Search code examples
csegmentation-faultdynamic-memory-allocationmatrix-multiplication

C - matrix multiplication segmentation fault


I get a segmentation fault when running this code. Anyone know why? Thanks.

#include <stdio.h>

int main()
{
    double **m1, **m2, **mr;
    int m1_rows, m1_cols, m2_rows, m2_cols, mr_rows, mr_cols;
    int i, j, k;

    printf("Enter number of rows for matrix 1: ");
    scanf("%d", &m1_rows);

    printf("Enter number of columns for matrix 1: ");
    scanf("%d", &m1_cols);

    printf("Enter number of rows for matrix 2: ");
    scanf("%d", &m2_rows);

    printf("Enter number of columns for matrix 2: ");
    scanf("%d", &m2_cols);

    //allocate memory for matrix 1 m1
    m1 = (double **) calloc(m1_rows, sizeof(double *));
    for (i = 0; i < m1_rows; i++) {
        m1[i] = (double *) calloc(m1_cols, sizeof(double));
    }

    //allocate memory for matrix 2 m2
    m2 = (double **) calloc(m2_rows, sizeof(double *));
    for (i = 0; i < m2_rows; i++) {
        m2[i] = (double *) calloc(m2_cols, sizeof(double));
    }

    //allocate memory for sum matrix mr
    mr = (double **) calloc(mr_rows, sizeof(double *));
    for (i = 0; i < mr_rows; i++) {
        mr[i] = (double *) calloc(mr_cols, sizeof(double));
    }

    //assign  mr_rows and mr_cols
    mr_rows = m1_rows;
    mr_cols = m2_cols;

    //initialize product matrix
    for (i = 0; i < m1_rows; i++) {
        for (j = 0; j < m2_cols; j++) {
            mr[i][j] = 0;
        }
    }

    //perform matrix multiplication
    for (i = 0; i < m1_rows; i++) {
        for (j = 0; j < m2_cols; j++) {
            mr[i][j] = 0;
            for (k = 0; k < m1_cols; k++) {
                mr[i][j] += m1[i][k] * m2[k][j];
            }
        }
    }

    //print result
    for (i = 0; i < mr_rows; i++) {
        for (j = 0; j < mr_cols; j++) {
            printf("%f\t", mr[i][j]);
        }
    }

    //free memory m1
    for (i = 0; i < m1_rows; i++); {
        free(m1[i]);
    }
    free(m1);

    //free memory m2
    for (i = 0; i < m2_rows; i++); {
        free(m2[i]);
    }
    free(m2);

    //free memory mr
    for (i = 0; i < mr_rows; i++); {
        free(mr[i]);
    }
    free(mr);

    return 0;
}

I ran using valgrind valgrind --tool=memcheck a.out for more info on the segmentation fault, but the result was over 30000 errors so it didn't print them out.


Solution

  • You are not assigning mr_rows and mr_cols. They need to be set like this:

    mr_rows = m1_rows;
    mr_cols = m2_cols;
    

    This line is no good:

    mr[i][j] += m1[i][k] * m2[k][j];
    

    That will be accessing elements out of bounds, not least because k is not initialized. You need that line of code inside three nested for loops. Indeed, you may as well roll the zeroising code into this too.

    for(i=0; i<m1_rows; i++){
        for(j=0; j<m2_cols; j++){
            mr[i][j] = 0;
            for(k=0; k<m1_cols; k++){
                mr[i][j] += m1[i][k]*m2[k][j];
            }
        }
    }
    

    Also, all your memory freeing loops are wrong. Instead of

    for(i=0; i<m1_rows; i++);{
        free(m1[i]);
    }
    free(m1);
    

    it should read

    for(i=0; i<m1_rows; i++){
        free(m1[i]);
    }
    free(m1);
    

    That stray semicolon was killing you.

    You also need to perform a check that the number of columns in m1 equals the number of rows in m2, i.e. test that m1_cols == m2_rows.

    One final point. You duplicate your code horribly here. Don't have three identical for loops to allocate a matrix and three identical for loops to deallocate. Extract those operations into helper functions which can be called from main.

    That's all that I can find!