Crazy stack trace

Question

I've added app hang detector to my Windows app recently. All it does - crashes my app in case it detects that app hang. In main window's procedure I update last time it was called. In a separate thread I check if the last time was not updated during 60 seconds. Note that I create this checking thread in main window's procedure. Not earlier.

Now I received the dump.

Main thread's stack is something crazy. All it shows is this:

ntdll.dll!_KiFastSystemCallRet@0()
ntdll.dll!_NtWaitForSingleObject@12()
ntdll.dll!_NtQueryInformationProcess@20()
kernel32.dll!@BaseThreadInitThunk@12()

BaseThreadInitThunk is present. But where is my code?

All symbols for my and system's modules are loaded.

If my code would not be called - app hung detector not crashed my app. So my code must be called. But where is it?

If you go looking for crazy ways in which an app can deadlock and hang then you're apt to find it. That thread didn't start executing your code yet. — Hans Passant, Dec 18 '14 at 11:43
It did. Because it's main thread and it created several service threads in my app, which works fine and stack trace for them is ok. — Alexander Dyagilev, Dec 18 '14 at 11:50
The kernel does play stack switching games with threads. Running a 32-bit app on a 64-bit version of Windows can be interesting as well. — Hans Passant, Dec 18 '14 at 12:10
Stack switching games with threads? Could you please provide some details? — Alexander Dyagilev, Dec 18 '14 at 12:40
Have you tried using _Application Verifier_ to trap the hang condition? — rrirower, Dec 18 '14 at 13:33
No. The problem is not in the hang. The problem is in this stack trace. I would like to understand the reason. — Alexander Dyagilev, Dec 19 '14 at 15:33
Note. This is just the one of the anonymous automatic reports users send to us. — Alexander Dyagilev, Dec 19 '14 at 17:56

Crazy stack trace

0 Answers0