Overview Features Coding Performance Forum Downloads Products OrderV4 Contact

Welcome to the Apollo Forum

This forum is for people interested in the APOLLO CPU.
Please read the forum usage manual.



All TopicsNewsPerformanceGamesDemosApolloVampireAROSWorkbenchATARIReleases
Information about the Apollo CPU and FPU.

Building a Nommu Linux Kernel for Amiga M68kpage  1 2 

Gunnar von Boehn
(Apollo Team Member)
Posts 5532
10 Mar 2021 19:17


Roland Engelen wrote:

  But with the Assembler thing. I don't know.
 

 
Lets say we talk about Horse.
Now imagine you did say:
"Yes we want more people to work with horses ... but the riding thing? I don't know."
 
 
I think for riders the "riding thing" is a major fun part of working with horses .... and just the same way, for people knowing ASM - the ASM coding on Amiga is maybe just the same..
 
So I would not see it as something "complicated" but as something "nice" and "enjoyable".
 
 
 
Roland Engelen wrote:

  And by the way very nice work you and your team has done there!

 
Many thanks!
 


Dj Up

Posts 37
10 Mar 2021 20:50


Of course I fall in a "camp" who would be able to see it from both sides in this matter. The "get a pi" argument is probably a valid one, but at the same time I don`t think most people who`d like to see how a Vampire would run a small linux-kernel thinks the Apollo-team is the one who should use time on it. Most people who find unix-like systems interesting is also supporters of its open-source-community making it so many can use whatever they got of free time to create and develop.

For me this is not something I`d be waiting for,or expecting to happen,and I would not at all want it to take time from the official team as Apollo OS and core is the most important.

All that being said,when I (as new to Amiga) realize how much more power this core got compared to what a good M68k Amiga over 20 years ago had, it feels tempting to see how a down-sized kernel could perform. A full on beast-kernel I agree would probably be so slow that it would not be a point to it.

The good thing is that there are alternatives for everyone anyway,as it seems the Amiga-community has had an uprising of many ongoing projects from all over the world. That can only be a good thing for the time ahead.



Harold Joseph Neufeldt

Posts 5
10 Mar 2021 23:29


Hi Nick,
I've been considering replying to your post for some time. I am doing some historical research on the Amiga and I've found out quite a bit about this.  You have toremember that the advanced OS was a plan by Commodore engineers to thoroughly redesign the Amiga. Initially it was opposed by Commodore management.  As far as I can tell the OS was completed by 1991 or 1992. It was designed to run on  a RISC processor only. In fact due to some features it could not have been backported to the 68K processor. It did take advantage of some  of thPA Risc features as well. It was a a fully compliant version ofSVR4 version  ofHPUnix. The Commodore engineers started work on this in the late l980's  It was a piece of a very large jigsaw puzzle. I've found a lot of the pieces but not all. There are quite a few pieces still missing and whether I'll find them remains tobe seen.  Classic Amiga apps would have been run on t6he Amiga-on-a-Chip hardware emulator. As I said the OS was a piece of a puzzle. I won't say anymore at this time


Vojin Vidanovic
(Needs Verification)
Posts 1896/ 1
10 Mar 2021 23:45


Story of hombre and pa RISC EXTERNAL LINK  Suggest it would be Windows NT, not some QNX or Unix like OS.
It seems like CBM invested in hardware, not software.
So from Amiga point of view, it's good it never came to be.

More of marriage was A3000 with optional Unix.
EXTERNAL LINK 
On Linux saga, in Debian 3.x Amiga Classic and m68k used to be well supported.


Gunnar von Boehn
(Apollo Team Member)
Posts 5532
11 Mar 2021 07:29


Talking about what all the misguided experiments which Commodore did, will be a something we can do over a drink in bar for fun for some evenings....

But doing misguided experiments is normal, every company does development-experiments of which some will turn out to be a surprising success and many will show to be nonsense and never go to public. In my experience this is just normal and every big company like SONY, IBM, SAMSUNG, HUAWEI are developing many times more stuff then the public knows and many of these crazy ideas die young for good reason.

Regarding Unix.
I can understand that people like Unix, and for sure this can work on the Vampires. I also ran Debian long time ago on my A4000. While we wish the people trying this all success and will support them this is for sure not an area we think to be a success.
We believe that the spirit of Amiga is in doing "Amiga stuff". And here is of course our focus.



Vojin Vidanovic
(Needs Verification)
Posts 1896/ 1
11 Mar 2021 08:46


Gunnar von Boehn wrote:

  Talking about what all the misguided experiments which Commodore did, will be a something we can do over a drink in bar for fun for some evenings....
 

 
  On sidenote, its great most of Hombre features, are revived by Vampire-080-SAGA featurelist :)
AAA was another route, but never came to be, until SAGA

So in one part AAA dream is here, even surpassed
EXTERNAL LINK 
 
Gunnar von Boehn wrote:

  We believe that the spirit of Amiga is in doing "Amiga stuff". And here is of course our focus.
 

 
  Its understandable that focus is on AmigaOS and AROS-ApolloOS support :) Its good to be able to use everything (hereby m68k Linux), when possible.


Carlos Milán

Posts 95
16 Mar 2021 09:08


When I started this thread it was never my intent to make a plea to the Apollo Core devs for supporting GNU/Linux. As I said earlier, I agree that their efforts are better put on AROS, that is a true Amiga experience.
 
  I started this thread to look if anyone was able to build a recent MMU-less m68k Linux kernel, that would run fine on the Vampire out of the box, but the MMU-less support for legacy m68k architecture was removed from vanilla, so I guess that this person used on the Atari ST a 2.6 version ( EXTERNAL LINK ).
 
  If there is documentation about Apollo Core MMU that could be published, maybe some kernel hacker can make the Vampire compatible with the kernel; but of course, I think everyone can live without it and just buy a Raspberry Pi :)


Stefan "Bebbo" Franke

Posts 139
03 May 2021 15:11


There is one thing to add:

Some people would like to have a gcc version which supports the 68080.

Unfortunately there are many tests of the gcc test suite which run only in Linux. There is qemu-system-m68k to do so, but it's slow, so maybe running on 68080 would improve the turn-around-cycles...


Nick Fellows

Posts 116
04 May 2021 15:19


From my perspective - I switched to Linux as soon as it became a possibility to do so. Thats how much I hated windows. When I first started to use it - there was an Amiga like familiarity with it. Decent multitasking , and excellent command line.

I realise if it wasnt for the Amiga those feelings wouldnt be there.

That said for the most part its the Linux or Open Source apps were really wanting here. Theres already 68k ports of some popular libraries. - look at what NovaCoder did with SDL. Maybe there are some easy wins out there.

Perhaps ill take a look when i've got my V1200 set up.




Vojin Vidanovic
(Needs Verification)
Posts 1896/ 1
04 May 2021 17:40


nick fellows wrote:

      From my perspective - I switched to Linux as soon as it became a possibility to do so. Thats how much I hated windows. When I first started to use it - there was an Amiga like familiarity with it. Decent multitasking , and excellent command line.
     

     
      Switchable GUI, RAM disk :)
     
      Surely, fixing GCC to support 080-AMMX would be fastest way to get a bit modern software. AmigaOS4 and MorphOS 3 live similar ways to day, but situation with PPC GCC support and even PPC Linuxes is far better.
     
      With any Vamp or your v1200 no m68k MMU is no go now even for old Debian 3 and so that used to have A1200 and m68k support inbuilt. m68k Linux needs MMU, unlike most of Amiga software.
         
      Some things are supported in m68k Linux thanks to Classic tradition
EXTERNAL LINK   
      So title of this thread "building a noMMU Kernel for m68k" is major step forward, if done. There were some Atari kernels and 68000 kernels before, as well as those for ColdFire, and if that could be fixed to work with 68080, would be great.
      Thus, ucLinux as it has latest noMMU m68k kernel.
     
      Best, if currently undocumented 68080 MMU could be supported in some late Linux 68030-68060 kernel.Like late old kernel 2.6.6 from 1999 or leeching last Debian 10 m68k kernel.
     
      But none is on horizon.
     
      No one is expecting full support from Vampire team,but Vamp PMMU docs or helping kernel fix.
 
 
    Gunnar once promised fixing MMU software for use on Vamp if its open source and bounty set. I nominate m68k Linux kernel MMU use.

posts 30page  1 2