Re: [hatari-devel] 68030 MMU work

[ Thread Index | Date Index | More Archives ]


The MMU for the 68030 should be complete when it comes to functionality (all functions emulated: PMOVE, PFLUSH, PTEST, PLOAD; working table search; working ATC with entry replacement algorithm, working transparent translation). There are some minor known bugs remaining with the PTEST instruction. I need to see what real software expects as a result in certain conditions. Anyway this should not cause problem unless using very advanced software.

The problems with connecting the MMU to the CPU remain.
I think now it would be a good time to test the existing code with Hatari, if you are interested.
All that needs to be done, is connecting the MMU to Hatari. Like before, all critical code is in cpummu030.c/h.



Am 20.08.2012 um 14:30 schrieb Andreas Grabher:

> Dear Hatari community!
> I've been working on adding 68030 MMU emulation to the Hatari-based "Previous" emulator (using new winuae cpu).
> I've succeeded to implement a table search procedure and an ATC. Looking up physical addresses from the ATC works. It seems to be OK for the early boot process of the mach kernel.
> The code is in SVN:
> In its current state, the MMU is still very incomplete. The actual problem is, that i don't know how to properly connect the MMU to the CPU code. At the moment it is "glued" on top of the 68040 MMU in a quite bad way.
> The second problem is, that i do not understand how the CPU emulation handles function codes. I'm confused with "super", "data", regs.sfc, regs.dfc, regs.s.
> Also i have some problems understanding all to the 68040 MMU code.
> Any help with these problems would be greatly appreciated!
> Feel free to use and modify the code. For testing, it should be easy to add to Hatari. You just need cpummu030.c and cpummu030.h. They are in the cpu directory (new winuae cpu).
> Some minor changes/additions to cpummu.h and newcpu.c/h are necessary (see sources).
> Regards,
> Andreas

Mail converted by MHonArc 2.6.19+