New multi-core kernel for AmigaOS 4 shown at Pianeta Amiga!

@khayoz

I’m grateful for what was done, I argued and defended AmigaOS4, gotten my share of bans on amigaworld.net, in retrospect I’m not sure was right about ArticaS, fixing it software was not correct way, the IRQ issues, a lot time was wasted because of buggy hardware, its of course, nice. They managed to get AmigaONE-XE to run stable, and got fpu emulation working on the A1222 but was this really good use of developers time? I think not.

@afxgroup
Quote:It's not limited to m68k software, AmigaOS 4.x PPC software using a crap C library like clib2 has to be restricted to core 0 as well.
https://github.com/adtools/clib2/blob/master/library/stdlib_main.c

@joerg

that is not the same.

what is correct why to do it? get it from the interface?

x

@LiveForIt
Quote:Sorry, I don't remember, it's been about 20-25 years ago since I did any serious AmigaOS 4.x development...
But one of the arguments the _start() function of PPC native executables, no matter if ELF REL (ancient AmigaOS 4.x versions, no idea if still supported) or ELF EXEC (current) called by dos.library, either gets a pointer to SysBase or to IExec in it's arguments (IIRC r4), and both are interchangeable and don't prevent SMP.
Accessing the unmaped zero page, incl. **4, causes an exception, and either isn't usable at all for SMP, or causes huge additional work.

@afxgroup

Thank you for working on this new multi-core kernel for AmigaOS4. I hope it works out well for everyone.

Cheers,
redfox

@LiveForItQuote:

There's this for IPv6 (I haven't tried it). It's for 68k but presumably porting it to OS4 will be easier than porting something from Linux (or more likely BSD):

https://github.com/tinic/AmiNetXDuo

AROS network stack received IPv6 support in recent months and it's based on existing AmiTCP so it should be straightforward port to AmigaOS4 IMHO.

@afxgroup: Thank you for explaining your motivation and give a little bit more inside into your thought process. I get your points.

The only thing I am still thinking of (and this is no offense):

I am not an attorney and the contract between Trevor and Hyperion about the kernel is very tight, but maybe there is a way to use an alternative kernel without breaking the contract. So asking them like: "Hey, I am experimenting with an alternative kernel and make good progress. Is there a legal way that you guys could use it with AOS4 instead of ExecSG?" Maybe the answer would have been: "No, sorry, not possible with the current contract." anyway, but now we don´t know.

But anyway: Good work, keep it going, who knows, what the future holds for us.

@LiveForItQuote:

I never spoke about A-EON but Hyperion.
As I said: I was in the AOS4-community for 20 years, mainly as editor for two German Amiga magazines. So I know the "drams" in the background too well. The point is: There are tensions between HJyperiopn and Trevor/AmigaKit for years. This is one of the reasons because there is still no port for Mirari. I know that Hyperion is looking for a solution for this situation, especially regarding the kernel. An alternative kernel could maybe have been a solution, done the right way (in a legal sense).

So a little talk before releasing the kernel wouldnt´t have hurt and maybe would have brought a solution. That´s all I say.

My guess is that afxgroup would like to keep the kernel open source and Hyperion would not. I can understand the reasoning from both sides, and it makes sense from each of their prospectives.