I'm confused about why an isa_parser dependency breaks compilation. It seems like you could specify an ISA to build (might not even matter which one, but why not do X86_FS), and let M5 compile the existing models as is, but then add in the Bochs model too as something that's independent of the parser. Sort of like how you need to specify an ISA to build M5 even if you're only going to run the ISA-independent memory tester. I know it's on Nate's to-do list to make a "no ISA" compilation mode possible, but I don't think that's necessary.
If that won't work, can you elaborate on why? Steve On Fri, Apr 30, 2010 at 2:15 PM, Gabriel Michael Black <[email protected]> wrote: > Which files specifically does it depend on? Rather than removing the things > that have dependencies, you might want to stub out the parser generated > files. Even though there could be a lot of .isa files, I think only > decoder.hh, decoder.cc, and *_exec.cc are generated, and those just rely on > there being a decode function with an appropriate signature. There's also a > file that describes the max number of source/destination operands, and in > your case it should be safe to just set those to 1 (or 0, but there may be > some gotchas with 0). I haven't ever considered in depth the implications of > a non isa_parser supported CPU model so I may be oversimplifying. > > Generally speaking, I think rather than try to remove the parts of M5 that > ask your code to do things you can't do or that might not make sense, you > should attempt to do them, or stub them out and make them panic if/when > they're used. That would be very important if you wanted to fold back into > the main code base since I don't think we're prepared to delete the simple > CPU. > > Gabe > > Quoting Rathijit Sen <[email protected]>: > >> Hi, >> >> We are trying to integrate Bochs x86 functional model into M5. In order >> to achieve that we are trying to have our own cpu model that does not use >> isa_parser generated files. In doing so, we had to remove other CPU models >> like AtomicSimpleCPU, TimingSimpleCPU etc, as they depend upon isa_parser >> generated files (and thus breaks compilation). But we found out that >> src/cpu/base.cc (which is the file for BaseCPU) also has dependencies on >> isa_parser generated files. If we take off BaseCPU files from compilation, >> then many other files like sim/pseudo_inst.cc etc are complaining as they >> have dependencies on BaseCPU files. >> >> We are a bit confused now. Can you please let us know on whether we should >> try to remove BaseCPU? >> >> Thanks >> >> Arka >> Rathijit >> >> _______________________________________________ >> m5-dev mailing list >> [email protected] >> http://m5sim.org/mailman/listinfo/m5-dev >> > > > _______________________________________________ > m5-dev mailing list > [email protected] > http://m5sim.org/mailman/listinfo/m5-dev > _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
