[Bug libgcc/56101] pthread program abort

[masahiko.kimoto at miraclelinux dot com via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56101

Masahiko Kimoto  changed:

   What|Removed |Added

 CC||masahiko.kimoto@miraclelinu
   ||x.com

--- Comment #6 from Masahiko Kimoto  
---
In case of removing GCC runtime libraries(libgcc_s.so, libstdc++.so) from root
filesystem, especially required in embedded environment, this constrainti
annoys us.

Is there any workaround without using dlopen(libgcc_s) ?

According to our examination, it does not abort when all libraries are linked
statically by -static option of GCC.

[Bug libgcc/56101] pthread program abort

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56101

--- Comment #5 from Andrew Pinski  ---
*** Bug 55779 has been marked as a duplicate of this bug. ***

[Bug libgcc/56101] pthread program abort

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56101

Andrew Pinski  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |WONTFIX

--- Comment #4 from Andrew Pinski  ---
This is the reason for a shared libgcc.  Not a bug.

[Bug libgcc/56101] pthread program abort

[siddhesh at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56101

--- Comment #3 from Siddhesh Poyarekar  ---
Here's the simplified non-preprocessed test, since it's not a problem with the
compiler proper:

$ cat > tmain.C
#include 
#include 
#include 

extern "C"
void *worker_thread(void *tap)
{
   int j[1] = { 0 }; 

   j[0] += *(int *)tap;

  pthread_exit(NULL);
  return NULL;
}

int main(void)
{
  pthread_t t;
  int targs;

  if (pthread_create(&t, NULL, worker_thread, &targs) != 0)
exit(-1);

  return(pthread_join(t, NULL));
}

[Bug libgcc/56101] pthread program abort

[siddhesh at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56101

Siddhesh Poyarekar  changed:

   What|Removed |Added

 CC||siddhesh at redhat dot com

--- Comment #2 from Siddhesh Poyarekar  ---
It's because libpthread dynamically loads libgcc_s.so.1.  As a result there are
two copies of dwarf_reg_size_table, where one (the dynamic one) is initialized
and the other (the static one) is used, resulting in this crash.

[Bug libgcc/56101] pthread program abort

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56101



Jonathan Wakely  changed:



   What|Removed |Added



 Status|UNCONFIRMED |NEW

   Last reconfirmed||2013-01-25

 Ever Confirmed|0   |1



--- Comment #1 from Jonathan Wakely  2013-01-25 
12:38:31 UTC ---

(In reply to comment #0)

> /remote/depotsrc/depotsrc/amd64-2.6/local_install/gcc-4.7.2_test2/bin/g++ -I.

> -Wall -Wextra -fno-strict-aliasing -fwrapv -m64 -c  -msse2 -mfpmath=sse 

> -fno-omit-frame-pointer -fno-dollars-in-identifiers -save-temps -pthread -o

> tmain.o tmain.C

> /remote/depotsrc/depotsrc/amd64-2.6/local_install/gcc-4.7.2_test2/bin/g++

> -static-libgcc -static-libstdc++  -o mtest tmain.o -pthread -lm -ldl



Most of these flags are irrelevant, only these are needed to reproduce it:



   -static-libgcc -static-libstdc++ -pthread



If either libgcc or libstdc++ is dynamically linked it doesn't abort