Re: [hatari-devel] Bcc target address in debugger

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 01/09/2021 à 23:13, Christian Zietz a écrit :
> Eero Tamminen schrieb:
> > Hi,
> >
> > On 29.8.2021 23.39, Miro Kropáček wrote:
> > > I've been playing with Hatari's debugger for the past few days and I 
> > > don't
> > > get one thing. Why does it always show $0000000 as the target 
> > > address? Like
> > > this:
> > >
> > > CPU=$1889a, VBL=972, FrameCycles=565296, HBL=276, LineCycles=48, DSP=N/A
> > > 0001889A 66fa                     BNE.B #$fa == $00000000 (T)
> > >
> > > There wasn't a single case where I'd see something else. The T/F 
> > > indicator
> > > works, the jump itself works, I just don't understand that that zero is
> > > supposed to mean. Latest Hatari, nothing strange, just stepping the 
> > > code.
> >
> > I think either Toni or Nicolas need to answer this...
>
> Well.... It seems that the UAE disassembler uses 0xFFFFFFFF as a special
> value to indicate whether a source or destination address is valid:
> https://github.com/tonioni/WinUAE/blob/1e81c84ed109b945dce6bdbba10eda9da3e35d64/disasm.cpp#L1743 
>
> Hatari, however, for reasons unknown to me, initializes these variables
> to 0 instead:
> https://git.tuxfamily.org/hatari/hatari.git/tree/src/cpu/disasm.c#n1754
> Unfortunately, when printing the target address, it checks for the
> special value to decide which address to use and ends up using deaddr2
> instead of seaddr2:
> https://git.tuxfamily.org/hatari/hatari.git/tree/src/cpu/disasm.c#n2190

Hi

good catch, thanks for pointing it.

it seems that at one point I did not fully merge some WinUAE's change 
into Hatari and this results in these difference (in that case, Hatari 
should use the same code as WinUAE)

I will do a diff between the 2 source trees to see if there're more 
point to fix.

Nicolas