IOMD development / issues

Thanks Jeffrey.

I’ve removed all of the podules, apart from the IDE so it can boot and softload. The bad news is that it does appear to be the IDE podule that’s causing this Data Abort. I’ve no idea what make the IDE podule is. It’s been many years since I’ve turned the RPC on. ”*Pod.” just describes it as “IDE & CDFS Expansion Card”.

If there are any details that I can provide to help or anything to test, let me know.

Can’t you just plug in the IDE cable direct to the motherboard? (You will lose any extra partitions of course, if they exist.)

Yes I can :) I’m assuming the IDE podule let you have partitions (and two extra IDE connections of course)? I honestly can’t remember, its been so long.

But plugging the HDD in to the onboard IDE does let me softload in to RISC OS 5. Although it doesn’t enter the desktop without a ”*Desktop”. I thought that had been fixed in Tom’s built? I’ll grab the Data Abort code a bit later on.

Is one of the problems associated with softloading, how to get !Boot to work correctly? Because I’m assuming the !Boot on the ROOL website isn’t compatible with RISC OS 3.7? If I use the Acorn universal boot structure I can softload but can’t configure the machine (another Data Abort error occurs).

Yes, there are problems with !Boot when softloading. It should be possible to merge the universal !boot with the ROOL !boot to produce one that works correctly with both OS versions, but I don’t think anyone’s attempted it yet.

Got even further and everything is starting to fall in to place:

1) Remove IDE podule and plug HDD in to onboard IDE and use ADFS as the filesystem. 2) Remove existing !Boot and replace with RO5 !Boot from ROOL site. 3) Copy over !System from Iyonix website. 4) Boot RPC and run Softload manually. 5) Machine boots in to RO5 and runs !Boot sequence as expected.

But I still can’t configure the machine. When I run !Boot it complains that the ProgInfo toolbox module is missing.

But the good news is that RO5 is finally softloading on my RPC without any errors during the process :)

Toolbox modules, including ProgInfo, can be downloaded from here . I presume they’re installed in !System.310.Modules.

I did come up with a solution to the !Boot problem. Rename existing !Boot to !Boot4, new boot to !Boot5, then make !Boot an Obey file containing the following :

Set Boot ADFS::4.$.!Boot5
RMEnsure UtilityModule 5.00 Set Boot ADFS::4.$.!Boot4
Run <Boot>

I’ve run into the podule issue with just about every podule I’ve tried, RO5 is meant to verify that the ROM loader and modules are all 32-bit but this doesn’t seem to work properly. My solution was a tad brute force, modify Podule to disable all podules!

Oh, and the !Boot from the self-extracting disc image, with no additions (ie the !System from the Iyonix site), works fine and can be configured.

Thanks Tom. Making much more sense now. The ProgInfo issue was because it was unplugged. RMReinit and I can now configure the machine.

When it softloads the machine reports 62Mb RAM at the PreDesk stage. Is this to do with how the softload stores the ROM image in memory?

I’ve run into the podule issue with just about every podule I’ve tried, RO5 is meant to verify that the ROM loader and modules are all 32-bit but this doesn’t seem to work properly. My solution was a tad brute force, modify Podule to disable all podules!

Are there any podules in existance that are actually 32-bit? It seems unlikely, because this would be the first system that actually uses them. (I know the Iyonix has a podule slot, but I never heard of it actually being implemented.

On a real IOMD machine, as far as I can see, RO5 would only be softloaded, so having to softload podule support too would not be a major disadvantage. (I can imagine this might be different on a virtual machine)

The Iyonix podule bus is fully functional, although I’m not aware of many cases of podule firmware being updated to be 32-bit compatible. I do remember running 32-bit versions of the Acorn SCSI podule driver when we were getting SCSIFS up and running again (as part of the development work towards USB mass storage). And I believe there was at least one of special-purpose Iyonix-only podules developed, using the faster E-H cycle timings, though they weren’t necessarily marketted to the public.

I did update the Atomwide serial card to work in the Iyonix and have been using it as the connection for 2 CID devices for !CallerID and a Pace modem used for !Voicemail. here

We did a 32 bit compatible podule for use in a RFID tester. You can find it here, (text in Dutch). NEDAP podule It was not Iyonix only, it can also be used in a RISC PC, but at lower speeds.

Hello,

I’m rather new to the risc community,as I bought from Ebay a Riscpc with RO4.02 and Strongarm, Ethernet, Castle USB podule and scsi podule . This two podules are not compatible with each other. I’m following this tread with interest. I’m now familiar with OS4, and the location/structure of the boot directory. However, I tried to install RO5 on that machine, but I never succeed. I find a lot of info in this thread, but for me it’s complicated to follow.

Is there some wiki where I can find the steps to install ROO on my machine, with the pitfalls and limitations ? Like a step/step installation.

Kind regards,

Mark

Hi. The IOMD port of the shared source RISC OS (i.e. the one for the RiscPC and A7000) is very much a work in progress rather than a finished product right now. You’re likely to experience problems with it. Installation should be as simple as installing the softload from here as per the instructions in the archive.

If you have specific problems, feel free to ask away on this thread.

Hello,

I’ve installed the softload. At start-up I can select RO5, but he still boots in RO4.02

Someone got some clues ?

Can I install it on a Riscstation with RO4.03 ?

Mark

I used version a while ago and I had problems, I had to manually load the shared c library from within the softload app. (I ended up loading it in the run file)

...doesn’t seem to work on any version of VirtualAcorn or Redsquirrel I’ve been able to try…

For info, I see that there’s some recent discussion about RO5 on VA over on the VA forum.

...doesn’t seem to work on any version of VirtualAcorn or Redsquirrel I’ve been able to try…

For info, I see that there’s some recent discussion about RO5 on VA over on the VA forum.

But RO5 does work on VA as far as it is possible without disc access

I’ve run into the podule issue with just about every podule I’ve tried, RO5 is meant to verify that the ROM loader and modules are all 32-bit but this doesn’t seem to work properly. My solution was a tad brute force, modify Podule to disable all podules!

This issue is caused by two little mistakes in the Podule module. The module jumps into a data area when it tries to call the dummy loader that replaces 26 bit loaders. Correcting that was enough to make VirtualRPC boot.

It works until more than one interrupt driver is installed on the podule interrupt vector. For all HAL builds apart from the Tungsten the code for interrupt driver chains is excluded. Including that code for generic 32 bit builds made it really work. But I don’t know if that is the way it should be done or if IOMD builds should have a seperate maschine type.

Back on the subject of development, I’ve confirmed that the IOMD port will not work with a 128mb SIMM in socket 2, but will if it’s in socket 1. This means everyone is okay except for RiscPC owners with 256mb RAM, though how many of those actually are there?

Is there any place to post code changes for others to verify and try before they are submitted to ROOL ?

FWIW, how many physical barriers are there to running the IOMD build in 26-bit mode?

Is there any place to post code changes for others to verify and try before they are submitted to ROOL ?

It sounds like your experimentation will be really useful :-) Anyway, I think code@riscosopen.org is what’s needed – sorry if I’m mistaken.

[Edit: That’s for RO5 code changes, of course.]

My Kinetic has 288MB, what would the situation be there?

I doubt RO5 would work with a Kinetic no matter how much RAM…

Being a Kinetic owner myself (although with <288MB), I’d be interested if you have any thoughts on this.

FWIW, how many physical barriers are there to running the IOMD build in 26-bit mode?

Probably quite a lot. While I think you can still make RO5 build as 26-bit, I suspect the amount of code that would now break would be biblical. Plus you’d have to change the logical addresses of everything, which would also undo a lot of the point of running RO5 in the first place.

Even if you could make it run in 26-bit mode, I don’t think it’s a particularly good idea. Having a consistent code model on all machines is much better than having some on 26-bit and some on 32-bit.

Being a Kinetic owner myself, I’d be interested if you have any thoughts on this.

I’d look into this more, except I don’t have a Kinetic. The main problem would be the memory map – where is the extra Kinetic RAM located? On top of that there’s register configuration for the CPLD - for which I don’t think any information is publically available.

Is there any place to post code changes for others to verify and try before they are submitted to ROOL ?

If it’s small then I’d just post the code here. If you want you can email the changes to me (tommowalker@gmail.com) and I can test them on RPCemu and on hardware.

except I don’t have a Kinetic

I’m happy to lend you mine if it’d help, and if you have time for this. I’ll need to hang on to it until I do the new C/C++ tools install from CD-ROM (not yet requested, in fact). After that, I can live without use of the RiscPC for a while. I could arrange to post it to you some time, or you’re welcome to collect it if you’re ever anywhere near Bristol.

That’s very kind of you, thanks. It’s probably best if we continue this via email – can you email me at the above address?