Dereferencing a pointer in a 'for' loop initializer creates a segmentation fault

Take a close look here:

for (int i=3, (*p)=0 ; i>=0; i--)

In the first part of the for you're defining a new pointer variable named p which shadows the p defined earlier and initializing it to NULL. You then dereference the NULL pointer in the loop which causes the segfault.

You can't have both a variable definition and an assignment to an existing variable together like that, so move the assignment of *p to before the loop:

*p = 0;
for (int i=3; i>=0; i--)

Or you can define i outside of the loop:

int i;
for (i=3, (*p)=0 ; i>=0; i--)

You could squeeze these together by abusing the comma operator:

for (int i=(*p=0,3) ; i>=0; i--)

Here the assignment to p happens as part of the initializer for i so it doesn't declare a new variable. But I wouldn't recommend this as it would make your code more difficult to read and understand.

Tip: Use -Wshadow to get a warning when a variable shadows another one.

[] $ gcc main.c -Wshadow
main.c: In function ‘main’:
main.c:13:21: warning: declaration of ‘p’ shadows a previous local [-Wshadow]
   13 |     for (int i=3, (*p)=0 ; i>=0; i--) // Here *p is a random number
      |                     ^
main.c:6:10: note: shadowed declaration is here
    6 |     int *p = NULL;
      |          ^

https://coliru.stacked-crooked.com/a/5de37f53cf0b094d

You are declaring a completely new variable called p:

for (int i=3, (*p)=0 ; i>=0; i--)

This is the same as:

for (int i=3, *p=0 ; i>=0; i--)

So you are creating an int i and an int *p, which points to address 0. This is not the same p as the one defined previously. It just shadows it. So when you dereference it, you get the segfault.