@verbatim
RTL-level loop invariant motion. Copyright (C) 2004-2013 Free Software Foundation, Inc.
This file is part of GCC.
GCC is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3, or (at your option) any later version.
GCC is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with GCC; see the file COPYING3. If not see http://www.gnu.org/licenses/.
This implements the loop invariant motion pass. It is very simple
(no calls, no loads/stores, etc.). This should be sufficient to cleanup
things like address arithmetics -- other more complicated invariants should
be eliminated on GIMPLE either in tree-ssa-loop-im.c or in tree-ssa-pre.c.
We proceed loop by loop -- it is simpler than trying to handle things
globally and should not lose much. First we inspect all sets inside loop
and create a dependency graph on insns (saying "to move this insn, you must
also move the following insns").
We then need to determine what to move. We estimate the number of registers
used and move as many invariants as possible while we still have enough free
registers. We prefer the expensive invariants.
Then we move the selected invariants out of the loop, creating a new
temporaries for them if necessary.
The data stored for the loop.