[zeromq-dev] libmzq crashes down in MSVCR100.dll

Tim tim.schultze at intelligrated.com
Mon Jun 10 17:57:32 CEST 2013


I have an app that is using clrmzq v2.2.5 (the latest stable release) and it 
appears that, after opening approximately 30 or 35 sockets (DEALER->REP) 
where there is no-one on the other end yet, the app crashes with:
Unhandled exception at 0x77e4bef7 in SimDashboard.exe: 0x40000015: %hs.
When I bring up the Visual Studio debugger as a result of this crash 
message, I'm placed in Microsoft's threadex.c file in teh 
_callthreadstartex(void) function (in what appears to be a _endthreadex 
function call.
The call stack is a little iffy, perhaps:
 	kernel32.dll!77e4bef7() 	
 	[Frames below may be incorrect and/or missing, no symbols loaded for 
kernel32.dll]	
 	kernel32.dll!77e4bef7() 	
 	libzmq.dll!100052ba() 	
 	libzmq.dll!1001792a() 	
 	libzmq.dll!1001cd07() 	
 	libzmq.dll!10010e41() 	
 	libzmq.dll!10010ede() 	
 	libzmq.dll!1001810e() 	
-->	msvcr100.dll!_callthreadstartex()  Line 314 + 0x6 bytes	C
 	msvcr100.dll!_threadstartex(void * ptd)  Line 292 + 0x5 bytes	C
 	kernel32.dll!77e6481f() 	

Details:
Targeting .NET 3.5 (but have also tested targeting .NET 4.0)
Tried the same using the latest pre-release of clrmzq (3.0.0.0 RC1) with 
similar results.
Modified the code to open socket, send msg, close socket with varying 
results. I'm a little shaky but recall that I got the same crash perhaps 
after a larger number of connections.
Added delays between initial open/send cycles. That made absolutely no 
difference (still dying after about 30 sockets opened and a message sent.
The one distinction between good/bad behavior seems to be Windows 2003 
Server 32-bit and Windows 2008 64-bit environments. I can't find any 
indication of threading issues in Server 2003 but perhaps that is it.




More information about the zeromq-dev mailing list