|
|---|
| | 
Aksel Andersen
Posts 120
11 Jul 2017 10:20
| Hence my reference to netbsd being the *nix ported to most platforms, but will not be able to run on the most advanced 68k ever. ;) (in the cores current status)
| |
| | 
Carlos Milán
Posts 95
12 Jul 2017 22:44
| NetBSD as most desktop or server oriented UNIXes requires compatible PMMU. They can use the Motorola 68851 or the integrated MMU in 68030, 68040 or 68060.
Apollo Core lacks compatible MMU so I think it is not more advanced in that sense. I wonder if it would be possible for Apollo Core to work together with a Motorola 68851 so it would have a compatible MMU without adding it the FPGA code.
EDIT: Oh btw, do not misunderstand me, I know Apollo Core has an advanced MMU; it is just happen to not be compatible with the Motorola ones.
| |
| | 
Lorenzo Pistone
Posts 22
19 Jul 2017 05:28
| Carlos Milán wrote:
|
NetBSD as most desktop or server oriented UNIXes requires compatible PMMU. They can use the Motorola 68851 or the integrated MMU in 68030, 68040 or 68060.
Apollo Core lacks compatible MMU so I think it is not more advanced in that sense. I wonder if it would be possible for Apollo Core to work together with a Motorola 68851 so it would have a compatible MMU without adding it the FPGA code.
EDIT: Oh btw, do not misunderstand me, I know Apollo Core has an advanced MMU; it is just happen to not be compatible with the Motorola ones.
|
So in the end mmu is only a matter of driver/module for the linux/bsd kernel, I think it will not be a big problem for someone expert in this kind of things.
| |
| | 
Stefan Niestegge
Posts 33
19 Jul 2017 16:41
| I'd be interested to run NetBSD as well as debian/m68k on my Vampired Amiga. I run both occasionally on my Falcon 060, too.
| |
| |
Carlos Milán
Posts 95
22 Jul 2017 12:57
| Lorenzo Pistone wrote:
|
So in the end mmu is only a matter of driver/module for the linux/bsd kernel, I think it will not be a big problem for someone expert in this kind of things. |
That's right. If someone were to patch the kernel in order to add Apollo Core MMU to Linux, then Linux/m68k would work and if the patch reaches Vanilla kernel, we would have the support right out of the box. The problem is that such code requires deep expertise in Linux kernel, m68k architecture and the Apollo Core MMU. It is not easy to find someone expert in these fields willing to do the job. Maintainers of the Debian Linux/m68k suggest that Apollo Core should be compatible with the standard Motorola MMUs. I wonder which path requires less effort: to add standard Motorola MMU support in the Apollo Core, to add a 68851 chip to the Vampire accelerators or to modify Linux/NetBSD and such to properly support Apollo Core MMU.
| |
| | 
Vojin Vidanovic
Posts 770
22 Jul 2017 13:02
| Carlos Milán wrote:
|
That's right. If someone were to patch the kernel in order to add Apollo Core MMU to Linux, then Linux/m68k would work and if the patch reaches Vanilla kernel, we would have the support right out of the box.
I wonder which path requires less effort: to add standard Motorola MMU support in the Apollo Core, to add a 68851 chip to the Vampire accelerators or to modify Linux/NetBSD and such to properly support Apollo Core MMU.
|
Exact dilemma we face to get to m68k Linux "out of box" (as at same principles, it worked with PPC AmigaNGs and PPC Linux distros aside from not reaching official level).
Hope it can gain entry to some "2018 and long stretch goals" to team to at least provide Apollo Core MMU and 080 expertise in getting some Linux kernel patch out that would abuse hopefully all advanced features of Apollo core, not just minimal.
Surely, at minimal level, just some "emu Mototola MMU" patch would be required.
Goal would be to have Debian 4.0 or later on Vampires, with such stuff as
Nautilus
EXTERNAL LINK
Totem
EXTERNAL LINK
OpenOffice
EXTERNAL LINK
IceWeasel
EXTERNAL LINK
Its so 2008, but yet beyond current AmigaOS 3.x/AROS software boundaries, so step forward.P.S. I am glad there is a Motorola hardware add on MMU, but I am not sure
does Apollo support it physically.
| |
| | 
M Rickan
Posts 177
22 Jul 2017 22:25
| Carlos Milán wrote:
|
I wonder which path requires less effort: to add standard Motorola MMU support in the Apollo Core, to add a 68851 chip to the Vampire accelerators or to modify Linux/NetBSD and such to properly support Apollo Core MMU.
|
Apollo support for Motorola MMUs would certainly make the most sense as other applications require it for compatibility.
| |
| | 
Niclas A
(Apollo Team Member)
Posts 219
23 Jul 2017 08:19
| m rickan wrote:
|
Carlos Milán wrote:
|
I wonder which path requires less effort: to add standard Motorola MMU support in the Apollo Core, to add a 68851 chip to the Vampire accelerators or to modify Linux/NetBSD and such to properly support Apollo Core MMU.
|
Apollo support for Motorola MMUs would certainly make the most sense as other applications require it for compatibility.
|
You guys don´t really understand what you are asking for. Gunnar has said on many occasions that the putting in a Motorola MMU in the Apollo core would be no problem technically but would come with a big performance hit. Instead he made a new MMU that works in the Apollo design. Nothing crazy about that as Motorola them selves changed design on the MMU on the 68030 vs 68040. What i don´t know is if the present MMU could be used to run Linux after support in the linux kernel was developed or not.
Maybe if a vampire card works on the Atari in the future that could become more likely. They seem to use the MMU alot more then the Amiga.
| |
| |
M Rickan
Posts 177
23 Jul 2017 19:38
| Niclas A wrote:
|
You guys don´t really understand what you are asking for.
|
I think we all appreciate the effort required and the trade-offs. It ultimately comes down to whether compatibility is a priority and if it can be preserved. It's great to have new and improved Apollo-specific features but they will require new software that won't work on other systems. By the same token, if certain legacy features won't be supported it means that older applications will need to be updated or simply can't be used.
| |
| | 
Keith Matthews
Posts 39
26 Jul 2017 07:00
| No intrest in Linux 68k whatsoever. It was a dog. To many overheads.
Hence why it was dropped as a supported Arch.I used to have Debian 68k on several machines. Nobody is going to keep the ports or the GCC etc up to scratch.
| |
| | 
Steve Ferrell
Posts 424
26 Jul 2017 08:58
| I have no interest in running 68K Linux at all. There are much better options for running Linux than a Vampire board or a classic 68k Amiga. If you guys want Linux, get a Pi or Panda board to run Linux or better yet, a cheap x86_64 system and STFU. Quit trying to bastardize this project into something it was never intended to be. The Vampire can't be all things to every one. The Vampire design goals never included Linux.
| |
| | 
Ian Parsons
Posts 230
26 Jul 2017 22:29
| Whilst I don't see Linux 68k as a priority for the Vampire, longer term for stand alone possibly ASIC based Apollo Core devices Linux would be almost essential for acceptance beyond the Amiga community. So having Linux available could be helpful in achieving progress to ASICs for Apollo.
| |
| |
Vojin Vidanovic
Posts 770
26 Jul 2017 23:14
| Steve Ferrell wrote:
| There are much better options for running Linux than a Vampire board or a classic 68k Amiga.
|
Well, I do understand we are here for AmigaOS, but NG "experience" has lead me to see how flexibile and usable Linux can be (when missing some important modern apps and drivers).
Linux works everywhere, so why not on modernized 68k, as it did in its beginning? The point is basically to enable use of nice Linux productivity apps we cant have in AOS.
Ian Parsons wrote:
|
Whilst I don't see Linux 68k as a priority for the Vampire, longer term for stand alone possibly ASIC based Apollo Core devices Linux would be almost essential for acceptance beyond the Amiga community. So having Linux available could be helpful in achieving progress to ASICs for Apollo.
|
Exactly, while we could start with old last supported Deb 4.0, once ASIC is done he could be a viable target for latest Deb, and as dual boot option might give it a viable choice for productivity.
AS with NGs, even most developed and matured hardware is great, but with no top noch software, its dudo. Linux provides modern, fast and user controlled OS with very nice apps.
And its all for free for Deb 4.0, beyond kernel and few drivers that needs to be sorted out. Porting to m68k of thousands of newer components would requiere a Vampire as new target m68k arch - hopefully a donation to a Deb team of a standalone or ASIC boards once done. But GCC recompiling is still possible.
| |
| | 
Obetto Sannala
Posts 61
05 Oct 2017 07:48
| I don't know if it has been already mentioned here: EXTERNAL LINK Any interest of running debian on the Vampire?
| |
| | 
Kresimir Lukin
Posts 65
05 Oct 2017 08:14
| Linux works everywhere, if Apollo-core will be produced as asic and board like pi will be available I am sure that linux will be ported.
I am using linux since 98 and witnessed big improvement and I hope I will witness that with Aros with Apollo-core, and no linux will be necessary.
| |
| | 
Tango One
Posts 102
05 Oct 2017 11:33
| love to have debian on amiga
| |
| | 
Gunnar von Boehn
(Apollo Team Member)
Posts 6214
05 Oct 2017 12:18
| The team is currently using actively on APOLLO
- OS 3
- AROS
- EMUTOS (ATARI)
- MAC OS (under FUSION/SHAPI)Linux is not on our todo
| |
| |
Obetto Sannala
Posts 61
05 Oct 2017 14:21
| It doesn't have to be someone of the Apollo Team, right?
| |
| | 
Captain Zalo
Posts 71
05 Oct 2017 18:14
| Obetto Sannala wrote:
|
It doesn't have to be someone of the Apollo Team, right?
|
If there's not a *BSD release for Vampire during the next 12 months, I'll be amazed. It seems some of the teams aim to make BSD run on anything remotely resembling computers. Apart from that, it'd be nice to have a stable OS for productivity running on Apollo hardware (partly to help fund the project, partly to piss my co-workers off).
| |
| | 
Kolbjřrn Barmen
(Needs Verification)
Posts 219/ 2
05 Oct 2017 21:48
| Captain Zalo wrote:
|
Obetto Sannala wrote:
|
It doesn't have to be someone of the Apollo Team, right?
|
If there's not a *BSD release for Vampire during the next 12 months, I'll be amazed.
|
And how are they supposed to figure out the MMU? Poke and peek? There is no way doing any BSD (or Linux) for Apollo Core without direct help from the Apollo Team, since they are only ones with knowledge about the MMU, it's not public information.
| |
|
Documentation about the Vampire hardware