G1-ATA Devices

[OzOnE]

OK, I've just done a few tests, and found that it does indeed lock out both PIO and DMA transfers...

I modified my old gd.c code so I can press the A button to do the gdrom_activate() thing manually.

I'm only accessing the HDD atm, and I can "fake" an incorrect BIOS checksum simply by flicking the BIOS select switch on my DC during the activate routine.

If I then read back via PIO, all I get is random junk data.
If I try reading via DMA, the DC will freeze completely.

I've just added a timeout to this too...

static inline int gd_wait_status(uint32 mask, uint32 val) {
int count = 200;
while( (*GD_ALTSTAT & mask) != val); // Wait until the status register has the appropriate value.
{
if(count == 0) return -1;
usleep(1000);
count--;
}
return 0;
}


I thought that would stop the DMA from freezing, but it still seems to do it? (I'm preventing it from setting SB_GDST if anything returns -1).
Not a problem though, I can see from using PIO when the checksum fails.

I've also learned a few things...

1. You CAN just re-activate the G1 port without needing to reset the DC.
2. When the HOLLY / BIOS check fails, it does block access to any non-ROM G1 devices (still not sure if it blocks the /CS signals, but doesn't matter).
3. The lockout definitely blocks it from even reading back the IDE status byte.
4. If the check fails, you can re-activate with a correct BIOS, and the port starts working again.

OK... Just added a quick routine to print the contents of a Word from the HDD - if the check fails, then it says "Activation FAILED!".

Now I can play with getting it to activate from a very small chunk of the BIOS instead.
Or, I can try incrementing through data in the BIOS until it passes the check.

If we can get that working, we can simply add that chunk of data to the custom BIOS.

OzOnE.