Name this Guru

[melott]

I'm tring to install OS 3.9 on my A2000.
I installed the 3.1 rom.
When I try to install OS 3.9 I get a Guru.
It says 8000003.
What does this guru mean??
I would think an A2000 can run OS 3.9.

Help ??

[mikeymike]

a google search appears to suggest a divide by zero error :-)

Beyond that I don't know.

[Cymric]

No, it's not a divide by zero error. If memory serves me, that is 80000005. The 80000003 is the dreaded 'unaligned address' Guru, indicating on the 68010- that a program has tried to access a program instruction at an odd address (Bad) and on the 68020+ that the stack was pointing at an odd address (VERY Bad). I do not know anything about support of the various OSes on the A2000 (so check with other Guru's ;-)) but here are a few possible causes:

- a defective SIMM
- an unpatched ramlib.handler (can be alleviated with SetPatch, IIRC)
- a too small a stack (can be alleviated with a 'stack' command prior to running a program)
- a very buggy program

Good luck in hunting down the cause!

[Piru]

@Cymric

'unaligned address' Guru, indicating on the 68010- that a program has tried to access a program instruction at an odd address (Bad) and on the 68020+ that the stack was pointing at an odd address (VERY Bad).

Not quite. On 68000 and 68010 this is caused by using WORD or LONG access on odd address or by jumping to odd address or by odd supervisor stackpointer on exception.

On 68020+ this is only caused by instruction prefetch attempt from odd address (jump/branch to odd address) or by odd supervisor stackpointer during exception.

- an unpatched ramlib.handler (can be alleviated with SetPatch, IIRC)

It's no handler and SetPatch doesn't fix it (at least upto AOS 3.5, IIRC OS 3.9 SetPatch does fix it).

MMULib.lha MMULib/Fixes/RamLibFix (or MCPRamLibPatch) fixes this problem.

[melott]

Ummm .........
OS 3.1 installed with no problems.

I tried installing OS 3.9 both with the
accellerator (68030) and without (68000).
Guru both ways.
I have 2 megs chip ram (Dkb MegaChip) and
a Supra 8 meg ram card (6 megs installed).
I've tried with and without the ram card.
I've tried all combinations.
Same results.......Guru

The machine runs fine on OS 3.1.

[mikeymike]

I don't know anything about installing one version of AmigaOS over another, but could you try 3.9 over the top of 3.1?

[Cymric]

Piru wrote:
Not quite. On 68000 and 68010 this is caused by using WORD or LONG access on odd address or by jumping to odd address or by odd supervisor stackpointer on exception.
On 68020+ this is only caused by instruction prefetch attempt from odd address (jump/branch to odd address) or by odd supervisor stackpointer during exception.

Thanks for the input, it's been a while. I must admit you mention things I cannot seem to remember, though. I was always told that the 68020 was indifferent to odd addresses and just used another memory access if it encountered one, save for odd stackpointers, be them user or supervisor. (In other words, SP must be even at all times, (SP) can be arbitrary.) But like I said, it's been a while.

[melott]

>by mikeymike on 2004/6/13 14:46:55

>I don't know anything about installing one version of
>AmigaOS over another, but could you try 3.9 over the top of 3.1?

Thats what I was tring to do.

Thats the way I did it on my A3k and it worked fine.

I tried installing over OS 3.1 on the A2000 and I get
the Guru. I just don't have any idea whats causing it.
It should work, the manual says it can be installed on
a clean HD or over OS 3.1 or OS 3.5.
I made an Emergency Boot disk (880k) on my A3k but
the A2000 won't even boot from that.
The floppy drive is good, no problems with any other disk.

[vic20owner]

I got this error after installing OS2.1 on my A3000 (it was running 2.04 when I got it).  Turned out to be a corrupt (or incompatible) library.  Figure out where it's happening (opening a file dialog for example) and then you'll have an idea whether it's a library or something else.

[melott]

@Vic20owner

Sounds like a place to start.........
It has to be something like that.

[Karlos]

Cymric wrote:

Piru wrote:
Not quite. On 68000 and 68010 this is caused by using WORD or LONG access on odd address or by jumping to odd address or by odd supervisor stackpointer on exception.
On 68020+ this is only caused by instruction prefetch attempt from odd address (jump/branch to odd address) or by odd supervisor stackpointer during exception.

Thanks for the input, it's been a while. I must admit you mention things I cannot seem to remember, though. I was always told that the 68020 was indifferent to odd addresses and just used another memory access if it encountered one, save for odd stackpointers, be them user or supervisor. (In other words, SP must be even at all times, (SP) can be arbitrary.) But like I said, it's been a while.

As I recall, one of the features of the 020 was that it could handle misaligned data access - but only data. Instruction addresses and the stack (which naturally may contain return addresses for code), however, must still be aligned.