This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Value profile based optimizations, part 1
> Hello,
>
> > I am not sure what to
> > do when it fails to validate - Richard previously mentioned that we can
> > probably spill the hard registers that is tricky in some cases. I would
> > be happy with the first solution that simply do validate and skip the
> > profiling when that fails.
>
> is there some standard way how to do it already? Or will I have to
> design it from scratch?
No, there is no way I am aware of, except for the code hoisting routines
I wrote some time ago. That ones do little bit more than you want (they
update the instructions to use new pseudo clobbers and so on, but you
can steal some code from that probably).
It is not that dificult - having bitmap of live registers you need to
figure out whether some of them will be modified or not and probably
be ready to deal with some special cases (not sure what right now)
>
> Zdenek
>
> > We already have the problem with inserting the profiling code on s390
> > where flags may be live across basic block boundaries. We can solve
> > this as side effect.