[Bug tree-optimization/19360] [4.0 Regression] ICE with -O1 -funit-at-a-time

2005-02-26 Thread cvs-commit at gcc dot gnu dot org 

--- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-02-26 
22:27 ---
Subject: Bug 19360

CVSROOT:/cvs/gcc
Module name:gcc
Branch: apple-ppc-branch
Changes by: [EMAIL PROTECTED]   2005-02-26 22:27:29

Modified files:
gcc: ChangeLog.apple-ppc tree-ssa-dom.c 

Log message:
2005-02-26  Dale Johanensen  [EMAIL PROTECTED]

Radar 3988125
PR 19360 (from mainline)
* tree-ssa-dom.c (tree_ssa_dominator_optimize): Clear SSA_NAME_VALUE
at the end of the main DOM loop rather than just before DOM exits.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.apple-ppc.diff?cvsroot=gcconly_with_tag=apple-ppc-branchr1=1.1.2.325r2=1.1.2.326
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-dom.c.diff?cvsroot=gcconly_with_tag=apple-ppc-branchr1=1.1.2.149.2.11r2=1.1.2.149.2.12



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19360

[Bug tree-optimization/19360] [4.0 Regression] ICE with -O1 -funit-at-a-time

2005-01-18 Thread jakub at gcc dot gnu dot org 

--- Additional Comments From jakub at gcc dot gnu dot org  2005-01-18 16:17 
---
Confirmed as fixed.  Thanks.

-- 
   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution||FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19360

[Bug tree-optimization/19360] [4.0 Regression] ICE with -O1 -funit-at-a-time

2005-01-17 Thread law at redhat dot com 

--- Additional Comments From law at redhat dot com  2005-01-17 20:45 ---
Subject: Re:  [4.0 Regression] ICE with -O1
-funit-at-a-time

On Tue, 2005-01-11 at 11:14 +, jakub at gcc dot gnu dot org wrote:
 --- Additional Comments From jakub at gcc dot gnu dot org  2005-01-11 
 11:14 ---
 tree-ssa-dom.c in this case uses a released SSA_NAME.
 The dom pass recorded a SSA_NAME X into another SSA_NAME (Y)'s SSA_NAME_VALUE,
 then tree_duplicate_bb called indirectly from thread_through_all_blocks
 marks X for rewrite and following rewrite_ssa_into_ssa releases X, but X's
 address is still in Y's SSA_NAME_VALUE.  Later on when Y's SSA_NAME_VALUE is
 used, it references a released SSA_NAME and the compiler crashes.
Yea.  I stumbled upon this before I went on vacation last week.  I've 
got a patch here that I'll get bootstrapped and regression tested.

jeff




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19360

[Bug tree-optimization/19360] [4.0 Regression] ICE with -O1 -funit-at-a-time

2005-01-11 Thread jakub at gcc dot gnu dot org 

--- Additional Comments From jakub at gcc dot gnu dot org  2005-01-11 11:14 
---
tree-ssa-dom.c in this case uses a released SSA_NAME.
The dom pass recorded a SSA_NAME X into another SSA_NAME (Y)'s SSA_NAME_VALUE,
then tree_duplicate_bb called indirectly from thread_through_all_blocks
marks X for rewrite and following rewrite_ssa_into_ssa releases X, but X's
address is still in Y's SSA_NAME_VALUE.  Later on when Y's SSA_NAME_VALUE is
used, it references a released SSA_NAME and the compiler crashes.


-- 
   What|Removed |Added

 CC||law at redhat dot com,
   ||dnovillo at redhat dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19360

[Bug tree-optimization/19360] [4.0 Regression] ICE with -O1 -funit-at-a-time

2005-01-10 Thread pinskia at gcc dot gnu dot org 

--- Additional Comments From pinskia at gcc dot gnu dot org  2005-01-10 
14:35 ---
Confirmed, here is the reduced testcase:
int bad[100];

int level;
int count;

static inline int g(int *idx)
{
int m;
int i;

do {
m=0;
i=*idx;
switch(i) {
case 0:
i++;
m=1;
break;
}
*idx=i;
} while(m0);
return m;
}

int f()
{
int i;

i=0;
for(;;) {
int m;
m=g(i);
if(m==0) break;
if(level==0  bad[m])
continue;
count++;
  }
return 0;
}

-- 
   What|Removed |Added

 Status|UNCONFIRMED |NEW
 Ever Confirmed||1
   Keywords||ice-on-valid-code
   Last reconfirmed|-00-00 00:00:00 |2005-01-10 14:35:50
   date||
Summary|ICE with -O1 -funit-at-a-   |[4.0 Regression] ICE with -
   |time|O1 -funit-at-a-time
   Target Milestone|--- |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19360

[Bug tree-optimization/19360] [4.0 Regression] ICE with -O1 -funit-at-a-time

2005-01-10 Thread kcwu at csie dot org 

--- Additional Comments From kcwu at csie dot org  2005-01-10 14:50 ---
(In reply to comment #2)
the complete list:

ICE with:
-O2
-O1 -funit-at-a-time
-O3 -fno-unit-at-a-time
-O1 -finline-functions

works well with:
-O0
-O1
-O2 -fno-unit-at-a-time
-O0 -finline-functions
-O3 -fno-unit-at-a-time -fno-inline-functions

With -O1 -finline-functions, it segfault at the same place(c-objc-common.c:273)

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19360