Lint masking bits and shift

[tesuji]

What it does

Consider this code snippet:

pub const fn to_4_be_nibles(op: u16) -> [u8; 4] {
    [
        (op & 0xF000 >> 12) as u8,
        (op & 0x0F00 >> 8) as u8,
        (op & 0x00F0 >> 4) as u8,
        (op & 0x000F) as u8,
    ]
}

We'd realize that the code is wrong. >> has stronger precedence than & operator.
So we has to use () to group (op & 0F000) and the like.
I'd like to have a lint about this.

Categories (optional)

• Kind: clippy::correctness

Drawbacks

None.

[llogiq]

This is not a correctness lint. It should not even lint like this, because shifting the mask may well be the intended outcome of the code.

I'm still somewhat sympathetic to introducing brackets here, and the lint should point out the current meaning and ask if that was really intended.

We actually have the precedence lint which could be extended to cover this case.

[liquid-dragons]

@rustbot claim

[Natsume-Neko]

@rustbot claim