[Bug middle-end/21847] New: [4.0/4.1 Regression] misscompiling of the following java code

[Bug middle-end/21847] New: [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

The following java code should print PASS but instead it prints FAIL at -O1 and above:
class t
{
public static void test()
{
int i =0;
int j = 1;
int k = j/i;

}
public static void main(String a1[])
{
try {
test();
}catch (ArithmeticException e)
{
  System.out.println("Pass");
  return;
}
  System.out.println("Fail");
}
}

-- 
           Summary: [4.0/4.1 Regression] misscompiling of the following java
                    code
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P2
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pinskia at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug middle-end/21847] [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-31 19:39 -------
DCE removes it even though it can throw.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |tree-optimization


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
 GCC target triplet|                            |i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-31 19:47 -------
Note if we move the division inside the throw block, this works just fine.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.0                       |4.0.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-05 23:10 -------
Compiling the test case on i686 -O with today's CVS, I get the following 
final_cleanup dump: 
 
;; Function t.main(java.lang.String[]) 
(_ZN1t4mainEP6JArrayIPN4java4lang6StringEE) 
 
t.main(java.lang.String[]) (a1D.407) 
Eh tree: 
   2 catch tree_label:<L0> 
   1 try 
{ 
  struct  * out.3D.749; 
  struct  * out.1D.741; 
 
  # BLOCK 0 
  # PRED: ENTRY [100.0%]  (fallthru,exec) 
  _Jv_InitClass (&t.classD.726); 
  test (); 
  goto <bb 2> (<L3>); 
  # SUCC: 1 (ab,eh,exec) 2 [100.0%]  (fallthru,exec) 
 
  # BLOCK 1 
  # PRED: 0 (ab,eh,exec) 
<L0>:; 
  _Jv_InitClass (&java.lang.System.classD.725); 
  out.1D.741 = outD.629; 
  *(((struct  *) out.1)->vtable + 120B) (out.1D.741, (struct  *) _CD_tD.617
[1]); 
  goto <bb 3> (<L6>); 
  # SUCC: 3 [100.0%]  (fallthru,exec) 
 
  # BLOCK 2 
  # PRED: 0 [100.0%]  (fallthru,exec) 
<L3>:; 
  _Jv_InitClass (&java.lang.System.classD.725); 
  out.3D.749 = outD.629; 
  *(((struct  *) out.3)->vtable + 120B) (out.3D.749, (struct  *) _CD_tD.617
[2]); 
  # SUCC: 3 [100.0%]  (fallthru,exec) 
 
  # BLOCK 3 
  # PRED: 1 [100.0%]  (fallthru,exec) 2 [100.0%]  (fallthru,exec) 
<L6>:; 
  return; 
  # SUCC: EXIT [100.0%] 
 
} 
 
 
When I disable DCE with -fno-tree-dce, I get: 
;; Function t.main(java.lang.String[]) 
(_ZN1t4mainEP6JArrayIPN4java4lang6StringEE) 
 
t.main(java.lang.String[]) (a1D.407) 
Eh tree: 
   2 catch tree_label:<L0> 
   1 try 
{ 
  booleanD.10 temp.9D.799; 
  booleanD.10 temp.8D.798; 
  booleanD.10 temp.7D.797; 
  struct  * eD.408; 
  booleanD.10 D.724; 
  struct  * out.4D.750; 
  struct  * out.2D.742; 
  booleanD.10 D.741; 
  voidD.8 * * D.738; 
 
  # BLOCK 0 
  # PRED: ENTRY [100.0%]  (fallthru,exec) 
  _Jv_InitClass (&t.classD.726); 
  D.741 = 0; 
  test (); 
  goto <bb 2> (<L3>); 
  # SUCC: 1 (ab,eh,exec) 2 [100.0%]  (fallthru,exec) 
 
  # BLOCK 1 
  # PRED: 0 (ab,eh,exec) 
<L0>:; 
  D.738 = <<<exception object>>>; 
  eD.408 = *(D.738 - 4); 
  temp.9D.799 = 0; 
  _Jv_InitClass (&java.lang.System.classD.725); 
  temp.7D.797 = 1; 
  out.2D.742 = outD.629; 
  *(((struct  *) out.2)->vtable + 120B) (out.2D.742, (struct  *) _CD_tD.617
[1]); 
  goto <bb 3> (<L6>); 
  # SUCC: 3 [100.0%]  (fallthru,exec) 
 
  # BLOCK 2 
  # PRED: 0 [100.0%]  (fallthru,exec) 
<L3>:; 
  temp.8D.798 = 0; 
  _Jv_InitClass (&java.lang.System.classD.725); 
  D.724 = 1; 
  out.4D.750 = outD.629; 
  *(((struct  *) out.4)->vtable + 120B) (out.4D.750, (struct  *) _CD_tD.617
[2]); 
  # SUCC: 3 [100.0%]  (fallthru,exec) 
 
  # BLOCK 3 
  # PRED: 1 [100.0%]  (fallthru,exec) 2 [100.0%]  (fallthru,exec) 
<L6>:; 
  return; 
  # SUCC: EXIT [100.0%] 
 
} 
 
With DCE enabled (compiled with gcj --main=t -O) the test case fails, but if 
DCE is disabeld (gcj --main=t -O -fno-tree-dce) the test case passes... 
 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-06-05 23:10:25
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-05 23:19 -------
outD.629 appears to be used uninitialized when DCE is enabled.  I don't want 
to jump to conclusions, but this smells like an alias-related problem. 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-06-05 23:21 -------
(In reply to comment #4)
> outD.629 appears to be used uninitialized when DCE is enabled.  I don't want 
> to jump to conclusions, but this smells like an alias-related problem. 

out is a global variable.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-05 23:35 -------
Ugh, been looking at the wrong dumps. 
Simple C test case: 
 
void 
test(void) 
{ 
  int i =0; 
  int j = 1; 
  int k = j/i; 
} 
 
gcc -O -fexceptions -fnon-call-exceptions should not optimize away the 
divide iiuc.  But it does. 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-05 23:42 -------
I believe something like this is necessary. 
 
Index: tree-ssa-dce.c 
=================================================================== 
RCS file: /cvs/gcc/gcc/gcc/tree-ssa-dce.c,v 
retrieving revision 2.44 
diff -u -3 -p -r2.44 tree-ssa-dce.c 
--- tree-ssa-dce.c      1 Jun 2005 02:50:59 -0000       2.44 
+++ tree-ssa-dce.c      5 Jun 2005 23:41:58 -0000 
@@ -279,6 +279,15 @@ mark_stmt_if_obviously_necessary (tree s 
   tree op, def; 
   ssa_op_iter iter; 
 
+  /* With non-call exceptions, we have to assume that all statements could 
+     throw.  If a statement may throw, it is inherently necessary.  */ 
+  if (flag_non_call_exceptions 
+      && tree_could_throw_p (stmt)) 
+    { 
+      mark_stmt_necessary (stmt, false); 
+      return; 
+    } 
+ 
   /* Statements that are implicitly live.  Most function calls, asm and 
return 
      statements are required.  Labels and BIND_EXPR nodes are kept because 
      they are control flow, and we have no way of knowing whether they can be 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-05 23:43 -------
In fact make that 
+      mark_stmt_necessary (stmt, true);  

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2005-06-06 05:58 -------
The patch looks sensible to me.

Andrew Haley, would you have time to test this patch?

Andrew (MacLeod), would you be able to approve this patch?  



-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bje at redhat dot com, aph
                   |                            |at redhat dot com, amacleod
                   |                            |at redhat dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[aph at redhat dot com]

------- Additional Comments From aph at redhat dot com  2005-06-06 10:44 -------
OK, I'm looking at it.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[aph at redhat dot com]

------- Additional Comments From aph at redhat dot com  2005-06-06 11:09 -------
Failure during bootstrap:

./xgcc -B./ -B/home/aph/gcc4/install/i686-pc-linux-gnu/bin/ -isystem
/home/aph/gcc4/install/i686-pc-linux-gnu/include -isystem
/home/aph/gcc4/install/i686-pc-linux-gnu/sys-include
-L/home/aph/gcc4/build/gcc/../ld -O2  -DIN_GCC    -W -Wall -Wwrite-strings
-Wstrict-prototypes -Wmissing-prototypes -Wold-style-definition  -isystem
./include  -fPIC -g -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED 
-I. -I. -I/home/aph/gcc4/gcc/gcc -I/home/aph/gcc4/gcc/gcc/.
-I/home/aph/gcc4/gcc/gcc/../include -I/home/aph/gcc4/gcc/gcc/../libcpp/include 
-DL_moddi3 -fvisibility=hidden -DHIDE_EXPORTS -fexceptions -fnon-call-exceptions
-c /home/aph/gcc4/gcc/gcc/libgcc2.c -o libgcc/./_moddi3.o
/home/aph/gcc4/gcc/gcc/libgcc2.c: In function '__divdi3':
/home/aph/gcc4/gcc/gcc/libgcc2.c:1056: error: Type mismatch between an SSA_NAME
and its symbol.
/home/aph/gcc4/gcc/gcc/libgcc2.c:1056: error: Missing definition
for SSA_NAME: D.4215_136in statement:
#   TMT.31_158 = V_MAY_DEF <TMT.31_154>;
*rp_35 = D.4215_136;
/home/aph/gcc4/gcc/gcc/libgcc2.c:1056: internal compiler error: verify_ssa failed.
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=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[aph at redhat dot com]

------- Additional Comments From aph at redhat dot com  2005-06-06 11:34 -------
I think you want mark_stmt_necessary (stmt, true)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[amacleod at redhat dot com]

------- Additional Comments From amacleod at redhat dot com  2005-06-06 12:07 -------
I agree, you should need 'mark_stmt_necessary (stmt, true)'

Presuming that it bootstraps and passes tests, its approved.

Andrew

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[aph at redhat dot com]

------- Additional Comments From aph at redhat dot com  2005-06-06 13:15 -------
With that change, passes bootstrap with c/c++/java, perfect i686-pc-linux-gnu
libgcj test result.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-07 07:48 -------
Re. comments #13 and #13, I _already_ said it should be "true" in comment #8  ;-)

I'll commit the patch tonight.  Mark, is this approval also for the 4.0 branch?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mark at codesourcery dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-07 07:49 -------
You, like many DCE bugs before you, are now MINE!

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |steven at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2005-06-05 23:10:25         |2005-06-07 07:49:25
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[mark at codesourcery dot com]

------- Additional Comments From mark at codesourcery dot com  2005-06-07 15:45 -------
Subject: Re:  [4.0/4.1 Regression] misscompiling
 of the following java code

steven at gcc dot gnu dot org wrote:
> ------- Additional Comments From steven at gcc dot gnu dot org  2005-06-07 07:48 -------
> Re. comments #13 and #13, I _already_ said it should be "true" in comment #8  ;-)
> 
> I'll commit the patch tonight.  Mark, is this approval also for the 4.0 branch?

Since I haven't built RC1 yet, yes, it is.  Please commit ASAP.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-06-07 16:49 -------
Subject: Bug 21847

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-4_0-branch
Changes by:	steven@gcc.gnu.org	2005-06-07 16:48:53

Modified files:
	gcc            : ChangeLog tree-ssa-dce.c 

Log message:
	PR tree-optimization/21847
	* tree-ssa-dce.c (mark_stmt_if_obviously_necessary): With
	-fnon-call-exceptions, also mark statements inherently
	necessary if they may throw.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.7592.2.281&r2=2.7592.2.282
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-dce.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.32.4.1&r2=2.32.4.2



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-06-07 16:54 -------
Subject: Bug 21847

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	steven@gcc.gnu.org	2005-06-07 16:54:16

Modified files:
	gcc            : ChangeLog tree-ssa-dce.c 

Log message:
	PR tree-optimization/21847
	* tree-ssa-dce.c (mark_stmt_if_obviously_necessary): With
	-fnon-call-exceptions, also mark statements inherently
	necessary if they may throw.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.9069&r2=2.9070
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-ssa-dce.c.diff?cvsroot=gcc&r1=2.44&r2=2.45



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847

[Bug tree-optimization/21847] [4.0/4.1 Regression] misscompiling of the following java code

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-06-07 16:55 -------
Should be fixed now.  Thanks all for testing/reviewing so quickly. 
 

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21847