[Bug c/30564] New: ice for legal code with -O3

[Bug c/30564] New: ice for legal code with -O3

[dcb314 at hotmail dot com]

I just tried to compile Suse Linux package lynx-2.8.6_rel2-23
with the GNU C compiler version 4.3 snapshot 20070119.

The compiler said

gcc -DLINUX  -D_GNU_SOURCE -DHAVE_CONFIG_H -I../.. -I../../src
-I../../src/chrtrans -I../../WWW/Library/Implementation -I../../   -O3
-fmessage-length=0 -D_FORTIFY_SOURCE=2 -pipe   -c ./makeuctb.c
./makeuctb.c: In function 'main':
./makeuctb.c:253: internal compiler error: in extract_range_from_assert, at
tree-vrp.c:820
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

Preprocessed source code attached. Flag -O3 required.


-- 
           Summary: ice for legal code with -O3
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dcb314 at hotmail dot com
  GCC host triplet: x86_64-suse-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug c/30564] ice for legal code with -O3

[dcb314 at hotmail dot com]

------- Comment #1 from dcb314 at hotmail dot com  2007-01-23 21:00 -------
Created an attachment (id=12941)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=12941&action=view)
C source code


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug c/30564] ice for legal code with -O3

[rguenth at gcc dot gnu dot org]

------- Comment #2 from rguenth at gcc dot gnu dot org  2007-01-23 21:29 -------
Nice one.

(gdb) up
#1  0x085aa055 in extract_range_from_assert (vr_p=0xbf806324, expr=0xb7ba1ee8)
    at /home/richard/src/trunk/gcc/tree-vrp.c:845
845       gcc_assert (limit != var);
(gdb) call debug_generic_expr (expr)
ASSERT_EXPR <i_657, i_657 != i_657>

reducing.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |tree-optimization
           Keywords|                            |ice-on-valid-code
            Summary|ice for legal code with -O3 |[4.3 Regression] ice for
                   |                            |legal code with -O3
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug c/30564] ice for legal code with -O3

[rguenth at gcc dot gnu dot org]

------- Comment #3 from rguenth at gcc dot gnu dot org  2007-01-23 21:55 -------
Honza, we're not folding the comparison on inlining and later ICE in tree-vrp
because of this (from the einline dump):

  # fp1_5 = PHI <fp1_7(5), fp1_4(8), fp1_4(9)>
<L7>:;
  i_30 = fp0_6;
  goto <bb 14> (<L10>);

<L18>:;
  D.1689_35 = unicount[i_3];
  D.1690_36 = D.1689_35 == 0;
  D.1691_37 = i_3 != i_3;
  D.1692_38 = D.1690_36 && D.1691_37;
  if (D.1692_38) goto <L15>; else goto <L17>;

<L15>:;
  RawOrEnc = 1;


look at the i_3 != i_3


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
          Component|tree-optimization           |c
     Ever Confirmed|0                           |1
           Keywords|ice-on-valid-code           |
   Last reconfirmed|0000-00-00 00:00:00         |2007-01-23 21:55:31
               date|                            |
            Summary|[4.3 Regression] ice for    |ice for legal code with -O3
                   |legal code with -O3         |
   Target Milestone|4.3.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug c/30564] ice for legal code with -O3

[rguenth at gcc dot gnu dot org]

------- Comment #4 from rguenth at gcc dot gnu dot org  2007-01-23 21:56 -------
Created an attachment (id=12942)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=12942&action=view)
somewhat reduced testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |tree-optimization
           Keywords|                            |ice-on-valid-code
            Summary|ice for legal code with -O3 |[4.3 Regression] ice for
                   |                            |legal code with -O3
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #5 from pinskia at gcc dot gnu dot org  2007-01-24 07:20 -------
Reduced testcase:
static int RawOrEnc = 0;
static inline void addpair(int fp, int un)
{
  if (RawOrEnc == 0 && fp != un)
    RawOrEnc = 1;
}
int f(int un0, char *a, unsigned int __s2_len)
{
  addpair(un0, un0);
  __s2_len < 4 ? __builtin_strcmp (a, "-") : 0;
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2007-01-24 07:24 -------
Remove the dead code and it works.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #7 from pinskia at gcc dot gnu dot org  2007-01-27 21:41 -------
This works with "4.3.0 20070127" on powerpc-darwin with -O3 and -O3 -m64.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #8 from pinskia at gcc dot gnu dot org  2007-01-27 22:30 -------
(In reply to comment #7)
> This works with "4.3.0 20070127" on powerpc-darwin with -O3 and -O3 -m64.
Ok, I had to change a paramater internal to GCC to get it to reproduce on
ppc-darwin but I am able to with today's compiler.

Looking into it.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #9 from pinskia at gcc dot gnu dot org  2007-01-27 23:31 -------
The problem is we are calling fold_marked_statements after renumbering the
basic blocks and such which causes us to get the wrong starting point.
patch which I am tesing:
Index: ../../gcc/tree-inline.c
===================================================================
--- ../../gcc/tree-inline.c     (revision 121236)
+++ ../../gcc/tree-inline.c     (working copy)
@@ -2658,6 +2658,10 @@
     gimple_expand_calls_inline (bb, &id);

   pop_gimplify_context (NULL);
+  
+  fold_marked_statements (last, id.statements_to_fold);
+  pointer_set_destroy (id.statements_to_fold);
+  
   /* Renumber the (code) basic_blocks consecutively.  */
   compact_blocks ();
   /* Renumber the lexical scoping (non-code) blocks consecutively.  */
@@ -2683,8 +2687,6 @@
      Kill it so it won't confuse us.  */
   cgraph_node_remove_callees (id.dst_node);

-  fold_marked_statements (last, id.statements_to_fold);
-  pointer_set_destroy (id.statements_to_fold);
   fold_cond_expr_cond ();
   /* We make no attempts to keep dominance info up-to-date.  */
   free_dominance_info (CDI_DOMINATORS);


-- 

pinskia at gcc dot gnu dot org changed:

           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=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #10 from pinskia at gcc dot gnu dot org  2007-01-28 07:58 -------
> The problem is we are calling fold_marked_statements after renumbering the
> basic blocks and such which causes us to get the wrong starting point.

We have to call verify_cgraph before calling fold_marked_statements, otherwise
we get an ICE about an function (a builtin) which we no longer call.  This
shows up in the Ada front-end in C code in which we call strlen in the function
which gets inlined with a constant string.

The patch is able to pass bootstrap but I still have another regression I need
to look into, dealing with an inline-asm and the testcase is x86 specific one
at that. (it does #ifdef __i386__).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #11 from pinskia at gcc dot gnu dot org  2007-01-28 19:10 -------
(In reply to comment #10)
> The patch is able to pass bootstrap but I still have another regression I need
> to look into, dealing with an inline-asm and the testcase is x86 specific one
> at that. (it does #ifdef __i386__).

SRA messes up somehow and it only happens with inlined functions so I am going
to look more into it.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   GCC host triplet|x86_64-suse-linux           |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #12 from pinskia at gcc dot gnu dot org  2007-01-28 22:57 -------
(In reply to comment #11)
> SRA messes up somehow and it only happens with inlined functions so I am going
> to look more into it.

I have a fix for the SRA issue now, SRA tries to look into 3rd/4th operand to
decide if the array is a non constant stride but in_array_bounds_p already
checks that so there is no need to check the 3rd/4th operand again.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #13 from pinskia at gcc dot gnu dot org  2007-05-22 23:10 -------
*** Bug 32033 has been marked as a duplicate of this bug. ***


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #14 from pinskia at gcc dot gnu dot org  2007-08-19 19:07 -------
The testcases here don't crash anymore but the one from PR 32033 does:


static int spready[] = {
};
void explosion_map (int y)
{
  for (int i = 0; i < 4; i++)
    if (y * spready[i] < 0)
      break;
}
void explosion (void)
{
  explosion_map (0);
  for (int i = 0; i < 2; i++)
    continue;
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #15 from pinskia at gcc dot gnu dot org  2007-08-19 19:19 -------
Testing a slight modifed version of the patch.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #17 from pinskia at gcc dot gnu dot org  2007-08-20 07:43 -------
Fixed, sorry it took me this long to test/submit/commit this patch, I had been
busy with other patches and other work.


-- 

pinskia at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564

[Bug tree-optimization/30564] [4.3 Regression] ice for legal code with -O3

[pinskia at gcc dot gnu dot org]

------- Comment #16 from pinskia at gcc dot gnu dot org  2007-08-20 07:43 -------
Subject: Bug 30564

Author: pinskia
Date: Mon Aug 20 07:42:55 2007
New Revision: 127638

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=127638
Log:
2007-08-20  Andrew Pinski  <andrew_pinski@playstation.sony.com>

        PR middle-end/30564
        * tree-inline.c (optimize_inline_calls): Move the cgraph checking
        code in front of the compacting of basic blocks.
        Move the folding of statements inbetween the cgraph checking
        and compacting of basic blocks.

2007-08-20  Andrew Pinski  <andrew_pinski@playstation.sony.com>

        PR middle-end/30564
        * gcc.c-torture/compile/pr30564-1.c: New test.
        * gcc.c-torture/compile/pr30564-2.c: New test.


Added:
    trunk/gcc/testsuite/gcc.c-torture/compile/pr30564-1.c
    trunk/gcc/testsuite/gcc.c-torture/compile/pr30564-2.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-inline.c


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30564