This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: using frame pointer + constant in reg_equiv_constant
- To: law at cygnus dot com
- Subject: Re: using frame pointer + constant in reg_equiv_constant
- From: Joern Rennecke <amylaar at cygnus dot co dot uk>
- Date: Mon, 30 Nov 1998 18:08:48 +0000 (GMT)
- Cc: amylaar at cygnus dot co dot uk, egcs-patches at cygnus dot com, wilson at cygnus dot com
> Don't be so sure. I've seen many cases were we make the substitution, then
> never validate the result. Over and over across many ports. I've always
> assumed that was the desired behavior for stuff in reg_equiv_constant. The
> most recent was about 2wks ago on the x86.
I have just done a
grep reg_equiv_constant */*.{h,c,md}
in the config subdirectory of gcc, without any result line.
So it's not port specific code that makes these substitutions.
AFAIK the problems are usually with constant that can't be reloaded easily,
e.g. stuff that needs extra registers for PIC.
But since the new invariants get reduced to sums like vanilla register
elimination, there shouldn't be any of those problems.