Detecting Deadlocks in C++ Using a Locks Monitor
By Tomer Abramson, April 01, 2003
In multithreaded programming, locking mechanisms are a must in order to keep data consistent. This article will discuss a technique for detecting and breaking deadlocks in multithreaded environment. It will also present a simple and powerful way to analyze why the deadlock occurred.
Detecting Deadlocks in C++ Using a Locks Monitor
Listing 4 Output file lockMonitor log
*** Detect deadlock. Dump stack of all threads in deadlock cycle ***
start stack trace for thread 1732:
address 2012751848
function: NtWaitForSingleObject
module: ntdll.dll
address 2011708251
function: ExitVDM
module: KERNEL32.dll
address 4241383
function: void __cdecl f1(void)
file: d:\locksmonitor\main.cpp
line: 20 + 14 bytes
module: LocksMonitor.exe
address 4241501
function: unsigned long __stdcall Thread_1(void *)
file: d:\locksmonitor\main.cpp
line: 29 + 0 bytes
module: LocksMonitor.exe
address 2011784483
function: SetThreadAffinityMask
module: KERNEL32.dll
finish stack trace for thread 1732:
start stack trace for thread 2032:
address 2012751848
function: NtWaitForSingleObject
module: ntdll.dll
address 2011708251
function: ExitVDM
module: KERNEL32.dll
address 4241223
function: void __cdecl f2(void)
file: d:\locksmonitor\main.cpp
line: 11 + 14 bytes
module: LocksMonitor.exe
address 4241565
function: unsigned long __stdcall Thread_2(void *)
file: d:\locksmonitor\main.cpp
line: 35 + 0 bytes
module: LocksMonitor.exe
address 2011784483
function: SetThreadAffinityMask
module: KERNEL32.dll
finish stack trace for thread 2032:
killed thread 1732 In Order to break deadlock
