[Bug lto/53780] New: [l4.7.1 lto] linker fails with lto and "standard" object file

[Bug lto/53780] New: [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

             Bug #: 53780
           Summary: [l4.7.1 lto] linker fails with lto and "standard"
                    object file
    Classification: Unclassified
           Product: gcc
           Version: 4.7.1
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: lto
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: vincenzo.innocente@cern.ch


here mostly for the record as works with 4.8.0 and it has an easy workaround
it maybe related to PR53337  

either on vanilla fedora 17 
or with
gcc version 4.7.2 20120615 (prerelease) [gcc-4_7-branch revision 188650] (GCC) 
GNU gold (GNU Binutils 2.22.52.20120515) 1.11

linker fails in the following example
c++ -fPIC -std=c++11 -fvisibility-inlines-hidden -O2 -flto -c boostLto.cc
c++ -fPIC -std=c++11 -fvisibility-inlines-hidden -Os  -c boostLto2.cc
echo link ok
c++  -flto -O2 -pthread boostLto2.o boostLto.o  -shared -fPIC -o bha.so
-fvisibility-inlines-hidden -std=c++11
echo link fails
c++  -flto -O2 -pthread boostLto.o boostLto2.o  -shared -fPIC -o bha.so
-fvisibility-inlines-hidden -std=c++11


fedora 17

link ok
link fails
`_ZTIN5boost6detail15sp_counted_baseE' referenced in section
`.data.rel.ro._ZTVN5boost6detail15sp_counted_baseE[_ZTVN5boost6detail15sp_counted_baseE]'
of boostLto2.o: defined in discarded section
`.gnu.linkonce.t._ZTIN5boost6detail15sp_counted_baseE' of boostLto.o (symbol
from plugin)
`_ZThn16_N5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEED1Ev'
referenced in section
`.data.rel.ro._ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEEE[_ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEEE]'
of boostLto2.o: defined in discarded section
`.gnu.linkonce.t._ZN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEED5Ev'
of boostLto.o (symbol from plugin)
`_ZThn16_N5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEED0Ev'
referenced in section
`.data.rel.ro._ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEEE[_ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEEE]'
of boostLto2.o: defined in discarded section
…..
`_ZThn56_N5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEED0Ev'
referenced in section `.data.rel.ro' of /tmp/ccnOkzqM.ltrans1.ltrans.o: defined
in discarded section
`.gnu.linkonce.t._ZN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_21thread_resource_errorEEEED0Ev'
of boostLto.o (symbol from plugin)
/bin/ld: bha.so: hidden symbol `.LTHUNK2.3978' isn't defined
/bin/ld: final link failed: Bad value
collect2: error: ld returned 1 exit status


=====================================

with
gcc version 4.7.2 20120615 (prerelease) [gcc-4_7-branch revision 188650] (GCC) 
GNU gold (GNU Binutils 2.22.52.20120515) 1.11

link ok
link fails
boostLto2.o:boostLto2.cc:function vtable for
boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::gregorian::bad_day_of_month>
>: warning: relocation refers to discarded section
/afs/cern.ch/user/i/innocent/w3/gcc47slc5/bin/ld: error: hidden symbol
'_ZThn16_N5boost16exception_detail10clone_implINS0_19error_info_injectorINS_9gregorian16bad_day_of_monthEEEED1Ev'
is not defined locally
…..
/tmp/innocent/ccZ6RxXH.ltrans1.ltrans.o:ccZ6RxXH.ltrans1.o:function vtable for
boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::gregorian::bad_day_of_year>
> [clone .local.282]: warning: relocation refers to discarded section
/afs/cern.ch/user/i/innocent/w3/gcc47slc5/bin/ld: error: hidden symbol
'_ZThn56_N5boost16exception_detail10clone_implINS0_19error_info_injectorINS_9gregorian15bad_day_of_yearEEEED0Ev'
is not defined locally
collect2: error: ld returned 1 exit status

==================
code
cat boostLto.cc
#include "boost/thread/mutex.hpp"
namespace edm0 {
    boost::mutex* getGlobalMutex();
}

boost::mutex* edm0::getGlobalMutex() { 
    static boost::mutex m_;
    return &m_;
}

[innocent@pcphsftkepler ctest]$ cat boostLto2.cc
#include "boost/thread/mutex.hpp"
namespace edm2 {
    boost::mutex* getGlobalMutex();
}

boost::mutex* edm2::getGlobalMutex() { 
    static boost::mutex m_;
    return &m_;
}

#include "boost/shared_ptr.hpp"
#include "boost/date_time.hpp"
#include "boost/date_time/gregorian/gregorian_types.hpp"
#include "boost/date_time/posix_time/posix_time_types.hpp"
#include "boost/date_time/c_local_time_adjustor.hpp"
typedef boost::posix_time::ptime TimePoint_t;
typedef boost::posix_time::time_duration Duration_t;

#include<string>
#include<sstream>
#include<iomanip>

#include<ctime>

struct A {
  A(){}
  virtual ~A(){}
  A(int j) : i(j){}
int i;
};

struct B {

  A & a() {return *pa;}
  boost::shared_ptr<A> pa;
  int j;

};


boost::shared_ptr<A> make2(int i) {
   return boost::shared_ptr<A>(new A(i));
}


int get2(boost::shared_ptr<A> a) {
  return (*a).i;

}


int get2(B b) {
  return b.a().i; 
}

 std::string dateStamp2(TimePoint_t theTime) {
       typedef boost::date_time::c_local_adjustor<boost::posix_time::ptime>
local_adj;
       tm ptm = boost::posix_time::to_tm( local_adj::utc_to_local(theTime) );
       std::ostringstream dateStampStr;
       dateStampStr << std::setfill('0') << std::setw(4) << ptm.tm_year+1900
                    << std::setfill('0') << std::setw(2) << ptm.tm_mon+1
                    << std::setfill('0') << std::setw(2) << ptm.tm_mday;
       return dateStampStr.str();
}

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #1 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-28 12:52:42 UTC ---
the failure on vanilla fedora may be related to PR53768

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #2 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 06:11:19 UTC ---
seems fixed by the back-port published in comment 4 PR53768

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #3 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 07:39:02 UTC ---
The patch  HAS a side effect that I observed also in 4.8 and had no time to
reduce yet.
(At least now I know the origin of it)
something pretty weird
plenty of 
error:
tmp/slc5_amd64_gcc470/src/FWCore/Modules/src/FWCoreModules/MulticoreRunLumiEventChecker.o:
multiple definition of
'edm::WorkerT<edm::EDAnalyzer>::implDoBegin(edm::EventPrincipal&,
edm::EventSetup const&, edm::CurrentProcessingContext const*)::{lambda(ed
m::Worker*)#1}::operator void (*)(edm::Worker*)() const'
/afs/cern.ch/user/i/innocent/w3/gcc47slc5/bin/ld:
tmp/slc5_amd64_gcc470/src/FWCore/Modules/src/FWCoreModules/EventContentAnalyzer.o:
previous definition he
re
/afs/cern.ch/user/i/innocent/w3/gcc47slc5/bin/ld: error:
tmp/slc5_amd64_gcc470/src/FWCore/Modules/src/FWCoreModules/Prescaler.o:
multiple definition of 
'edm::WorkerT<edm::EDFilter>::implDoBegin(edm::EventPrincipal&, edm::EventSetup
const&, edm::CurrentProcessingContext const*)::{lambda(edm::Worker*)#1}::_FUN(
edm::Worker*)'
/afs/cern.ch/user/i/innocent/w3/gcc47slc5/bin/ld:
tmp/slc5_amd64_gcc470/src/FWCore/Modules/src/FWCoreModules/LogErrorFilter.o:
previous definition here

ld is 
GNU gold (GNU Binutils 2.22.52.20120515) 1.11

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #4 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 10:34:22 UTC ---
take back: (never do two updates at once)
in SOMETHING else committed between
gcc version 4.7.2 20120615 (prerelease) [gcc-4_7-branch revision 188650] (GCC) 
and
gcc-4_7-branch revision 189063

that causes the regression in comment 3

the patch above revision 188650 is ok and does not cause regression

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #5 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-06-29 10:50:48 UTC ---
Candidates are

+2012-06-19  Jason Merrill  <jason@redhat.com>
+
+       PR c++/52637
+       * dwarf2out.c (modified_type_die): Use scope_die_for.
+       (gen_type_die_with_usage, dwarf2out_finish): Likewise.
+       (uses_local_type_r, uses_local_type): New.
+       (scope_die_for): Keep a type that uses a local type in local scope.
+       Use get_context_die for namespace and type scope.


+2012-06-19  Richard Guenther  <rguenther@suse.de>
+
+       PR middle-end/53470
+       * tree.c (free_lang_data_in_type): Do not clear TYPE_CONTEXT but
+       replace it with the first non-BLOCK context.

if you are using -g and all C++ changes.

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #6 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 13:20:31 UTC ---
the regression in comment 3 happens with 
gcc version 4.7.2 20120620 (prerelease) [gcc-4_7-branch revision 188811] (GCC) 
(and was a fix required by myself)
will open a new bug report  (btw no -g involved!)
is connected to "sigcpp": will try to reproduce with something small

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

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

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

--- Comment #7 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-06-29 13:45:50 UTC ---
Hmm.  Does

Index: gcc/tree.c
===================================================================
--- gcc/tree.c  (revision 189046)
+++ gcc/tree.c  (working copy)
@@ -4852,7 +4852,15 @@ find_decls_types_r (tree *tp, int *ws, v
       fld_worklist_push (TYPE_MAIN_VARIANT (t), fld);
       /* Do not walk TYPE_NEXT_VARIANT.  We do not stream it and thus
          do not and want not to reach unused variants this way.  */
-      fld_worklist_push (TYPE_CONTEXT (t), fld);
+      if (TYPE_CONTEXT (t))
+       {
+         tree ctx = TYPE_CONTEXT (t);
+         /* We adjust BLOCK TYPE_CONTEXTs to the innermost non-BLOCK one.
+            So push that instead.  */
+         while (ctx && TREE_CODE (ctx) == BLOCK)
+           ctx = BLOCK_SUPERCONTEXT (ctx);
+         fld_worklist_push (ctx, fld);
+       }
       /* Do not walk TYPE_CANONICAL.  We do not stream it and thus do not
         and want not to reach unused types this way.  */

fix it?

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #8 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 15:03:17 UTC ---
NOT when applied on the top of
gcc version 4.8.0 20120629 (experimental) [trunk revision 189073] (GCC) 
svn diff
Index: gcc/tree.c
===================================================================
--- gcc/tree.c    (revision 189073)
+++ gcc/tree.c    (working copy)
@@ -4910,7 +4910,15 @@
       fld_worklist_push (TYPE_MAIN_VARIANT (t), fld);
       /* Do not walk TYPE_NEXT_VARIANT.  We do not stream it and thus
          do not and want not to reach unused variants this way.  */
-      fld_worklist_push (TYPE_CONTEXT (t), fld);
+      if (TYPE_CONTEXT (t))
+       {
+         tree ctx = TYPE_CONTEXT (t);
+         /* We adjust BLOCK TYPE_CONTEXTs to the innermost non-BLOCK one.
+            So push that instead.  */
+         while (ctx && TREE_CODE (ctx) == BLOCK)
+           ctx = BLOCK_SUPERCONTEXT (ctx);
+         fld_worklist_push (ctx, fld);
+       }
       /* Do not walk TYPE_CANONICAL.  We do not stream it and thus do not
      and want not to reach unused types this way.  */

I'll try also with 4.7

did not manage yet to reproduce with simple examples using sigcpp...

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #9 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 15:31:12 UTC ---
in any case the "regression" disappear if I declare inline the template methods
in question
  template<typename T>
+  inline
   bool 
   WorkerT<T>::implDoBegin(EventPrincipal& ep, EventSetup const& c,
CurrentProcessingContext const* cpc) {

etc

so I would suggest not to touch anything until is not understood If it a real
compiler bug or a "user" code mistake…

v.



On 29 Jun, 2012, at 5:03 PM, vincenzo.innocente at cern dot ch wrote:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780
> 
> --- Comment #8 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-29 15:03:17 UTC ---
> NOT when applied on the top of
> gcc version 4.8.0 20120629 (experimental) [trunk revision 189073] (GCC) 
> svn diff
> Index: gcc/tree.c
> ===================================================================
> --- gcc/tree.c    (revision 189073)
> +++ gcc/tree.c    (working copy)
> @@ -4910,7 +4910,15 @@
>       fld_worklist_push (TYPE_MAIN_VARIANT (t), fld);
>       /* Do not walk TYPE_NEXT_VARIANT.  We do not stream it and thus
>          do not and want not to reach unused variants this way.  */
> -      fld_worklist_push (TYPE_CONTEXT (t), fld);
> +      if (TYPE_CONTEXT (t))
> +       {
> +         tree ctx = TYPE_CONTEXT (t);
> +         /* We adjust BLOCK TYPE_CONTEXTs to the innermost non-BLOCK one.
> +            So push that instead.  */
> +         while (ctx && TREE_CODE (ctx) == BLOCK)
> +           ctx = BLOCK_SUPERCONTEXT (ctx);
> +         fld_worklist_push (ctx, fld);
> +       }
>       /* Do not walk TYPE_CANONICAL.  We do not stream it and thus do not
>      and want not to reach unused types this way.  */
> 
> I'll try also with 4.7
> 
> did not manage yet to reproduce with simple examples using sigcpp...
>

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[vincenzo.innocente at cern dot ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

--- Comment #10 from vincenzo Innocente <vincenzo.innocente at cern dot ch> 2012-06-30 15:24:24 UTC ---
reduced testcase for comment 3 regression as PR53821

[Bug lto/53780] [l4.7.1 lto] linker fails with lto and "standard" object file

[matt at use dot net]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53780

Matt Hargett <matt at use dot net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |matt at use dot net

--- Comment #11 from Matt Hargett <matt at use dot net> 2012-10-11 19:10:00 UTC ---
I am also seeing the message:
boost::exception_detail::clone_impl<boost::exception_detail::error_info_injector<boost::gregorian::bad_day_of_year>
 warning: relocation refers to discarded section

in my final link of a large C++ program compiled with LTO (with boost not
compiled as LTO). I tried compiling boost with LTO to work around the problem,
but got an ICE instead (will file that later).

Do the patches mentioned below (still) fix the problem in 4.7.x?