missing OTIR

[jariseon]

hi ivan. i've been evaluating cpu emulators and was able to get your z80 up and running in an hour or so. thanks for great work.

in one of my tests i hit an assert. i suppose that was due to a missing OTIR handler. added the code below, though i'm unsure if it is entirely correct. no idea about tick count.

void on_otir() {
  fast_u16 hl = self().on_get_hl();
  fast_u16 bc = self().on_get_bc();
  fast_u8 b = bc >> 8;
  fast_u8 c = bc & 0xff;
  while (b--) {
    fast_u8 t = self().on_read_cycle(hl);
    self().on_output(c, t);
    hl = inc16(hl);
  }
  self().on_set_bc(make16(0,c));
  self().on_set_hl(hl);
  fast_u8 f = self().on_get_f();
  f |= zf_mask;
  f &= ~nf_mask;
  self().on_set_f(f);
}

[kosarev]

Thanks for reporting this, @jariseon. The code looks reasonable. I will need to take a closer look and check it against the specs I have. Should be able to it tomorrow evening. Thanks again!

[kosarev]

@jariseon Addressed in https://github.com/kosarev/z80/commit/8de084ac927a7cc972767c8389c7813b73b81db3. Can you check if the fix works for you, please?

[jariseon]

yes, it works! thanks a million.

btw, do the fast int types provide substantial performance improvements? for some reason i have totally missed them, and only used the basic types like uint8_t in my projects.

[kosarev]

Sure, no problem. Re fast_* types: they are primarily to avoid unnecessary truncations and extra warnings about integer promotions from uint8_t. I don't have any specific numbers, but logic suggests the less the number of truncations, the faster the code.

Closing this.