[Bug c++/59066] New: Segmentation fault.

[Bug c++/59066] New: Segmentation fault.

[alexandre.hamez at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

            Bug ID: 59066
           Summary: Segmentation fault.
           Product: gcc
           Version: 4.8.2
            Status: UNCONFIRMED
          Severity: blocker
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: alexandre.hamez at gmail dot com

Created attachment 31188
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=31188&action=edit
Preprocessed sources.

G++ crashes with the following backtrace :

g++ -std=c++11  ./bug.ii -O0
./bug.ii: In member function 'SDD<C> homomorphism<C>::operator()(SDD<C>) [with
C = conf0]':
./bug.ii:1203:60: internal compiler error: Segmentation fault
     return apply_binary_visitor(evaluation<C>(), **this, *x);

0x8a386f crash_signal
    ../.././gcc/toplev.c:332
0xa37012 make_decl_rtl(tree_node*)
    ../.././gcc/varasm.c:1197
0x64ac37 rtx_for_function_call
    ../.././gcc/calls.c:1675
0x64ac37 expand_call(tree_node*, rtx_def*, int)
    ../.././gcc/calls.c:2983
0x6f8f4b expand_expr_real_1(tree_node*, rtx_def*, machine_mode,
expand_modifier, rtx_def**)
    ../.././gcc/expr.c:10246
0x655e20 expand_call_stmt
    ../.././gcc/cfgexpand.c:2114
0x655e20 expand_gimple_stmt_1
    ../.././gcc/cfgexpand.c:2152
0x655e20 expand_gimple_stmt
    ../.././gcc/cfgexpand.c:2304
0x65668d expand_gimple_basic_block
    ../.././gcc/cfgexpand.c:4138
0x6586e6 gimple_expand_cfg
    ../.././gcc/cfgexpand.c:4657

G++ is called as follow:
g++ -std=c++11  ./bug.ii -O0

[Bug c++/59066] Segmentation fault.

[alexandre.hamez at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

--- Comment #1 from Alexandre Hamez <alexandre.hamez at gmail dot com> ---
I managed to reduce the code causing the segfault of gcc to ~1200 lines, but
there are certainly still a lot of useless lines (coming from <type_traits>).

[Bug c++/59066] Segmentation fault.

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|blocker                     |normal

[Bug c++/59066] C+11, 'using' instead of 'typedef' causes a segmentation fault.

[alexandre.hamez at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

Alexandre Hamez <alexandre.hamez at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Segmentation fault.         |C+11, 'using' instead of
                   |                            |'typedef' causes a
                   |                            |segmentation fault.

--- Comment #2 from Alexandre Hamez <alexandre.hamez at gmail dot com> ---
I've found what causes the crash. On line 1199 of bug.ii, if 'using data_type =
variant<identity<C>>' is changed to 'typedef variant<identity<C>> data_type',
the segmentation no longer occurs [1].
I changed the summary to reflect that.

[1] Note that this particular code will fail to compile because of a missing
member.

[Bug c++/59066] C++11, 'using' instead of 'typedef' causes a segmentation fault.

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

--- Comment #3 from Paolo Carlini <paolo.carlini at oracle dot com> ---
Doesn't look like a front-end issue.

[Bug c++/59066] [C++11] 'using' instead of 'typedef' causes a segmentation fault.

[reichelt at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

Volker Reichelt <reichelt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-02-14
                 CC|                            |reichelt at gcc dot gnu.org
            Summary|C++11, 'using' instead of   |[C++11] 'using' instead of
                   |'typedef' causes a          |'typedef' causes a
                   |segmentation fault.         |segmentation fault.
     Ever confirmed|0                           |1
      Known to fail|                            |4.7.0, 4.8.0, 4.9.0

--- Comment #4 from Volker Reichelt <reichelt at gcc dot gnu.org> ---
Confirmed. Reduced testcase that crashes since GCC 4.7.0 when template
alias (aka 'using') was introduced:

============================================================================
template<typename T, template <typename> class U> void foo(const U<T>&) {}

template<typename> struct A
{
  using B = A;
  B bar();
};

void baz()
{
  foo(A<int>().bar());
}
============================================================================

Stack trace with current 4.8 branch:

bug.cc: In function 'void baz()':
bug.cc:11:22: internal compiler error: Segmentation fault
   foo(A<int>().bar());
                      ^
0xad469f crash_signal
        ../../gcc/gcc/toplev.c:332
0xcff5e2 tree_check
        ../../gcc/gcc/tree.h:3683
0xcff5e2 make_decl_rtl(tree_node*)
        ../../gcc/gcc/varasm.c:1197
0x7d72f3 rtx_for_function_call
        ../../gcc/gcc/calls.c:1675
0x7d72f3 expand_call(tree_node*, rtx_def*, int)
        ../../gcc/gcc/calls.c:2983
0x8aadcd expand_expr_real_1(tree_node*, rtx_def*, machine_mode,
expand_modifier, rtx_def**)
        ../../gcc/gcc/expr.c:10263
0x7e82fd expand_call_stmt
        ../../gcc/gcc/cfgexpand.c:2114
0x7e82fd expand_gimple_stmt_1
        ../../gcc/gcc/cfgexpand.c:2152
0x7e82fd expand_gimple_stmt
        ../../gcc/gcc/cfgexpand.c:2304
0x7e9162 expand_gimple_basic_block
        ../../gcc/gcc/cfgexpand.c:4138
0x7eacdc gimple_expand_cfg
        ../../gcc/gcc/cfgexpand.c:4657
Please submit a full bug report, [etc.]


Stack trace with trunk:

cc1plus: internal compiler error: Segmentation fault
0xba523f crash_signal
        ../../gcc/gcc/toplev.c:337
0x87e004 tree_check
        ../../gcc/gcc/tree.h:2918
0x87e004 decl_assembler_name_hash
        ../../gcc/gcc/symtab.c:98
0x87e46f insert_to_assembler_name_hash
        ../../gcc/gcc/symtab.c:209
0x87e5c1 symtab_initialize_asm_name_hash
        ../../gcc/gcc/symtab.c:432
0x88cda2 analyze_functions
        ../../gcc/gcc/cgraphunit.c:1124
0x88e2a5 finalize_compilation_unit()
        ../../gcc/gcc/cgraphunit.c:2320
0x689cae cp_write_global_declarations()
        ../../gcc/gcc/cp/decl2.c:4448
Please submit a full bug report, [etc.]

[Bug c++/59066] [C++11] 'using' instead of 'typedef' causes a segmentation fault.

[reichelt at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

Volker Reichelt <reichelt at gcc dot gnu.org> changed:

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

--- Comment #5 from Volker Reichelt <reichelt at gcc dot gnu.org> ---
Jason, your patch for PR60182 seems to have fixed this one, too.
Do you want to add the reduced testcase (in comment#4) to the testsuite?

[Bug c++/59066] [C++11] 'using' instead of 'typedef' causes a segmentation fault.

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59066

Jason Merrill <jason at gcc dot gnu.org> changed:

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

--- Comment #6 from Jason Merrill <jason at gcc dot gnu.org> ---
I don't think the testcase is sufficiently different.  Thanks.

*** This bug has been marked as a duplicate of bug 60182 ***