Hi all, I've been playing with CGEN the last couple of weeks and have finally got a recent version of binutils building with my CGEN-generated output and it sort-of looks like it's working.
I came across this post from Jeremy on how we might want to name the toolchain in future, http://opencores.org/forum,OpenRISC,0,3478 and have followed his lead on the CGEN stuff. I've got my work-in-progress stuff up here: https://github.com/juliusbaxter/cgen (note: be sure to select the or1k branch, and the interesting stuff are the or1k-* files under cpu/ ) I have binutils building fine with this, and have changed the port so that it's now named or1k, and uses the CGEN-generated source in opcodes/. I'm yet to do any extensive testing, but it compiles and I thought I'd bring it to people's attention in case they're interested in playing with it, and to double check that the naming of things is correct. My binutils sources are here: https://github.com/juliusbaxter/binutils-or1k The next steps are to get some gas testsuite stuff up and running and then get the simulator working. But I'd just like to check that this looks OK so far, and that the naming and my organisation in the CGEN stuff is appropriate (arch: or1k, isas: ORBIS32, ORBIS64, ORFPX32 etc., machine: or32, or64 etc. and model: or1200 etc.) Something I'm curious about is how the different machines might be selected when using the compiler, so it knows not to emit the 64-bit data instructions when using or32, for instance. Thanks for any feedback, Julius _______________________________________________ OpenRISC mailing list [email protected] http://lists.openrisc.net/listinfo/openrisc
