Re: LisaEm 1.0.0-RC2 Available for download

From: Ray Arachelian <ray_at_email.domain.hidden>
Date: Mon, 09 Jul 2007 11:04:28 -0400

gilles wrote:
> the most precise 68k core is the musashi (this is the core I choosed
> after 2 tries (first one was the Castaway 68k core, then I tried
> scream)). But It's not easy to compare since status bits are not
> handled in the same way.

What other cores have you tried?

> I did this kind of job in the past to debug a java 6809 core. I traced
> all the registers at every instruction. I then compared the results.

It's a bit more complex in this case - I suspect I may have addressing mode issues, so in this case, I either have to maintain two address spaces, or trap the memory fetch/store requests from one core and
compare it to the results in the other.

It's probably easier to maintain two address spaces, and then compare them after both instructions execute, but of course that's much slower. Alternatively, some sort of small table that has enough space to handle enough accesses for a full MOVEM instruction would work, and clear the table inbetween executed opcodes.

The reason that I suspect this is because I've tested the Generator core against a real 68040 - so either I'm running into flag differences between the 68040 and the 68000, or it's something that's not tested, which means branching, or addressing modes.
> I've noticed a problem with ROM selection in your new version. I had
> to edit the conf file. This affects the Win32 version (I run XP pro).


What's the issue you've encountered? Received on 2007-07-09 11:04:31

This archive was generated by hypermail 2.4.0 : 2020-01-13 12:15:14 EST