Re: [hatari-devel] tracing of cpu (dsp) registers

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 24/01/2018 à 23:43, Miro Kropáček a écrit :
> On 25 January 2018 at 09:41, Miro Kropáček <miro.kropacek@xxxxxxxxx 
> <mailto:miro.kropacek@xxxxxxxxx>> wrote:
>
>
>     On 25 January 2018 at 06:30, Eero Tamminen <oak@xxxxxxxxxxxxxx
>     <mailto:oak@xxxxxxxxxxxxxx>> wrote:
>
>         Just to verify, do you mean running "registers" command after
>         *each* executed CPU instruction?
>
>     Yes. Literally a command like "trace cpu_reg" (dsp_reg wouldn't hurt
>     either). Sometimes it's more useful to see the flow of register
>     changes than the instructions alone.
>
>
> Just a suggestion, don't do this trace output so huge -- maybe just 
> basic registers (Dn, An, PC, SR) to keep the output simple and easy to 
> follow.

Hi

I sometimes does this myself when I want to trace a problematic 
emulation issue : I add a call to m68k_dumpstate after each call to 
disasm at current PC.

This makes a pretty big output file in the end, but I think that's 
something you have to accept in case you want to get such low level of 
details. Outputting the result of several seconds of emulation can 
easily gives some text file of 1 or 2 GB (when combined with "--trace all")

Then you can filter this output file with some dedicated tools / scripts 
(grep, awk, perl, whatever you like), but I don't think the overhead of 
maintaining a simpler dump function with basic registers is worth the effort

Nicolas