fix undocumented-unsafe-blocks false positive

[llogiq]

This fixes #9142 by iterating over the parent nodes as long as within a block, expression, statement, local, const or static.

---

changelog: none

[rust-highfive]

r? @giraffate

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

[llogiq]

I now no longer look at blocks.

This would of course still allow e.g. foo(bar(Baz(unsafe { .. }))) including possible newlines in between and accept the safety comment before foo.

[llogiq]

Ping @giraffate

Or @Jarcho do you want to take over the review?

[Jarcho]

I guess I can take over.

I'm still not sure about walking up expressions in general. Method and let chains can result in some pretty crazy expressions. Clippy in particular is pretty bad about having 40-50 line let chains, although I assume that's unusual in other projects.

I think the best plan here would be to fix the current regression by only walking up a single level if the parent is a let statement (and in the same SyntaxContext) and checking there. We can then figure out what forms should be allowed by the lint and deal with that in a different PR.

[llogiq]

@Jarcho Ok, this now only looks one parent upwards. r?

[rustbot]

Error: Parsing assign command in comment failed: ...'' | error: specify user to assign to at >| ''...

Please file an issue on GitHub at triagebot if there's a problem with this bot, or reach out on #t-infra on Zulip.

[llogiq]

That comment was meant for Jarcho, not Rustbot. But nice error 😉.

[Jarcho]

Can you change this to be on two lines. e.g.

// SAFETY: ok
let _ =
    unsafe {}

[llogiq]

Ah, rustfmt pulled this together.

[Jarcho]

All looks good now. Thank you.

@bors r+

[bors]

📌 Commit e19fe89 has been approved by Jarcho

It is now in the queue for this repository.

[bors]

⌛ Testing commit e19fe89 with merge fdaa425...

[bors]

☀️ Test successful - checks-action_dev_test, checks-action_remark_test, checks-action_test
Approved by: Jarcho
Pushing fdaa425 to master...