215. next_multiple_of — Round Up to a Multiple Without the +m-1 Dance

Padding a length up to the next multiple of 8? The classic (n + 7) / 8 * 8 works right up until it overflows or you fat-finger the constant. next_multiple_of says exactly what you mean.

The hand-rolled version shows up everywhere alignment matters — buffer sizes, page rounding, table padding:

1
2
3
4

// Easy to get subtly wrong, and `n + m - 1` can overflow near the top of the range
fn round_up(n: usize, m: usize) -> usize {
    (n + m - 1) / m * m
}

Every integer type has next_multiple_of: the smallest value >= self that is a multiple of the argument. Already a multiple? It’s returned unchanged.

1
2
3
4

assert_eq!(13u32.next_multiple_of(8), 16); // round up
assert_eq!(16u32.next_multiple_of(8), 16); // already aligned, untouched
assert_eq!(0u32.next_multiple_of(8), 0);
assert_eq!(23u32.next_multiple_of(10), 30);

When the rounded value would overflow, next_multiple_of panics — but checked_next_multiple_of hands you an Option instead, so you stay in control:

1
2

assert_eq!(250u8.checked_next_multiple_of(8), None); // 256 won't fit in u8
assert_eq!(13u8.checked_next_multiple_of(8), Some(16));

One method, the intent on the page, and no off-by-one waiting to bite you.