SpritesMind.Net

But if you're building a clean toolchain for cross-compiling, you almost certainly don't have the needed newlib. That's the point - you probably only have the "native" newlib for your system, which will NOT make a proper build. Of course. But after you've done it once, there's usually no need to sp...

The two-stage build is so that newlib is used when building the standard c/c++ libs. You build binutils first, build the plain C compiler with no libs, use that to build the newlibs, then use BOTH of those to build the FULL gcc suite of compilers. GCC only really needs the headers from newlib when ...

Maybe it's the difference between using gcc-core + g++ instead of gcc + g++; gcc-core has problems even building with the ARM processor, so maybe it also has problems with the 68000 as well. Nah, I've been using the gcc-core and gcc-g++ packages for quite some time now. Just to test, I built 4.5.2 ...

I've built GCC 4.5.1 for m68k, and I don't recall having to do anything special to get C++ support working. I had to apply a patch so MPFR 3.0.0 would build, but that has apparently been fixed in 4.5.2. The configure command I used was just $ ../../gcc-4.5.1/configure --prefix=/Users/anders/local --...

No I haven't, thanks for the heads up. I'll keep it in mind if this thing ever proceeds beyond the paper-plan stage.

That is true, but I like the chip due to the reduced part count. I do wonder why they removed the flush pin though.

The FT2232D's CPU FIFO mode is interesting too, as it would simplify the logic for the comms interface. The other interface could be used in 245 mode for loading data.

I plan to make cards without the ROM. Concept: RAM, CPLD and supervisor with a memory card or PC link. The procedure is the following: when powerup the supervisor fills the primary firmware into RAM and makes nMRES. At rest of time, it handles requests from M68K (memory card or PC link access). Thu...