Re: egcs-1.1b alpha-dec-osf3.2 internal error in change_address()

[Hans-Peter Nilsson <hp at bitrange dot com>]

On Thu, 8 Apr 1999, Richard Henderson wrote:
> And so if the programmer knows that they will never use alloca, and
> that the code will always run only on Alpha, then they should be able
> to assume they can usurp $15.

I guess the -ffixed etc. frame-pointer-checking stuff in regclass.c needs
some tweaking then?

> So your whole debugging session starts way too late.  You
> would have needed to find out why fp_is_frame_pointer is
> true at all.

The short answer is line 3782 in alpha.c:alpha_expand_epilogue

  fp_is_frame_pointer = ((TARGET_OPEN_VMS && vms_is_stack_procedure)
			 || (!TARGET_OPEN_VMS && frame_pointer_needed));

But I guess you'll bounce right back and ask me why frame_pointer_needed
is 1?  If it's not the obvious answer that "it's the default for
non-optimizing compilations", I'll have to look some more.

Thanks so far.
brgds, H-P