large partitions (which differ from MSDOS
large partitions).
More internal consistency checks in mksfs, mountsfs, chsfs.
Major rewrite of SFS library internals.
Use of XMS buffering or more conservative write strategy to fix occasional
copy-on-write buffering problem in newer versions of DOS.
Support for multiple mounted SFS volumes
Support for batch mode operation for WinSFS
WinSFS - Windows front-end for SFS
Fixed (hopefully) problem with race condition of hotkey or timed unmount
during disk access
Improved overwriting of SFS driver data on unmount
Better self-checking in the SFS driver during operation
1.00 Public release
Fixed some problems in the technical section of the docs
Added the Quick-start section to the docs
Precomputed the encryption key in SFS.SYS and made assorted other small
changes. This makes the driver run approximately 1.7 times faster, and
gives a decrease of about 1K in the overall size.
Even better data overwriting based on an analysis of disk data encoding
schemes.
Added a quick introduction to encryption issues section to the docs.
mksfs now asks for a new disk when encrypting floppies, like the DOS
format program.
Footnote
--------
Conversation (via email) with a beta-tester:
Him: The disk mount failed
Me : How
Him: It just gives garbage
Me : What sort of garbage, precisely
Him: Well, it mounts OK but then I get garbage
Me : I'm writing the definitive work on garbage, so I want you to tell me
everything you experience. And remember, this is for posterity, so be
precise.