Ufdjrw

I have this code but it has an error: Segmentation Fault(core dumped) and it doesn't work with more the 2 threads. Any idea of what am i doing wrong?

This code is for calculate pi by Leibniz formula

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


#define NUM_HILOS 2

struct datos

 int inicio;
 int fin;
 float *pi;


*calcPi (void *datos)
 struct datos *datos_proceso;
 datos_proceso = (struct datos *) datos;
 int i = datos_proceso -> inicio;
 int end = datos_proceso -> fin;
 printf("inicio %d n", i);
 printf("fin %d n", end); 
 float *pi = datos_proceso -> pi;
 int signo = 1;
 do 
 *pi = *pi +(signo*4.0)/((2*i)+1);
 i++;
 signo *= -1;
 //printf("%f n", *pi);
 while(i<end);



int main()

 int error, i;
 float *pi;
 int j = -1;
 /*variable para hilos*/

I think that the error is over here but i don't know how to fix it

 struct datos hilo_datos[NUM_HILOS];
 pthread_t idhilo[NUM_HILOS];
 //printf("este es pi %f n", *pi);
 for(i=0; i<NUM_HILOS; i++)
 
 hilo_datos[i].inicio =j+1;
 hilo_datos[i].fin =j+1000;
 hilo_datos[i].pi = pi;
 printf("%d n", hilo_datos[i].inicio);
 printf("%d n", hilo_datos[i].fin);
 j += 1000;
 

 for(i=0; i<NUM_HILOS; i++)
 
 error=pthread_create(&idhilo[i], NULL, (void *)calcPi, &hilo_datos[i]);


 

 for(i=0; i<NUM_HILOS; i++)
 
 pthread_join(idhilo[i], NULL);
 
 printf("este es pi %f n", *pi); 
return 0;

Your errors were mainly forgetting simple things like variable initialization. When a pointer such as float *pi; is accessed before being initialized it will almost always cause problems. At the very least it should have raised a compiler warning. By the way, turn on all your compiler warnings. GCC options

Here are a few specifics to get a clean build...

1 add return statement to calcPi function

 ...
 return 0;
}

2 terminate struct datos with a ;

struct datos 
...
;
 ^

3 function:

* calcPi (void *datos)...

should be:

void * calcPi (void *datos)...

Or better:

void calcPi (struct datos *d)... //passing a struct pointer 

4 Initialize your variables before using them. for example:

float *pi; //uninitialized
float *pi = NULL;//initialized pointer
pi = malloc(sizeof(float)*NUM_HILOS);//array of pi with NUM_HILOS elements

then, in following assignment statements use pi[i], ...

hilo_datos[i].pi = pi[i];

Or, just create a simple float: (you do not need a pointer in this case)

float pi = 0;//works just fine for what you are doing
 //no further initialization is needed

Other problems include mis-application of threads, creating inappropriate variable types (i.e. for the way you are using it, float *pi; could just be float pi; negating the need for malloc())

A very simple example of using this algorithm to compute pi (without threading) is included below for illustration:

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

#define NUM_HILOS 10 //array elements (no threads)

struct datos //modified for use as array, no threads

 float pi; //PI
;

void calcPi (struct datos *d)

 struct datos *datos_proceso = d;
 float pi = datos_proceso[0].pi;
 int signo = 1;
 int i;

 for(i=0;i<NUM_HILOS;i++)
 
 pi = pi + (signo*4.0)/((2*i)+1);
 signo *= -1;
 d[i].pi = pi; //change values for NUM_HILOS to see how 
 //values for pi converge here. 
 


int main()

 int error, i;
 float pi = 0;
 int j = -1;

 //your comment: I think that the error... 
 //... (The error you were seeing here was
 //caused by an uninitialized float *pi; which has been
 //changed to float pi = 0; in this example)

 struct datos hilo_datos[NUM_HILOS];
 for(i=0; i<NUM_HILOS; i++)
 
 hilo_datos[i].pi = 0; //initialize all to zero
 

 calcPi(hilo_datos);//send array of struct, check all elelments when returned

 for(i=0; i<NUM_HILOS; i++)
 
 printf("este es pi %f n", hilo_datos[i].pi); 
 
 getchar();
return 0;