This is the mail archive of the
mailing list for the GCC project.
Re: New lang hook: ok_for_sibcall
Fergus Henderson writes:
> On 21-May-2003, Andrew Haley <email@example.com> wrote:
> > Fergus Henderson writes:
> > > On 21-May-2003, Andrew Haley <firstname.lastname@example.org> wrote:
> > > > Sibcalls break Java programs in interesting ways. To fix this while
> > > > still allowing sibcalls to be used where it won't hurt, I need to be
> > > > able to control when a sibcall is allowed.
> > >
> > > What about sibcalls via function pointers?
> > > For those, fndecl in expand_call() will be a null pointer.
> > > Your patch doesn't seem to handle that case.
> > It doesn't need to. The test 'fndecl == NULL_TREE' is before the call
> > to ok_for_sibcall.
> Which test `fndecl == NULL_TREE'? Specifically, which source file,
> revision number, and line number?
> I don't see any such test in calls.c in the current mainline.
> I think you may be looking at an old version of calls.c?
Indeed I have; I was fixing a bug in gcc 3.3. I'll tweak my patch for
mainline and resubmit.