Re: [hatari-devel] Data access breakpoints?

[ Thread Index | Date Index | More lists.tuxfamily.org/hatari-devel Archives ]


Le 08/09/2018 à 10:00, Miro Kropáček a écrit :
On Fri, 7 Sep 2018 at 21:49, Nicolas Pomarède <npomarede@xxxxxxxxxxxx <mailto:npomarede@xxxxxxxxxxxx>> wrote:

    unfortunately it's not supported yet, as this requires to intercept all
    read/write acceses and can have much more impact on emulatoin speed
    than
    just checking PC value for example.

Trust me, it can't be slower than trace cpu_disasm ;-)

    So, you can't detect read access, but you can detect write accesses if
    the value is changed. For example :
    b (0xff8260.b) ! (0xff8260.b)
    will break each time resolution is changed in $ff8260.

Yes, I'm aware of this. However as I mentioned, if I'm interested where in code a certain operation happens (accessing Videl for instance), sometimes it writes the same value again and again so this wont work.

I was thinking about something like

b (0xff8260.b) ! (0xff8260.b) || (0xff8260.b) = (0xff8260.b)

but I guess that would trigger every single cycle, not only when written.

Yes, this would not work as expected.

I'm also using "--trace all,-int" when I want to precisely track a memory acces, or sometimes I temporary add my own code into get_byte/get_word/get_long, which can be a temporary solution if you're building hatari from sources.

In the meantime, there's no easier solution :(

Nicolas




Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/