Re: [hatari-devel] asm56000.ttp problem |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/hatari-devel Archives
]
Hi,
Laurent, please also update Hatari. With yesterday's Hatari
version there were "infinite" bombs after asm56000.ttp terminated,
with Nicolas last changes, Hatari detects double bus error and
shows a dialog about it.
- Eero
On perjantai 09 tammikuu 2015, Eero Tamminen wrote:
> On torstai 08 tammikuu 2015, Laurent Sallafranque wrote:
> > I've just tested with hatari 1.4 and the behaviour is the same.
> > At first glance, I would suspect either a EOF problem (the line not
> > working is the last one) or the parameter string used into tomshell.tos
> >
> > If I call directly asm56000.ttp, I've got another behaviour.
> > If I call asm56000.ttp from tosshell.tos without arguments, I've got
> > the same behaviour as above
> > If I call asm56000.ttp from tosshell.tos with the following arguments
> > (-A -B decoder.asm >make\debug.txt), I get the error message.
>
> If you just want to use the program instead of debugging problems
> in running tosshell, I suggest giving the arguments to the program
> with Hatari facilities:
> hatari-prg-args.sh --conout 2 -- ./asm56000.ttp -A -B decoder.asm
>
> ("hatari-prg-args.sh" is script for tools/ directory, latest version
> of that and of Hatari need to be in your $PATH for above to work.
> Or you can hardcode your latest Hatari build path to the script.)
>
>
> Alternatively, you could use "hconsole", the Hatari remote control shell.
> It requires a bit more effort, but you automate a lot more with it and
> it should work already with Hatari v1.6 release (and later).
>
> Here's how you would automate same with "hconsole" & EmuTOS shell:
> $ hconsole.py commands.txt --exit -- -d tmp/ -m --tos etos512k.img
>
> (i.e. run hconsole commands from "commands.txt" and exit after
> giving them Hatari, start Hatari with "-d tmp/ -m --tos etos512k.img"
> options, where "tmp" is the directory where asm56000.txt is.)
>
> ----- commands.txt content ------
> # without this EmuTOS boot takes 10s more
> setopt --fastfdc on --fast-forward on
>
> # wait for EmuTOS to boot
> sleep 3
>
> # Invoke EmuCON with ^Z: Control down, press Z, Control up
> keydown 29
> keypress Z
> keyup 29
>
> # press Return
> keypress 28
>
> # output command to execute
> text asm56000.ttp -A -B decoder.asm > debug.txt
> keypress 28
>
> # wait for build to finish
> sleep 3
>
> # show saved command output
> text type debug.txt
> keypress 28
> ---------------------------------
>
> (Files in tools/hconsole/ have more info on hconsole.)
>
> > I would say that the >make\debug.txt parameter is not correctly
> > interpreted, but I have no clue how to verify this.
>
> In debugger doing at program start (pc=text):
> info basepage
> info gemdos
>
> Tells about command line parameters given to program and
> about the GEMDOS open file descriptors state. Can you
> provide that with full GEMDOS trace?
>
>
> - Eero
>
> PS. Are you using Miro's Qt/C++ based LOD -> binary converter?
>
> I've converted that into plain ANSI-C so that it can be run also
> under Hatari (as part of my native BadMood builds under Aranym/MiNT),
> in case you're interested...
>
> > Regards
> > Laurent
> >
> > Le 08/01/2015 21:21, Nicolas Pomarède a écrit :
> > > Le 08/01/2015 21:17, Laurent Sallafranque a écrit :
> > >> Hi all,
> > >>
> > >> It seems that there's a bug into hatari with asm56000.ttp.
> > >> I tried to recompile the mpeg2 library for some tests and it doesn't
> > >> work.
> > >> Mikro has tried on his falcon and it works.
> > >>
> > >> The thread for more infos is there :
> > >>
> > >> http://dhs.nu/bbs-coding/index.php?request=4814
> > >>
> > >>
> > >> It seems that something goes wrong when using hatari, but I don't
> > >> know exactly where to start from.
> > >> I've activated the gemdos traces but I don't see anything that jumps
> > >> on me.
> > >
> > > Hi,
> > >
> > > did you try with some older Hatari versions ? With old cpu core or
> > > new cpu core ? I guess it should not require cycle exactness, so it
> > > should work with any cpu config.
> > > It could help to track regression.
> > >
> > > Nicolas