while_immutable_condition false positive with mut pointers

[Deewiant]

Code like the following triggers clippy::while_immutable_condition:

static mut GLOBAL: *mut i8 = 0 as *mut _;

fn main() {
    let mut x = 0;
    let p: *mut i8 = &mut x;
    unsafe {
        GLOBAL = p;
        while *p == 0 {
            *GLOBAL += 1;
        }
    }
}

The exact error message is:

error: Variable in the condition are not mutated in the loop body. This either leads to an infinite or to a never
running loop.
 --> src/main.rs:8:15
  |
8 |         while *p > 0 {
  |               ^^^^^^
  |
  = note: #[deny(clippy::while_immutable_condition)] on by default
  = help: for further information visit https://rust-lang-nursery.github.io/rust-clippy/master/index.html#while_immutable_condition

There are clearly two mut pointers involved above, but it's also possible for this to happen with e.g. external functions:

extern {
    fn stash_the_pointer(x: *mut i8);
    fn use_the_pointer();
}

fn main() {
    let mut x = 0;
    let p: *mut i8 = &mut x;
    unsafe {
        stash_the_pointer(p);
        while *p == 0 {
            use_the_pointer();
        }
    }
}

Clippy complains also about the above, although in reality, stash_the_pointer could've saved the pointer somewhere, allowing use_the_pointer to update it.

$ cargo clippy --version
clippy 0.0.212 (2e26fdc 2018-11-22)

Having the lint disregard mut pointers entirely would be a simple way of papering over this, but a more accurate fix might be something along the lines of:

• If the while condition is based on a mut pointer;
• And the pointer has been assigned to a static global or passed to a function;
• And there is a use of a static global or a function call in the loop;
• Then the lint shouldn't fire.

But maybe it's not that simple, I didn't really think about it very hard.