Set last `n` bits in unsigned int

If you meant the least-significant n bits:

((uint32_t)1 << n) - 1

On most architectures, this won't work if n is 32, so you may have to make a special case for that:

n == 32 ? 0xffffffff : (1 << n) - 1

On a 64-bit architecture, a (probably) faster solution is to cast up then down:

(uint32_t)(((uint64_t)1 << n) - 1)

In fact, this might even be faster on a 32-bit architecture since it avoids branching.

Here's a method that doesn't require any arithmetic:

~(~0u << n)