--- Comment #13 from jakub at gcc dot gnu dot org 2008-06-23 11:58 ---
Subject: Bug 36508
Author: jakub
Date: Mon Jun 23 11:57:19 2008
New Revision: 137037
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=137037
Log:
PR tree-optimization/36508
* tree-ssa-pre.c (com
--- Comment #14 from jakub at gcc dot gnu dot org 2008-06-23 11:58 ---
Fixed.
--
jakub at gcc dot gnu dot org changed:
What|Removed |Added
Status|NEW
--- Comment #12 from jakub at gcc dot gnu dot org 2008-06-23 11:52 ---
Subject: Bug 36508
Author: jakub
Date: Mon Jun 23 11:51:34 2008
New Revision: 137036
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=137036
Log:
PR tree-optimization/36508
* tree-ssa-pre.c (com
--
mmitchel at gcc dot gnu dot org changed:
What|Removed |Added
Priority|P3 |P1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36508
--- Comment #11 from rguenther at suse dot de 2008-06-12 20:17 ---
Subject: Re: [4.3 Regression] ICE in compute_antic
On Thu, 12 Jun 2008, dberlin at gcc dot gnu dot org wrote:
> --- Comment #10 from dberlin at gcc dot gnu dot org 2008-06-12 14:52
> ---
> FWIW, the comment r
--- Comment #10 from dberlin at gcc dot gnu dot org 2008-06-12 14:52
---
FWIW, the comment right above the assert has proven to be true.
In a few years and releases, this is only the second time anyone has ever hit
it :)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36508
--- Comment #9 from dberlin at gcc dot gnu dot org 2008-06-12 14:51 ---
Subject: Re: [4.3 Regression] ICE in compute_antic
The assert is there because often when people break PRE, it goes into
infinite loops due to non-convergence, and eats all memory and CPU
very very very quickly.
It
--- Comment #8 from rguenth at gcc dot gnu dot org 2008-06-12 14:41 ---
I guess the assert is just bogus. But of course maybe Danny wants to have a
look?
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--
--- Comment #7 from rguenth at gcc dot gnu dot org 2008-06-12 14:39 ---
The following fails with -O -ftree-pre:
void
foo (short *sp)
{
int k;
k = 1;
#define SP0 *sp++ = 1;
while (1)
{
if (k > 6)
break;
SP0
k++;
}
k = 1;
while (1)
{
i
--- Comment #6 from rguenth at gcc dot gnu dot org 2008-06-12 14:21 ---
Created an attachment (id=15757)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=15757&action=view)
testcase that fails on x86_64 as well
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36508
--- Comment #5 from pinskia at gcc dot gnu dot org 2008-06-12 14:20 ---
Well not branch cost differences but LOGICAL_OP_NON_SHORT_CIRCUIT is 0. Now
LOGICAL_OP_NON_SHORT_CIRCUIT should be 1 on PPC but there needs some expand
support for getting the bools using crand/crand instructions (w
--- Comment #4 from rguenth at gcc dot gnu dot org 2008-06-12 14:15 ---
We get different gimplification on x86_64 and ppc due to branch-cost
differences
(appearantly):
- :;
- D.1604 = i <= 125;
- D.1605 = k <= 11;
- D.1606 = D.1604 && D.1605;
- if (D.1606)
+ :;
+ if (i > 125)
--- Comment #3 from rguenth at gcc dot gnu dot org 2008-06-12 14:04 ---
It works for me on x86_64. Confirmed on ppc.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
---
--- Comment #2 from rguenth at gcc dot gnu dot org 2008-06-12 14:00 ---
Probably related to PR36439.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
14 matches
Mail list logo
