------- Comment #6 from andreast at gcc dot gnu dot org 2009-11-14 20:57 ------- I tried on another core2duo (x86_64-apple-darwin9), same issue. W/o extra_gij_ldflags=-Wl,-allow_stack_execute for ecjx_LINK I fail with the same entry in Library/Logs/CrashReporter:
ecj1_2009-11-14-202538_imac-catherine.crash ... Thread 0 Crashed: 0 libSystem.B.dylib 0x00007fff83de6f16 __kill + 10 1 libgcj.11.dylib 0x00000001000129db _Jv_Throw + 75 (exception.cc:128) 2 libgcj.11.dylib 0x00000001000050fd __ZL10catch_segvi + 61 3 libSystem.B.dylib 0x00007fff83de53fa _sigtramp + 26 4 ??? 0x0000000103f057a0 0 + 4361050016 Thread 1: 0 libSystem.B.dylib 0x00007fff83d86dc2 __semwait_signal + 10 1 libgcj.11.dylib 0x0000000100060bd4 _Jv_CondWait(_Jv_ConditionVariable_t*, _Jv_Mutex_t*, long long, int) + 404 ... Adding '-Wl,-allow_stack_execute' to the ecjx_LINK stage, I can build _and_ execute the resulting class (with gij) and the resulting binary w/o CrashReporter call. Also, I removed the '-Wl,-allow_stack_execute' invocation in testsuite/lib/libjava.exp and ran the testsuite. I then found exactly the same pattern as from the above pattern from CrashReporter for several test cases. >From my POV, the failure has nothing todo with unwinder issues. I'd like to know what CPU you're running the build/install/test on. Is your extra_gij_ldflags equal to '-Wl,-allow_stack_execute'? Do you clean the install directory before installing a new version? And yes, it is not gcj which fails but ecj. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41991
