exec: avoid possible undefined behavior in count()

The tricky problem is this check:

	if (i++ >= max)

icc (mis)optimizes this check as:

	if (++i > max)

The check now becomes a no-op since max is MAX_ARG_STRINGS (0x7FFFFFFF).

This is "allowed" by the C standard, assuming i++ never overflows, because signed integer overflow is undefined behavior. This optimization effectively reverts the previous commit 362e6663ef ("exec.c, compat.c:
fix count(), compat_count() bounds checking") that tries to fix the check.

This patch simply moves ++ after the check.

---
Not sure how many people are using Intel's icc to compiled the kernel. Some projects like LinuxDNA did.

The kernel uses gcc's -fno-strict-overflow to disable this optimization. icc probably doesn't recognize the option.

To illustrate the problem, try this simple program:

int count(int i, int max)

{
        if (i++ >= max) {
                __builtin_trap();
                return -1;
        }
        return i;
}

#include <stdio.h>
#include <stdlib.h>

int main(int argc, char **argv)

{
        int x = atoi(argv[1]);
        int max = atoi(argv[2]);
        printf("%d %d %d\n", x, max, count(x, max));
}

$ gcc -O2 t.c
$ ./a.out 2147483647 2147483647
Illegal instruction (core dumped)

$ icc -O2 t.c
$ ./a.out 2147483647 2147483647
2147483647 2147483647 -2147483648

There's no difference whether we add -fno-strict-overflow or not.

---
 fs/exec.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

index 18c45ca..20df02c 100644
@@ -434,8 +434,9 @@ static int count(struct user_arg_ptr argv, int max)
 			if (IS_ERR(p))
 				return -EFAULT;
 
-			if (i++ >= max)
+			if (i >= max)
 				return -E2BIG;
+			++i;
 
 			if (fatal_signal_pending(current))
 				return -ERESTARTNOHAND;

1.7.10.4

Thanks!

The icc developers have confirmed this bug and filed a defect ticket. I'll let you know if there's any further update.

- xi