Overview Features Instructions Performance Forum Downloads Products Reseller Contact

Welcome to the Apollo Forum

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



All TopicsNewsPerformanceGamesDemosApolloVampireCoffinReleases
The team will post updates and news here

AROS 68k for 68080page  1 2 3 4 5 6 7 8 

Wawa T

Posts 473
03 Jul 2018 02:58


errm, no i dont really think that works.
 
  however what might work is loading a kickstart module with arosbootstrap insted of loadmodule. it doesnt then need to fit within kickstart and ata and scsi device might actually not conflict with each other. it might be achieved just placing it in the same line in the s-s behind it along with the path. however the path needs to be accessible at (aros) boot time, so it needs not to be on a device to be accessed with. which is a dead circle, except we have any third party device accessible at all times, such as some flash drive or something.


Vojin Vidanovic

Posts 348
03 Jul 2018 14:22


wawa t wrote:

  ks module with arosbootstrap 
 

 
  Since new v4 Vamps are supposed to have dual KS space onboard, could an AROS KS Vamp module be made with Remus or similar? If not acessable from ROM/menu, then at least soft kickable, where it would first reboot the machine. Like the old days.
 


Wawa T

Posts 473
03 Jul 2018 14:39


i dont know.


Michael R

Posts 255
06 Jul 2018 04:07


It's a positive sign that substituting scsi.device for ata.device along with a few other tweaks gets AROS to work. That's a step in the right direction. Perhaps once AROS 68k gets to a point where it boots and runs fairly well on Vampire accelerators that may drive development further. But it is very interesting so far.


Gunnar von Boehn
(Apollo Team Member)
Posts 3431
06 Jul 2018 06:12


Vojin Vidanovic wrote:

Since new v4 Vamps are supposed to have dual KS space onboard, could an AROS KS Vamp module be made with Remus or similar?

Also Vampires V2 can directly boot AROS.



Olaf Schoenweiss

Posts 427
06 Jul 2018 11:06


can they?

I asked some time ago if boot problems are solved and got no answer


Vojin Vidanovic

Posts 348
07 Jul 2018 11:09


Olaf Schoenweiss wrote:

can they?
 
  I asked some time ago if boot problems are solved and got no answer

I wonder can Simo tell us how he used to boot AROS

AROS 68k on core 2.7 AROS Sep 2 2017 snap with OWB 1.00
EXTERNAL LINK


Samuel Crow

Posts 290
07 Jul 2018 12:02


Olaf Schoenweiss wrote:

can they?
 
  I asked some time ago if boot problems are solved and got no answer

It's got a Heisenbug.  If you've got the debugger running it boots every time.  If you're running release code it usually doesn't but sometimes does.  Mike suspects a race condition.


Wawa T

Posts 473
07 Jul 2018 12:35


its been ages..
  perhaps something like that here, just the opposite:
 
  Write log message before register write, write may cause interrupt and another log message while previous is still being written to serial port.
 
  EXTERNAL LINK 
what if writing debugs to serial causes an interrupt, ata/ide is idle waiting for in case of apollo hardware?


Michael Ness

Posts 1
08 Jul 2018 05:44


The boot problems are not solved completely.  Like Sam said, it's a Heisenbug, and it can be worked around by enabling debug output at startup.  So as far as I know, this is how the Vampire currently boots AROS.
 
  A couple months ago I was sure that I had found the root of the problem after several weeks of testing.  I got it to a place where it would boot consistently...if you patiently waited for more than a minute with a blank screen after the bootstrap.  A minute or more wait before booting may as well be an eternity.
 
  Anyway, I've been waiting for a bit of inspiration to continue this.  Maybe this thread is that inspiration.
 
  The good news is that time should be freeing up in the next week or so for more Vampire work.  I'll take another crack at the Heisenbug.
 
  Thanks wawa, for the link to that diff.  Could you offer any more detail about it? 
 
  The frustrating thing for me is that the things I really want to do for AROS, like a new vampire graphics HIDD, are basically stalled until this IDE business gets sorted out.  If I can't boot reliably, and relatively quickly, it makes development a real pain in the ass.
 
 


Vojin Vidanovic

Posts 348
08 Jul 2018 11:09


Michael Ness wrote:

  The good news is that time should be freeing up in the next week or so for more Vampire work.  I'll take another crack at the Heisenbug.
 
  The frustrating thing for me is that the things I really want to do for AROS, like a new vampire graphics HIDD, are basically stalled until this IDE business gets sorted out.  If I can't boot reliably, and relatively quickly, it makes development a real pain in the ass.
 

Thank you for "taking the pain" and taking your shot. May your skills and best of luck serve you. Improved graphics speed would be a great next step.


Stefano Briccolani

Posts 199
08 Jul 2018 15:45


I understood there's a problem with the ide, but it's this a problem with v500v2 integrated ide or a general aros68k problem with amiga ide (an a600 with vampire still uses the original gayle and his standard ide)?



Vojin Vidanovic

Posts 348
08 Jul 2018 16:14


Stefano Briccolani wrote:

I understood there's a problem with the ide, but it's this a problem with v500v2 integrated ide or a general aros68k problem with amiga ide (an a600 with vampire still uses the original gayle and his standard ide)?
 

There were reports it affects all 68k hardware


Wawa T

Posts 473
08 Jul 2018 19:11


Michael Ness wrote:

  Thanks wawa, for the link to that diff.  Could you offer any more detail about it? 

best source of information in this case would be the person who commited it, toni wilen. you are on aros dev ml, but you can usually mail him directly to cut the corners, especially if it wont interest anyone else on the list.

as to the bug itself, as you remember we were hunting a similar issue for ages with a4000 internal ide with toni. afair this was due to an interrupt that hasnt been triggered or missed under certain conditions. i neither remember nor understand this enough to give an explanation in depth, but for me it looks extremely similar, as this part of code is extremely fragile what concerns interrupt timings.

the code in
AROS/trunk/AROS/arch/m68k-amiga/hidd/gayle_ata
where i pointed to will obviously affect a600 as well, since gayle is responsible for ide controllers there and in a1200 (which was fine to begin with, but its another matter), but it shoulnt influence a500, that only has gary and ide is an own extension of vampire board (does aros boot there?)

specifically pay attention to both commits by toni 13 months ago:
r54754 -> ata_gayle.conf
r54755 -> interface_pio.c

alas, would you want to make any changes, i now wont be able to cross test on gayle hardware without vampire (a4000, a1200) for a few weeks, to ensure no regression, but since its easy to revert, that shouldnt be a problem.

also, what about commiting vampire gfx hidd to aros repo? is it all asm or c with inlines?


Vojin Vidanovic

Posts 348
08 Jul 2018 19:27


wawa t wrote:

  alas, would you want to make any changes, i now wont be able to cross test on gayle hardware without vampire (a4000, a1200) for a few weeks, to ensure no regression, but since its easy to revert, that shouldnt be a problem.
 

 
  Since this is largerly Gayle related bug (as it seems) could usage of  V500 IDE interfrace on Vampire card (as well as on all v4 models) solve this issue by using this IDE and attached SD Card/hard drive as boot device? (note Vamp  would need to support boot from  there too, which I hope will done by v4 days)


Wawa T

Posts 473
08 Jul 2018 19:53


i cant answer that. i dont have a vampire and i dont even have an a500 or a600 (anymore). so i cannot practically test anything related.


Vojin Vidanovic

Posts 348
08 Jul 2018 20:05


wawa t wrote:

i cant answer that. i dont have a vampire and i dont even have an a500 or a600 (anymore). so i cannot practically test anything related.

But that could be a potentaial "way out"? Hope you will get that standalone once done, I  do  have same problem as you :-)


Wawa T

Posts 473
08 Jul 2018 22:02


Vojin Vidanovic wrote:

  But that could be a potentaial "way out"? Hope you will get that standalone once done, I  do  have same problem as you :-)
 

 
  if i was to spend money on amiga id have supported kal (in fact i was about to). there is enough demand for vampire as far as i can judge. anyway my assistance would probably be limited. to date i prefer to invest into support for genuine systems, with better support for extended ones, such as vampire, as consequence.
 
  but we will see.


Vojin Vidanovic

Posts 348
08 Jul 2018 22:57


wawa t wrote:

    anyway my assistance would probably be limited. to date i prefer to invest into support for genuine systems, with better support for extended ones, such as vampire, as consequence.
   
    but we will see.
 

 
  Aiding AROS 68k fixing to at least work and upstreaming fixes to future AROS releases would be more then enough.Even  Vamp4+ could bring a potential hardware to make sense to update and use AROS 68k in something that  is  not UAE. Vamps on other hands need an OS that is copyright free (as in KLF  tunes  :-) to have something shipped with it without much legal mumbo jumbo. So is a win win alliance.
 
  I hope Vamp standalone will qualify as "genuine"  :-)


Wawa T

Posts 473
08 Jul 2018 23:35


as you see i cant fix much myself, my programming experience is lesser than the vast number if not majority of people in amiga scene.

posts 142page  1 2 3 4 5 6 7 8