Dear Expert, I have a simple C program with 'linklist' data structure processing as below. ////////////////////////////////////// #include <stdio.h> #include <stdlib.h> typedef struct node{ unsigned int id; struct node * next; }; void main(void){ struct node * first; struct node * n; unsigned int i; printf("/////////////////\r\n"); first->next = (struct node *) malloc(sizeof(struct node) * 10); if(first->next == NULL)printf("** Not enough memory space\r\n"); printf("first->next %X\r\n",first->next); first = NULL; //no first link for(i=1;i<=10;i++){ n->id = i; n->next = first; first = n; }; while(first != NULL){ printf("n->id = %X , n->next %X\r\n",first->id,first->next); first = first->next; } free(first); } ///////////////////// When I run this program , the result is unterminate loop. But I try to test this program on Turbo C++ when including "malloc.h" , it is work and result is correct! What wrong with this codeing? Then,I wonder does C51 support link list data processing? I will looking forward your reply and thank you for all suggestion. Birdy.
Hi, can you please extend and get into details, about why malloc use is not recommended ?
"Hi, can you please extend and get into details, about why malloc use is not recommended ?" Additional overhead in terms of code size, speed and RAM usage. If none of these things are an issue in your project then by all means use dynamic memory allocation, but I think it is rare to find an 8051 type project that isn't constrained by at least one of those things. Similar arguments apply when considering whether to use an operating system or to just code as god intended (I stole that quote from somewhere). A final point is that dynamic memory allocation pretty much enforces the use of pointers which offer the less experienced an excellent opportunity to introduce lots of hard to find bugs into their code.
"A final point is that dynamic memory allocation pretty much enforces the use of pointers which offer the less experienced an excellent opportunity to introduce lots of hard to find bugs into their code." Especially on the 8051, where pointers are more complicated due to having all those different address spaces... Having to access everything via pointers may also hit efficiency...
"why malloc use is not recommended ?" In addition to what Stefan's already said, think about the issue of fragmentation; ie, when your memory gets left with lots of little allocated areas scattered about...
"why malloc use is not recommended ?" Put it the other way around: for what reason would you actually need to use it? Unlike a PC, where the programmer usually has no idea how much RAM the final user's machine might have available, embedded systems usually have a well-defined, fixed amount of RAM (especially the type of systems appropriate to 8051s). Therefore, you are going to have to calculate the required size of your pool area, and set aside that much RAM at design time. So, if you have to lock-up that RAM at design time anyway, why bother with the overhead of dynamic allocation?! Of course, you might have a good reason to use it in your specific application - but that's why embedded systems usually avoid it.