3000 with 8000 0004 gurus!

[tuf]

I've got a 3000 that I've been working on for weeks. In a nutshell, it throws 8000 0004 repeatedly when i do a couple of things.

The 3000 has 3.1.4 ROMs, 16MB RAM, SCSI2SD, Ariadne_II network and a GVP Spectrum video card.

1) If I ftp using Roadshow (via shell ftp OR AmiFTP) it'll crash - it crashes a little less if I download to RAM:, but it'll still crash
2) When I switch modes with P96, it crashes regularly trying to switch

Both are totally repeatable and throw the 8000 0004.

I've tried everything I can think of:

I switched the SCSI2SD with a real drive, I replaced the 3000s SCSI chip, I've tried 3.1.4 and 3.2 (w/3.1.4 roms), I've tried different TCPIP stacks, I've tried different filesystems on the SCSI2SD, I've messed with every setting possible with the SCSI2SD and the 3000's setbatt settings. I can even get the error with both the network and video card out of the machine if I repeatedly do a 'addnetinterface ariadne_ii' in the shell.

I'm kind of out of ideas - from what I've read the error has to do with the processor libraries... snoopdos doesn't really tell me much. All I know is it's nearly always the 0004 error. I've scoured the net for weeks and tried everything.

Does anyone have any ideas? Thanks!!

[jbenam]

Hi,

I have the same issue (kinda?) on my A4000D. It's been two years since I noticed it, but back then I though it was an issue of my PCI bridgeboard, instead it looks like it's a wider hardware issue:

https://eab.abime.net/showthread.php?p=1497072

Pretty sure it's a hardware issue in your case as well

If you look at it, it happens when you do Zorro stuff - using RTG, using network.

Of this being a Zorro-related issue I am certain. As it only happens when I use Zorro cards, otherwise my A4000D is rock solid.

Sadly no one has been able to give me any solid leads, so I have started desoldering and testing anything that sits on the Zorro III bus.

One interesting test you could do: check if one of your cards has Force Zorro II mode (it's usually a jumper) and see if the issue happens in Z2 mode. From what I gather, this looks like a Zorro3-only issue.

[tuf]

Originally Posted by jbenam

Hi,

I have the same issue (kinda?) on my A4000D. It's been two years since I noticed it, but back then I though it was an issue of my PCI bridgeboard, instead it looks like it's a wider hardware issue:

Well I read your thread and it doesn't sound good

I hope you can get that worked out without having to rip the whole 4000 down to bits!

Your advice is sound - so far I've only noticed the problem when doing something that has to do with Zorro cards. I found it interesting I could reproduce the error without any cards installed by running some of the Roadshow utils.

I wish I knew someone else with a 3000 close by I could borrow a few parts from

[jbenam]

Originally Posted by tuf

I found it interesting I could reproduce the error without any cards installed by running some of the Roadshow utils.

That's probably because it's trying to access a non-existing card. As far as I can remember, 8000 0004 stands for illegal instruction.

I have found some more battery acid damage under one of the SMD 74 logic I have desoldered. Hopefully it's just a matter of a trace being broken somewhere...

[stevelord]

Originally Posted by tuf

1) If I ftp using Roadshow (via shell ftp OR AmiFTP) it'll crash - it crashes a little less if I download to RAM:, but it'll still crash

I've had this for a couple of years now. I find IBrowse and ATC more stable. For a while I thought it was my A3660 but I recently switched to a Cyberstorm and can get it to guru on command by running roadshow's shell FTP client.

It's been like this since somewhere around the time I upgraded the firmware on my ZZ9000 and some time after I installed the A3660. Unfortunately my change logging wasn't as good as it is now. Everything was rock solid with the picasso and A3640, but something changed.

I also had a full Zorro bus at one point (Repulse, ZZ9000, FastATA and BigRAMPlus). I think the A3660's lack of on-board RAM meant I was reliant on Z3 Ram and that usually was when instability started. Now I have a Cyberstorm and Deneb I'm not using Ethernet and USB on the ZZ9000 and things are a lot smoother but I can still trigger gurus (e.g. Roadshow FTP).

It's possible some software might be a bit sketch, there could be an issue with the FTP, it could be interrupts or possibly a buster bug. I'm now leaning towards a mix of memory leaks and buster issues, but if you'd have asked me a year ago I'd flit between ZZ9k and A3660.

[mousehouse]

I've had a ton of these errors with different 68040 / 68060 CPU libraries, it's a nightmare. I have a 4000D (stock except for a SCSI2SD) that does this occasionally and then also refuses to run WHDload.

My best resolve is/was to install the OS from scratch - which kinda solved the problem for a while until it comes back.

Seems like some sort of corruption happening, but I'm clueless as to where, when and how.

[brett71]

Originally Posted by tuf

I've got a 3000 that I've been working on for weeks. In a nutshell, it throws 8000 0004 repeatedly when i do a couple of things.

The 3000 has 3.1.4 ROMs, 16MB RAM, SCSI2SD, Ariadne_II network and a GVP Spectrum video card.

1) If I ftp using Roadshow (via shell ftp OR AmiFTP) it'll crash - it crashes a little less if I download to RAM:, but it'll still crash
2) When I switch modes with P96, it crashes regularly trying to switch

Both are totally repeatable and throw the 8000 0004.

I've tried everything I can think of:

I switched the SCSI2SD with a real drive, I replaced the 3000s SCSI chip, I've tried 3.1.4 and 3.2 (w/3.1.4 roms), I've tried different TCPIP stacks, I've tried different filesystems on the SCSI2SD, I've messed with every setting possible with the SCSI2SD and the 3000's setbatt settings. I can even get the error with both the network and video card out of the machine if I repeatedly do a 'addnetinterface ariadne_ii' in the shell.

I'm kind of out of ideas - from what I've read the error has to do with the processor libraries... snoopdos doesn't really tell me much. All I know is it's nearly always the 0004 error. I've scoured the net for weeks and tried everything.

Does anyone have any ideas? Thanks!!

Just a random thought and question.....what revision Super Buster are you using in the 3000?

[tuf]

Originally Posted by stevelord

I've had this for a couple of years now. I find IBrowse and ATC more stable. For a while I thought it was my A3660 but I recently switched to a Cyberstorm and can get it to guru on command by running roadshow's shell FTP client
....

It's possible some software might be a bit sketch, there could be an issue with the FTP, it could be interrupts or possibly a buster bug. I'm now leaning towards a mix of memory leaks and buster issues, but if you'd have asked me a year ago I'd flit between ZZ9k and A3660.

Thanks for the info - and interesting you had problems with the RS FTP client. I too thought the culprit must be memory, but I replaced the SIMMS (I have a SIMM converter in there), tried it with nothing but chip memory, etc.

For the Buster mine says xxxxxxx-11, so I assume this is the 'good' Buster.

Thanks!

[tuf]

Originally Posted by mousehouse

I've had a ton of these errors with different 68040 / 68060 CPU libraries, it's a nightmare. I have a 4000D (stock except for a SCSI2SD) that does this occasionally and then also refuses to run WHDload.

My best resolve is/was to install the OS from scratch - which kinda solved the problem for a while until it comes back.

I'm not an expert with CPU libraries, but I've re-installed the OS for testing probably 20 times by now, with our without the 3.1.4 A3000 modules installed, MMULibs from 3.2, random 030 libs from here and there, no CPU libs etc. I don't know enough about how to actually install and manage the CPU libs for the 3000.

I'm an absolute expert on doing an OS install and installing Roadshow from floppy though. I've done it so often it takes me like 5 minutes tops

[tuf]

Originally Posted by brett71

Just a random thought and question.....what revision Super Buster are you using in the 3000?

It's got a '-11' on the chip - I assume that means v11?

[stevelord]

Originally Posted by tuf

I'm not an expert with CPU libraries, but I've re-installed the OS for testing probably 20 times by now, with our without the 3.1.4 A3000 modules installed, MMULibs from 3.2, random 030 libs from here and there, no CPU libs etc. I don't know enough about how to actually install and manage the CPU libs for the 3000.

I'm an absolute expert on doing an OS install and installing Roadshow from floppy though. I've done it so often it takes me like 5 minutes tops

I feel seen but for the 4000. I've been learning what I can about the internals but if I didn't know better I'd say there was some form of bus contention or freeze mucking up timing. Guaranteed crashes at 1920x1080x16-bit with Ethernet and USB on the ZZ9000 and the FastATA. But run it in 1280x720x8-bit with just Ethernet and FastATA and it was mostly fine. 800x600x16-bit with Ethernet, FastATA and Repulse was fine too.

[tuf]

Originally Posted by stevelord

Guaranteed crashes at 1920x1080x16-bit with Ethernet and USB on the ZZ9000 and the FastATA. But run it in 1280x720x8-bit with just Ethernet and FastATA and it was mostly fine. 800x600x16-bit with Ethernet, FastATA and Repulse was fine too.

I guess it's fairly common for people to have wildly unreliable Amigas. Maybe that's part of the experience!

I was thinking of ordering an X-Surf (or any other network card I can find) and trying that, or maybe I'll try a Plipbox I have around somewhere and see if that helps. I can live without the video card but I really need some sort of networking - especially b/c I don't have any sort of SD card solution for the 3000 to get files on it.

[slaapliedje]

Originally Posted by jbenam

That's probably because it's trying to access a non-existing card. As far as I can remember, 8000 0004 stands for illegal instruction.

I have found some more battery acid damage under one of the SMD 74 logic I have desoldered. Hopefully it's just a matter of a trace being broken somewhere...

I always thought this is a memory or 'ramlib' error, as I used to have these all the time until I got my motherboard recapped.

[jbenam]

Originally Posted by slaapliedje

I always thought this is a memory or 'ramlib' error, as I used to have these all the time until I got my motherboard recapped.

I don't know enough of the Zorro architecture to be sure of what I am saying, but I think that ramlib is used by the system to access the memory-mapped Zorro cards. I hope I didn't say something very stupid

[gdonner]

It might(?) be worth checking your A3000 motherboard RAM. I was having some very annoying crashes with my A3000 until I replaced the ZIP chips, and since then no crashes.