Home Page Forums MAME MAME ex- USSR and Soviet Bloc hardware

I should have crossposted this (http://forums.bannister.org/ubbthreads.php?ubb=showflat&Number=46766#Post46766) -- I've emulated the clone keyboard, and the video board driver is mostly usable (as long as you don't press Caps Lock . It's essentially an embedded terminal -- takes power from Q-bus, keyboard input, talks to host over serial port, and outputs composite video.

The keyboard uses nearly identical protocol and scancode set as LK201 -- f.e. it ignores the keyboard mode commands (LK_CMD_MODE in http://www.netbsd.org/docs/Hardware/Machines/DEC/lk201.html) and there is an extra key to switch layout (Latin/Cyrillic).

Keyboard driver needs more work though -- bidirectional mode is missing at the moment since video board does not use it.

Some progress -- the board is talking to SIMH running Ultrix, and the shifted lines at the bottom are likely because of core bugs.



... and to find them I need to boot RT-11 and run tests from FTMON (XXDP workalike), but that is not easy -- in this screenshot RT-11 is running off virtual disk but not talking to the console for some reason:

Nice. BTW, emulated LK-201 should send fine now.

What is the good way to cause a bus error on access to unmapped memory? I have added a fake input line to T11 core and assert it from AM_READWRITE handlers, but this does not seem the right way.

That's exactly how a bunch of 680x0-based drivers do it.

Yeah, I stole idea from those

But it's rather clunky -- have to recalc am_readwrite windows manually, every time I add a new device + dunno how to convert 'offset' passed to handlers to offset from start of RAM.

Bus error usually *is* an input pin on the processor, so that's correct how to do it.

offset is simply the number of data bus units. For 8-bit data buses, offset = byte address, minus the start of the handler. For 16-bit, offset is the number of 16 bit words, and for a 32-bit data bus it's the number of 32-bit words.

So for this memory range on a CPU with a 32-bit data bus:

...a CPU access to address 0x3010 would be an offset of 4 (0x3010 - the base 0x3000 = 0x10. Converted to 4-byte words = 4).

Elektronika MS1201.02 (a Q-bus board, basis of some desktops in DVK series) WIP -- showing ODT prompt The cpu on this board (K1801VM2) has ODT not in firmare but in 'shadow' ROM and has extra instructions to switch between 'user' and ODT ('halt') modes.

with bankdev plugged in, performance doesn't suck so much and now I see there actually is a core bug

Bankdev is definitely your friend