G1-ATA Devices

[OzOnE]

Hi,

I have quite a good test app for the G1 port lock now.

It just checks some known data on the HDD attached to the DC.

I can do the "gdrom_activate" test, and it lets me read the data back fine with the stock BIOS or SWAT's "sd_loader_with.bios".

If I flick the Flash BIOS switch during the test (only takes about 500ms), it corrupts the BIOS and the test fails.

I can do the test again without changing the switch and the test passes fine.

If I try a PIO transfer with G1 locked out, it only returns floating data, or 0x7F7F.

If I try a DMA transfer with G1 locked out, the transfer freezes the DC.

But, I now have a way to test it, so just need to send it data instead of the BIOS.

I will have to hook up the rest of the G1 address signals (about 11 lines) to the FPGA soon.
Of course, most of the signals are already on the G1 port, but it needs the extra ones for direct access to the full BIOS.

That will work in a similar way to what I did on the NAOMI - I can program the Flash on the FPGA board with the stock DC BIOS, then modify say one Word of data at a time.

I should be able to increment to the next value when the DC tries to read a specific location in the "fake" BIOS until it passes the checksum.
(try in a loop, like you say. )

I don't think it locks out access to the BIOS, only to the ATA / ATAPI devices.
It probably just inhibits the /CS1 and /CS0 signals on the G1 port, and blocks the data?

EDIT: Actually, it can't be locking access to the BIOS, otherwise it wouldn't be possible to retry the test and unlock the G1 port.

It will be a bit of a brute-force hack, but at least we'd have a way of "signing" the custom BIOS images.

I even tried testing many different parts of the BIOS files earlier to see if there was a standard checksum algo it uses.
I'm not a cryptographer though, so that didn't go too well. lol

I tried reading only half of the BIOS through the magic 0xa05f74e4 register, but that failed the check too?

I have a hunch that it does the checksum on "blocks" of the BIOS at a time.
That way, they can still allow for different BIOS images, but just add some padding bytes to make the checksum pass?

OzOnE