Thread: NULL-pointers

Hey!
I have just malloc'd a chunk of memory for a struct with various members.
Some of those members is pointers.

I want to check if the pointer isnt given a value by doing so:

Code:

if(structPtr->ptr == NULL) ...;

Is it enough to zero-fill the memory, and assume that a zerofilling means the ptr will be NULL?

In other words, is it safe to assume that NULL == 0(or something similar, like (void *) 0)?
Or can NULL be implemented to be -1 or something like that?

Originally Posted by Drogin

Is it enough to zero-fill the memory, and assume that a zerofilling means the ptr will be NULL?

Yes. As you know, you cannot do this:

Code:

typedef struct {
	char *ptr = NULL;
} this;

But this will set eg.ptr to NULL:

Code:

this eg = {0};

Beware King Mir's caveat tho.

Originally Posted by Drogin

Hey!
I have just malloc'd a chunk of memory for a struct with various members.
Some of those members is pointers.

I want to check if the pointer isnt given a value by doing so:

Code:

if(structPtr->ptr == NULL) ...;

Is it enough to zero-fill the memory, and assume that a zerofilling means the ptr will be NULL?

In other words, is it safe to assume that NULL == 0(or something similar, like (void *) 0)?
Or can NULL be implemented to be -1 or something like that?

NULL can theoretically be implemented as something like -1, so using memset or similar functions will technically not guarantee null initialization. However, if you explicitly set the pointers to 0, then it is guaranteed to work, because a 0 in a pointer context will always be the same as NULL.

Code:

So initializing a struct with {0} will always work.

Practically I've never heard of a compiler that does not implement NULL as 0. Checking for 0 is such a common operation even not counting pointers, that architectures tend to make a separate optimized instruction. Checking for -1 is not so common.

The same rule applies to floating point numbers.