Re: [hatari-devel] Falcon demos tests : part 1

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On 10/31/20 12:49 AM, Laurent Sallafranque wrote:
> I've tested Bound Falcon without DSP and it works perfectly for me.
>
> (MMU ON if this is useful)

I've tried it with and without:
- MMU
- VGA
- DSP

And all I get is a black screen
(also with Hatari v2.2.0).
bound.prg doesn't seem to be frozen as it
eventually returns to desktop, graphics
just aren't visible.

Could you send me your config?

Also, are you using Hatari version built from
latest Git commit, without any patches on top?


> What is the bug here.
>
> If you've found something, I think it's very interresting to share with
> us the problem and the solution.
>
> I love this EKO demo so much...

As I said, Hatari crashed once for me with it.

Because of the warning message before the crash,
there could be some hard to trigger issue with
Falcon crossbar handling. If you run EKO in Hatari
under Valgrind, it might report something.

(Valgrind slows things tens of times, so it's
going to take a while to run demo from start to
end under it.)

Just "apt install valgrind" and prefix your Hatari command with "valgrind ".


	- Eero

> Laurent
>
>
> Le 29/10/2020 à 11:25, Eero Tamminen a écrit :
> > Hi,
> >
> > On 10/29/20 12:34 AM, Laurent Sallafranque wrote:
> > > For EKO system demo, I launched it 3 times. Once worked perfectly, 
> > > for the 2 others, I had the sound broken until the final tridi world 
> > > (the music is resetted here).
> > >
> > > Only sound is broken, the demo continues.
> >
> > I tried another run, and couldn't reproduce it
> > either.
> >
> > -> I guess there is na issue with crossbar DMA
> >    handling, but triggering it is somehow timing
> >    dependent.
> >
> > Running Hatari with Valgrind could find out
> > the bug, but I'm still busy testing other things.
> > Would you have time for that?
> >
> >
> > > This demo seems to be at the limits of the DSP <-> CPU timings, but 
> > > works, which is quite a good news.
> > >
> > > I'll test for bound.
> >
> > Thanks!
> >
> >     - Eero
> > > Le 28/10/2020 à 23:11, Eero Tamminen a écrit :
> > > > Hi,
> > > >
> > > > On 10/27/20 1:12 AM, Eero Tamminen wrote:
> > > > > "Bound" (uses no DSP) seems to have
> > > > > also regressed, I get now only a black screen.
> > > > >
> > > > > Do you get same?
> > > >
> > > > Laurent?
> > > >
> > > >
> > > > > > EKO Systems:  perfectly works (except a little transition in the 
> > > > > > betty boo image).
> > > > > >        Music plays well until the 3D scifi scene and plays well 
> > > > > > again in the tridi world.
> > > >
> > > > Close to end I got Hatari crash:
> > > > -----------------------------------
> > > > WARN : crossbar DMA Play: Illegal buffer size (from 0x0d84e8 to 
> > > > 0x0d84bc)
> > > > Segmentation fault
> > > > -----------------------------------
> > > >
> > > >
> > > > > > EKO epidemic music disk : no change
> > > > > >
> > > > > > underscore demo : no change
> > > > > >
> > > > > > Hmmm demo : better than before (runs with MMU  + cycle exact + 
> > > > > > prefetch) (not tested to the end)
> > > >
> > > > Verified.
> > > >
> > > >
> > > > I've updated now info for all demos that have had
> > > > issues earlier.  Ones that didn't work with CPU
> > > > exact emulation, do work now.
> > > >
> > > > Except for earlier and above mentioned issues,
> > > > there were no regressions in them.
> > > >
> > > >
> > > > I will next go through some Falcon games & demos
> > > > that haven't had problems with TOS4 in earlier
> > > > Hatari versions, but had problems with EmuTOS.
> > > >
> > > > If they don't work with EmuTOS, I'll check whether
> > > > they still work with TOS4.
> > > >
> > > >
> > > >     - Eero