Re: [hatari-devel] MEMWATCH freezes Hatari

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 16/10/2018 à 18:47, Toni Wilen a écrit :
> > > In that case, ps=00003f00 oc=00004211 and mmu030_opcode=4211
> >
> > This looks perfectly fine. 4211 = CLR.B (A1) which is the instruction
> > that was supposed to be retried.
>
> To make this clearer:
>
> if mmu030_opcode=-1: instruction opcode fetch caused fault and whole
> instruction fetch needs to be retried, including nasty opcode stagea/b
> prefetch stuff if more compatible.
>
> if mmu030_opcode>=0: value equals opcode that caused data fault. Retry
> by jumping back to "regs.opcode = regs.irc = mmu030_opcode;". (Not using
> complete fetch retry that -1 case uses).

Thanks for the clarification, I also saw those unused fields in 
Exception_build_stack_frame ; so my bad, unlike what I said to you 
earlier in another mail, my disasm traces are not correctly placed, I 
misread the "for (;;)" loop and didn't see the retried instruction :(

I will make further traces/test, because there was still some errors 
when CE mode is on (maybe a cache issue)

Nicolas