http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55617
--- Comment #7 from Alexander Potapenko <glider at google dot com> 2013-01-29 11:56:02 UTC --- Here's the dump of __mod_init_func (the static ctors array): =================================================== Disassembly of section __DATA.__mod_init_func: 0000000100001040 <__DATA.__mod_init_func>: 100001040: 5c pop %rsp 100001041: 0d 00 00 01 00 or $0x10000,%eax 100001046: 00 00 add %al,(%rax) 100001048: 88 0d 00 00 01 00 mov %cl,0x10000(%rip) # 10001104e <_ret+0xff6e> =================================================== -- Looks like both __GLOBAL__sub_I_00099_1_cov.cc (0000000100000d88, which is the analog of _asan.module_ctor in Clang instrumentation) and __GLOBAL__sub_I_cov.cc (0000000100000d5c, the original module ctor) are present in __mod_init_func, but are ordered incorrectly. I've fixed the order using bvi for OS X: =================================================== 0000000100001040 <__DATA.__mod_init_func>: 100001040: 88 0d 00 00 01 00 mov %cl,0x10000(%rip) # 100011046 <_ret+0xff66> 100001046: 00 00 add %al,(%rax) 100001048: 5c pop %rsp 100001049: 0d 00 00 01 00 or $0x10000,%eax =================================================== and the resulting binary didn't segfault for me.
