This is the mail archive of the 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]


On Wed, 2004-05-19 at 16:24, Steven Bosscher wrote:
> On Wednesday 19 May 2004 22:55, wrote:
> >  >CALL_PLACEHOLDERs are now obsolete since we expand
> >  >tail calls properly from calls.c.
> >  >Bootstrapped and tested on amd64 and i686.  OK?
> >
> > Err, don't you have some cleanup in calls.c to do?  I don't see any changes
> > which eliminate the creation of the 2 call streams and all the related
> > gunk.  Or is that still necessary for some reason?
> Oh, ehr, ehm...  It it still there?
Certainly pieces are still there.  For example, we still create two
insn streams, one for tailcalls one for normal calls.  I don't think
we attach them to CALL_PLACEHOLDERS anymore.

> I thought that "gunk" was already killed because it was not necessary
> anymore anyway (since it originally was only necessary because when we
> got to calls.c we had not identified tail calls yet).
> Let me have a look.  In the mean time, what about the current patch?
My only concern was the possibility that there may be some
dependencies between the code you're zapping and the remaining
multiple stream expansion stuff we do in calls.c.  If after reviewing
calls.c you don't expect any problems, then go ahead with your 
original cleanup of CALL_PLACEHOLDER and we'll deal with whatever
cleanups we can do in calls.c separately.


ps.  I hope this isn't a duplicate -- I've had some outgoing email
trouble lately.

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