This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: regparm
james.g.mansion@hsbcgroup.com wrote:
>Now, would it be possible (in the case of register parameter passing only)
>to extend the life of the temporary/pseudo until ALL arguments to the
>called function are completely ready, and then assign into hard registers?
A.S. Forgive me if what I'm saying is not exactly current anymore, but I've
been out of the loop on gcc development for more than three years now.
To a certain extent this is already happening: the call insn is where
the life ends because of the reg-use-marks for regparm arguments.
The only problem at the moment is that the call-insn itself can contain
an expression (to lead the address of the function). In the current code,
it forgets to avoid the regparm registers since their life extended till
briefly in front of the call-insn.
Anyway, I've currently reintegrated the most important pending patches with
egcs for regparm support for x86. The compiler still compiles itself using
stackparm. Sadly, a regression test using a complete regparm environment
is not possible (yet), because that would require more patches (also rejected
by Kenner in the past; but I still have them, and will reintegrate those
too as soon as this batch is in).
I'll try to gift-wrap the things I have now later today, then the rest of
the community can take a stab at it.
--
Sincerely, srb@cuci.nl
Stephen R. van den Berg (AKA BuGless).
To the Java programmers:
Due to the holiday next Monday, there will be no garbage collection.