Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!decwrl!labrea!eos!ames!killer!texbell!uhnix1!sugar!ssd
From: ssd@sugar.hackercorp.com (Scott Denham)
Newsgroups: comp.sys.amiga
Subject: Re: 5 1/4" Floppy
Summary: RawRead, Error -4, BugBugBug
Message-ID: <3685@sugar.hackercorp.com>
Date: 3 Apr 89 07:46:10 GMT
References: <192@tlvx.UUCP> <611@jc3b21.UUCP> <194@tlvx.UUCP>
Organization: Sugar Land Unix - Houston
Lines: 50

In article <194@tlvx.UUCP>, sysop@tlvx.UUCP (SysOp) writes:
> 
> What happens (specifically), in case someone is interested in the problem,
> PC Copy will access the 5.25" drive (the light blinks and the head moves), but
> it will not bring up a directory.  If I try to copy an Amiga file to IBM format,
> it says the directory is full.
> 
> I had simply assumed it was an expansion memory thing (like some old software),
> especially since I wasn't the only one with the problem.  Now I'm really curious
> as to what the problem might be.... and I'm out of ideas!  But, it's no big
> deal really; just barely a nuisance (I'm starting to feel guilty about taking
> up net-time now :-).
> 
Well, this sounds a lot like some ongoing problems I have had with the
5 1/4" drive. I also have a homebrew drive that I built for my 1000. 
Upon upgrading to a 2000, the drive quit working in IBM mode, though I
could still mount it and use it quite happily all day long in Amiga mode.
I moved the drive BACK to the old 1000, and it still works great there.
I labored under the assumption that it was my hardware hack for some
months, until someone else on the net reported the same problem. We got
a quick response from Commodore explaining that there was a
newly-discovered bug in trackdisk.device (both 1.2 and 1.3) whereby it
is checking a random place for the length of a RawRead/RawWrite request
and thus may invalidly return a length error (-4). This was explained to
be most likely to happen on a machine with no DF1: (e.g. a 2000 without
a second internal drive). I sadly accepted this explanation and the 
promise of a fix in 1.4. Since then I have come into posession of a 500,
and thought "A-HA - now the 5 1/4" can be on DF1 like it is on the 1000,
and I can go back to using disks to get stuff back and forth instead of
relying on a rather unwieldy chain of modem connections. No such luck -
said drive behaves exactly the same way on the 500 as it does on the
2000. The most obvious symptom, other than the ones you described, is
the posting of "Error -4!" if you attempt to use PCFormat.  If it really
is checking a bad location in memory, I'd guess that the presence/absence
of your Starboard memory moves things around enough to circumvent the
problem. I've found that running a simple program that uses the same
RawRead request packet, I can get it to read every once in a while when
running under a debugger; apparently something gets stuck in memory that
it likes. 
 Hopefully someone from C-A will respond and clarify this again - perhaps
now they know more about the problem or (against all hope) have a 
SetPatch sort of fix to get around it.  Anyway, any other of you 
potential 5 1/4" users out there should be aware of this lest you fritter
away as much time trying to figure it out as I have. On the bright side,
I now probably have the best-shielded, most carefully wired drive gate
hack around - I rebuilt it twice thinking it was my quick-and-dirty 
wire-wrap job or 7' unshielded ribbon cable that was killing it :->
 
    Scott Denham 
     Houston, TX