A major announcement: IRIX-32 Kernel Project - Printable Version
+- IRIX Network Forums (//forums.irixnet.org)
+-- Forum: Site-Related (//forums.irixnet.org/forum-39.html)
+--- Forum: Site Discussion (//forums.irixnet.org/forum-29.html)
+--- Thread: A major announcement: IRIX-32 Kernel Project (/thread-3941.html)
+- IRIX Network Forums (//forums.irixnet.org)
+-- Forum: Site-Related (//forums.irixnet.org/forum-39.html)
+--- Forum: Site Discussion (//forums.irixnet.org/forum-29.html)
+--- Thread: A major announcement: IRIX-32 Kernel Project (/thread-3941.html)
RE: A major announcement: IRIX-32 Kernel Project - nuclear - 06-03-2023
Ok, yes I understand that the first step is to document the existing kernel, I'm just asking what the ultimate goal is. You answered most of my questions, and reusing existing pieces is a good way to get there much faster. My only remaining question is about target architecture. Are you making this to replace IRIX with a free version on SGI systems, or to have something like IRIX on modern systems?
RE: A major announcement: IRIX-32 Kernel Project - Raion - 06-03-2023
Our primary goal is to remake it for SGI systems. Inevitably there might be ports to other architectures and that's one of the reasons that I would like to focus on 32-bit first besides the cost/engineering challenges of later kernels. Something I've been bike shedding is that if we manage to get a successful drive with this and in 3-4 years we have progress under way for a functional 32-bit IRIX, we can start seeing what pieces we can reuse for an RE of 64-bit IRIX.
In the meantime though there still will be developments. TruHobbyist is still working on a FAT32/VFAT driver last I spoke to him and he's using my incomplete libdisk/mount(1) code to help with that. Having a driver like that released will help with say adding other FS to IRIX such as ext2. It's also a pretty important step to getting mass storage for IRIX.
I'm also still working on my IRIXCE project, which is to patch security issues with the base OS, and eventually offer a repack of the existing OS with all new features. The subproject Neweoe already offers many new versions of commands. Eventually we should have something that is enticing enough there for us to start retargeting some of our third party software distributions. And hey, if people disagree with my IRIXCE's goals or anything I do here, never fear I plan to make sure that everything is as documented as possible so that people can go off and start their own projects with blackjack and hookers if they prefer.
RE: A major announcement: IRIX-32 Kernel Project - Raion - 06-08-2023
Most crowdfunding platforms are not seeming to offer what we need. In particular, we obviously don't have a prototype of the resultant documentation, because of the nature. That takes a few of the big players out.
I'm going to take my time and find the right platform for us. If anyone has any suggestions, by all means.
RE: A major announcement: IRIX-32 Kernel Project - jwhat - 06-15-2023
Hi Raion & others,
on IRIX-32, can you advise on:
Why 32-bit ?
- One of fundemental leap-frog moves that SGI made with its MIPS line was to be first out the door with 64-Bit Architecture
- Seems that targeting 32-bit OS is ignoring the vast majority of SGI hardware being used by the SGI community
Why start clean room ?
- Surely it would be better to start with OpenBSD/sgi (which can boot on SGI machine I believe) or some other BSD kernel variant which has already resolved the legal/licensing issues, is proven to work and be portable.
- Then focus exclusively on the SGI/OS HW/SW interface areas and IRIX compatable ABI so exisitng Applications can run.
Best wishes for your endevours and apologies in advance if I have mis-interpreted what you are trying to achieve.
Greetings from Oz,
jwhat/John.
RE: A major announcement: IRIX-32 Kernel Project - Raion - 06-16-2023
> Why 32-bit ?
Because simpler. Doing 6.5.x is VASTLY more expensive, and I'm not only uncomfortable of asking that amount but unsure if the spark would remain here.
> Surely it would be better to start with OpenBSD/sgi[...]
No. It would not save any time, and the end result would not be the same thing. Solaris is closer to IRIX than BSD, GNU/Linux or any other OS, but it's still miles apart and illumos is much newer.
To advise on the difficulty of what you're asking, what you're suggesting is instead of driving across the rural highways of the Outback to get from Canberra to Perth, you instead suggest taking Quantas to Madagascar, modifying your vehicle to be amphibious and then sail across the Indian Ocean in the middle of typhoon season while flying the flag of ISIS.
It's going to be difficult doing it the way that I have proposed but any other way is way too much time, money and such. Have you even used a BSD, much less on MIPS?
> Then focus exclusively on the SGI/OS HW/SW interface areas and IRIX compatable ABI so exisitng Applications can run.
This in particular is the problem. It's not possible to write new graphics drivers for SGI hardware once you get past the 2D boards like the Newport. They are not like a modern graphics driver but instead use a form of microcode with an opengl interface over it that passes commands to the card. The problem is you can't decompile microcode. So throw out graphics drivers. It's not gonna work. The only systems with PCI interfaces to use very ancient graphics cards from 15 to 20 years ago is a small fraction of the entire community.
And trying to reconstruct the IRIX ABI is a lot like asking for someone to clone the Empire State Building down to the exact rivet with just a copy of Warhol's film Empire which only shows the very top of the building anyways. It's not impossible but it is very very difficult.
So no, jwhat, I understand your concerns but we're not changing course. There is no other practical way to do this and that's the simplest way I can explain to somebody without a modicum of background in systems programming. The only other ways that would be feasible would be of questionable legality, be illegal to distribute under US and EU law, and still require an army of programmers who would have to at that point work for free because when you commit a crime and profit off of it, you end up having stricter penalties including charges of racketeering.
RE: A major announcement: IRIX-32 Kernel Project - Raion - 06-30-2023
Small update:
I'm still searching for a crowdfunding website and I've started reaching out to a few of them to see which would permit our project being hosted. I'm not having much luck but I will figure out something I'm sure. The big ones are off the table even after reaching out personally.
There was a concern raised by someone on Reddit and I wanted to clarify some past comments:
1. Yes, I do have a copy of IRIX 6.5's kernel source. Unfortunately I'm unwilling to leak it because of legal counsel's prior suggestions. No amount of internet lawyering or offers to take my copy will change my mind. Nothing I have is particularly novel, as to my recollection it was just a partial copy of a later kernel with no graphics or other code that is of value.
2. We are not looking at this time to do anything with 6.5 it's simply not financially viable. The scope of the operating system is far wider than 5.3's and simply not viable.
3. This project is focused on doing things the right way. I'm not here to do things in a half-assed manner that will get me or anyone else prosecuted for copyright infringement and I want to ensure the future of IRIX.
RE: A major announcement: IRIX-32 Kernel Project - Raion - 08-10-2023
Here's a final update for the time being:
We have received a lucrative offer of private funding so I am probably going to do that. People who wish to contribute to the project financially are still permitted to do so but there is no longer a necessary need to do so, as everything will be covered within the next 6-8 months by me, our businesses that donated, and a potential few private donors who have offered their support. Anybody who still wishes is more than welcome and it would still benefit the project because we can use this for stretch goals. Any unused funds will be, at discretion of the original donor returned to them or reinvested into the next leg of the project which is building a new kernel based on the documentation that will be published.
Expect me to advise further in the coming months.
RE: A major announcement: IRIX-32 Kernel Project - jan-jaap - 05-27-2024
It's been a year since the original announcement, how's progress on this project?
RE: A major announcement: IRIX-32 Kernel Project - Raion - 05-27-2024
The original intention was to start the project in May however we have faced some delays. There will be updates when I have them however. But there is nothing of substance to talk about right now and as everything was privately funded between me and businesses involved, we do not have a deadline to adhere to.
RE: A major announcement: IRIX-32 Kernel Project - Raion - 07-09-2024
I needed to see where things were over the next 2 months but for the time being, we have had to postpone the project. This is not due to lack of funding, it was entirely because the development staff I had chosen for the analysis side had personal issues that they had to back out of. It's unfortunate, because these people are extremely talented and have already proven to me that it is possible for us to do this but unfortunately there was a significant amount of internal friction as there tends to be with such projects.
As a result of this I have consulted with those who donated (entirely private businesses/individuals) and we are focusing on smaller projects. Some of the funding will be released to the D2 project with the blessing of those involved, and the rest will be split between smaller projects that I will be funding.
I needed to see where the dust settled before I called the project shelved for the time being but I have known about this since April. It was exactly this sort of scenario which is why I did not want to take community money after talking it over with involved businesses.
One business, the oceanography company based in VA who contributed money, asked me to post this:
"It is a shame that the project fell through but we do not have any doubts about the intentions of the organizer behind IRIX-32, the development team or their expertise, leadership, and ambitions. Using IRIX isn't a hobbyist application for us but an integral part of our business as it serves as the rendering system for our ocean surveys. We have known Mr. Kazuo K. since about 2014 when he began doing work for us on our Onyx2 cluster. We will continue to support the endeavors of the community and contribute funding where possible especially if it will address problems in the long term that we are facing."
It ultimately does not bother me that this fell through although I feel bad for getting the community hopes up that this would be a major project in the next year or two. We will however continue to release information about IRIX as the remaining team members are willing to work on smaller projects off the retained budget. You learn every time you attempt something.