22183 2008-11-11 13:14:39 -0800 Crashes in Leopard build bot in LayoutTests/storage and LayoutTests/fast/js/exceptions-thrown-in-callbacks.html 2009-07-31 15:35:05 -0700 1 1 1 Unclassified WebKit Tools / Tests 528+ (Nightly build) Mac OS X 10.5 RESOLVED INVALID P2 Normal --- 1 pol webkit-unassigned bolsinga darin simon.fraser oldest_to_newest 98333 0 pol 2008-11-11 13:14:39 -0800 For some time, there has been 10 crashes in the Leopard build bot in LayoutTests/storage: http://build.webkit.org/results/trunk-mac-intel-pixel/197/results.html 98393 1 pol 2008-11-11 16:47:13 -0800 Here's the backtrace (apparently the same for all crashes): Exception Type: EXC_BAD_ACCESS (SIGBUS) Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000000000000 Crashed Thread: 0 Thread 0 Crashed: 0 libSystem.B.dylib 0x92efcd9e pthread_mutex_lock + 18 1 com.apple.WebCore 0x01455061 WebCore::guidForOriginAndName(WebCore::String const&, WebCore::String const&) + 529 (Database.cpp:368) 2 com.apple.WebCore 0x01455347 WebCore::Database::Database(WebCore::Document*, WebCore::String const&, WebCore::String const&) + 375 (Database.cpp:135) 3 com.apple.WebCore 0x0145667b WebCore::Database::openDatabase(WebCore::Document*, WebCore::String const&, WebCore::String const&, WebCore::String const&, unsigned long, int&) + 123 (Database.cpp:101) 4 com.apple.WebCore 0x0144eb1f WebCore::DOMWindow::openDatabase(WebCore::String const&, WebCore::String const&, WebCore::String const&, unsigned long, int&) + 143 (DOMWindow.cpp:814) 5 com.apple.WebCore 0x015f8e88 WebCore::jsDOMWindowPrototypeFunctionOpenDatabase(JSC::ExecState*, JSC::JSObject*, JSC::JSValue*, JSC::ArgList const&) + 664 (JSDOMWindow.cpp:4230) Thread 1: 0 libSystem.B.dylib 0x92ef54a6 mach_msg_trap + 10 1 libSystem.B.dylib 0x92efcc9c mach_msg + 72 2 com.apple.CoreFoundation 0x92cc00ce CFRunLoopRunSpecific + 1790 3 com.apple.CoreFoundation 0x92cc0cf8 CFRunLoopRunInMode + 88 4 com.apple.CFNetwork 0x948aaeca CFURLCacheWorkerThread(void*) + 396 5 libSystem.B.dylib 0x92f266f5 _pthread_start + 321 6 libSystem.B.dylib 0x92f265b2 thread_start + 34 Thread 2: 0 libSystem.B.dylib 0x92ef54a6 mach_msg_trap + 10 1 libSystem.B.dylib 0x92efcc9c mach_msg + 72 2 com.apple.CoreFoundation 0x92cc00ce CFRunLoopRunSpecific + 1790 3 com.apple.CoreFoundation 0x92cc0cf8 CFRunLoopRunInMode + 88 4 com.apple.Foundation 0x913cf100 +[NSURLConnection(NSURLConnectionReallyInternal) _resourceLoadLoop:] + 320 5 com.apple.Foundation 0x9136bbad -[NSThread main] + 45 6 com.apple.Foundation 0x9136b754 __NSThread__main__ + 308 7 libSystem.B.dylib 0x92f266f5 _pthread_start + 321 8 libSystem.B.dylib 0x92f265b2 thread_start + 34 Thread 3: 0 libSystem.B.dylib 0x92f455e2 select$DARWIN_EXTSN + 10 1 libSystem.B.dylib 0x92f266f5 _pthread_start + 321 2 libSystem.B.dylib 0x92f265b2 thread_start + 34 Thread 0 crashed with X86 Thread State (32-bit): eax: 0x00000000 ebx: 0x92efcd9d ecx: 0x00000000 edx: 0x00000000 edi: 0x00000000 esi: 0x16ed4180 ebp: 0xbfffdc18 esp: 0xbfffdbe0 ss: 0x0000001f efl: 0x00010282 eip: 0x92efcd9e cs: 0x00000017 ds: 0x0000001f es: 0x0000001f fs: 0x00000000 gs: 0x00000037 cr2: 0x00000000 98395 2 pol 2008-11-11 16:47:41 -0800 *** Bug 22184 has been marked as a duplicate of this bug. *** 98396 3 pol 2008-11-11 16:48:41 -0800 There's a 11th crash with the same backtrace here: LayoutTests/fast/js/exceptions-thrown-in-callbacks.html 98397 4 simon.fraser 2008-11-11 16:48:51 -0800 This looks a bit like one of Greg's: static Mutex& stringIdentifierMutex = *new Mutex; ... static HashMap<String, int>& stringIdentifierToGUIDMap = *new HashMap<String, int>; 98398 5 bolsinga 2008-11-11 16:51:13 -0800 I'll pretty sure this is covered by the latest patch in Bug 21810. 98400 6 simon.fraser 2008-11-11 16:53:00 -0800 Darin committed that. It's related to bug 22061 and bug 22061. I'm guessing we're seeing the same compiler issue discussed in 21810. 135950 7 bolsinga 2009-07-30 12:54:33 -0700 Is this applicable anymore? 136270 8 bolsinga 2009-07-31 15:23:29 -0700 I do not see LayoutTests/storage nor LayoutTests/fast/js/exceptions-thrown-in-callbacks.html in today's Mac Leopard skip list, so these tests are now successfully running. So this can be closed.