[Bug debug/45865] New: [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[Bug debug/45865] New: [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

           Summary: [4.6 regression] Failed to build 403.gcc in SPEC CPU
                    2006
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hjl.tools@gmail.com


On Linux/ia32, revision 164827 gave:

gcc -m32 -c -o jump.o -DSPEC_CPU -DNDEBUG -I.  -O2 -msse2 -mfpmath=sse
-ffast-math                 jump.c
[hjl@gnu-16 build_base_lnx32-gcc.0000]$ grep jump.c make.err 
jump.c: In function 'reversed_comparison_code_parts':
jump.c:761:1: internal compiler error: in dwarf2out_cfi_begin_epilogue, at
dwarf2out.c:2930
[hjl@gnu-16 build_base_lnx32-gcc.0000]$

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

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

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-02 13:06:39 UTC ---
It is caused by revision 164552:

http://gcc.gnu.org/ml/gcc-cvs/2010-09/msg00849.html

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.0

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #3 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-06 01:11:30 UTC ---
A simple testcase:

[hjl@gnu-32 rrs]$ cat pr45865.c
typedef struct rtx_def *rtx;
enum machine_mode {
  VOIDmode,
  CCFPmode,
  CCFPUmode,
  MAX_MACHINE_MODE
};
enum mode_class {
  MODE_CC,
  MODE_FLOAT,
  MODE_COMPLEX_FLOAT,
  MODE_VECTOR_FLOAT
};
extern const enum mode_class mode_class[(int) MAX_MACHINE_MODE];
enum rtx_code {
  UNKNOWN,
  GEU,
  ORDERED,
  CONST_INT
};
struct rtx_def {
  unsigned int code: 16;
  unsigned int mode : 8;
};
extern enum rtx_code reverse_condition (enum rtx_code);
enum rtx_code
reversed_comparison_code_parts (enum rtx_code code, rtx insn, rtx arg0,
                rtx arg1)
{
  enum machine_mode mode;
  mode = (enum machine_mode) (arg0)->mode;
  if (mode == VOIDmode)
    mode = (enum machine_mode) (arg1)->mode;
  if ((mode_class[(int) (mode)]) == MODE_CC)
    return (mode != CCFPmode && mode != CCFPUmode
        ? reverse_condition (code)
        : reverse_condition_maybe_unordered (code));
  switch (code) 
    {
    case GEU:
      return reverse_condition (code);
    case ORDERED:
      return UNKNOWN;
    }
  if (((enum rtx_code) (arg0)->code) == CONST_INT
      || (((enum machine_mode) (arg0)->mode) != VOIDmode
      && ! ((mode_class[(int) (mode)]) == MODE_FLOAT
        || (mode_class[(int) (mode)]) == MODE_COMPLEX_FLOAT
        || (mode_class[(int) (mode)]) == MODE_VECTOR_FLOAT)))
    return reverse_condition (code);
  return UNKNOWN;
}
[hjl@gnu-32 rrs]$ /export/gnu/import/rrs/164914/usr/bin/gcc -O2 -S -m32
pr45865.c
pr45865.c: In function \u2018reversed_comparison_code_parts\u2019:
pr45865.c:52:1: internal compiler error: in dwarf2out_cfi_begin_epilogue, at
dwarf2out.c:2930
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
[hjl@gnu-32 rrs]$

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2010-10/msg00417.htm
                   |                            |l

--- Comment #4 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-06 02:49:56 UTC ---
A patch is posted at

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg00417.html

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-09 13:06:13 UTC ---
Created attachment 22006
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22006
A testcase for -fcompare-debug failure

[hjl@gnu-6 gcc]$ ./xgcc -B./ -O2 -g -fcompare-debug foo.i -S -m32
xgcc: error: foo.i: -fcompare-debug failure (length)
[hjl@gnu-6 gcc]$

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #6 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-09 22:45:44 UTC ---
A small testcase

--
[hjl@gnu-6 gcc]$ cat /export/gnu/import/delta/foo.c
typedef union tree_node *tree;
enum ix86_builtin_type {
  IX86_BT_LAST_VECT,
  IX86_BT_LAST_PTR
};
extern const enum ix86_builtin_type ix86_builtin_type_ptr_base[];
extern tree build_qualified_type (tree, int);
extern tree build_pointer_type (tree);
tree
ix86_get_builtin_type (enum ix86_builtin_type tcode, unsigned int index)
{
  tree type, itype;
  int quals;
  if (tcode <= IX86_BT_LAST_PTR)
    quals = 0x0;
  else
    quals = 0x1;
  itype = ix86_get_builtin_type (ix86_builtin_type_ptr_base[index],
                 index);
  if (quals != 0x0)
    itype = build_qualified_type (itype, quals);
  type = build_pointer_type (itype);
  return type;
}
[hjl@gnu-6 gcc]$ ./xgcc -B./ -O2 -g -fcompare-debug  -S -m32
/export/gnu/import/delta/foo.c -Wall
xgcc: error: /export/gnu/import/delta/foo.c: -fcompare-debug failure (length)
[hjl@gnu-6 gcc]$ 
----

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #7 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-10 00:48:54 UTC ---
(note 139 66 113 NOTE_INSN_EPILOGUE_BEG)

(insn 113 139 114 (set (reg:SI 3 bx)
        (mem/c:SI (plus:SI (reg/f:SI 7 sp)
                (const_int 4 [0x4])) [6 S4 A8])) y.i:54 63 {*movsi_internal}
     (nil))

dwarf2out_cfi_begin_epilogue failed to handle:

(code_label 130 27 128 13 "" [1 uses])

(note 128 130 115 [bb 6] NOTE_INSN_BASIC_BLOCK)

(insn/f:TI 115 128 67 (parallel [
            (set (reg/f:SI 7 sp)
                (plus:SI (reg/f:SI 7 sp)
                    (const_int 12 [0xc])))
            (clobber (reg:CC 17 flags))
            (clobber (mem:BLK (scratch) [0 A8]))
        ]) y.i:54 868 {pro_epilogue_adjust_stack_si_add}
     (expr_list:REG_UNUSED (reg:CC 17 flags)
        (expr_list:REG_CFA_ADJUST_CFA (set (reg/f:SI 7 sp)
                (plus:SI (reg/f:SI 7 sp)
                    (const_int 12 [0xc])))
            (expr_list:REG_CFA_RESTORE (reg:SI 4 si)
                (expr_list:REG_CFA_RESTORE (reg:SI 3 bx)
                    (nil))))))

(call_insn/j:TI 67 115 68 (set (reg:SI 0 ax)
        (call (mem:QI (symbol_ref:SI ("reverse_condition") [flags 0x41] 
<function_decl 0x7ffff18bb100 reverse_condition>) [0 S1 A8])
            (const_int 4 [0x4]))) y.i:52 881 {*call_value_0}
     (nil)
    (nil))

(barrier 68 67 39)
...
(insn:TI 114 113 142 (set (reg:SI 4 si)
        (mem/c:SI (plus:SI (reg/f:SI 7 sp)
                (const_int 8 [0x8])) [6 S4 A8])) y.i:54 63 {*movsi_internal}
     (nil))

(jump_insn 142 114 143 (set (pc)
        (label_ref 130)) 605 {jump}
     (nil)
 -> 130)

(barrier 143 142 70)

(code_label 70 143 71 7 "" [3 uses])

(note 71 70 133 [bb 13] NOTE_INSN_BASIC_BLOCK)

(insn:TI 133 71 140 (parallel [
            (set (reg/i:SI 0 ax)
                (const_int 0 [0]))
            (clobber (reg:CC 17 flags))
        ]) y.i:54 54 {*movsi_xor}
     (expr_list:REG_UNUSED (reg:CC 17 flags)
        (nil)))

(note 140 133 107 NOTE_INSN_EPILOGUE_BEG)

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #8 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-10 01:12:37 UTC ---
Does this patch

---
diff --git a/gcc/dwarf2out.c b/gcc/dwarf2out.c
index af5f809..14d4f9b 100644
--- a/gcc/dwarf2out.c
+++ b/gcc/dwarf2out.c
@@ -2872,7 +2872,7 @@ dwarf2out_cfi_begin_epilogue (rtx insn)
     continue;

       /* Look for both regular and sibcalls to end the block.  */
-      if (returnjump_p (i))
+      if (returnjump_p (i) || any_uncondjump_p (i))
     break;
       if (CALL_P (i) && SIBLING_CALL_P (i))
     break;
---

make any senses?

[Bug debug/45865] [4.6 regression] Failed to build 403.gcc in SPEC CPU 2006

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #9 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-12 21:59:12 UTC ---
A patch is posted at

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg00800.html

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
                URL|http://gcc.gnu.org/ml/gcc-p |
                   |atches/2010-10/msg00417.htm |
                   |l                           |
   Last reconfirmed|                            |2010.10.16 17:50:48
          Component|debug                       |rtl-optimization
            Summary|[4.6 regression] Failed to  |[4.6 Regression]
                   |build 403.gcc in SPEC CPU   |ifcvt/crossjump failed to
                   |2006                        |mark return jump
     Ever Confirmed|0                           |1

--- Comment #10 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-16 17:50:48 UTC ---
The function has multiple epilogues:

 epilogue 1
....
label1:
 ...
 return

 epilogue 2
 ...
 jump label1
 epilogue 3

According to

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg00817.html

"jump label1" should be masked as a return jump.

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jgu222 at gmail dot com

--- Comment #11 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-18 04:37:11 UTC ---
*** Bug 46062 has been marked as a duplicate of this bug. ***

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2010-10/msg01858.htm
                   |                            |l

--- Comment #12 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-21 17:47:33 UTC ---
A patch is posed at

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg01858.html

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

Richard Henderson <rth at gcc dot gnu.org> changed:

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

--- Comment #13 from Richard Henderson <rth at gcc dot gnu.org> 2010-10-23 17:25:44 UTC ---
(In reply to comment #10)
> The function has multiple epilogues:
> 
>  epilogue 1
> ....
> label1:
>  ...
>  return
> 
>  epilogue 2
>  ...
>  jump label1
>  epilogue 3
> 
> According to
> 
> http://gcc.gnu.org/ml/gcc-patches/2010-10/msg00817.html
> 
> "jump label1" should be masked as a return jump.

That's incorrect, it is not a return jump.

This bug happens much earlier than you're looking.  We've
cross-jumped in between epilogues and we've produced a 
situation in which the unwind info is incorrect.

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
                 CC|                            |jakub at gcc dot gnu.org

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[hjl at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

--- Comment #14 from hjl at gcc dot gnu.org <hjl at gcc dot gnu.org> 2010-11-03 17:07:08 UTC ---
Author: hjl
Date: Wed Nov  3 17:07:04 2010
New Revision: 166259

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=166259
Log:
Revert revision 164552.

gcc/

2010-11-03  H.J. Lu  <hongjiu.lu@intel.com>

    PR rtl-optimization/45865
    * Makefile.in (df-problems.o): Revert revision 164552.
    * basic-block.h (enum bb_flags): Likewise.
    * cfgcleanup.c (block_was_dirty): Likewise.
    (try_forward_edges): Likewise.
    (try_crossjump_bb): Likewise.
    (try_head_merge_bb): Likewise.
    (try_optimize_cfg): Likewise.
    (cleanup_cfg): Likewise.
    * df-core.c (df_set_bb_dirty): Likewise.
    * df-problems.c: Likewise.
    (df_simulate_find_uses): Likewise.
    (MEMREF_NORMAL, MEMREF_VOLATILE): Likewise.
    (find_memory, find_memory_store): Likewise.
    (can_move_insns_across): Likewise.
    * df.h (can_move_insns_across): Likewise.
    * ifcvt.c (find_memory): Likewise.
    (dead_or_predicable): Likewise.

gcc/testsuite/

2010-11-03  H.J. Lu  <hongjiu.lu@intel.com>

    PR rtl-optimization/45865
    * gcc.dg/pr45865.c: New.
    * gcc.dg/torture/pr45865.c: Likewise.

    * gcc.target/arm/headmerge-1.c: Revert revision 164552.
    * gcc.target/arm/headmerge-2.c: Likewise.
    * gcc.target/i386/headmerge-1.c: Likewise.
    * gcc.target/i386/headmerge-2.c: Likewise.

Added:
    trunk/gcc/testsuite/gcc.dg/pr45865.c
    trunk/gcc/testsuite/gcc.dg/torture/pr45865.c
Removed:
    trunk/gcc/testsuite/gcc.target/arm/headmerge-1.c
    trunk/gcc/testsuite/gcc.target/arm/headmerge-2.c
    trunk/gcc/testsuite/gcc.target/i386/headmerge-1.c
    trunk/gcc/testsuite/gcc.target/i386/headmerge-2.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/Makefile.in
    trunk/gcc/basic-block.h
    trunk/gcc/cfgcleanup.c
    trunk/gcc/df-core.c
    trunk/gcc/df-problems.c
    trunk/gcc/df.h
    trunk/gcc/ifcvt.c
    trunk/gcc/testsuite/ChangeLog

[Bug rtl-optimization/45865] [4.6 Regression] ifcvt/crossjump failed to mark return jump

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45865

H.J. Lu <hjl.tools at gmail dot com> changed:

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

--- Comment #15 from H.J. Lu <hjl.tools at gmail dot com> 2010-11-05 11:52:51 UTC ---
Fixed.