Rule Definition
Using those functions slow down application. It might be necessary to use them however. In addition it is recommended to centralize their utilization to: - use of well-proven memory management primitives that are easier to control, test, etc., when they are grouped in a well-identified component than when they are distributed all over the source code, - create the opportunity to build a custom memory management layer that will replace all individual small-size system "malloc" by individual small-size custom "malloc" that will rely on a centralized large-size system "malloc"; this would increase performance

Remediation
Use a centralized memory allocation function.

Violation Code Sample

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

#define MY_SIZE 3000000
clock_t start, finish;
struct toto
{ int i, j; };

toto * my_array[MY_SIZE];
toto * first = NULL;

toto * MyMalloc( int i)
{
   if (0 == i )
      first = (toto *)malloc( sizeof( toto) * MY_SIZE);
   return (first+i);
}

int main(int argc, char* argv[])
{
   double duration;
   /* Standard allocation */
   start = clock();
   for (int i = 0; i<MY_SIZE; i++)
   { my_array[i] = (toto *)malloc( sizeof( toto)); }
   finish = clock();
   duration = (double)(finish - start) / CLOCKS_PER_SEC;
   printf( "%2.5f seconds\n", duration );

   /* Optimized allocation */
   start = clock();
   for (i = 0; i<MY_SIZE; i++)
   { my_array[i] = MyMalloc( i);  }
   finish = clock();
   duration = (double)(finish - start) / CLOCKS_PER_SEC;
   printf( "%2.5f seconds\n", duration );
   return 0;
}

Fixed Code Sample

The output of the sample program is the following:
1.69200 seconds
0.21000 seconds