Home   News   Concept   AMIGA-Compatible   Hardware   Forum   Questions+Answers   Pictures   Contact & Team

Welcome to the Natami / Amiga Forum

This forum is for AMIGA fans interested in the new NATAMI platform.
Please read the forum usage manual.



All TopicsNewsQAFeaturesTalkTEAMLogin to post    Create account
Do you have questions about the Natami?
Post it here and we will answer it!

AROS for 68050/70 NatAmipage  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
SID Hervé
France

Posts 668
03 Sep 2010 00:58


Will it be easy and fast to backport AROS to NatAmi (Is it started?) (with full support of 68050/70 and SuperAGA)?

Should we really say a backport for NatAmi ?

Mercis

Samuel D Crow
USA
(Natami Team)
Posts 1304
03 Sep 2010 03:50


From what I've seen posted elsewhere on the forums here, the NatAmi will be shipped running an AmigaOS distribution running AfaOS which is a hybrid of AmigaOS and AROS.  In other words, it won't be a full port, only certain features will have been backported.

Marcel Verdaasdonk
Netherlands

Posts 4045
03 Sep 2010 04:53


Some of the team and some forum members are looking into it.
  But i am sure if more would work on it it could be done faster at this moment. ;)
 
  Besides as Samuel said it would come with AOS by default.

I Immortal
Netherlands

Posts 67
03 Sep 2010 07:17


Hi,

Yes there is some work being done in this area. Although it is going verry slowly due to a shortage of time and knowledge on my account :).
what we want to do first is get hosted AROS building on 68k-linux. To do this we are using
- aranym (an atari emulator supporting mmu and linux)
- debian sarge distro with some updated packages
Aranym emulates a 040 so compilation is pretty slow. I just got my system fully setup last weekend and i think i found the first real build error that is not related to having outdated buildtools. I did not have the time yet to look into what the real problem is. After getting AROS hosted build the next step probably would be getting it run "nativly" on aranym. And after that going to Amiga. although i guess at that time the natami will already be released :)

I did not yet think it was time to spread the word about this as it is not an official project(i and one of the teammebers were interested in the subject of porting AROS back to 68k) and we still do not have an overview of the amount of work that needs to be done. But seeing other people having an interrest in the subject is great. Maybe we could make it a little more official. When you are interrested in join our efforts just drop a post here or join on IRC (irc.melodic.de #natami).

I am able to supply the aranym configuration and disk images if people are interrested.

Loïc Dupuy
France

Posts 253
03 Sep 2010 07:36


A naïve question, will UAE + linux x86 and cross compilation tools can do the trick to develop AROS for 68k ?

Gunnar von Boehn
Germany
(Moderator)
Posts 5795
03 Sep 2010 08:30


AROS is great!
AROS is one of the three AMIGA OS reimplemetations.
AROS could be the future for AMIGA.

AROS is open source - which is great!

But if we are really honest AROS is as of today not in all regards the better choice for us than the original Commodore AMIGA OS.

The Commodore AMIGA OS and the Commodore AMIGA Chipset were a perfect symbosies. The OS was perfectly using the special features of the OS (DMA, Blitter, Copper, you name them).

None of the three AMIGA OS Reimplementatino (MOS/OS4/AROS) has this ability. This means the once perfect symbiosis of chipset and OS - is lost. Only the real Commodoro OS can do this.

IF in the future the ability to use the AMIGA chipset features (Copper, DMA etc) will be re-added to AROS then AROS will become the perfect OS.



I Immortal
Netherlands

Posts 67
03 Sep 2010 09:39


Loïc Dupuy wrote:

A naïve question, will UAE + linux x86 and cross compilation tools can do the trick to develop AROS for 68k ?

There was a reason for it that i do not remember anymore. I will ask and get back to you.

I Immortal
Netherlands

Posts 67
03 Sep 2010 09:44


Gunnar von Boehn wrote:

AROS is great!
  AROS is one of the three AMIGA OS reimplemetations.
  AROS could be the future for AMIGA.
 
  AROS is open source - which is great!
 
  But if we are really honest AROS is as of today not in all regards the better choice for us than the original Commodore AMIGA OS.
 
  IF in the future the ability to use the AMIGA chipset features (Copper, DMA etc) will be re-added to AROS then AROS will become the perfect OS.

This is ofcause the endgoal of our "project". But first lets start with getting the whole thing building for 68k again :)

SID Hervé
France

Posts 668
03 Sep 2010 10:29


Therefore, the only thing which is a formality, it is to recompile a program 0S 3.1 for AROS. The reverse is not true.

«Donc, la seule chose qui soit une formalité, c'est de recompiler un programme OS 3.1 pour AROS. L'inverse n'est pas vrai.»

I Immortal
Netherlands

Posts 67
03 Sep 2010 10:38


SID Hervé wrote:

Therefore, the only thing which is a formality, it is to recompile a program 0S 3.1 for AROS. The reverse is not true.
 

AROS is supposed to be AOS3.1 api and amiga binary compatible. How this worksout in practice is probably a different story.

EXTERNAL LINK

SID Hervé
France

Posts 668
03 Sep 2010 10:59


If I understood your answer and the answer of Gunnar von Boehn :
  AROS for NatAmi will not be as easy that to recompile a program OS3.1
  AROS for NatAmi will not be a backport.
 
  Mercis
 
  «Si j'ai bien compris votre réponse et celle de Gunnar von Boehn :
  AROS pour NatAmi ne sera pas aussi facile que de recompiler un programme OS 3.1
 
  AROS pour NatAmi ne sera pas un rétroportage.
 
  Mercis»

Olaf Schoenweiss
Germany

Posts 812
03 Sep 2010 11:53


it would be really great. it would be perfect for natami. The same spirit and vision

I Immortal
Netherlands

Posts 67
03 Sep 2010 12:11


SID Hervé wrote:

If I understood your answer and the answer of Gunnar von Boehn :
    AROS for NatAmi will not be as easy that to recompile a program OS3.1
    AROS for NatAmi will not be a backport.
   

Depends on the deffinition i guess. Yes like Gunnar said hardware specific code needs to be implemented to get AROS working on the Natami. But this is also the case for the other platforms that are still supported. And even the x86 version needs new drivers everytime they want to support a different gfx card or other piece of hardware.
I don't know if getting AROS running on the Natami can be called a backport because the Natami was not around when the AROS project was started. There is however an 68K tree that is nolonger maintained.
The AROS sources contain specific files for different systems. And during the prebuild configuration parameters are used to know for what system one is compiling.

If AOS3.1 applications will run without a recompilation i can not awnser. i did not look into that yet as it is not one of my priorities. i do not know if amiga binary and api competibility will be enough.

SID Hervé
France

Posts 668
03 Sep 2010 12:35


I really hope that your work will fully exploit the NatamI. This would provide a new benefit to both projects. But I guess that will not be easy.
 
«Je souhaite vraiment que votre travail exploitera complètement le NatAmi. Cela apporterait un bénéfice aux deux projets. Mais je suppose que cela ne va pas être facile.»

André Jernung
Sweden
(MX-Board Owner)
Posts 989
03 Sep 2010 12:40


Getting the AROS 68k port up and running again would be beneficial for the entire Amiga community. Even people not using a Natami could emulate an Amiga for free using a free and open AROS kickstart instead of the old kickstart images.
 
The current 68k AROS tree does not build at all at the moment. It hasn't been maintained for a long time. However, the AROS devs take great care not to mess things up for the 68k port in the future, so it is still in the back of their heads.
 
Only a few tools build at the moment. Getting it up and running requires quite a bit of knowledge of how AROS is constructed.
That's why I think we would need at least a couple of seasoned AROS devs to get this going, realistically.
 
 
Loïc Dupuy wrote:

A naïve question, will UAE + linux x86 and cross compilation tools can do the trick to develop AROS for 68k ?
 

 
UAE + linux 68k does not work properly yet. It boots halfway, but there is no network functionality, etc. There is an effort to get Debian working on WinUAE, documented in a thread on the EAB boards if I recall correctly.
Therefore, the most logical path for a linux-68k crossdev environment is the Atari emulator Aranym. In this early stage the Amiga chipset is not vital in any way for porting.
 
The logical steps would be like this, I think:
1. Getting AROS-hosted to build on linux-m68k on Aranym.
2. Getting AROS-hosted to build on linux-m68k on a real classic Amiga.
3. Start working on native AROS-68k while patching back all Amiga chipset calls in the code.
4. When AROS boots native on classic Amiga, move on to Natami. Add support for SAGA/NCS chipset calls on Natami.
 
There are two AROS bounties for making this happen with quite a lot of money donated. People have picked them up and quit after a while. So it is a big task, and I think it needs quite a lot of people working on it. But with Natami coming up soon, we now have a good incentive.
 
EXTERNAL LINK 
EXTERNAL LINK

Loïc Dupuy
France

Posts 253
03 Sep 2010 12:51


André Jernung wrote:

UAE + linux x86 does not work properly yet. It boots halfway, but there is no network functionality, etc. There is an effort to get Debian working on WinUAE, documented in a thread on the EAB boards if I recall correctly.

I do not want to run linux IN UAE. It was a more hardcore idea.

On a x86 linux, use cross compilation to generate os or kickstart roms, then use UAE to test these roms.
It takes less than a second to boot in a basic pal laced WB on my computer. i can write directly in the amiga system directory tree from linux.
So we can compile, restart uae (quicker than launch a modern executable or a java program), have an error, look at uae debug, correct the source and do the cycle again.
After the library in the distributed files, go to replace one by one functions in the kickstart rom, perhaps even reducing rom code just to boot on HD to load everything else and have more flexibility for updates.

Is there any technical hurdle that prevent this with UAE (JIT disabled) ??

Wawa Tk
Germany

Posts 581
03 Sep 2010 13:13


i dont know much about programming and do not like to spread fud as i was accused to do when i said something like that about a year ago but given the project current dynamics i do not believe we will see results within our lifetime.

i am totally supporting the idea of backporting aros to 68k amiga but i start to doubt if this is realistic.
think most difficult will be replacing the kickstart in particula bootstrap files, where even a dedicted amateur but without original concept will not be enough. i assume the whole new approach will be neccessary as it would be completely different to how pc hardware boots. besides i dont know if uae was suitable to test it. maybe you need a real hardware with a kickstart flash (there are some solutions).

btw. i still do not underdstand why not use an amidevcpp cross compiler on a fast pc.

and last but not least i have been looking at some aros code lately in order to see if it ports and it seems aros api in fact differs from aos3.x. as example


const APTR DevFuncTable[] =
{
    &AROS_SLIB_ENTRY(devOpen, dev),
    &AROS_SLIB_ENTRY(devClose, dev),
    &AROS_SLIB_ENTRY(devExpunge, dev),
    &AROS_SLIB_ENTRY(devReserved, dev),
    &AROS_SLIB_ENTRY(devBeginIO, dev),
    &AROS_SLIB_ENTRY(devAbortIO, dev),
    (APTR) -1,
};

well it might be pretty easy to replace it with an os3 counterpart, there is some guide how to do the opposite on aros site, but still it makes a difference.

Marcel Verdaasdonk
Netherlands

Posts 4045
03 Sep 2010 13:15


plenty of variable on UAE, so want to test them all?

oh and how it the compatibility on UAE, sure on WINUAE it's fine, but UAE and winUAE are two different matters all together.

SID Hervé
France

Posts 668
03 Sep 2010 14:16


(Correct me if I'm wrong)

Something like the use of the MMU by Debian (and Linux m68k) generates somes problems for WinUAE and the current developer focuses on the support of the Amiga.

Sorry, I do not fully understand the previous answers so
Can we hope that AROS is specifically written to take advantage of cores?

Mercis

«(Sauf erreur)

Quelque chose comme l'usage de la MMU par Debian (et linux m68k) génère d'obscures problèmes pour WinUAE et l'actuel développeur se concentre sur le support de l'Amiga.

Désolé, Je ne comprend pas complètement les réponses précédentes donc
Peut-on espérer que AROS soit spécialement écrit pour tirer partie des cores ?»

Matt Hey
USA

Posts 783
03 Sep 2010 14:16


wawa tk wrote:

  ...think most difficult will be replacing the kickstart in particular bootstrap files...

The low level OS libraries will probably end up being optimized disassembled->reassembled libraries anyway. It's only the high level parts of the OS that need to be back ported.


  and last but not least i have been looking at some aros code lately in order to see if it ports and it seems aros api in fact differs from aos3.x. as example
 
 

  const APTR DevFuncTable[] =
  {
      &AROS_SLIB_ENTRY(devOpen, dev),
      &AROS_SLIB_ENTRY(devClose, dev),
      &AROS_SLIB_ENTRY(devExpunge, dev),
      &AROS_SLIB_ENTRY(devReserved, dev),
      &AROS_SLIB_ENTRY(devBeginIO, dev),
      &AROS_SLIB_ENTRY(devAbortIO, dev),
      (APTR) -1,
  };
 

That looks like a standard Amiga library + device function table to me. The "dev" will have to be replaced by the individual 32 bit function addresses of the functions. This is done by the relocatable loader on the Amiga as these values are relocated 32 bit values. The &AROS_SLIB_ENTRY() is probably a macro that takes care of all this. The AROS_ at the start could have been left off as this is probably nothing specific to AROS.


posts 311page  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16