[kaffe] PATCH: consistently separate jitter and backend register ID
stack at cs.utah.edu
Mon Mar 15 09:50:02 PST 2004
> > Working on the MIPS JIT3 I have made some small changes to the
> > architechture-independent register code for JIT3 that at least allow
> > the MIPS JIT to pass all the tests in test/internal. To summarize,
> > slotRegister now returns the `jitter ID' of the assigned register (the
> > index into the reginfo array), instead of the `backend ID'. All code
> > that depends on the former is unchanged, and all code that depends on
> > the latter now uses a new macro, `slotRegisterRegister', that is
> > equivalent to slotRegister, but returns the backend ID.
> Okie, I'll take a look at this and check it in tonight/tomorrow morning.
I looked at the patch, but I'm not totally convinced its the right thing
to do... I'd really prefer something that had stronger typing so that
these mistakes don't continue to happen. For example, the x86 backend
mixes up uses of the jitter ID for a register and the backend ID. Its a
little trivial, but still...
More information about the kaffe