[Bug middle-end/15014] New: [tree-ssa] [regression] labels after are removed even though they are used

[Bug middle-end/15014] New: [tree-ssa] [regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

Try to run the following code -O1 or above, you will either get a link error or an as error:
int a;
int
main()
{
l1:
  a= &&l1-&&l2;
  return 0;
l2:;
}

I first reported it here: <http://gcc.gnu.org/ml/gcc/2004-04/msg00556.html> but the 
descussion went into wrong direction, I just wantted to know if someone had noticed this 
before.

-- 
           Summary: [tree-ssa] [regression] labels after are removed even
                    though they are used
           Product: gcc
           Version: tree-ssa
            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=15014

[Bug middle-end/15014] [tree-ssa] [regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-19 15:42 -------
I forgot to say that converting to CFG based IR is causing this failure.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |tree-ssa


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [tree-ssa] [regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-19 15:55 -------
This is causes all of the regressions on powerpc-apple-darwin which do not show up on 
anyother target.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [tree-ssa] [regression] labels after are removed even though they are used

[bangerth at dealii dot org]

------- Additional Comments From bangerth at dealii dot org  2004-04-19 16:03 -------
Confirmed. With out address-of-label extension, this code is  
definitely valid. Interestingly, I get a linker error for  
a label "L3" :-) 
 
g/x> /home/bangerth/bin/gcc-tree-ssa/bin/gcc -O1 x.c 
/tmp/cclM7GY7.o(.text+0x12): In function `main': 
: undefined reference to `.L3' 
collect2: ld returned 1 exit status 
 

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2004-04-19 16:03:48
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [tree-ssa] [regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-19 22:50 -------
Here is some information, the main problem is that BB 1 is not being marked as 
reachable, so it is being removed, we need to mark it as reachable some how.

;; basic block 0, loop depth 0, count 0
;; prev block -1, next block 1
;; pred:       ENTRY (fallthru)
;; succ:       EXIT
l1:;
l1.0 = (int)&l1;
l2.1 = (int)&l2;
T.2 = l1.0 - l2.1;
a = T.2;
return 0;

$22 = (struct basic_block_def *) 0x4152007c
(gdb) p debug_bb_n(1)
;; basic block 1, loop depth 0, count 0
;; prev block 0, next block -2
;; pred:      
;; succ:       EXIT
l2:;
return;

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [tree-ssa] [regression] labels after are removed even though they are used

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2004-05-04 12:05 -------
Tree SSA produces the following .s file at -O:

.L2:
        movl    $.L2, %eax
        movl    $.L3, %edx
        subl    %edx, %eax
        movl    %eax, a(%rip)
        movl    $0, %eax
        ret

And obviously the problem is that .L3 is never defined.

What are the semantics of this test case, is it allowed
to move the label around to another place?  GCC 3.3.1
clearly does, and it produces this (also at -O):

.L2:
        leal    .L2(%rip), %eax
        leal    .L3(%rip), %edx
        subl    %edx, %eax
        movl    %eax, a(%rip)
.L3:
        movl    $0, %eax
        ret



-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2004-04-19 16:03:48         |2004-05-04 12:05:25
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug rtl-optimization/15014] [3.5 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-06-08 13:49 -------
After RTH's change for having only one RETURN_EXPR, the problem is now on rtl level.


2004-06-07  Richard Henderson  <rth@redhat.com>
        
        * gimple-low.c (struct lower_data): Add the_return_label and
        one_return_stmt. 
        (lower_function_body): Initialize and use them.
        (lower_return_expr): New. 
        (lower_stmt): Call it.
        * gimplify.c (gimplify_return_expr): Force the argument to be either
        null or a result_decl.
        * tree-gimple.c: Update gimple grammer to match.
        * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Deny
        coalescing of result_decls.


And here is the new test which fails:
#ifndef NO_LABEL_VALUES
int main ()
{
  void *x = &&L2;
  if (&&L3 - &&L1 > 1)
    abort();
 L1: return 0;
 L2: abort ();
 L3:;
}
#else
int main(){return 0;}
#endif

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|tree-optimization           |rtl-optimization
   Last reconfirmed|2004-05-04 12:05:25         |2004-06-08 13:50:09
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [3.5 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-06-20 20:37 -------
Now the orginal testcase fails now after some changes from RTH and at -O0 after CFG is always built 
now.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu dot
                   |                            |org
          Component|rtl-optimization            |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-04 14:29 -------
Mine I will send it out soon.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-04 14:46 -------
Patch here: <http://gcc.gnu.org/ml/gcc-patches/2004-10/msg00246.html>.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-07 17:19 -------
Since Jeff says this is the wrong approach I am not going to update my patch because I have better 
things to do with my time than fix cfgcleanup.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|pinskia at gcc dot gnu dot  |unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW
           Keywords|patch                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-10 05:26 -------
I have a new fix which should be more acceptable, testing right now.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |pinskia at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-10 14:20 -------
Plus I am fixing up some of the tree optimizations for multiple entry point at the same time.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-10 17:53 -------
Patch which fixes the problem and is a better fix than my first patch which does not work all ways:
<http://gcc.gnu.org/ml/gcc-patches/2004-10/msg00860.html>.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-18 19:30 -------
Fixed, since the testcases are already there, there is no need to apply any more.

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014

[Bug middle-end/15014] [4.0 regression] labels after are removed even though they are used

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-10-18 19:29 -------
Subject: Bug 15014

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	pinskia@gcc.gnu.org	2004-10-18 19:29:53

Modified files:
	gcc            : ChangeLog tree-cfg.c 

Log message:
	2004-10-18  Andrew Pinski  <pinskia@physics.uc.edu>
	
	PR middle-end/15014
	PR middle-end/16973
	* tree-cfg.c (remove_bb): If we have a label expression in the
	basic block and the label we have taken the address, move the
	label expression to the basic block which is previous in the
	linked list.
	(tree_verify_flow_info): Fix printing out the label name of the
	problematic label expression.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.5932&r2=2.5933
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/tree-cfg.c.diff?cvsroot=gcc&r1=2.79&r2=2.80



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15014