CODEWITHSUNDEEP
cstrict-aliasingrestrict
NeilB
NeilB

Reputation: 357

Understanding restrict keyword

To understand strict-aliasing and the usage of restrict keyword, I am trying the program shown below. In this case, same memory location is referenced by two pointers. Hence without user explicitly telling, compiler can't do any optimization (i.e. without using restrict keyword). Hence my understanding was that without restrict keyword, output of the program will be 40 and with 'restrict' output will be 30 (as 'a' does not need to be read from memory after "*b += *a"). However even with restrict, output is 40. why the optimization is not happening?

I am following "What is the strict aliasing rule?" for understanding. 

#include <stdio.h>

void merge_two_ints(int * restrict a, int * restrict b) {
  *b += *a;
  *a += *b;
}

int
main(void)
{
    int x = 10;
    int *a = &x;
    int *b = &x;
    merge_two_ints(a, b);
    printf("%d\n", x);
    return 0;
}

bash-3.2$ gcc -Wall -O3 -fstrict-aliasing -std=c99 strict_alias2.c

bash-3.2$ ./a.out 40

Upvotes: 1

Views: 549

Answers (1)

John Bollinger
John Bollinger

Reputation: 180978

You observe:

In this case, same memory location is referenced by two pointers. [...] my understanding was that without restrict keyword, output of the program will be 40 and with 'restrict' output will be 30 (as 'a' does not need to be read from memory after "*b += *a").

But your expectation is directly contrary to the C standard, which specifies:

The intended use of the restrict qualifier [...] is to promote optimization, and deleting all instances of the qualifier from all preprocessing translation units composing a conforming program does not change its meaning (i.e., observable behavior).

(C2011 6.7.3/8; emphasis added)

Furthermore, 6.7.3/8 also says,

An object that is accessed through a restrict-qualified pointer has a special association with that pointer. This association, defined in 6.7.3.1 below, requires that all accesses to that object use, directly or indirectly, the value of that particular pointer.

and without going into the details, 6.7.3.1 explicitly specifies that when its provisions are not satisfied by a given program, the behavior is undefined. Your program triggers this provision, and you are anticipating a particular manifestation of the resulting undefined behavior. Nothing in the C language justifies that.

However even with restrict, output is 40. why the optimization is not happening?

The C language never requires either optimization in general or any specific optimization to be performed.

Upvotes: 5

Related Questions