[Bug c++/19108] New: [4.0 regression] ICE initializing arrays

[Bug c++/19108] New: [4.0 regression] ICE initializing arrays

[reichelt at gcc dot gnu dot org]

The following valid code snippet causes an ICE on mainline when
compiled with "-O":

==================================
struct A
{
    int i[6];
    A () : i() {}
};

struct B
{
    A a;
    B(const A& x) : a(x) {}
};

B b=A();
==================================

  bug.cc: In function 'void __static_initialization_and_destruction_0(int,
int)':
  bug.cc:13: internal compiler error: in sra_hash_tree, at tree-sra.c:383
  Please submit a full bug report, [etc.]

This seems to be a recent regression (within the last week).
The problem might be related to PR 18793.

-- 
           Summary: [4.0 regression] ICE initializing arrays
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code, monitored
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: reichelt at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org,mark at codesourcery dot
                    com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-21 15:30 -------
This is more likely related to a bug which I filed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
          Component|c++                         |tree-optimization
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2004-12-21 15:30:54
               date|                            |
   Target Milestone|---                         |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-21 16:10 -------
sra_hash_tree does not handle RANGE_EXPRs.

This implements them but it might not be the correct approach though:
Index: tree-sra.c
===============================================================
====
RCS file: /cvs/gcc/gcc/gcc/tree-sra.c,v
retrieving revision 2.47
diff -u -p -r2.47 tree-sra.c
--- tree-sra.c	10 Dec 2004 21:54:42 -0000	2.47
+++ tree-sra.c	21 Dec 2004 16:09:09 -0000
@@ -378,6 +378,11 @@ sra_hash_tree (tree t)
       h = iterative_hash_expr (DECL_FIELD_OFFSET (t), 0);
       h = iterative_hash_expr (DECL_FIELD_BIT_OFFSET (t), h);
       break;
+    
+    case RANGE_EXPR:
+      h = sra_hash_tree (TREE_OPERAND (t, 0));
+      h ^= sra_hash_tree (TREE_OPERAND (t, 1));
+      break;
 
     default:
       gcc_unreachable ();


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rth at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2004-12-22 12:43 -------
Looks like fall-out from PR18191.  I'll try to take care of this.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |steven at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2004-12-21 15:30:54         |2004-12-22 12:43:56
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2004-12-31 12:13 -------
I cannot reproduce this (it is monitored??). 
 
I still think something like Andrew's patch is necessary, 
but without a test case, well, how can we be sure?? 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2004-12-31 16:08 -------
I can still reproduce it with the above testcase:
gcc version 4.0.0 20041230 (experimental) on i686-pc-linux-gnu.

Maybe it is target dependant?
Did you try larger array sizes than 6?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-12-31 16:10 -------
(In reply to comment #5)
> I can still reproduce it with the above testcase:
> gcc version 4.0.0 20041230 (experimental) on i686-pc-linux-gnu.
> 
> Maybe it is target dependant?

It is. You might have to use -march=i686.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0 regression] ICE initializing arrays

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-01-01 16:28 -------
Yup, -march=i686 does the trick for me: 
 
$ ./cc1plus t.C -O -m32 -march=i686 
 A::A() 
 A::A() 
 A::A() 
 B::B(const A&) 
 B::B(const A&) 
 B::B(const A&) 
 void __static_initialization_and_destruction_0(int, int) 
 void _GLOBAL__I_b() 
 
Analyzing compilation unit 
Performing intraprocedural optimizations 
Assembling functions: 
 void __static_initialization_and_destruction_0(int, int) 
 
t.C: In function 'void __static_initialization_and_destruction_0(int, int)': 
t.C:13: internal compiler error: in sra_hash_tree, at tree-sra.c:384 
Please submit a full bug report, 
with preprocessed source if appropriate. 
See <URL:http://gcc.gnu.org/bugs.html> for instructions. 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[reichelt at gcc dot gnu dot org]

------- Additional Comments From reichelt at gcc dot gnu dot org  2005-03-16 20:52 -------
Are there any news, Steven?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[sayler at thewalrus dot org]

------- Additional Comments From sayler at thewalrus dot org  2005-03-18 15:49 -------
working to replicate this in current CVS.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sayler at thewalrus dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[sayler at thewalrus dot org]

------- Additional Comments From sayler at thewalrus dot org  2005-03-18 23:00 -------
Bug still present.  The patch in #2 seems to fix the issue, though.

I can't say that the hash function chosen is the most efficient, but it's surely
better than an ICE.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2005-03-24 17:57 -------
My concern with this patch is that SRA may need other updates to full take into
account RANGE_EXPR.  RTH, comments?

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2005-03-28 19:13 -------
Indeed, SRA *does* need to be updated to handle the RANGE_EXPR.  And not in the
hash routine at all -- it needs to happen before that.  Consider the following
modified test case:

struct A
{
    int i[6];
    A () : i(1) {}
};

struct B
{
    A a;
    B(const A& x) : a(x) {}
};

B b=A();

int main()
{
  return b.a.i[0] == 1 ? 0 : 1;
}

Steven, if you don't have time to work on this, reassign the bug to me.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-03-28 21:04 -------
Indeed I do not have time to work on this.  I should have unassigned 
this bug long ago, sorry about that. 
 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|steven at gcc dot gnu dot   |rth at gcc dot gnu dot org
                   |org                         |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-03-29 22:44 -------
Subject: Bug 19108

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	rth@gcc.gnu.org	2005-03-29 22:43:48

Modified files:
	gcc            : ChangeLog tree-sra.c 
Added files:
	gcc/testsuite/g++.dg/opt: pr19108.C 

Log message:
	PR tree-opt/19108
	* tree-sra.c (generate_element_init_1): Handle RANGE_EXPR.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.8019&r2=2.8020
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-sra.c.diff?cvsroot=gcc&r1=2.54&r2=2.55
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/opt/pr19108.C.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-03-29 22:56 -------
Subject: Bug 19108

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-4_0-branch
Changes by:	rth@gcc.gnu.org	2005-03-29 22:56:05

Modified files:
	gcc            : ChangeLog tree-sra.c 
Added files:
	gcc/testsuite/g++.dg/opt: pr19108.C 

Log message:
	PR tree-opt/19108
	* tree-sra.c (generate_element_init_1): Handle RANGE_EXPR.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.7592.2.89&r2=2.7592.2.90
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-sra.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.53&r2=2.53.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/opt/pr19108.C.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.2.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108

[Bug tree-optimization/19108] [4.0/4.1 regression] ICE initializing arrays

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2005-03-29 22:58 -------
Fixed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19108