Upstorm: A distro

Software & Operating Systems Linux
1

So for a long time I have wanted to have a PowerMac G5. Well, now I have one, but looking at linux options mate up and abandoned PPC [buncha losers] and the only up-to-date options that are around are morph and BSD. Ya know what? I don’t feel like paying 115 bucks for an amigaOS that I don’t even like very much that doesn’t have a good web browser or abiword (Like Icaros does), and I don’t feel like screwing with BSD because I’m a fat lazy duck (not fuck, that’d be rood). So the easiest route is going to be to make a distro.

At least for me, and it sounds fun.

Now I’m also probably crazy… No. No I am crazy. But I need something to work on and FolloRec is still in design stages and research stages. This I have somewhat of an idea.

So the biggest problems any PPC based linux faces is abandonment. The Dev’s get busy, or they’re just really stupid, or they have lives, or they’re just stupid, but something is always in the way. So what I want to do is hopefully end up with something that you can hand-run to a new base and redistribute the built packages to other users sorta like peer to peer updates. I’d like to use xbps for this as well as it has built in build tools for building up either from source to pump right into the system, or making new xbps packages.

The only real problems I can forsee is getting users to try it / help, figuring out how C works in total, and trying to learn the nitpicks of IBM chips vs freescale chips. I imagine some nut with an AmigaOne X5000 is gunna want to run this OS, which is perfect, but fuck me if I’m going to afford a 2000 dollar toy like that. I’d have better luck getting a G5 quad that works than having money for that. IDK maybe I’ll set up a donation thing for that in the future. You can let me know ideas for that, or if you, for some reason, have a spare X1/5/000 then I guess let me know??

Main Goals:
To not be called Upstorm OS or Upstorm linux, its just Upstorm like Ubuntu is Ubuntu or Fedora is Fedora

To be up-to-date. No kernel 4.4 here. 4.14 will be shipped at base to start, or whatever is a better standard at the time, and then hopefully users can build their own kernel updates / spread them around.

To be independent. Not out of spite, I just don’t want to work off of a base like mate and not have over all control of it. Yeah I COULD maintain it I guess, but I don’t like APT that much and systemD slows PPC down a lot.

To have a choice in Init. I like Runit, a lot, so I’ll start with that, but SysD should be available. I’d probs want upstart available too.

To /potentially/ let users run mac apps. A lot of the OS VM stuff is getting worked on on these intel and AMD systems, but what about a 10.4 VM to play Age Of Empires? That’d be handy. IDEK if anyone is doing that on the freescale machines. Or at least what the efforts are there.

To have Up-To-Date drivers. IDK what nvidia would do, but AMD wouldn’t care. Perhaps some work to nouveau is in order?

And finally, to be accessible. If theres a way to make as good of documentation for dev’ing on PPC as the archwiki has for documenting everything, I want that written up. Its a good platform and people spit on it alot. So anything from devs to noobs. These G5’s and their earlier bretheren are everywhere and they are cheap. If theres something not ancient to run on them that at least something like newpipe and a word processor, maybe even zsnes, can be run then some uni student who can barely afford to be there will be happy I think. Or someone else. IDK, but you get me.

The first steps I need to take, I think, are learning C and Python. I’m not a programmer, I can never keep interest unless I have a real world project that I will end up using, but hopefully this will keep me going on my learnins. Probably the hardest way, but once I get it I get it completely. Its just how I learn. I’ll have to take my G5 that I have here and get Ubu Mate running on it and go from there. Haven’t built from source for a while, aside from having programs do it for me like pacman or xbps, so maybe theres something different.

And if anyone has any interest or major suggestions here then PM me or start a PM group with me and we’ll add people and all chat.

3 Likes
2

You’re looking at huge amounts of work preparation and all else that comes in between for building an operating system from scratch (independent of all others). Keep your goals realistic from the very start and plan well ahead of time. The best laid plan is the one that adapts to the changing circumstances.
Also, I believe there is an operating system with [a] Developer[s] of the same that you know as Void Linux (operating system itself) that you should get in contact with to obtain a working knowledge of where to start and how to proceed accordingly. I believe that Community would be instrumental for you at least in those regards.

3

Yeah I know. I already got people in a telegram group telling me I’m going to fail. But, its the only thing I want to really do with these powerpc machines that I got a few days ago.

Yup. Its all prelim stage for now. Probably will be for a month or two. Gotta get my space set up to even do this sorta thing first on top of figure out contacts. Then I need something running on the machine I’m deving for, which annoyingly enough I think has to be ubuntu mate. Maybe lubuntu if they have an 18.04 candidate out but I haven’t checked. But I know my goals for it and I want to hold them.

It would be, and I know a few people on an FB PPC linux page are ready to test anything I throw at them. Next would be getting contacts over in europe that dev on the freescale machines that are out. Maybe people like Dan Wood would know who I can talk to about that.

1 Like
4

Alright, good fucking god if people could leave me alone and I could quit my BS job I feel like this would be going somewhere already.

So my first step is to prep my workspace. Need the machine running something and to be hooked up to keyboard/mouse/monitor.

Side note I’m selling an apple cinema display for 200 bucks. 1920X1200, no damage or anything. Maybe it’ll just go to craigslist. IDK.

I think the best place to start, at the moment, is running Ubu Mate. As much as I am annoyed that the devs started something to immediately abandon it (in my space of time, a year is like a month, so basically they quit too fast in my head), its my best option at the moment. Need to play with build tools and maybe just mainline the machine for a while and get a feel for how it works. Only problem is its 2GB of ram will get to be a bit of a pain in the ass, but I think thats pretty realistic to what people have at home so fuck it it’ll work.

Then I need to get some research in on build tools. Is it just a bunch of makefiles? Is there an actual toolkit? The most I have done actual code wise thus far is mcguyver some shit to make hyperthreading work on pentium 4’s and patching little particulates of code to make a GPU not be fucky. So this’ll be fun.

Then, after all that shenanigans, I need to look at what my actual features are going to be. I don’t want to big of a list starting out, just what I need. Optimally? 4.12 as kernel, xfs and ext4 progs, xorg, xfce, a network stack, runit, and xbps. Xbps doesn’t even need to actually connect to anything just try and poll a repo. I feel like thats a good start.

When I get to the actual OS building and testing the mini shit I feel like LFS is a good reference guide? But I really don’t know. I’m still doing hardware research and trying to see what I even have available. What I have will work for now, but if theres a way to get non-openfirmware GPU’s to work in powermac’s that’d be fuggin sweet. If not, I guess I’m learning about GPU Bios flashing too.

IDK how long this first step will take. Hopefully I can get my job changed out here soon and I can have more loose time than I currently have. That’d be nice at least. And if theres better references than LFS let me know, otherwise thats where I am going to learn from for now.

1 Like
5

Awesome !

But I somehow get a lot of vibes about “cross-compilation” and “test automation” , which also sounds awesome when someone else is doing it, not so much when you’re the one.

Good luck!

2 Likes
6

Well, I’ve just ordered a dual processor PowerMac G5 for cheap and been able to get LFS up and running a couple of times so I may be of help.
A couple of questions though: why kernel 4.12? Why not a longterm release like 4.14?
Also, I’ve heard good things about pkgsrc as a package manager. It’s originally from NetBSD but works quite well on different OS’s and architectures apparently.

2 Likes
7

I thought 4.12 was the LTS. Well, shows how much I’ve been able to keep track lately.

pkg is actually what XBPS is half based on, with the other half being pacman. It kind of functions as both with the ability to build from a tar or from base instructions in a .xbps package.

Lol well thank you. I don’t have much time right now between hay, work trying to get me to cover everyone, and sleeping for 4 hours when I get home. Annoying isn’t a description.

What machine’d ya get? Hope its not a 1.8!

8

someone may have already mentioned this but I’m pretty sure the “alternative” init system with the widest support is openRC

1 Like
9

why not have a look at a project like Yocto instead?
its not the easiest to get a hold off, but a whole lot easier then what you’e describing.
basically choose a branch and add a new kernel/architechture, then compile and voila.
pretty sure you can add package managers and what not to your distro here aswell.

1 Like
10

I’m aware of this. Runit is so stupidly simple in comparison though. Its literally folders, 2 scripts, and symlinks. ANYONE can learn how it works. The only thing I want to know is if theres something like journal from systemd that can be run with runit.

I agree, OpenRC is popular, but I’m after simplicity and complete independence. So if one dev fell out, someone else could easily move in to replace them. If package maintenance fell out because the devs get busy for 6 months, I’m very likely to do that, anyone can build the newest LTS kernel or version of runit, bump it to the test repo, and anyone else can tell the OS to search the test repo for a kernel version they want and depending on age… Actually I just had an idea.

There could be a package management app specifically built to interact with the repo’s. It connects and keeps a counter to how many polls and pulls there are to the repo’s available packages and you can look at the tester’s comments for the packages to know what ones to use.

The point is infrastructure as well as OS. Make it easy for the user whether they are a pro or home user. More than likely most users will be home users, but I want it all to be straight forward and easy to pick up.

Read above. PPC OS’s get abandoned all the time because the user base isn’t huge. Its a matter of infrastructure as well as durability. Theres nothing in the works that works the way I am describing as far as I can tell. Yocto is close, sure, but thats better for embedded computers where you tool one thing at a time, not a distro that thousands will use.

11

It’s a dual 2.0ghz: https://everymac.com/ultimate-mac-lookup/?search_keywords=a1047 the guy only said it’s an A1047 with dual 2ghz. I’ll find out in a few days. linked to this on the listing: https://everymac.com/systems/apple/powermac_g5/specs/powermac_g5_2.0_dp.html

12

noice lemme know when you got it!

13

lol theres a lot more people with these machines than you think. This isn´t just for G5´s.

14

The primary target architectures of LFS are the AMD/Intel x86 (32-bit) and x86_64 (64-bit) CPUs. On the other hand,the instructions in this book are also known to work, with some modifications, with the Power PC and ARM CPUs.

Good to know.

15

I’ve got a G5, if you put something together I’ll install it.

1 Like
16

PowerMac update: Shipped, should be here around the 6th.
Also, this might be handy: CLFS-ppc64.pdf
Hasn’t been updated since 2014 but should be a good starting place.

1 Like
17

I have just the basic LFS manual. It says in there that the standard book should work, for the most part, with ARM, SPARC, and PowerPC, but that there will be some stuff missing OOTB because a lot of the parts of those platforms keep the secrets for themselves. I hope to rip that from, at least, the hands of ppc and make it more open like IBM wanted it to be originally.

Also, cool. When you get it I don’t think I’ll be ready to start on actual code yet. I had 3 days in a row off this week from work and I was sick. I didn’t want to work on anything. Hoping to put some real work into this soon.

1 Like
18

The CLFS book shows you how to set up and use cross compilation. It’s a fair bit faster than compiling on the Mac.

19

Well I’m not going to run OSX on main on my G5 possibly ever. My ibook is for that.


https://amigakit.amiga.store/product_info.php?products_id=1303

Thats the machine I need to get for this though. G5’s are what is easily available, thats my prime.

If I got one of those I would literally sell all of my shit (except a choice few machines).

20

@stenstorp Imma be honest I forgot who you were for a little bit haha.

So where do you want to start?