kpicott@alias.UUCP (Socrates) (09/06/90)
I thought I'd post a short summary of what has happened so far with the hard
drive from hell. I'll have more detail after I actually solve the problem,
but if anyone has any ideas that are not here I would greatly appreciate
hearing about them:
(1) Bad connections.
All connections have been checked, rechecked, reseated and tested for
continuity. No problems.
(2) Move bad files into a separate directory.
Unfortunately when things start to go bad the drive has a "death
touch". Any file I access after problems start is immediately
trashed. Text files lose their second block and executables are
not recognizable as such anymore.
(3) Use a bunch of small partitions instead of a couple of big ones.
When I create 6 small partitions, FFS does not validate anything
beyond the second one.
(4) Virus.
I applied KV to everything I have, and nothing shows as bad.
Problems start before any program runs (ie. during the restore
process I do incremental checks and the problem has popped up
halfway into a restore).
(5) MountList problems.
This holds the most promise because a lot of the entries are
not thoroughly understand (by myself and most people responding).
My mountlist entries for the big partition setup are below.
Does anybody who has a 1.3 Enhancer manual see anything amiss?
The disk has 1243 cylinders, 15 heads and 35 BlocksPerTrack.
RES3: Device = hddisk.device
Priority = 10
Unit = 3
Flags = 0
Surfaces = 15
BlocksPerTrack = 35
Reserved = 2
Interleave = 0
LowCyl = 0
HighCyl = 1
Buffers = 1
BufMemType = 0
Globvec = -1
Mask = 0xFFFFFF /* One person had the experience that an
incorrect value in this field would trash
a 5 1/4" drive. He finally decided on
0x7FFF. As I now understand it this
value controls the memory locations used
for data transfer, thus any values that
are less than or equal to the amount of
memory you have is valid. I have a 4M
expansion so this should be okay. */
MaxTransfer = 0xFE00
BootPri = 0
DosType = 0x444F5301
FF1: Device = hddisk.device
FileSystem = L:FastFileSystem
Priority = 10
Unit = 3
Flags = 0
Surfaces = 15
BlocksPerTrack = 35
Reserved = 2
PreAlloc = 0
Interleave = 0
LowCyl = 81
HighCyl = 1240
Buffers = 35
BufMemType = 1
Globvec = -1
Mask = 0xFFFFFF
MaxTransfer = 0xFE00
BootPri = 0
Mount = 1
DosType = 0x444F5301
(6) Old FFS bug which trashed large partitions.
I borrowed FFS 2.0 from a friend of mine and the same problem came
back. Not to mention that my version (1.3.2) is supposed to have
this bug fixed.
(7) Reserved entry set to 0.
As you can see below, this is not the case.
(8) Program accidentally corrupting blocks.
The sequence of corruption is too orderly for this. Plus I get
the problem w/o running anything (see (4)).
(9) Large text files grouped together are percieived by the Amiga (or the
controller) as a single file and this starts trashing.
This is rather interesting. I do have a subdirectory that contains
a lot of doc files (my :man directory of course). I haven't tried
this yet but it seems rather strange that this would do anything bad to
the drive.
(10) Generic hardware problems (bad chips etc...)
I did have the 2090A tested at one time but it was never tested on a
large disk. Possibly the interaction causes some sort of problems.
I'm now searching for a dealer who will let me beg, borrow or steal
a 2091 with HDToolBox to test out this hypothesis.
(11) Bad prep information.
The only thing in prep that I am not absolutely positive is correct
is the maximum cylinder. Prep seems to start with cylinder 0, yet
when I tell it that my drive has 1243 cylinders it asks for
"write pre-comp cylinder" with a default of 1243 (which seems 1 too
large). There is no indication as to what this cylinder is for, and
I have tried smaller values (1239 to 1242).
As you can see I've gone through quite a lot of testing to find out what is
going wrong. Unfortunately none of it has borne any fruit. In fact, now
the disk drive is having trouble formatting partitions after prepping.
(The guru pays a visit after formatting is done and Initialization starts.)
I enjoy the flexibility the Amiga offers, when it's working properly, but I
really don't want to work off floppies for the rest of its life. If you
have any new ideas for what could be going wrong, please let me know. My
original offer of choice of my games still holds.
Thanks to all who have replied so far. If nothing else, I have gained a
better knowledge of how the disk is supposed to be working.
--
Kevin Picott aka Socrates aka kpicott%alias@csri.toronto.edu
Alias Research Inc. 110 Richmond St. E. Toronto, Ontario... M5C 1P1
(416) 362-9181 x336