Mount new disc?

[gusmanb]

Hi.

I'm trying to be able to mount a disc after an eject. I have created a callback when the eject happens and then I select a new image and have tried to initialize the disc using disc_init with no luck.

After that I have checked the basiliskII code and I see that in control 65 it checks for devices that must be mounted and then it mounts them by calling Execute68kTrap, but I can't find an equivalent in umac (the trap, not code 65, I see that is still not supported).

Is it possible to mount new disks with the current code or the trap system is missing?

Sorry if anything that I say is too obvious or absurd but I have zero clue on how the Mac works and I'm mostly doing this blindly 😆

BTW. the pico-mac is really cool, I have modified it to work with DVI and included a file browser to select the images from SD, that's why I would like to be able to change the inserted image 😄

Cheers.

[gusmanb]

I've been digging more and I found the m68ki_exception_trap/m68ki_exception_trapN functions and I think that I mostly replicated the code od Basilisk, but when the mount function is called the emulator hangs up. Any hint on what could be happening?

This is the code I have implemented:

bool SonyMountVolume()
{
	sony_drinfo_t* info = &drives[0];
	if(info->to_be_mounted)
	{
		WriteMacInt8(info->status + dsDiskInPlace, 1);	// Inserted removable disk
		WriteMacInt8(info->status + dsWriteProt, info->read_only ? 0xff : 0);
		return true;
	}

	return false;
	
}

static void mount_mountable_volumes(void)
{
	if(drives[0].to_be_mounted)
	{
		SonyMountVolume();
		m68k_set_reg(M68K_REG_D0, drives[0].num);
		m68k_set_reg(M68K_REG_A0, 7);
		m68ki_exception_trapN(0xa02f); //Tried "trap" and "trapn"
		drives[0].to_be_mounted = 0;
	}

}

//This is called in the 1Hz function
void SonyInterrupt(void)
{
	if (!acc_run_called)
		return;

	mount_mountable_volumes();
}

int16_t SonyControl(uint32_t pb, uint32_t dce)
{
	uint16_t code = ReadMacInt16(pb + csCode);
	DDBG("SonyControl %d\n", code);

	// General codes
	switch (code) {
		case 1:		// KillIO (not supported)
			return set_dsk_err(-1);

		case 9:		// Track cache control (ignore, assume that host OS does the caching)
			return set_dsk_err(noErr);

		case 65: {	// Periodic action (accRun, "insert" disks on startup)
				mount_mountable_volumes();
				WriteMacInt16(dce + dCtlFlags, ReadMacInt16(dce + dCtlFlags) & ~0x2000);
				acc_run_called = true;
				// The original emulation code hooked this to mount_mountable_volumes,
				// which called back to PostEvent(diskEvent).
				return set_dsk_err(-1);
		}
	}
 
      /******/
}

[evansm7]

Hi @gusmanb -- sorry for the the very late reply. (I read this originally on my phone, was super hard to look at code, and then ... it went on the pile.). In case you are still working on this here are my thoughts:

Well done, first of all, looks like good progress. One thing that springs to mind is that if you invoke mount_mountable_volumes() from the emulator 1Hz interrupt, it in effect is making the 68k take a random trap no matter what it's currently doing. There will be times when it's not OK to take a trap (e.g. when in the middle of handling one already).

I while back I noticed the minivmac disc support installs a periodic timer task in the 68k .Sony driver replacement (i.e. the equivalent of macsrc/sonydrv.S) and I think it is that task that invokes the mount trap. That means it's 68k code doing the trap at a convenient time to MacOS.

I'm pretty much a beginner in 68k and MacOS programming, but I think this is the right approach: make a timer task call into the emulator (by accessing the faulting_address in a similar way to the other calls) and, if a return value says to remount (based on to_be_mounted) then on return the task does the trap.