Programming

Hello All :
I write a .c program to test the exactually resource the memory leak
as follows：

Code:

  1 #include <stdio.h>
  2 #define NUM  100000 
  3 void *Thread_Run(void * arg){
  4     //TODO
  5     //pthread_datch(pthread_self());
  6     int socket= (int)arg;
  7     close(socket);
  8     return NULL;
  9 }
 10 int main(void){
 11     
 12     int i,j;   
 13     pthread_t  pid[NUM];  
 14     for( i = 0; i < NUM ; i++){
 15         j = 0;
 16         j = socket(AF_INET, SOCK_DGRAM, 0);
 17         pthread_create(&pid[i],NULL,Thread_Run,(void *)j);
 18     
 19     }
 20     while(1){
 21         
 22         ;
 23     }
 24     return 0 ;
 25 }

I create NUM threads and let them not detached .so their must memory leak , the question is how I know exactly what resource the memory leaks ?

Best regard

aobai

---------- Post updated at 09:38 PM ---------- Previous update was at 07:04 AM ----------

Maybe it's nothing to do with memory , I let the created threads joinable , and return implicitly (return NULL) ,but don't use pthread_join() to release the exited thread ID , if I create too many threads, It's must beyond the nums of system allowed

consider this :

first: creating a lot of threads does not mean necessarily that a lot of memory will be consumed ;

(that's because the amount of memory used just to create and manage a thread is rather small)

-- then you would have to make your thread function "use" (really use) some - or a lot of - memory in order to actually see a system memory shortage ;

second: by the code you provided, I suspect you will run out of file descriptors way much sooner than any memory shortage can be perceived ;

and last, but not least: creating a very large number of threads may result in an overall system response time degradation, due to the scheduling of so many threads, depending on the thread support implementation on your host ;

(that's because "N:1" or "pure user mode" threads would not harm your system's performance, for the operating system would not perceive that many threads as they are all in userland, as well as the thread scheduling ;
but if, on the other hand, your thread implementation is "1:1" or "pure kernel threads", then this code of yours is sure to slow all of your system down to a near halt - unless you have a plethora of cores and a really fast clock)

good luck, and success !
alexandre botao
<< botao {dot} org >>