Question

what’s the difference in
[ insert_node(&(tmp->left),value);] VS [ tmp=tmp->right; insert_node(&(tmp),value);]

void insert_node(struct btree **bt,int value)
{
     btree *tmp= *bt, *r ;

     if(*bt==NULL)// first node 
     {
          (*bt)=(struct btree *)malloc(sizeof(btree));
          (*bt)->data=value;
          (*bt)->left=NULL;
          (*bt)->right=NULL;           
     }       
     else
     {
         if(value > tmp->data)
            insert_node(&(tmp->right),value);   
         else
            insert_node(&(tmp->left),value);                 
     }

      #if 0  //start
           OR  /** COMMENT START

         earlier I had below piece of code but not working 
         can any please explain what's the difference in  
         insert_node(&(tmp->left),value); VS [ tmp=tmp->right; insert_node(&(tmp),value);]
         COMMENT END **/ 

      else
     {
         if(value > tmp->data)
            tmp=tmp->right;  
         else
            tmp=tmp->left ;
            insert_node(&tmp,value);                 
     }
   #endif //end

}

Answer 1

In the non-working case, you are giving it the address of your tmp variable, and therefore updating that pointer, which is on your stack and not going to be used for anything.

In the working case, you are giving the address of the actual pointer in your bt node which is what you want.