Dag Lem

I noticed that Denise now bails out on unknown options, and that the option -cia-6526a is changed to -cia-8521. The attached patch makes the testbench run for Denise again.

CIA ICR bug when switching serial port direction

It does not :-/

... and here is the patch :-)

See https://sourceforge.net/p/vice-emu/bugs/1219/?page=3#d27d/7f92 above. As far as I can tell, Denise has the exac same problem. The only reason why the tests seemingly pass for Denise, is that the options to choose between old an new CIA are wrong. See attached patch.

As far as I can tell, only the 0_7f tests are really needed, as they go through all the baud rates. The other tests can be useful to hunt for bugs, but they can always be assembled and started manually.

Based on your findings, the code path taken for CIA_MODEL_6526 in ciacore.c:strange_extra_sdr_flags is really for the 6581 4485 chips, for which there is no separate model in the emulator. The attached patch fixes this issue by commenting out the code path for the outlier 4485 chips, allowing the tests to pass for both (normal) old and new chips.

@gpz What I'm wondering is how come the (original) cia-sdr-icr tests do not take the CIA model into consideration, while VICE does show different behavior depending on the model. It seems like what is supposed to emulate an 8521 chip in VICE actually emulates an 6526, since the tests pass on real hardware with 6526 chips, but only pass in VICE with emulation of 8521 chips. What is your take on this? Unfortunately the C64Cs I have lying around (which possibly have 8521 chips ) both have broken power...