This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Register Allocator query.


> Michael Matz <matz@suse.de> wrote:
> Hi,
> On Fri, 15 Nov 2002,Leon Taylor wrote:
> > I am doing some jugglery with code and want to
> > to do this with an arbitrary pseudo in new reg-
> > allocator. Is it possible to do this in a
> > convenient
> way without modifying too much code?
> 
> No, the new RA assigns hardregs to every pseudo it
> can find ;-)  I.e. you
> would need to change some code. 

Is it possible *not* to make a "web" corresponding 
to a pseudo so that its assignment doesn't take place.
Or do I need to describe those pseudos as hard regs.
Or some other tweak is required.
Actually I don't want to explore a wrong direction
here
so your comments would be valueable.

> Stack pseudos (on
> HEAD they are
> identified by having a reg number >=
> max_normal_pseudo) sometimes are not
> colored.  But in the final rewriting phase they are
> replaced by stack slots.  This means, that after 
> the new RA run no pseudos without assignment are 
> left.

A related query. In a comment in "web" structure
in gcc/ra.h
/* While spilling this is the rtx of the home of
spilled webs.It can be a mem ref (a stack slot), or a
pseudo register.  */
  rtx stack_slot;

So does this means that it is mem only after final
re-write and it is necessarily a pseudo during 
the one_pass () function ?

Thanks,
Leon.
----------------------------

________________________________________________________________________
Missed your favourite TV serial last night? Try the new, Yahoo! TV.
       visit http://in.tv.yahoo.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]