--- Comment #7 from rguenth at gcc dot gnu dot org 2008-03-10 17:15 ---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #6 from rguenth at gcc dot gnu dot org 2008-03-10 17:15 ---
Subject: Bug 34677
Author: rguenth
Date: Mon Mar 10 17:14:45 2008
New Revision: 133081
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133081
Log:
2008-03-10 Richard Guenther <[EMAIL PROTECTED]>
PR
--- Comment #5 from rguenth at gcc dot gnu dot org 2008-03-10 14:39 ---
I have a patch for the non-PRE-of-globals part.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
-
--- Comment #4 from dberlin at gcc dot gnu dot org 2008-01-04 17:31 ---
Subject: Re: PREs insert_fake_stores is mostly useless
FRE will handle DECL's, and VN will value number them.
I've made small attempts at make PRE work with globals, but i'm lazy
and haven't done it for real yet :)
--- Comment #3 from rguenth at gcc dot gnu dot org 2008-01-04 17:03 ---
I think the other PR is only about VN global constants. But also an incoming
pointer to a structure is not handled:
struct X { int i; };
int foo(struct X *a, int argc)
{
int b;
int c;
int i;
int d, e;
fo
--- Comment #2 from rguenth at gcc dot gnu dot org 2008-01-04 17:00 ---
The plain DECL does not work because
1) we don't do the storetmp
2) can_PRE_operation and can_value_number_operation return false (so we
do not enter a into EXP_GEN
3) create_value_expr_from cannot handle
--- Comment #1 from pinskia at gcc dot gnu dot org 2008-01-04 16:57 ---
Isn't there already a bug about load PRE with global variables?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34677
