CD32Load

[Retro1234]

Thanks for the Reply

[jotd]

Toni you could reproduce it with your CD32+floppy? but maybe pluggin the floppy overrides the CIA behaviour (SX32 overrides it: cool spot worked before I found the level 6 issue).

I want to test it on my terrible fire + compact flash. If the old version of cool spot doesn't work, then bingo too.

[Amigajay]

Originally Posted by jotd

@Amigajay: I checked and the interrupt is properly handled (and I'm pretty sure I tested GG on real HW and it worked, maybe a slave update issue got to check that).

Tried 1.2 with the same result so the slave update didnt change anything.

[Toni Wilen]

Originally Posted by jotd

Toni you could reproduce it with your CD32+floppy? but maybe pluggin the floppy overrides the CIA behaviour (SX32 overrides it: cool spot worked before I found the level 6 issue).

I can test it with and without floppy. But I really want 100% confirmation it is level 6 interrupt -> different color0 writes in each interrupt handler required. I don't do assumptions!

[Amigajay]

I made a quick single game iso just to show the problem i uploaded to the zone, using MASKINT6 on the cd32load command line is needed to make the game run on real hardware but has no sfx or music, whilst on winuae it has sfx but no music.

Without MASKINT6 the game has sfx and music on winuae but doesnt load on real hardware.

[Toni Wilen]

So it must be Akiko built-in CIA-B chip difference vs real CIA chips or some unconnected CIA line is hardwired to active state. Which would also explain why floppy adapter "fixes" it.

Does this game (accidentally) enable some other CIA interrupt source than timer (?) used for music/sfx?

[Galahad/FLT]

Does Putty Squad have the exact same issue? Obviously I have the source code to it, and if it does, then it might be a bit easier to locate directly from the source code.

[jotd]

Yes, Putty Squad has the issue. Using MASKINT6 works, else, no way. But I'm sure GG was 1) fixed for this issue, and 2) ran on my CD32 before the drive broke

[Galahad/FLT]

What does MASKINT6 do exactly? Replace an existing level6 or remove it?

[jotd]

since VBR is relocated, MASKINT6 doesn't redirect VBR-relocated level 6 interrupt to the zero page vector that the game installed (much easier to do than to track SR or INTENA for level 6 enable!)

[Toni Wilen]

Where are the files/image that is confirmed having this issue? Or does all have it?

[Amigajay]

Originally Posted by Toni Wilen

Where are the files/image that is confirmed having this issue? Or does all have it?

I posted the iso in the zone, it’s got 47 days left

[jotd]

ok, good news (and bad news): I could test

- cool spot (old slave)
- cool spot (new slave)
- global gladiators (new slave)

on my CD32 using the excellent Terrible Fire + CF

As opposed to CD32 SX1, the CIA B interrupt isn't overridden, so:

- cool spot (old): locks up with checkered screen
- cool spot (new): works (yay)
- global gladiators (new): locks up with black screen

So now I'm puzzled since I'm sure it ran. Or did I persuade myself that it did?

Good news is: I can use HRTMon & WHDLoad to debug that from my console, even if the CD drive is dead (also saves me the hassle to burn CDs to test)

so big thanks again to Supaduper who provided me the TF accelerator for free.

@Toni if you want me to perform some more tests, it's very easy for me: take the CF in Winuae, copy exes, put the CF (carefully) in the CD32, boot, test (and can even write to disk big chunks of data!). When my CD drive is fixed, I'll be able to run games from CD & even debug CD32load for the remaining non working games like Chaos Engine 2 AGA (which works from TF/CF with JST/Whdload with fastmem)

[jotd]

Good news: there was another level 6 interrupt. Has been fixed & tested on my CD32 & TerribleFire. Now the game runs.

update sent to whdload page.

[Amigajay]

Originally Posted by jotd

Good news: there was another level 6 interrupt. Has been fixed & tested on my CD32 & TerribleFire. Now the game runs.

update sent to whdload page.

Great news and good job JOTD i will burn another disc of my megadrive disc and test over the weekend!

Another one ticked off!

[Toni Wilen]

Originally Posted by jotd

Good news: there was another level 6 interrupt. Has been fixed & tested on my CD32 & TerribleFire. Now the game runs.

update sent to whdload page.

I think I have already asked but here it is again: exactly which CIA interrupt mask bit?

[jotd]

Toni I'm not an expert at this. We discussed that earlier. Seems that CD32 (without SX32) triggers some level 6 interrupts from CIA-B. Your floppy adapter changes the behaviour, though.

that's how I fixed that on Cool Spot / GG:

; avoids crashes on CD32
mask_int_6
	move.w	#$2000,$DFF09C
	btst.b	#0,$BFDD00		; acknowledge CIA-B Timer A interrupt
	RTE

[Toni Wilen]

Can you check what $bfdd00.b contains? (Only if bit 7 is set)

Also could you check how many times the interrupt triggers? (for example by writing to COLOR0 in interrupt handler). Is it only every few seconds or every frame or something else?

btw, just reading the register will clear all interrupts, btst #0 here looks quite weird (EDIT: and it is good idea to read $bfdd00 first because now you get two interrupts back to back each time)

[jotd]

Ok Toni, I created the special slave, will boot my CD32 + RAM + HRTMon tomorrow and will report my findings. If you can add this CIA timer feature, that will be cool, although now with the TerribleFire I can do a lot of things directly on the CD32. That thing rocks!

[Retro1234]

Would CD32load work from floppy/df0? I once had Mutantology + Mutation Gold CD that I think is based on RNC loader running from floppy.