Apache OpenOffice (AOO) Bugzilla – Issue 111271
Mac OS X: crash during shutdown, __tcf_2 on stack
Last modified: 2017-05-20 11:33:12 UTC
Building smoketestoo_native of DEV300_m77 based CWS sb120 on unxmacxi non-pro once failed with a crashing soffice, build output ---8<--- /net/so-cwsserv02/export/cws/sb120/DEV300/ooo/test/source/cpp/officeconnection.cxx:149: Assertion Test name: (anonymous namespace)::Test::test tearDown() failed - equality assertion failed - Expected: 0 - Actual : 4294967295 Failures !!! Run: 1 Failure total: 1 Failures: 1 Errors: 0 dmake: Error code 1, while making 'cpptest' ---8<--- and a new soffice instance starting up (with the first start wizard, as smoketestoo_native calls soffice with -nofirststartwizard), even though smoketestoo_native calls soffice -norestore. Thanks to ulimit -c unlimited, a /cores/core.59712 was generated, which does not look to helpful though: (gdb) info threads 15 core thread 14 0x9630a688 in objc_msgSend () 14 core thread 13 0x96e54b96 in recvfrom$NOCANCEL$UNIX2003 () 13 core thread 12 0x96e0c2ae in semaphore_wait_signal_trap () 12 core thread 11 0x96e7918e in wait4 () 11 core thread 10 0x96e0c266 in mach_msg_trap () 10 core thread 9 0x96e0c266 in mach_msg_trap () 9 core thread 8 0x96e0c266 in mach_msg_trap () 8 core thread 7 0x96e0c266 in mach_msg_trap () 7 core thread 6 0x96e0c2a2 in semaphore_wait_trap () 6 core thread 5 0x96e0c266 in mach_msg_trap () 5 core thread 4 0x96e0c266 in mach_msg_trap () 4 core thread 3 0x96e0c266 in mach_msg_trap () 3 core thread 2 0x96e0c266 in mach_msg_trap () 2 core thread 1 0x96e0c2c6 in semaphore_timedwait_signal_trap () * 1 core thread 0 0x2a75a7f6 in __tcf_2 () (gdb) thread 1 [Switching to thread 1 (core thread 0)] 0x2a75a7f6 in __tcf_2 () (gdb) where #0 0x2a75a7f6 in __tcf_2 () #1 0x96e34d9c in __cxa_finalize () #2 0x96e34c90 in exit () #3 0x96f7927a in NSApplicationMain () #4 0x01f3ce0e in ImplSVMainHook () #5 0x01c592cb in SVMain () #6 0x000d5f2c in soffice_main () #7 0x00002f0e in main () (gdb) info sharedlibrary 0x2a75a7f6 217 libbf_frmmxi.dylib - 0x2a6ce000 dyld Y Y /net/so-cwsserv02/export/cws/sb120/DEV300/unxmacxi/installation/opt/OpenOffice.org.app/Contents/basis-link/program/libbf_frmmxi.dylib at 0x2a6ce000 (offset 0x2a6ce000) (objfile is) [memory object "/net/so-cwsserv02/export/cws/sb120/DEV300/unxmacxi/installation/opt/OpenOffice.org.app/Contents/basis-link/program/libbf_frmmxi.dylib" at 0x2a6ce000] (gdb) disas 0x2a75a7f6 Dump of assembler code for function __tcf_2: 0x2a75a7f6 <__tcf_2+0>: push %ebp 0x2a75a7f7 <__tcf_2+1>: mov %esp,%ebp 0x2a75a7f9 <__tcf_2+3>: call 0x2a7bed64 <__i686.get_pc_thunk.cx> 0x2a75a7fe <__tcf_2+8>: mov 0xf90f2(%ecx),%eax 0x2a75a804 <__tcf_2+14>: mov %eax,0x8(%ebp) 0x2a75a807 <__tcf_2+17>: pop %ebp 0x2a75a808 <__tcf_2+18>: jmp 0x2a85405f <dyld_stub_rtl_uString_release> 0x2a75a80d <__tcf_2+23>: nop End of assembler dump. (gdb) info reg eax 0x0 0 ecx 0x0 0 edx 0x0 0 ebx 0x96e34cb6 -1763488586 esp 0xbfffa6fc 0xbfffa6fc ebp 0xbfffa758 0xbfffa758 esi 0x437e5a8 70772136 edi 0x19 25 eip 0x2a75a7f6 0x2a75a7f6 <__tcf_2> eflags 0x10246 66118 cs 0x17 23 ss 0x1f 31 ds 0x1f 31 es 0x1f 31 fs 0x0 0 gs 0x37 55 (gdb) x/16b 0xbfffa6f0 0xbfffa6f0: 0xa8 0xe5 0x37 0x04 0x19 0x00 0x00 0x00 0xbfffa6f8: 0x58 0xa7 0xff 0xbf 0x9c 0x4d 0xe3 0x96 (I had also occasionally experienced similar scenarios during shutdown when running the unoapi tests in the same setting, also with __tcf_2 on top of the crashing(?) stack, but that symbol belonging to various libraries, not necessarily libbf_frmmxi.dylib.)
No idea, only that __tcf_ seem to be destructors for static objects; see e.g. this bug report http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17977 That issue itself is of course not the problem here; might be loading/unloading order. Me no likey static destructors :-(
raising prio, as this affects build stability (see <http://tools.openoffice.org/servlets/ReadMsg?list=tinderbox&msgNo=398> for enabling subsequenttests on buildbots)
.
Reset assigne to the default "issues@openoffice.apache.org".