tracetool - a configurable and efficient logging framework for C++

tracetool is a framework created by froglogic for tracing the execution of a C or C++ program and inspecting its state. This is achieved by instrumenting the source code of the target program and linking the recompiled sources against a shared

tracelib

License

tracetool is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

Installation

Building the framework requires

• A C++ compiler, e.g. GCC or Microsoft Visual Studio 2008 or newer
• CMake 3.0.0 or newer
• Qt 5.6
• For building the documentation, Doxygen is needed.

The actual build is the merely a matter of running the commands

cmake -DCMAKE_PREFIX_PATH=
make

On Windows, you may want to pass

-G "NMake Makefiles"

nmake

Quick Start

Here's a quick step by step guide on how to instrument a basic program so that it generates trace information. Here's the initial source code of the sample program:

#include 
#include 

int main()
{
    using namespace std;
string name;
cout << "Please enter your name: ";
cin >> name;

cout << "Hello, " << name << "!" << endl;
}

First, instrument the above source code so that the

tracelib.h

#include 
#include 

#include "tracelib.h"
int main()
{
    fTrace(0) << "main() entered";
using namespace std;

string name;
cout &lt;&lt; "Please enter your name: ";
cin &gt;&gt; name;

fWatch(0) &lt;&lt; fVar(name);

cout &lt;&lt; "Hello, " &lt;&lt; name &lt;&lt; "!" &lt;&lt; endl;

fTrace(0) &lt;&lt; "main() finished";
}

Save the resulting file to e.g.

hello\_instrumented.cpp

Here's a sample compile line for use with

cl.exe

cl /EHsc hello_instrumented.cpp /I \include\tracelib
    \lib\tracelib.lib
    ws2_32.lib

The resulting binary (

hello\_instrumented.exe

Creating a configuration file is a matter of writing some XML. Save the following file as

tracelib.xml

hello\_instrumented.exe

  hello_instrumented
  
  

Now, running the program will generate quite a bit of additional output:

03.09.2010 16:00:56: Process 2524 [started at 03.09.2010 16:00:56] (Thread 468): [LOG] 'main() entered' hello_instrumented.cpp:8: int __cdecl main(void)
Please enter your name: Max
03.09.2010 16:00:57: Process 2524 [started at 03.09.2010 16:00:56] (Thread 468): [WATCH] hello_instrumented.cpp:16: int __cdecl main(void)
Hello, Max!
03.09.2010 16:00:57: Process 2524 [started at 03.09.2010 16:00:56] (Thread 468): [LOG] 'main() finished' hello_instrumented.cpp:20: int __cdecl main(void)
03.09.2010 16:00:57: Process 2524 [started at 03.09.2010 16:00:56] finished

Components

The tracetool framework consists of multiple components:

• tracelib.dll

  (resp.

  libtracelib.so

  on Linux) is a dynamic library (plus accompanying header files) which has to be linked into any application which wishes to use the tracing functionality.

• tracegui

  is a GUI for reviewing previously recorded traces as well as recording and watching the trace generated by running applications live.

• traced

  is a daemon process which collects and stores trace data in the background without running a GUI.Traced applications can be configured to send their output over a network connection to

  traced.exe

  . The recorded traces can be sent to other people and reviewed later using

  tracegui

  .

• trace2xml

  is a utility program for dumping a trace database generated by

  tracegui

  or

  traced

  into an XML file which can then be processed by other scripts.

• xml2trace

  performs the reverse operation of

  trace2xml

  : given an XML file, a

  .trace

  file is generated which can be loaded by

  tracegui

  .

• convertdb

  is a helper utility for converting earlier versions of databases with tracelib traces.

Acknowledgements

This project is really just standing on the shoulders of giants. It was made possible by and takes advantage of a few other open source projects:

• The PCRE library is used for matching function names to include in the execution trace via regular expressions.
• For matching functionn names to include in the execution trace via wildcard patterns, the wildcmp function is used.
• Reading the

  tracelib.xml

  file is done using the TinyXML parser.
• Stack traces logged for selected trace points are generated by StackWalker.
• Efficient storage of execution traces is implemented using SQLite.