[Bug middle-end/56964] ICE with -fno-sync-libcalls when target lacks atomic operations

[rwahl at gmx dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56964

rwahl at gmx dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rwahl at gmx dot de

--- Comment #1 from rwahl at gmx dot de ---
I can reproduce this too when building multilib cross compiler:

cat
gcc-4.8.3/build-arm-pp-linux-uclibcgnueabi/arm-pp-linux-uclibcgnueabi/libatomic/config.log
...
configure:12468: checking for __atomic_test_and_set for size 1
configure:12487:
/home/nb/builds/toolchain-010/tmp_cross/build/gcc-4.8.3/build-arm-pp-linux-uclibcgnueabi/./gcc/xgcc
-B/home/nb/builds/toolchain-010/tmp_cross/build/gcc-4.8.3/build-arm-pp-linux-uclibcgnueabi/./gcc/
-B/home/nb/builds/toolchain-010/tmp_cross/build_root_2//bin/
-B/home/nb/builds/toolchain-010/tmp_cross/build_root_2//lib/ -isystem
/home/nb/builds/toolchain-010/tmp_cross/build_root_2//arm-pp-linux-uclibcgnueabi/include
--sysroot=/home/nb/builds/toolchain-010/tmp_cross/build_root_2//arm-pp-linux-uclibcgnueabi
  -o conftest -Os -fno-sync-libcalls    conftest.c  >&5
conftest.c: In function 'main':
conftest.c:40:27: internal compiler error: Segmentation fault
      __atomic_test_and_set(x, 0);
                           ^
Please submit a full bug report,
...

Under different build circumstances the build of conftest.c even just hangs
which was my original problem I was trying to solve and it took me half a day
to dig around and find that bug report.