Memory Inspector show nothing on GB or GBC games about ralibretro HOT 8 CLOSED

The Memory Inspector defaults to displaying the first address in a Note. I assume that most Notes are based on offset memory (based on VBA?). Clear the value being watched or compensate for the 0x00c000? offset to view the correct ones.

In any case, that's a UI issue for RAIntegration.

from ralibretro.

Seems like this happen when the displaying note is trying to locate an address outside of the 0x7FFF.
Though GB/GBC have a range up to 0xFFFF

from ralibretro.

I'm guessing this is a duplicate of #19, which was supposedly fixed in 1.0.5. What version shows up when you click "About"?

from ralibretro.

nm, you said 1.0.7 in your first post.

from ralibretro.

Also RALibretro commit hash is 7584575 in case this help.
Its the latest build we have in our download page.

from ralibretro.

I took a look at this, and while I understand what's happening, I'm not sure how to fix it. Someone more familiar with the libretro cores will have to address this.

The basic problem is that the GB/GBC cores only map the System RAM to the integration DLL as $0000-$7FFF.

This is wrong. RAVBA-M actually registers the full address space for GB/GBC, so the System RAM should be mapped from $C000-$DFFF and echoed at $E000-$FDFF. I'm not sure why _core.GetMemorySize(RETRO_MEMORY_SYSTEM_RAM) is returning 32K, I think it should only be 8K.

Additionally, the Save RAM is completely ignored by RALibRetro and needs to be mapped to $A000-$BFFF. And I have no idea what needs to be mapped from $0000-$9FFF.

The solution to this seems to be to replace the mapping with a raw read for GB/GBC:

      RA_InstallMemoryBank(0, (void*)::memoryReadRaw, (void*)::memoryWriteRaw, 0x10000);

unsigned char Application::memoryReadRaw(unsigned addr)
{
  const struct retro_memory_map* mmap = _core.getMemoryMap();
  const retro_memory_descriptor *desc = mmap->descriptors;
  const retro_memory_descriptor *end = desc + mmap->num_descriptors;
  while (desc < end)
  {
    if (desc->start <= addr && desc->start + desc->len > addr)
    {
      unsigned char* memory = (unsigned char*)desc->ptr + desc->offset;
      return memory[addr - desc->start];
    }

    ++desc;
  }

  return 0;
}

This seems to mostly work with the Gambatte core, but reads between $E000 and $FDFF fail because the echo ram isn't registered in the mmap.

The mGBA core doesn't provide a mmap, so fails miserably (does it only support GBA?).

from ralibretro.

The mGBA core doesn't provide a mmap, so fails miserably (does it only support GBA?).

Yes.

https://docs.libretro.com/guides/retroachievements/#game-boy-game-boy-color

from ralibretro.

solved in eee25ff

from ralibretro.