Dealing with Segmentation Faults

What is a Segmentation Fault

Question: I’m getting a Segmentation Fault when I run my program. What does this mean and how do I deal with it?

A Segmentation Fault is a runtime error that occurs when you access memory that you do not own. The following situations cause “Seg-faults”

We make a pointer/array but forget to malloc

float *A; // make a pointer 
A[3] = 4;

We go outside the bounds of an array

int n = 4;
float *A = malloc(n * sizeof(float));
A[n] = 10; // we own the bytes up to A[n-1] but not A[n]

We make a pointer but forget to allocate space

Point p; // forget "= mkPoint(0, 0);"
p->x = 3;

We malloc space for a struct but don’t malloc space for the pointers that it contains

Array A = malloc(sizeof(struct array));
A->data[3] = 5; // data is a float * that has not yet been malloced

Finding a Segmentation Fault

Question: I want to find the line where this happens but this error doesn’t have a line number. What’s going on?

Seg-faults are “runtime errors” rather than “compile-time errors.” This means that the compiler gcc isn’t smart enough to find them for us and we need to run the program before we run into the problem. gcc was the program that was helping us with line numbers and cryptic error messages before. Now we really appreciate how useful the compiler is.

Here are some methods for tracking down a Seg-fault.

Using gdb

First compile your program with the -g flag. I.e.

gcc mycode.c -o mycode.exe -g

Then run the gdb program with your executable as an input

gdb mycode.exe

And then in the gdb prompt type in run and press enter. Your program should run and stop at the line that causes the segfault. You can print the value of any active variable in your program by typing

print variable_name

And you can what function called this one by typing up