[Bug regression/61510] New: [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[Bug regression/61510] New: [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[hp at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

            Bug ID: 61510
           Summary: [4.10 Regression]:
                    20_util/scoped_allocator/requirements/explicit_instant
                    iation.cc and
                    tr1/6_containers/tuple/requirements/explicit_instantia
                    tion.cc
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Keywords: ice-on-valid-code
          Severity: normal
          Priority: P3
         Component: regression
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hp at gcc dot gnu.org
                CC: hubicka at gcc dot gnu.org
              Host: x86_64-unknown-linux-gnu
            Target: cris-axis-elf

Created attachment 32940
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=32940&action=edit
cc1plus -fpreprocessed explicit_instantiation.ii -O2 -std=gnu++11

These tests previously passed, now they fail.
A patch in the revision range (last_known_working:first_known_failing)
(211433:211435) exposed or caused these regressions.  Since then they fail as
follows:

Running
/tmp/hpautotest-gcc1/gcc/libstdc++-v3/testsuite/libstdc++-dg/conformance.exp
...
FAIL: 20_util/scoped_allocator/requirements/explicit_instantiation.cc (test for
excess errors)
...
FAIL: tr1/6_containers/tuple/requirements/explicit_instantiation.cc (test for
excess errors)


The messages in libstdc++.log are similar:

...
output is:
cc1plus: internal compiler error: Segmentation fault
0xaf51c5 crash_signal
        /tmp/hpautotest-gcc1/gcc/gcc/toplev.c:337
0x7e6b42 analyze_functions
        /tmp/hpautotest-gcc1/gcc/gcc/cgraphunit.c:1054
0x7e6f64 finalize_compilation_unit()
        /tmp/hpautotest-gcc1/gcc/gcc/cgraphunit.c:2339
0x5bf9ac cp_write_global_declarations()
        /tmp/hpautotest-gcc1/gcc/gcc/cp/decl2.c:4647
Please submit a full bug report,
with preprocessed source if appropriate.
Please include the complete backtrace with any bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.

FAIL: 20_util/scoped_allocator/requirements/explicit_instantiation.cc (test for
excess errors)

Preprocessed code for the above FAIL is attached.

Author of suspect patches in revision range CC:ed.

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[hubicka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

--- Comment #1 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
What happens at the line 1054 of cgraphunit.c.
                  = cgraph_get_node (DECL_ABSTRACT_ORIGIN (decl));
                  origin_node->used_as_abstract_origin = true;
Do you really get into case where abstract origin is NULL?

Honza

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Version|4.7.0                       |4.10.0
   Target Milestone|---                         |4.10.0

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

jgreenhalgh at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|cris-axis-elf               |cris-axis-elf,
                   |                            |arm-none-eabi,
                   |                            |aarch64-none-elf
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-06-19
                 CC|                            |jgreenhalgh at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #2 from jgreenhalgh at gcc dot gnu.org ---
Confirmed across bare metal arm/aarch64 targets.

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

--- Comment #3 from jgreenhalgh at gcc dot gnu.org ---
Created attachment 32975
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=32975&action=edit
Reduced testcase.

Adding a reduced testcase for this bug. 

Honza: To answer your question, if I add gcc_assert (origin_node) before the
dereference, it triggers - so in my case we definitely do have a NULL
origin_node.

If I dump decl, DECL_ABSTRACT_ORIGIN (decl) the results look "sane" (I
think)...

decl:

    full-name "std::scoped_allocator_adaptor<_OuterAlloc,
_InnerAllocs>::scoped_allocator_adaptor() [with _OuterAlloc =
std::allocator<short int>; _InnerAllocs = {}]"
    pending-inline-info 0x7f32a2507ab0 template-info 0x7f32a2362680

DECL_ABSTRACT_ORIGIN:

    full-name "std::scoped_allocator_adaptor<_OuterAlloc,
_InnerAllocs>::scoped_allocator_adaptor() [with _OuterAlloc =
std::allocator<short int>; _InnerAllocs = {}]"
    pending-inline-info 0x7f32a2507990 template-info 0x7f32a2362680

cgraph_get_node (DECL_ABSTRACT_ORIGIN (decl)) == (nil)

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

--- Comment #4 from jgreenhalgh at gcc dot gnu.org ---
Obviously something simple like this will fix the problem, but I don't know the
area to say that it doesn't just mask a deeper bug.

Honza?


---

2014-06-19  James Greenhalgh  <james.greenhalgh@arm.com>

    * cgraphunit.c (analyze_functions): Check we have an origin
    node before dereferencing it.


diff --git a/gcc/cgraphunit.c b/gcc/cgraphunit.c
index
1b7ab330a3e51a9bc3336f64998f4d89b45d342b..82e5a6838e36c464191c533e163a00c825265c84
100644
--- a/gcc/cgraphunit.c
+++ b/gcc/cgraphunit.c
@@ -1051,7 +1051,8 @@ analyze_functions (void)
         {
           struct cgraph_node *origin_node
               = cgraph_get_node (DECL_ABSTRACT_ORIGIN (decl));
-          origin_node->used_as_abstract_origin = true;
+          if (origin_node)
+            origin_node->used_as_abstract_origin = true;
         }
         }
       else

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[trippels at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

Markus Trippelsdorf <trippels at gcc dot gnu.org> changed:

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

--- Comment #5 from Markus Trippelsdorf <trippels at gcc dot gnu.org> ---
Might be related to PR61558.

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[jgreenhalgh at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

--- Comment #6 from jgreenhalgh at gcc dot gnu.org ---
Author: jgreenhalgh
Date: Fri Aug  1 08:56:05 2014
New Revision: 213425

URL: https://gcc.gnu.org/viewcvs?rev=213425&root=gcc&view=rev
Log:
[Patch] Not very subtle fix for pr61510

gcc/

    PR regression/61510
    * cgraphunit.c (analyze_functions): Use get_create rather than get
    for decls which are clones of abstract functions.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cgraphunit.c

[Bug regression/61510] [4.10 Regression]: 20_util/scoped_allocator/requirements/explicit_instantiation.cc and tr1/6_containers/tuple/requirements/explicit_instantiation.cc

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61510

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Author: jakub
Date: Wed Oct  1 14:46:32 2014
New Revision: 215767

URL: https://gcc.gnu.org/viewcvs?rev=215767&root=gcc&view=rev
Log:
    PR c++/63306
    Backported from mainline
    2014-08-01  James Greenhalgh  <james.greenhalgh@arm.com>

    PR regression/61510
    * cgraphunit.c (analyze_functions): Use get_create rather than get
    for decls which are clones of abstract functions.

    * g++.dg/ipa/pr63306.C: New test.

Added:
    branches/gcc-4_9-branch/gcc/testsuite/g++.dg/ipa/pr63306.C
Modified:
    branches/gcc-4_9-branch/gcc/ChangeLog
    branches/gcc-4_9-branch/gcc/cgraphunit.c
    branches/gcc-4_9-branch/gcc/testsuite/ChangeLog