This code generates a segfault. Where have I gone wrong?

Question

The goal of this code is to quickly read some data into memory from fields in a tab delimited file and sort them. I have found that when I run this code, I get a segmentation fault. I assume it's something to do with my limited knowledge of strtok. I know it would be easier to use some c++ functions for tokenizing strings, however, I would like to have this code run as fast as possible. It seems like most c++ code would have me unnecessarily allocating space for new objects. Ideally, the code will be run on files containing 100's of millions of lines. So, it needs to be fast.

    #include <stdlib.h>
    #include <stdio.h>

    #include <string>
    #include <iostream>
    #include <vector>
    #include <algorithm>

    using namespace std;

    class Node
    {
      public:
      string name;
      int position1;
      int position2;
      string desc;
      float value;

      bool operator<(const Node& T) const;
    };

    bool Node::operator<(const Node &T) const
    {
      int result;
      result = name.compare(T.name);
      if (result !=0) return(result);

      if (position1 != T.position1) return(position1 < T.position1);

      if (position2 != T.position2) return(position2 < T.position2);

      return(false);

    }

    class NodeList
    {
      public:
      vector<Node> nodes;
    };


    int main(void)
    {
      string filename = "table.txt";
      FILE* infile = fopen(filename.c_str(), "r");

      int buflen = 1000;
      char buffer[buflen];

      NodeList K;
      Node T;


      while(fgets(buffer,buflen,infile) != NULL)
      {
         cout<< buffer << endl;

         T.name      = string(strtok(buffer, "\t\n"));
         T.position1 = atoi  (strtok(NULL  , "\t\n"));
         T.position2 = atoi  (strtok(NULL  , "\t\n"));
         T.desc      = string(strtok(NULL  , "\t\n"));
         T.value = atof  (strtok(NULL  , "\t\n"));

         K.nodes.push_back(T);
      }

      sort(K.nodes.begin(),K.nodes.end());

      return(0);

}

EDIT: The segfault occurs in the sort command. Without the sort command the code runs normally. Edited to take comments into account. Here is the output from the debugger:

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0xffffffffffffffe8 0x00007fff83a078bb in std::string::compare ()
(gdb) bt
#0  0x00007fff83a078bb in std::string::compare ()
#1  0x0000000100001333 in Node::operator< (this=0x7fff5fbfeef0, T=@0x1001fffe0) at test.cpp:27
#2  0x000000010000274e in std::__unguarded_linear_insert<__gnu_cxx::__normal_iterator<Node*, std::vector<Node, std::allocator<Node> > >, Node> (__last={_M_current = 0x100200000}, __val=@0x7fff5fbfeef0) at stl_algo.h:2309
#3  0x0000000100003f28 in std::__unguarded_insertion_sort<__gnu_cxx::__normal_iterator<Node*, std::vector<Node, std::allocator<Node> > > > (__first={_M_current = 0x100200200}, __last={_M_current = 0x1002581e0}) at stl_algo.h:2406
#4  0x000000010000437b in std::__final_insertion_sort<__gnu_cxx::__normal_iterator<Node*, std::vector<Node, std::allocator<Node> > > > (__first={_M_current = 0x100200000}, __last={_M_current = 0x1002581e0}) at stl_algo.h:2439
#5  0x0000000100004422 in std::sort<__gnu_cxx::__normal_iterator<Node*, std::vector<Node, std::allocator<Node> > > > (__first={_M_current = 0x100200000}, __last={_M_current = 0x1002581e0}) at stl_algo.h:2831
#6  0x00000001000019e8 in main () at test.cpp:76

If I go up one level and look at the values, I get this:

(gdb) print T
$1 = (const Node &) @0x1001fffe0: {
  name = {
    _M_dataplus = {
      <std::allocator<char>> = {
        <__gnu_cxx::new_allocator<char>> = {<No data fields>}, <No data fields>},
      members of std::basic_string<char,std::char_traits<char>,std::allocator<char> >::_Alloc_hider:
      _M_p = 0x0
    }
  },
  position1 = 0,
  position2 = 0,
  desc = {
    _M_dataplus = {
      <std::allocator<char>> = {
        <__gnu_cxx::new_allocator<char>> = {<No data fields>}, <No data fields>},
      members of std::basic_string<char,std::char_traits<char>,std::allocator<char> >::_Alloc_hider:
      _M_p = 0x0
    }
  },
  value = 0
}

The values for this.name etc look like they come from the file, but whatever it is being compared to has values that are all 0's or NULL.

Answer 1

Compiling with g++ -Wall -g , I see that you need to include string.h to get strtok , and your operator< needs to return something if none of the earlier if statements were true. After that...

1. You aren't checking the return value of fopen , so the first segfault I found was when I hadn't created a table.txt to test with.

2. You aren't checking the return value of strtok either, so if no matching column exists, then you can pass NULL to atoi , and get a segfault there.

You need to use gdb 's bt command when your program crashes to find out what line triggered the crash.