home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
FishMarket 1.0
/
FishMarket v1.0.iso
/
fishies
/
201-225
/
disk_224
/
xebec
/
xebec.readme
< prev
next >
Wrap
Text File
|
1992-05-06
|
5KB
|
133 lines
PAUL KIENITZ'S STUFF FOR SUCKERS WITH XEBEC HARD DISKS
I have an excuse for why I got mine: it was part of a used system at a good
price. What's your excuse?
If you have a Xebec 9720H or other Xebec hard disk on your Amiga, these two
programs may be helpful to you. The first one, called Scum (short for scsi
mount) is for getting the Fast File System to work on the Xebec. The problem
is, even with the best Mountlist entry I can come up with, Mount will not
successfully mount a FFS partition on a Xebec unless first you either mount
another partition with the Slow File System, or run Scum instead. Scum is
an extremely simple program, and I have no idea either why it works or why
it's necessary. Note that the Xebec provided SCSIMOUNT program can't use
FFS. Using Scum is several seconds faster than mounting another partition
from floppy disk, so even if you have such a partition (as I do; I use it as
a sort of experimental animal) Scum is still worth having. It returns error
code 10 if it fails (if it can't find devs:scsi.device).
EVERY ONCE IN A WHILE, something seems to hang up in my startup sequence. I
don't know whether it's Scum or Mount or what, but it seems safe to assume
that this program, which never should have been necessary in the first place,
is not a perfect solution. If your startup sequence freezes up when it's
just getting going, simply give it the old triple finger to start over.
Here is a working Mountlist entry for a Xebec 9720H with no partitions:
DH0: Device = scsi.device
Unit = 0
Flags = 0
Surfaces = 4
BlocksPerTrack = 17
Priority = 10
Reserved = 1
LowCyl = 0
HighCyl = 614 /* Don't believe the manual when it says 611 */
Buffers = 30 /* or whatever */
BufMemType = 3
DosType = 0x444F5301
StackSize = 4000
GlobVec = -1
FileSystem = L:FastFileSystem
Mount = 1
#
And here is a pair of mountlist entries for one with a large FFS partition
and a minimum size slow partition:
DH0: Device = scsi.device
Unit = 0
Flags = 0
Surfaces = 4
BlocksPerTrack = 17
Priority = 10
Reserved = 1
LowCyl = 2 /* number of cylinders used by slow partition */
HighCyl = 614
Buffers = 30 /* whatever */
BufMemType = 3
DosType = 0x444F5301
StackSize = 4000
GlobVec = -1
FileSystem = L:FastFileSystem
Mount = 1
#
SLOW: Device = scsi.device
Unit = 0
Flags = 0
Surfaces = 4
BlocksPerTrack = 17
Priority = 10
Reserved = 1
LowCyl = 0
HighCyl = 1 /* last cylinder before first FFS cylinder */
Buffers = 4
BufMemType = 3
DosType = 0x444F5300
StackSize = 1000
Mount = 1
#
(Actually, I put my own slow partition at the high cylinder end.)
Typical use in a startup-sequence:
setpatch ; if you've got release 1.3
scum
mount dh0:
movesys dh0: ; reassigns SYS:, C:, DEVS:, etc -- see MoveSys
; probably in the CLImax directory on this disk
If you want to create some new arrangement of partitions, you don't need
(in fact you should certainly avoid) the Xebec MAKEPART program. Just get
the mountlist entries right and then FORMAT QUICK should do ya just fine.
If you used MAKEPART before, I think you'll have to reformat all partitions
even if none of them are different, because Xebec puts the partition
information on the disk and Mount doesn't. You don't need HDPARMS any more.
----------------
The other program is a replacement for the bloated and cumbersone Park
program provided by Xebec. It assumes the disk is unit zero; this can be
easily changed by anyone with a C compiler (Aztec, hopefully) in the
unlikely event that you have some other device on the SCSI chain before it.
You just type Parx or double click the picnic table icon and the disk parks.
It's one twentieth the size of their Park program. This leaves you with no
need for any Xebec software except scsi.device, and HDINST if you ever need
a low-level SCSI format (I think that's what you use to map out bad sectors).
All in all, I'm extremely unimpressed with Xebec "the zero defect company".
Though I must admit that the hardware has never had trouble (why should it?
it's mature, well-understood (i.e. obsolete (85ms)) technology) the software
is dog meat. I once started disassembling the device driver (while figuring
out how to write Parx) and within minutes found two busy-wait loops, as well
as ridiculous code which essentially replaced the Exec message passing
mechanism with a strange ring-queue and signal arrangement with no discern-
ible improvement over regular messages. And I doubt that it's really
necessary for the disk unit to weigh eleven pounds and take up as much desk
space as the keyboard does.
This stuff is public domain; any feedback is appreciated. Like, if you want
a Parx not restricted to unit 0, let me know.
Paul Kienitz
6430 San Pablo ave
Oakland CA 94608
USA
(°¿°) <- Official Seal of Dorkiness certifies that Scum and Parx were written
without the use of ROM Kernel Manuals.