Closed Bug 763081 Opened 9 years ago Closed 9 years ago
OOP <iframe mozbrowser> tests crash on Windows
See also bug 776832 comment 44.
Bug 776832 comment 38 gives these two stack traces: Thread 40 (crashed) 0 crashinjectdll.dll!CrashingThread(void *) [crashinjectdll.cpp:cfbb2b8a8728 : 17 + 0x0] eip = 0x03511000 esp = 0x08ffffb8 ebp = 0x08ffffec ebx = 0x00000000 esi = 0x8f8e8d8c edi = 0x8b8a8988 eax = 0x00000000 ecx = 0x08ffffb0 edx = 0x7c90eb94 efl = 0x00010246 Found by: given as instruction pointer in context 1 kernel32.dll + 0xb50a eip = 0x7c80b50b esp = 0x08ffffbc ebp = 0x08ffffec Found by: stack scanning Thread 3 (crashed) 0 ntdll.dll + 0x7f392 eip = 0x7791f392 esp = 0x029afbd0 ebp = 0x029afbd0 ebx = 0xfffffffb esi = 0x08fe463c edi = 0x08fe4640 eax = 0xc0000024 ecx = 0x029afbbc edx = 0x778e64f4 efl = 0x00000286 Found by: given as instruction pointer in context 1 ntdll.dll + 0x2fd0f eip = 0x778cfd10 esp = 0x029afbd8 ebp = 0x029afbe8 Found by: previous frame's frame pointer 2 nspr4.dll!_PR_MD_UNLOCK [w95cv.c:cfbb2b8a8728 : 312 + 0x6] eip = 0x7354a3a0 esp = 0x029afbf0 ebp = 0x029afc0c Found by: previous frame's frame pointer 3 nspr4.dll!PR_Unlock [prulock.c:cfbb2b8a8728 : 315 + 0x8] eip = 0x73546472 esp = 0x029afbf8 ebp = 0x029afc0c Found by: stack scanning 4 xul.dll!mozilla::ipc::AsyncChannel::ProcessLink::OnMessageReceived(IPC::Message const &) [AsyncChannel.cpp:cfbb2b8a8728 : 667 + 0x7] eip = 0x6cc97caf esp = 0x029afc00 ebp = 0x029afc0c Found by: stack scanning 5 xul.dll!IPC::Channel::ChannelImpl::ProcessIncomingMessages(base::MessagePumpForIO::IOContext *,unsigned long) [ipc_channel_win.cc:cfbb2b8a8728 : 333 + 0xb] eip = 0x6cd7795c esp = 0x029afc14 ebp = 0x029afc4c Found by: call frame info 6 xul.dll!IPC::Channel::ChannelImpl::OnIOCompleted(base::MessagePumpForIO::IOContext *,unsigned long,unsigned long) [ipc_channel_win.cc:cfbb2b8a8728 : 427 + 0x13] eip = 0x6cd77b9d esp = 0x029afc54 ebp = 0x029afc68 Found by: call frame info 7 xul.dll!base::MessagePumpForIO::WaitForIOCompletion(unsigned long,base::MessagePumpForIO::IOHandler *) [message_pump_win.cc:cfbb2b8a8728 : 521 + 0xb] eip = 0x6cd7526b esp = 0x029afc70 ebp = 0x029afc90 Found by: call frame info 8 xul.dll!base::MessagePumpForIO::WaitForWork() [message_pump_win.cc:cfbb2b8a8728 : 500 + 0x9] eip = 0x6cd7532c esp = 0x029afc98 ebp = 0x029afcb4
I cannot reproduce this failure on my Windows 7 machine, by the way. At least not with a regular --disable-debug --enable-optimize build, running mochitests from the content/base/test directory.
I appreciate that you guys are looking at this, but why do we care? The failures on Windows don't seem to suggest failures on the device...
Probably for similar reasons as those for enabling OOP tests on other platforms.
(In reply to Justin Lebar [:jlebar] (PTO 9/27 - 10/1) from comment #4) > I appreciate that you guys are looking at this, but why do we care? The > failures on Windows don't seem to suggest failures on the device... I'm just afraid of a bug lurking somewhere that's just not as frequent on Gonk, but nevertheless present. But maybe you're right and it's just Windows IPC.
> I'm just afraid of a bug lurking somewhere that's just not as frequent on Gonk, but nevertheless > present. Ah, I see. That's entirely possible, although I don't have any reason to believe that it's true, at this point.
Testing these on Windows w/latest mc in bug 820200 I didn't see any problems. I'm tempted to enable these when I push the fix for 820200.
(In reply to Jim Mathies [:jimm] from comment #8) > Testing these on Windows w/latest mc in bug 820200 I didn't see any > problems. I'm tempted to enable these when I push the fix for 820200. If it works, that's great. If it doesn't work (or starts going even more randomorange than these tests normally go), I'm not going to lose sleep over keeping them disabled on Windows.
We've only seen one windows test failure (bug 782633) on elm across a series of runs. So I think these are safe to turn on.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.