--- Comment #6 from rguenth at gcc dot gnu dot org 2009-02-17 11:29 ---
Reducing on the a-i branch.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from rguenth at gcc dot gnu dot org 2009-02-17 12:17 ---
Created an attachment (id=17309)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17309action=view)
reduced testcase
More reduced testcase. Still sensitive to UID changes.
--
--- Comment #8 from rguenth at gcc dot gnu dot org 2009-02-17 13:11 ---
The change that the UID variations cause is ordering of two PHI-nodes which
result in
Value numbers:
-line_37 = line_18
+line_18 = line_37
from
bb 7:
# line_23 = PHI line_37(9), 0B(6)
# line_32 = PHI
--- Comment #9 from rguenth at gcc dot gnu dot org 2009-02-17 13:23 ---
This happens to fix it:
Index: tree-ssa-pre.c
===
--- tree-ssa-pre.c (revision 144226)
+++ tree-ssa-pre.c (working copy)
@@ -3537,7 +3537,10
--- Comment #10 from jakub at gcc dot gnu dot org 2009-02-17 13:39 ---
Unfortunately it doesn't, it seems to fix just the rh485708.3.3.3.3.3.i
testcase, but not the rh485708.i testcase, nor the original one. I'll attach
the full testcase soon.
--
--- Comment #11 from jakub at gcc dot gnu dot org 2009-02-17 13:45 ---
Created an attachment (id=17310)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17310action=view)
rh485708.orig.i
Original, unreduced, testcase.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39204
--- Comment #12 from rguenth at gcc dot gnu dot org 2009-02-17 13:47
---
Or rather
Index: tree-ssa-pre.c
===
--- tree-ssa-pre.c (revision 144227)
+++ tree-ssa-pre.c (working copy)
@@ -1704,7 +1704,7 @@
--- Comment #13 from rguenth at gcc dot gnu dot org 2009-02-17 13:49
---
This one seems to fix it
Index: tree-ssa-pre.c
===
--- tree-ssa-pre.c (revision 144227)
+++ tree-ssa-pre.c (working copy)
@@ -1707,6
--- Comment #1 from jakub at gcc dot gnu dot org 2009-02-16 16:38 ---
Created an attachment (id=17306)
-- (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17306action=view)
rh485708.i
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39204
--- Comment #2 from jakub at gcc dot gnu dot org 2009-02-16 16:40 ---
./cc1 -O2 -m32 rh485708.i triggers this in 32-bit HWI build (but both in stage1
built by 4.3.x and later stages), in x86_64 gcc it isn't reproduceable with -O2
-m32.
--
--
jakub at gcc dot gnu dot org changed:
What|Removed |Added
Target Milestone|--- |4.4.0
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39204
--- Comment #3 from hjl dot tools at gmail dot com 2009-02-16 18:57 ---
This is caused by revision 142617:
http://gcc.gnu.org/ml/gcc-cvs/2008-12/msg00315.html
--
hjl dot tools at gmail dot com changed:
What|Removed |Added
--- Comment #4 from jakub at gcc dot gnu dot org 2009-02-16 19:12 ---
This testcase is so sensitive to any uid changes (just removing unused u_long
typedef makes the bug go away), so I think doing a binary search is useless
here.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39204
--- Comment #5 from hjl dot tools at gmail dot com 2009-02-16 19:33 ---
Do we get overflow with 32-bit HWI in this testcase?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39204
--
mmitchel at gcc dot gnu dot org changed:
What|Removed |Added
Priority|P3 |P1
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39204
15 matches
Mail list logo
