https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60244
--- Comment #9 from Vladimír Čunát <vcunat at gmail dot com> ---
More information about my case: it's x86_64, Linux, gcc-5.3.0 from tarball
being built by gcc-4.8.3. I re-tried with --enable-debug and ran the
segfaulting command in gdb (I filtered seemingly uninteresting lines):
$ gdb -- /tmp/nix-build-gcc-5.3.0.drv-1/build/./gcc/xgcc
(gdb) run -B/tmp/nix-build-gcc-5.3.0.drv-1/build/./gcc/ -dumpspecs
Program received signal SIGSEGV, Segmentation fault.
(gdb) bt full
#0 0x000000000047ea8c in atexit ()
No symbol table info available.
#1 0x000000000040fd9a in driver::global_initializations (this=0x7fffffffd9f0)
at ../../gcc-5.3.0/gcc/gcc.c:6985
No locals.
#2 0x000000000040fb81 in driver::main (this=0x7fffffffd9f0, argc=3,
argv=0x7fffffffdaf8) at ../../gcc-5.3.0/gcc/gcc.c:6900
early_exit = false
#3 0x0000000000416958 in main (argc=3, argv=0x7fffffffdaf8) at
../../gcc-5.3.0/gcc/gcc-main.c:45
d = {explicit_link_files = 0x47ea00 <__libc_csu_init>
"AWA\211\377AVI\211\366AUI\211\325ATL\215%0\033&", decoded_options = 0x6e6d70,
decoded_options_count = 3}
It really puzzles me that the segfault occurs *during* a call to atexit. It
uses glibc-2.23. Any idea about what could be wrong? Thanks.
