Re: [hatari-devel] IKBD regression with EmuTOS

[Thomas Huth <th.huth@xxxxxx>]

 Hi!

Am Sun, 10 Feb 2013 15:16:50 -0500
schrieb "Roger Burrows" <rfburrows@xxxxxxxxx>:

> According to the IKBD specs from Atari, sending 0x80 0x01 resets the
> keyboard, which performs a self-test.  If the test is successful, the
> ikbd will return a code of 0xF0 within 300ms.

That must be an old spec file... The lowest nibble in that byte
actually tells you the version number of the IKBD ROM. Only very early
STs (I guess the 520 series?) returned 0xF0, all newer versions return
0xF1 instead (IIRC I also saw this on my 1040 STf).

> So I would say that if Hatari returns 0xF1, it is  returning the
> wrong code.

There are some games or demos which expect 0xF1 (e.g. Dragonnels demo),
so Hatari is always using this value.

> And EmuTOS should be waiting for a 0xF0 with a 300msec
> timeout.

I think the cleanest solution would be to mask the received byte with
0xF0 to ignore the version nibble.

 Thomas