[Bug middle-end/45230] New: gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[Bug middle-end/45230] New: gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[howarth at nitro dot med dot uc dot edu]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3217 bytes --]

Current gcc trunk with the proposed graphite merge
(http://gcc.gnu.org/ml/gcc-patches/2010-08/msg00080.html) still ICEs the
compiler when the gcc.c-torture/execute/strncmp-1.c testcase is compiled with
-fgraphite-identity. The error appears as...

gcc-4 strncmp-1.c -w -Os -lm -m32 -fgraphite-identity -o strncmp-1.x7
strncmp-1.c: In function ‘main’:
strncmp-1.c:44:1: internal compiler error: in rename_uses, at sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

The ICE doesn't backtrace in gdb...

 gdb /sw/lib/gcc4.6/libexec/gcc/x86_64-apple-darwin10.4.0/4.6.0/cc1
GNU gdb 6.3.50-20050815 (Apple version gdb-1469) (Wed May  5 04:36:56 UTC 2010)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "x86_64-apple-darwin"...Reading symbols for shared
libraries ............. done

(gdb) r -fpreprocessed strncmp-1.i -fPIC -quiet -dumpbase strncmp-1.c
-mmacosx-version-min=10.6.4 -m32 -mtune=generic -auxbase strncmp-1 -Os -w
-version -fgraphite-identity -o strncmp-1.s
Starting program:
/sw/lib/gcc4.6/libexec/gcc/x86_64-apple-darwin10.4.0/4.6.0/cc1 -fpreprocessed
strncmp-1.i -fPIC -quiet -dumpbase strncmp-1.c -mmacosx-version-min=10.6.4 -m32
-mtune=generic -auxbase strncmp-1 -Os -w -version -fgraphite-identity -o
strncmp-1.s
Reading symbols for shared libraries .++++++++++++..... done
GNU C (GCC) version 4.6.0 20100807 (experimental) (x86_64-apple-darwin10.4.0)
        compiled by GNU C version 4.6.0 20100807 (experimental), GMP version
4.3.2, MPFR version 2.4.2-p3, MPC version 0.8.1
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
GNU C (GCC) version 4.6.0 20100807 (experimental) (x86_64-apple-darwin10.4.0)
        compiled by GNU C version 4.6.0 20100807 (experimental), GMP version
4.3.2, MPFR version 2.4.2-p3, MPC version 0.8.1
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
Compiler executable checksum: 51323d20c79050973b53377b407f3f68
strncmp-1.c: In function ‘main’:
strncmp-1.c:44:1: internal compiler error: in rename_uses, at sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

Program exited with code 04.
(gdb) bt
No stack.

A log stepping from the sese.c:534 breakpoint up to the crash is attached.


-- 
           Summary: gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-
                    identity
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: howarth at nitro dot med dot uc dot edu
 GCC build triplet: x86_64-apple-darwin10
  GCC host triplet: x86_64-apple-darwin10
GCC target triplet: x86_64-apple-darwin10


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[howarth at nitro dot med dot uc dot edu]

------- Comment #1 from howarth at nitro dot med dot uc dot edu  2010-08-08 01:56 -------
Created an attachment (id=21435)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=21435&action=view)
gdb log stepping from sese.c:534 breakpoint until crash


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[zsojka at seznam dot cz]

------- Comment #2 from zsojka at seznam dot cz  2010-08-15 23:23 -------
Created an attachment (id=21486)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=21486&action=view)
reduced testcase

Crashes at x86_64-linu as well.
Valgrind output (r163261):
$ valgrind --trace-children=yes -q
/mnt/svn/gcc-trunk/binary-163261-lto-fortran-checking-yes-rtl-df/bin/gcc -Os
-fgraphite-identity testcase.c
testcase.c: In function 'foo':
testcase.c:2:1: internal compiler error: in rename_uses, at sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu dot org]

------- Comment #3 from spop at gcc dot gnu dot org  2010-08-20 23:50 -------
Subject: Bug 45230

Author: spop
Date: Fri Aug 20 23:49:58 2010
New Revision: 163432

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=163432
Log:
Add testcase for PR45230.

2010-08-20  Sebastian Pop  <sebastian.pop@amd.com>

        PR middle-end/45230
        * gcc.dg/graphite/id-pr45230.c: New.

Added:
    branches/graphite/gcc/testsuite/gcc.dg/graphite/id-pr45230.c
Modified:
    branches/graphite/gcc/ChangeLog.graphite


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #4 from t66667 at gmail dot com  2010-08-26 06:04 -------
*** Bug 45406 has been marked as a duplicate of this bug. ***


-- 

t66667 at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |t66667 at gmail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #5 from t66667 at gmail dot com  2010-09-09 04:55 -------
What is the status of this PR ?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[zsojka at seznam dot cz]

------- Comment #6 from zsojka at seznam dot cz  2010-09-09 20:02 -------
The fix has been commited to the graphite branch and will be commited to trunk
later.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #7 from t66667 at gmail dot com  2010-09-10 05:47 -------
Thank you very much for fixing this bug, could you please close this after it
is fixed on trunk ?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #8 from t66667 at gmail dot com  2010-09-14 22:55 -------
No idea, what is going on but I saved your testcase.c as pr45230 and it doesn't
produce the ICE, target gcc is x86_64-w64-mingw32.

bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-gcc  -w -Os -lm -m32
-fgraphite-identity -c pr45230 
x86_64-w64-mingw32-gcc: warning: pr45230: linker input file unused because
linking not done

bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-gcc -r -nostdlib -O1
-fgraphite-identity -c pr45230 
x86_64-w64-mingw32-gcc: warning: pr45230: linker input file unused because
linking not done
bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-g++ -r -nostdlib -O1
-fgraphite-identity -c qdrawhelper_mmx.ii 
qdrawhelper_mmx.ii: In function 'void comp_func_Source(uint*, const uint*, int,
uint) [with MM = QMMXIntrinsics, uint = unsigned int]':
qdrawhelper_mmx.ii:45806:41: internal compiler error: in rename_uses, at
sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[howarth at nitro dot med dot uc dot edu]

------- Comment #9 from howarth at nitro dot med dot uc dot edu  2010-09-14 23:27 -------
ICEs are atill present for both strncmp-1.c and reduced testcase on
x86_64-apple-darwin10 at r164287.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #10 from t66667 at gmail dot com  2010-09-15 00:17 -------
(In reply to comment #9)
> ICEs are atill present for both strncmp-1.c and reduced testcase on
> x86_64-apple-darwin10 at r164287.
> 

You are right. I was suspicious about the pr45230 having no .c extension in the
file name (weird). So I renamed the file and test it again.
bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-gcc  -w -Os -lm -m32
-fgraphite-identity -c pr45230.c 
pr45230.c: In function 'foo':
pr45230.c:2:1: internal compiler error: in rename_uses, at sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.                                        
See <http://gcc.gnu.org/bugs.html> for instructions.

Did few more tests and found the g++ failed at -O1 not -Os for
qdrawhelper_mmx.ii
The gcc failed at -Os and not -O3 for pr45230.c

bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-g++ -O1
-fgraphite-identity -c qdrawhelper_mmx.ii 
qdrawhelper_mmx.ii: In function 'void comp_func_Source(uint*, const uint*, int,
uint) [with MM = QMMXIntrinsics, uint = unsigned int]':
qdrawhelper_mmx.ii:45806:41: internal compiler error: in rename_uses, at
sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-g++ -Os
-fgraphite-identity -c qdrawhelper_mmx.ii 
bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-gcc  -w -O3 -lm -m32
-fgraphite-identity -c pr45230.c 
bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-gcc  -w -O0 -lm -m32
-fgraphite-identity -c pr45230.c 
bash-4.1$ ./xmingw-trunk-w64-sjlj/bin/x86_64-w64-mingw32-gcc  -w -Os -lm -m32
-fgraphite-identity -c pr45230.c 
pr45230.c: In function 'foo':
pr45230.c:2:1: internal compiler error: in rename_uses, at sese.c:534
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #11 from t66667 at gmail dot com  2010-09-15 00:23 -------
FYI, I applied these to gcc-trunk for testing purposes.
[PATCH 1/6] Factor out is_gimple_reg calls.
[PATCH 2/6] A function is affine when CHREC_RIGHT is invariant.
[PATCH 3/6] Fix chrec_contains_symbols_defined_in_loop.
[PATCH 4/6] Also handle ARRAY_REFs in instantiate_scev_r.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[zsojka at seznam dot cz]

------- Comment #12 from zsojka at seznam dot cz  2010-09-15 18:39 -------
You are right, it still fails in the graphite branch, though with different
ICE:
(r164299, x86_64-linux)

$ gcc gcc/testsuite/gcc.c-torture/execute/strncmp-1.c -Os -m32                  
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c: In function 'test':
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:37:5: warning: incompatible
implicit declaration of built-in function 'abort' [enabled by default]
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c: In function 'main':
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:139:3: warning: incompatible
implicit declaration of built-in function 'exit' [enabled by default]
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:44:1: error: Invalid first
operand of MEM_REF.
&u1.buf
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:79:12: note: in statement
# .MEM_120 = VDEF <.MEM_12>
MEM[(unsigned char *)&u1.buf] = 0;

gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:44:1: error: Invalid first
operand of MEM_REF.
&u2.buf
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:80:12: note: in statement
# .MEM_119 = VDEF <.MEM_120>
MEM[(unsigned char *)&u2.buf] = 0;

gcc/testsuite/gcc.c-torture/execute/strncmp-1.c:44:1: internal compiler error:
verify_stmts failed
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

------- Comment #13 from t66667 at gmail dot com  2010-09-20 04:00 -------
Ok this ticket is getting very confusing now and no development recently.
 ------- Comment #12 From Zdenek Sojka  2010-09-15 18:39  [reply] -------  
You are compiling it without graphite and yet you are using graphite branch to
test. Then you are getting another type of ICE.

------- Comment #10 From t66667@gmail.com  2010-09-15 00:17  [reply] -------  
Based on my observations the bug in PR45406 is triggered differently but with
the same ICE and different language C C++.

------- Comment #3 From Sebastian Pop  2010-08-20 23:50  [reply] -------  
Can you tell us a little more about what is going on in this PR45230.
Are you still working on the problem ?

If not I will detach PR45406 from this ticket.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[zsojka at seznam dot cz]

------- Comment #14 from zsojka at seznam dot cz  2010-09-20 04:57 -------
(In reply to comment #13)
>  ------- Comment #12 From Zdenek Sojka  2010-09-15 18:39  [reply] -------  
> You are compiling it without graphite and yet you are using graphite branch to
> test. Then you are getting another type of ICE.

-fgraphite-identity is enabled with -O[23s] in the graphite branch


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

--- Comment #24 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-16 22:30:35 UTC ---
*** Bug 45406 has been marked as a duplicate of this bug. ***

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

Sebastian Pop <spop at gcc dot gnu.org> changed:

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

--- Comment #23 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-08 23:04:56 UTC ---
Fixed.

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

--- Comment #22 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-08 23:01:46 UTC ---
Author: spop
Date: Wed Dec  8 23:01:40 2010
New Revision: 167609

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167609
Log:
Fix PR45230, PR45231, and PR45370: fold_stmt_inplace after replace_exp.

2010-12-08  Richard Guenther  <rguenther@suse.de>
        Sebastian Pop  <sebastian.pop@amd.com>

    PR tree-optimization/45230
    PR tree-optimization/45231
    PR tree-optimization/45370
    * sese.c (rename_uses): Returns a bool.  Call
    recompute_tree_invariant_for_addr_expr only on the RHS of a
    GIMPLE_ASSIGN.
    (graphite_copy_stmts_from_block): Call fold_stmt_inplace when
    rename_uses returns true.
    * tree-ssa-copy.c (replace_exp): Add a comment about calling
    fold_stmt_inplace after replace_exp.

    * gcc.dg/graphite/id-pr45230-1.c: New.
    * gcc.dg/graphite/id-pr45231.c: New.
    * gfortran.dg/graphite/id-pr45370.f90: New.

Added:
    trunk/gcc/testsuite/gcc.dg/graphite/id-pr45230-1.c
    trunk/gcc/testsuite/gcc.dg/graphite/id-pr45231.c
    trunk/gcc/testsuite/gfortran.dg/graphite/id-pr45370.f90
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/sese.c
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree-ssa-copy.c

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[howarth at nitro dot med.uc.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

--- Comment #21 from Jack Howarth <howarth at nitro dot med.uc.edu> 2010-12-08 02:15:28 UTC ---
(In reply to comment #20)
> Patch here:
> http://gcc.gnu.org/ml/gcc-patches/2010-12/msg00629.html

The proposed patch applied to r167573 as well as...

Index: opts.c
===================================================================
--- opts.c      (revision 167318)
+++ opts.c      (working copy)
@@ -462,6 +462,9 @@
     { OPT_LEVELS_1_PLUS, OPT_fcombine_stack_adjustments, NULL, 1 },

     /* -O2 optimizations.  */
+#ifdef HAVE_cloog
+    { OPT_LEVELS_2_PLUS, OPT_fgraphite_identity, NULL, 1 },
+#endif
     { OPT_LEVELS_2_PLUS, OPT_finline_small_functions, NULL, 1 },
     { OPT_LEVELS_2_PLUS, OPT_findirect_inlining, NULL, 1 },
     { OPT_LEVELS_2_PLUS, OPT_fpartial_inlining, NULL, 1 },

eliminates the ICEs compiling the gcc.c-torture/execute/strncmp-1.c testcase on
x86_64-apple-darwin10.

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

--- Comment #20 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-07 23:09:22 UTC ---
Patch here:
http://gcc.gnu.org/ml/gcc-patches/2010-12/msg00629.html

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

--- Comment #19 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-07 19:45:19 UTC ---
With this patch:

diff --git a/gcc/sese.c b/gcc/sese.c
index 65f8556..140b16f 100644
--- a/gcc/sese.c
+++ b/gcc/sese.c
@@ -547,8 +547,8 @@ rename_uses (gimple copy, htab_t rename_map,
gimple_stmt_iterator *gsi_tgt,
       gsi_insert_seq_before (gsi_tgt, stmts, GSI_SAME_STMT);
       replace_exp (use_p, new_expr);

-
-      if (TREE_CODE (new_expr) == INTEGER_CST)
+      if (TREE_CODE (new_expr) == INTEGER_CST
+         && gimple_code (copy) == GIMPLE_ASSIGN)
        {
          tree rhs = gimple_assign_rhs1 (copy);

I now see the same error as in comment #12

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

Sebastian Pop <spop at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |ASSIGNED
   Last reconfirmed|                            |2010.12.07 19:34:49
         Resolution|FIXED                       |
         AssignedTo|unassigned at gcc dot       |spop at gcc dot gnu.org
                   |gnu.org                     |
     Ever Confirmed|0                           |1

--- Comment #18 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-07 19:34:49 UTC ---
Mine.  On trunk r167554 I see this failing:
./xgcc -B. -fgraphite-identity -Os -m32
gcc/testsuite/gcc.c-torture/execute/strncmp-1.c

in sese.c:

      tree rhs = gimple_assign_rhs1 (copy);

and here copy is a call statement.

[Bug middle-end/45230] gcc.c-torture/execute/strncmp-1.c ICEs with -fgraphite-identity

[t66667 at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45230

--- Comment #17 from t66667 at gmail dot com <t66667 at gmail dot com> 2010-10-01 03:24:52 UTC ---
(In reply to comment #16)
> Fixed.

sorry you only introduced the problem into another error.
this maybe multiple problems.
what did you fix, which revision?
should i un-mark duplicate to pr45406 ?
since pr45406 and the testcase which zdenek attached in comment #2 is still
valid.

so i will reopen the problem.

do you want to fix pr45406 ?
it is not a good sign you do not say nothing.