Softloading 5.15 on risc PC???

Ok so thought I should try this… as my risc os 4 roms chips died so stuck on 3.7

I downloaded the softload and followed instructions, then downloaded 5.15 rom renamed as per softload readme and restarted machine.

I get message asking if I want to softload to which I chose Y.

next line is an error message saying shared C library is out of date (error number &800E91)

Made sure I had latest C lib in my system and tried again but still the same error. I know there roms where supposed to be a little rough but would at least like to get it to boot.

Any help greatly appreciated.

John

The 5.15 IOMD ROM is very much a work in progress. Demonstrated at Wakefield, it showed various missing features:

• Display problems (shifted a few columns horizontally out of place, but the mouse isn’t, so the two don’t line up)
• No networking
• No ADFS filing system (so no hard disc access)
• Floating point won’t work as the wrong FPEmulator module is present
• We were only able to make it work at all on ARM 7 class processors; StrongARM failed completely

In short, it’s not just rough, it’s unusable! That’s why we’re only talking about the Iyonix 5.14 official release in our recent press release material – the Risc PC / A7000 ROM is a fair distance away from being usable yet.

When we eventually get a stable ROM the next step will be to figure out how to make it work with the hundreds of different, customised, RISC OS 3, 4 and 6-style !Boot structures out there – it’s likely that the best we could do is just release an approved !Boot structure and leave it up to end users to use this, then merge in whatever customisations they wanted to keep from their old boot system.

Actually I’m quite surprised you got as far as an error about the C library if you’re using a StrongARM – what processor is in your RIscPC?

Strongarm

Thaks for reply. thought it was getting to WIMP even though unstable.

Just wanted a play as I am stuck with 3.7. Will keep my eyes open for developments.

Think an approved !boot sounds like a good plan.

John

Well, it sounds like you got it running better than we did :-)

Could it have anything to do with the version of risc os in ROM on my RPC compared to yours?

I have Risc OS 3.7, SA233, 128Mb, 2Mb VRAM…

Or does it load so little of RISC OS before softload it makes no diffrence?

Anyway looking forward to developments. I just can’t afford at moment to replace my ROM’s with 4.02 or 4.39 (wich is going to be about £50) then pay another £50 for Risc OS 6. So any other route to somthing better than 3.7 is going to be most welcome.

So if it gets as far as an error about CLib being out of date (assuming it is working otherwise) how would I fix this? Just thought I may as well keep trying if it is getting further than you saw.

Think I may have an ARM 600 or 700 somewhere so may try that as well.

John

Since the ROM is not finished and not even intended to work properly, there’s no need to spend time testing different processor architectures – that’ll get done in due course. That said, if/when a final version of 5.15 IOMD gets released, it would be useful to have a few people around with things like ARM 6 and 7 cards to check it works OK on those (unless we know it won’t because the code requires CPU features absent in the older architectures).

Presently I wouldn’t worry about the CLib problem. The ROM doesn’t even have a working ADFS or working floating point emulator yet so you would have no hard disc access and even ROM applications like !Draw wouldn’t launch.

No worries.

Was just doing it to see if it works and fill and evening with somthing that interested me.

Will wait till some more progress is made

Thanks Again

John

PS Yeah think I have both ARM6 and ARM7 cards so once somthing is working I will happilly post some results on diffrent cards.

There is a long list of known issues:

• The softload tool needs you to modify your boot sequence to RMEnsure and RMLoad the CLib from your system in order to run – our supplied package missed that fact

• The dev ROM will reset your CMOS every time it boots

• If you have NVRAM, it will probably fail to boot – you’ll need to remove it

• You will need to reconfigure your mouse type because it defaults to the wrong value

• We accidentally included the IYONIX FPEmulator in the build so it will randomly crash a lot

• As Andrew said, it doesn’t have ADFS or networking support yet

So, best to steer very clear unless you really like pain.

LOL,

I like pain…

Not going to do anything with it. Just filling a spare hour or two having a play at softloading. Thanks for all responses. Will give it another go but will wait for later version before worrying too much about it

John

next line is an error message saying shared C library is out of date (error number &800E91)

So finally we’ve found something that StubsG might be good for. ;)

it’s likely that the best we could do is just release an approved !Boot structure and leave it up to end users to use this, then merge in whatever customisations they wanted to keep from their old boot system.

I may have something that could help with this: On my website at http://www4.webng.com/resurgam/ is a program called MulBoot. I originally wrote this so I could easily run both RO4 and RO6 on my machine. MulBoot both swaps the !Boot sequences and maintains the seperate CMOS settings that both OSes require.

It occurs to me that a copy of MulBoot supplied with the RO5 boot sequence & softloader within it, would make it simple to try out RO5 on a RPC while maintaining the users original !Boot sequence.

I’ve uploaded a new A7000/RiscPC softload archive onto our site. This should remove a lot of the problem ARM instructions which were too new for the hardware, including having the correct FPEmulator.

I’ve also made some attempt to improve the softload process a bit, although it’s untested because I don’t have any pre-IYONIX hardware myself. Let me know how you get on with it.

Note: it is still VERY work in progress – no hard drive, no network, etc, etc.

Hello everyone!

I was just thinking that in case your only rom chips do break, you’ll be left with an unusable system.

So, would it be possible to write the 5.x rom into a couple of eproms,put them into an actual machine and make it work?

I’ve seen in different places that the actual rom in the rom chips is in a different format than what an emulator would use.

Do you know also what type of eprom would be compatible with an A7000+? It would be nice to test the roms the hard way… XD

The situation is that at the moment, the 5.xx IOMD ROM images assume that another RISC OS ROM has already booted on the machine (and in particular, that it has already set up the RAM controller). Since a softloaded ROM is in reality executing from RAM, one thing it mustn’t do is reset the RAM controller!

Traditionally, ROMs for IOMD machines were intelligent enough to spot whether they were executing from ROM or RAM and only initialise the RAM controller in the former case. Since the IOMD HAL has only been used in a softload capacity to date, nobody has (yet) reinstated the RAM controller initialisation code.

So the answer is: no, not at the moment, though it is possible that this might work in future.

This may be a silly question (noting that I haven’t been able to open the main source archive under Linux), but…

The IOMD RO 5.15 lacks supports for VRAM, RAM other than 8MB, ADFS, etc. I was wondering – is there access anywhere to either the original RO 3.7 code for these features, or detailed specifications regarding how these should be set-up? Without either of these, it would be difficult to make progress on the port!

Are you for real? This is about the third time you’ve berated the fact that you can’t open the main source archive but you still haven’t answered my questions on the subject…

You can’t open the main source archive? Which one? Was this related to the post I answered about the archives at one point being called .tar.bz2 when in fact they were just tar archives?

• What archive are you talking about?
• When did you download it?
• How big is it?
• What command line are you using to extract its contents?

Hello :)

It was the second time I had mentioned that fact I couldn’t open the main source archive (and rather than berating it, I said that I will try to open it on RISC OS as it will probably work on there).

Sorry for missing your first post 3 days ago, I guess it had to do with coming off of nightshift and back on to days :). Anyway, in answer to your questions:

It was on http://www.riscosopen.org/content/downloads/other-zipfiles

And was the CVS source repository (I note it was rebuilt daily, I got it maybe about 2 weeks ago). The extension was .tar.bz2

I was using the Ubuntu GUI to explore the contents, but it wouldn’t open it.

I think my error was assuming that, because the Tungsten/IOMD/Beagleboard sources were .tar.bz2 and were labelled .tar.bz2, then the other sources were also going to be .tar.bz2 if they were labelled as such. I did try renaming the large source archive to .zip, but that didn’t work. I guess my plan of going into RISC OS and setting filetype to ‘Archive’, then loading it into the tar/bz2 dearchiving tool, would have worked, but I haven’t got round to doing that yet.

So, sorry again for missing that post (and, is it possible to put a small label on the other-zipfiles page, under ‘description’, to note that the filetypes are set incorrectly?) as it may stop people new to the project asking the same question :)

Matt

The IOMD RO 5.15 lacks supports for VRAM, RAM other than 8MB, ADFS, etc. I was wondering – is there access anywhere to either the original RO 3.7 code for these features, or detailed specifications regarding how these should be set-up?

It’s a bit dated now, but this wiki page gives some of the information you need:

http://www.riscosopen.org/wiki/documentation/pages/Building+RISC+OS+for+the+Risc+PC

Specifically on ADFS, this is the modern ATA API that the IOMD port is lacking:

http://www.riscosopen.org/viewer/view/castle/RiscOS/Sources/HAL/Tungsten/doc/ATA

No problem – maybe the other post was someone else asking the same question and then not replying to my questions, I definitely seemed to be answering the same question a lot! :)

What happened was this: I was implementing a set of automated features, including rebuilding those archives daily. I found a bug in one of my scripts which meant that a set of archives had been built as tar files but were called .tar.bz2 so they looked like compressed tar files, even though they weren’t.

The archives on the web site now are (I believe) genuine compressed tar files of the sources in question so your simplest solution would be to rename your existing download from .tar.bz to .tar. Or you could just download it again.

Let me know if there are any further problems with these tarballs.

I am going to sell my Risc PC because I am using other RISC OS computers now. I have RISC OS 4 ROMs inside and I thought it might be a good idea to install RISC OS 5 since I got this CD at the Wakefield Show and RISC OS 4 does not support DHCP etc.

Ok, I failed of corse but I found this thread. Because most of the thread is from April I wonder whether the IOMD version is now closer to being finisched? Or to ask differently: How long do you estimate will it take to bring it to a state where it is actually useable?

Thanks, Michael

Steve : I renamed it to tar, and it opens now :) So problem solved! Thanks for your help! I hadn’t wanted to keep downloading it as it is a significant chunk of bandwidth for your site. I will use this file for my development, until I get the Tools CD which I will order soon, then install CVS :)

Michael : I think it is going to take quite a while before RISC OS 5 is available to soft-load properly on the RISC PC. I know the display out-of-line problem has been fixed, as has the wrong FPEmulator problem.

I think getting a usable RISC OS 5 will involve recognising memory/ADFS, then recognising networking. That is quite a lot of work, but I am going to have a go at it (not sure if I have the skill, but we will see!)

Matt