audio3d: sceAudio3dPortGetAttributesSupported() - init num_capabilities variable to zero

[mailwl]

when capabilities is nullptr, sceAudio3dPortGetAttributesSupported() should return count of available capabilities
this PR initialized num_capabilities variable to zero, to return empty capabilities

[georgemoralis]

is that tested with specific game?

[StevenMiller123]

I don't think we should reduce the log severity to debug when the function is still mostly stubbed out.

[georgemoralis]

else { //move error log here }

[mailwl]

looks like returning zero make audio3d off, and no need to fill zero-size array

[mailwl]

is that tested with specific game?

saw it in CyberPunk, but there is setted to 0 before calling. But if it not set, we potentially return garbage:

__int64 __fastcall sub_3F89F20(__int64 a1)
{
  unsigned int v1; // r14d
  char *v2; // rbx
  __int64 v3; // rcx
  __int64 v5; // [rsp+0h] [rbp-40h] BYREF
  unsigned int count; // [rsp+Ch] [rbp-34h] BYREF
  __int64 v7; // [rsp+10h] [rbp-30h]

  v1 = 0;
  v7 = *(_QWORD *)_stack_chk_guard;
  count = 0;
  sceAudio3dPortGetAttributesSupported(a1, 0, &count);
  v2 = (char *)&v5 - ((4LL * count + 15) & 0xFFFFFFFFFFFFFFF0LL);
  sceAudio3dPortGetAttributesSupported((unsigned int)a1, v2, &count);
  if ( count )
  {
    v3 = 0;
    while ( *(_DWORD *)&v2[4 * v3] != 3 )
    {
      if ( ++v3 >= (unsigned __int64)count )
        return 0;
    }
    LOBYTE(v1) = 1;
  }
  return v1;
}

[mailwl]

looking at source, one capability in use, so i return it to caller

[squidbus]

Instead of casting you can use fmt::ptr