Homebrew GameYob 3DS - GB/C emu

[Idaho]

Here to report an issue I'm having with pokemon crystal:

The game runs flawlessly but sometimes when I open a menu or change from an area to another, the screen goes blank and I can't do anything, not a big problem for me as I save very often but still I think it might require investigation.

Thank you for your dedication to this project guys

 

[ShadowOne333]

If I download one of the newest build CIAs, should I uninstall the older CIA first from my 3DS or can I simply install over the currently installed one?

 

[MarkDarkness]

The scaling filter option is there, however apparently it doesn't work properly. Fixed now. Don't know why you'd need a clock though...?

Additionally, I re-implemented autosaving from the DS version, fixed a few misc things, and made a bunch of changes internally.

Download: https://www.dropbox.com/s/ms1kes4jhrb3meg/gameyob.zip?dl=0

EDIT: If you downloaded, please redownload. Had to make a few adjustments to rendering code.

Thank you, I'm going to test it tonight. The clock is just a nice feature to have, I guess... and since the code is already there in theory. Dunno how different the 3DS libraries are in relation to the DS ones to justify tinkering around with it, though.

One thing that I do still miss though is the line highlight when selecting games or a coloring of the currently selected ROM instead of just an asterisk. It does make a big difference when viewing conditions are not ideal (like the bus I'm playing in right now).

 

[davhuit]

If I download one of the newest build CIAs, should I uninstall the older CIA first from my 3DS or can I simply install over the currently installed one?

You should uninstall the old one first. I think the id is 1927 or something like that.

 

[ShadowOne333]

You should uninstall the old one first. I think the id is 1927 or something like that.

Would it work if I delete the title from Data management and reinstall the CIA?

 

[Deleted User]

I just noticed the single screen option is doing the same as the "Game Screen" one (Top/Bottom).

Normal behavior?

Seems like it was intended, but honestly it doesn't really make sense to me. I'll probably stop that from happening.

Here to report an issue I'm having with pokemon crystal:

The game runs flawlessly but sometimes when I open a menu or change from an area to another, the screen goes blank and I can't do anything, not a big problem for me as I save very often but still I think it might require investigation.

Thank you for your dedication to this project guys

Noted.

Thank you, I'm going to test it tonight. The clock is just a nice feature to have, I guess... and since the code is already there in theory. Dunno how different the 3DS libraries are in relation to the DS ones to justify tinkering around with it, though.

One thing that I do still miss though is the line highlight when selecting games or a coloring of the currently selected ROM instead of just an asterisk. It does make a big difference when viewing conditions are not ideal (like the bus I'm playing in right now).

I could probably add that back in, won't take too long.

EDIT: Okay, two changes made. I changed the single screen option into a pause on menu option, because honestly that was the only potentially useful thing single screen did. The menu is always hidden while playing anyway. I also added line highlighting to the file selector.

Download: https://www.dropbox.com/s/ms1kes4jhrb3meg/gameyob.zip?dl=0

 

[Drenn]

EDIT: Okay, two changes made. I changed the single screen option into a pause on menu option, because honestly that was the only potentially useful thing single screen did. The menu is always hidden while playing anyway. I also added line highlighting to the file selector.

I only added the single screen option because someone was using a DS where one of the screens didn't work properly. I never got around to implementing the feature properly on a 3DS. I wouldn't worry about it though.

 

[Deleted User]

I only added the single screen option because someone was using a DS where one of the screens didn't work properly. I never got around to implementing the feature properly on a 3DS. I wouldn't worry about it though.

Ah, okay. I didn't want to have to mess with screen formats a bunch, since the ctrulib console changes the console screen format to single buffered RGB565, and it didn't really seem that useful to me anyway.

On another note, do you have any idea what in this commit might have broken any games? Specifically, Pokemon Pinball (MBC5).

 

[Drenn]

Ah, okay. I didn't want to have to mess with screen formats a bunch, since the ctrulib console changes the console screen format to single buffered RGB565, and it didn't really seem that useful to me anyway.

On another note, do you have any idea what in this commit might have broken any games? Specifically, Pokemon Pinball (MBC5).

Crazy idea, but could it be something about the rumble feature?

It was in the middle of my overhaul so it could be any number of things. But I recall people saying it only crashed after hitting the bumper, or something like that...

 

[Deleted User]

Crazy idea, but could it be something about the rumble feature?

It was in the middle of my overhaul so it could be any number of things. But I recall people saying it only crashed after hitting the bumper, or something like that...

I did notice that the hasRumble function was using MBC instead of the actual mapper value, but fixing that didn't fix the issue. Also, from my further tests, it seems to happen for the most part at random.

EDIT: Oh, and it still happens with rumble turned off in the game's options.

 

[Drenn]

I did notice that the hasRumble function was using MBC instead of the actual mapper value, but fixing that didn't fix the issue. Also, from my further tests, it seems to happen for the most part at random.

EDIT: Oh, and it still happens with rumble turned off in the game's options.

If it really is random then I'd suspect something is wrong with the interrupts. I'm grasping at straws though...

 

[Deleted User]

If it really is random then I'd suspect something is wrong with the interrupts. I'm grasping at straws though...

I'll take a look. In case I forgot to mention, what happens is the game locks up momentarily, sprites like the flippers disappearing, and it resets to the beginning of execution with a lot of graphical glitches. This only occurs during a pinball game.

 

[Drenn]

I'll take a look. In case I forgot to mention, what happens is the game locks up momentarily, sprites like the flippers disappearing, and it resets to the beginning of execution with a lot of graphical glitches. This only occurs during a pinball game.

Yeah, what's happening is code execution somehow goes back to the very beginning. It forgets that it's running on a GBC, and thinks it's now running on a GB, which is the cause of the graphical glitches.

 

[Deleted User]

Yeah, what's happening is code execution somehow goes back to the very beginning. It forgets that it's running on a GBC, and thinks it's now running on a GB, which is the cause of the graphical glitches.

I put some debug into handleInterrupts to print IRQ information:

printf("Triggered IRQ %d, %hu\n", irqNo, isrVectors[irqNo]);

And when it happens, it prints this out:

Triggered IRQ -1, 65535

Seems like it might be interrupts after all?

 

[Drenn]

I put some debug into handleInterrupts to print IRQ information:

printf("Triggered IRQ %d, %hu\n", irqNo, isrVectors[irqNo]);

And when it happens, it prints this out:

Triggered IRQ -1, 65535

Seems like it might be interrupts after all?

Oh boy. Now, if I understand correctly, this line

    int irqNo = __builtin_ffs(interruptTriggered) - 1;

is returning -1? And this can only happen if interruptTriggered is 0?
The culprit must be this:

        if(interruptTriggered) {
            /* Hack to fix Robocop 2 and LEGO Racers, possibly others. 
             * Interrupts can occur in the middle of an opcode. The result of 
             * this is that said opcode can read the resulting state - most 
             * importantly, it can read LY=144 before the vblank interrupt takes 
             * over. This is a decent approximation of that effect.
             * This has been known to break Megaman V boss intros, that's fixed 
             * by the "opTriggeredInterrupt" stuff.
             */
            if(!halt && !opTriggeredInterrupt)
                extraCycles += runOpcode(4);
            extraCycles += handleInterrupts(interruptTriggered);
            interruptTriggered = ioRam[0x0F] & ioRam[0xFF];
        }

interruptTriggered must be unset when it does that extra runOpcode call. This is indeed a weird hack... perhaps simply adding another interruptTriggered check before calling handleInterrupts will fix it?

 

[Deleted User]

Oh boy. Now, if I understand correctly, this line

    int irqNo = __builtin_ffs(interruptTriggered) - 1;

is returning -1? And this can only happen if interruptTriggered is 0?
The culprit must be this:

        if(interruptTriggered) {
            /* Hack to fix Robocop 2 and LEGO Racers, possibly others.
            * Interrupts can occur in the middle of an opcode. The result of
            * this is that said opcode can read the resulting state - most
            * importantly, it can read LY=144 before the vblank interrupt takes
            * over. This is a decent approximation of that effect.
            * This has been known to break Megaman V boss intros, that's fixed
            * by the "opTriggeredInterrupt" stuff.
            */
            if(!halt && !opTriggeredInterrupt)
                extraCycles += runOpcode(4);
            extraCycles += handleInterrupts(interruptTriggered);
            interruptTriggered = ioRam[0x0F] & ioRam[0xFF];
        }

interruptTriggered must be unset when it does that extra runOpcode call. This is indeed a weird hack... perhaps simply adding another interruptTriggered check before calling handleInterrupts will fix it?

That does indeed seem to fix it! At least, the save state I use to trigger it doesn't do so anymore. Thanks!

Oh, and to that guy who wanted a clock, I added and touched up the clock printing code from the DS version. Check the "Console Output" option.

Fixed Download: https://www.dropbox.com/s/ms1kes4jhrb3meg/gameyob.zip?dl=0

EDIT: Thinking about it... the problem probably started with that commit because of how interruptTriggered was changed to a global variable. Before, it wasn't able to be changed by runOpcode, I guess.

 

[ShadowOne333]

Damn... This updates are AMAZING!
I might wait a few days before I install the CIA though, in case another commit is uploaded.

 

[davhuit]

Would it work if I delete the title from Data management and reinstall the CIA?

Yeah, it's the same.

 

[Deleted User]

By request, I've re-added BIOS support for those who like to see the boot screen before games. I've also made sure it doesn't interfere with palette options.

Download: https://www.dropbox.com/s/ms1kes4jhrb3meg/gameyob.zip?dl=0

 

[Smoker1]

What should the BIOS be named and where should they be placed?

Edit: I dont know where I put them, but hey, I got it to work. Also gbc_bios.bin works for the File Name.
Now just need to figure out how to get the Borders working.