#include<stdio.h>
void foo(int **arr) {
arr[1][1]++;
}
main() {
int arr[20][20];
printf("%d\n",arr[1][1]);
foo((int**)arr);
printf("%d\n",arr[1][1]);
}
views:
311answers:
6Because foo expect a pointer to a pointer to int and you are passing it a pointer to an array of 20 int. Casting it won't change the fact that it isn't the correct type.
foo
needs to know the array size (well, at least the second array dimension, first isn't needed), otherwise it can't do the necessary pointer arithmetic for the [1][1]
.
Problem is that int arr[20][20]
for 2d array means that this array is stored as 1d array, and lines are stored one after other. when you do indexing to int **arr
you actually take 2nd element from first line of array, then you dereference it and take first element there.
Suppose you declare: int arr[ 10 ][ 20 ] ;
What type is arr?
You may think that it'sint **
, but that's incorrect.Its actually of type
int (*)[20]
when it decays (like when you pass it to a function);
Array decaying applies only once.
Now consider the following,
#include<stdio.h>
#include<stdlib.h>
void foo(int arr[][20]) {
arr[1][1]++;
}
main() {
int (*arr)[20];
arr = malloc(sizeof(int (*)[]) * 2); //2 rows & malloc will do implicit cast.
printf("%d\n",arr[1][1]);
foo(arr);
printf("%d\n",arr[1][1]);
}
Output :
$ gcc fdsf.c && ./a.out
0
1
arr and arr+1 are pointing to array of 20 integers.
arr + 0 --> int int int ... int (20 ints, contiguous)
[0][0] [0][1]
arr + 1 --> int int int ... int (20 ints, contiguous)
[1][0] [1][1]
Here's what an int[2][2]
looks like in memory:
int[2] int[2]
That is, an array immediately followed by another array.
Here's what an int[2]
looks like in memory:
int int
That is, an int immediately followed by another int.
So, here's also what an int[2][2]
looks like in memory:
int int int int
^ ^
| |___ this is arr[1][1]
|
|____ this is p[1], assuming sizeof(int*) == sizeof(int)
If you cast arr
to an int**
, I'm going to call the result p
. Then it points to the same memory. When you do p[1][1]
you don't get arr[1][1]
. What the program does instead is, it reads the value at p[1]
, adjusts that up by the size of an int, and dereferences it. If that second int contained, say, the value "21" then you have just tried to dereference the pointer "25" (if int
is 4 bytes). That ain't right.
Arrays are not the same as pointers, and 2-D arrays are certainly not the same thing as pointers-to-pointers.
If you change it like this, you get the expected result:
#include<stdio.h>
void foo(int arr[][20]) {
arr[1][1]++;
}
int
main() {
int arr[20][20];
arr[1][1] = 1;
printf("%d\n",arr[1][1]);
foo(arr);
printf("%d\n",arr[1][1]);
}