Home Page Forums Other Emulator Message Boards Modeler and M1 Any future updates to M1?

Ok, thank you.

Apologies if this question is stupid... If SDL is used (on a Mac) won't it talk directly to the sound hardware via Core Audio, so the SDL routines on their own should be sufficient? The point I'm trying to make is: can I just take the oss.cpp file and remove most of the code apart from the SDL-specific routines and that should get me fairly close to a working solution?

In theory, but the SDL routines there almost certainly aren't for SDL 2.x as far as I remember (I'm not sure 2.x *existed* at that time) so they won't build against current SDL 2.30 (or the 3.0 preview).

Basically what I'm saying is that the source drop you found is self-service. That was true in 2010 when I still remembered any of it, but it's way more true 14 years later.

Ok, understood and thank you again for your help - I appreciate you are busy with new projects and M1 was a very long time ago. I did think SDL 2 would be too recent, and using SDL 1 probably won't work on the later OSs so I will have to rewrite the SDL routines.

If I make any meaningful progress I will post here again. If any Mac users are interested in working on this please PM me.

Update: I've managed to get the command line version built and working, sort of:



I've tested with Hot Chase, Cyberbots, Aggressors of Dark Kombat and Godzilla (pinball machine) so far and only the first songs as the key commands don't appear to be working, but I definitely hear some sound! I had to comment-out some code in roms.c that was causing an illegal hardware instruction error, and also remove the reference to the lasound library in the make file before it would build. I only have SDL 2 on my machine but this doesn't seem to be an issue. Using pulseaudio didn't work even though I have the library installed, so I'll stick with SDL for the moment.

Batman - Revenge of the Joker (Genesis game) gives me a segmentation fault. The fm2612.c file does not appear to be being included in the build, so I'll ignore these games for the moment as I presume it's related to this file.

I've identified the cause of the segfaulting on the Genesis games, which is due to the YM2151 driver code when run on a 64-bit system (I think). This also affects Atari System 1 boards and various other games. It appears that the key code (KC) channel/register needs to be initialized when the chip is started/reset - I found this out by looking at the MAME 0.100 source code.

In the M1 code, open sound/ym2151.c and find the YM2151ResetChip function, then where you have this line

chip->oper[i].volume = MAX_ATT_INDEX;

immediately after it add

chip->oper[i].kc_i = 768; /* min kc_i value */

Hope that helps someone. I do not claim to understand anything about the YM2151 chip, I just copied-and-pasted the code.

This fixes YM2151 games on my Android port also. I can listen to MERCS and Magic Sword again. Thanks!

Thanks for your feedback! I'm still getting intermittent crashes on some games, so I think there may be other values that need to be initialized but I haven't had time to look at M1 recently. The crashes usually happen once you've switched between different games. There is a newer version of the Genesis driver in the source code but when I swapped it with the old driver I didn't get any sound. Arcade games that use the YM2151 chip usually play fine now, it's the Genesis games that are not working reliably.

I'm wondering if R. Belmont and Richard Bannister would consider open sourcing M1 and especially the Mac port to Github to allow others to take on continuing it.

I don't know how many times I need to say this – I get someone asking about it every few months – but the Mac port of M1 was open-sourced a very long time ago.

You can download it here.

The code isn't massively useful, mind, insofar as it is written in Carbon and that doesn't build or run on current machines.