Add global_variables lint

[EFanZh]

changelog: [global_variables]: check static variables with interior mutability.

[rustbot]

r? @Jarcho

rustbot has assigned @Jarcho.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[Jarcho]

This looks like it's the same lint as declare_interior_mutable_const but for static items. For consistency this should be named declare_interior_mutable_statics (plural, the other one violates the naming guide) and should share the same implementation.

[EFanZh]

@Jarcho: I have moved the implementation together with declare_interior_mutable_const, but I didn’t change the name because global_variables seems to be easier to understand than declare_interior_mutable_statics. If you still prefer your proposed name, I am OK to change it.

Also, I am not able to exclude statics with interior mutability but with “frozen” initializers. For example: A static using None::<AtomicU32> as the initializer can be excluded. This detection used for consts uses the is_value_unfrozen_poly function, which panics for statics. What should I do from here?

• Ignore this false positive, since using a interior mutable static that isn’t actually mutable does not make much sense, the impact could be minor, or
• Continue trying to figure out how to detect the case with non-mutable value case.

[Jarcho]

@EFanZh Sorry for the delay. The lint pass has been completely rewritten so you will need to rebase. You still won't be able to use the MIR value, but you can now use is_init_expr_freeze to evaluate the HIR expression used to initialize the static.

[github-actions]

Lintcheck changes for b438103

Lint	Added	Removed	Changed
clippy::global_variables	88	0	0

This comment will be updated if you push new changes

[rustbot]

This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.