Why GDB always jumps in and jumps out and seems uncontrollable when debug c++?

Why GDB always jumps in and jumps out and seems uncontrollable when debug c++?

[Erming Pei]

Hi,
Sorry for bothering.
Could anyone help me with a problem of debugging with GDB on LINUX?
If you have no time dealing with this problem, would you recommand some 
webpage or some mail addresses to me? Thanks very much in advance.

When I debug a program made by c and c++ code using gdb, it always 
enters into the inner side of standard funtions and even their libraries;
That didnt happened when I only debug program that only made by c. (not 
c/c++)
e.g.
#1 the following is a part of my program.
....
118 char *homeDir;
119 if ( (homeDir=getenv("ATLPROD_ROOT_DIR"))==NULL) ////
120   {
121     *messages << MSG::ERROR << "ATLPROD_ROOT_DIR not set !" << endreq;
122     exit(-1);
123   }
124                                                                   // 
remove lock files
125 string pipo(homeDir);
126 string com="rm "+pipo+"/jobs/lock*";
127 system(com.c_str());
128                                                              
129 asite.connexionToServer("localhost",const_cast<char 
*>((theParameters->getPort()).c_str()));
130                                                           
131 int SERV=asite.getDist();

#2 the scence of debugging:
#from line 119,  when input  "next" command, it entered into 
/usr/include/c++/3.2.3/bits/char_traits.h. but I dont want to see that, 
I only want to see the line 120!!
......
119     if ( (homeDir=getenv("ATLPROD_ROOT_DIR"))==NULL) //<---main.cpp
(gdb)n
664       allocator() throw() {}         //<---Why this?
(gdb)n
956       {
(gdb)n
957       basic_string<_CharT, _Traits, _Alloc> __str(__lhs);
(gdb)n
135       { return strlen(__s); }
(gdb)n
957       basic_string<_CharT, _Traits, _Alloc> __str(__lhs);
(gdb)n
135       { return strlen(__s); }
(gdb) bt
#0 main (argc=1, argv=0xbfff9b80)
 at /usr/include/c++/3.2.3/bits/char_traits.h:135
(gdb) n
229       { return _M_dataplus._M_p; }
(gdb)n
668       ~allocator() throw() {}
(gdb)n
229       { return _M_dataplus._M_p; }
(gdb)n
665       allocator(const allocator&) throw() {}
(gdb)n
38     {
(gdb)n
40     __asm__ __volatile__ ("lock; xaddl %0,%2"
(gdb)n
38     {
(gdb)n
229       { return _M_dataplus._M_p; }
(gdb)n
119       { __c1 = __c2; }
(gdb)n
781       {
(gdb)n
Detaching after fork from child process 13210.
229       { return _M_dataplus._M_p; }
(gdb)n
119       { __c1 = __c2; }
(gdb)n
781       {
(gdb)n
229       { return _M_dataplus._M_p; }
(gdb)n
668       ~allocator() throw() {}
(gdb)n
229       { return _M_dataplus._M_p; }
(gdb)n
38     {
(gdb)n
40     __asm__ __volatile__ ("lock; xaddl %0,%2"
(gdb)n
38     {
(gdb)n
131     int SERV=asite.getDist(); //<---- It jumped to line 131 of 
main.cpp when input "next" continously. it skiped the midlines(119-130)
(gdb)
133     string temp = theParameters->getPassword();

It always happens when I try to debug the program.  it jumps in and out. 
Furthermore, when I want to stop the program at a line e.g. 
Core.cpp:437. Gdb is not able to stop at this line.
But it can execute this line when I run the program normally. It's so 
strange.
Besides, I've set the step-mode off. And before debug, I set CFLAGS=-g 
-O2 CXXFLAGS= -g -O2 CC = gcc  CPP = gcc -E  CXX = g++ CPPFLAGS = -Wall 
-g -Wno-deprecated.

Is that any relations with STL or multi-threads or multi-processes or 
else??
The gdb version is: GNU gdb Red Hat Linux (6.3.0.0-0.30.1rh)

Any hint from you would be greatly appreciated.

Best Regrads,
Erming.

Re: Why GDB always jumps in and jumps out and seems uncontrollable when debug c++?

[Joel Brobecker]

> Besides, I've set the step-mode off. And before debug, I set CFLAGS=-g 
> -O2 CXXFLAGS= -g -O2 CC = gcc  CPP = gcc -E  CXX = g++ CPPFLAGS = -Wall 
> -g -Wno-deprecated.

I think that it will be easier to debug if you remove the -O2.
What you are seeing in the jumping around is compiler optimizing
your code and doing what I think compiler experts call instruction
scheduling.

-- 
Joel