Make weird name lints trigger behind cfg_attr

[est31]

The weird name lints (unknown_lints, renamed_and_removed_lints), the lints that lint the linting, were previously not firing for lint level declarations behind cfg_attr, as they were only running before expansion.

Now, this will give a unknown_lints warning:

#[cfg_attr(all(), allow(this_lint_does_not_exist))]
fn foo() {}

Lint level declarations behind a cfg_attr whose condition is not applying are still ignored. So this still won't give a warning:

#[cfg_attr(any(), allow(this_lint_does_not_exist))]
fn foo() {}

Furthermore, this PR also makes the weird name lints respect level delcarations for them that were hidden by cfg_attr, making them consistent to other lints. So this will now not issue a warning:

#[cfg_attr(all(), allow(unknown_lints))]
mod foo {
    #[allow(does_not_exist)]
    fn foo() {
    }
}

Fixes #97094

[rust-highfive]

r? @lcnr

(rust-highfive has picked a reviewer for you, use r? to override)

[lcnr]

Do we want that, considering that in the future one might write #[cfg_attr(version = "1.70", allow(some_new_lint))]? Or am I misunderstanding something here

[est31]

@lcnr it will only lint if the condition inside the cfg evaluates to true. If the condition is false, it won't lint. See the last entry in the added test. I've also edited the PR description to make this clearer. Making it peek inside all cfg_attr regardless of whether they apply or not is a possibility, but I agree that it's not a good idea, which is why I didn't chose it for this PR.

[est31]

I'm now using i == 0 because it still works, and have changed the test to ensure that it works with that debug flag enabled. I've also reordered the bodies as requested. re-r? @lcnr

[lcnr]

one nit, then r=me

[est31]

@lcnr your understanding is correct, yes. To make sure, I've removed the == 0 check locally and tested it, and indeed it emitted the warnings multiple times, so the == 0 is needed. I've added your suggestion to the PR.

[est31]

re-r? @lcnr

[lcnr]

@bors r+

[bors]

📌 Commit 7a36d24d93d00569c7907228f96d259f38a6f571 has been approved by lcnr

[est31]

Sorry for pinging you again, @lcnr I've improved the last two tests a little to not have a cfg_attr inside the module but only a cfg. Now they read #[allow(nonex_lint_fn)]. I think this improves test readability because the thing being tested in those two last tests is the cfg_attr of the module outside. re-r? @lcnr

[lcnr]

@bors r+

[bors]

📌 Commit 2a8b60f has been approved by lcnr