needless_return: do not lint on proc macros

[blyxyas]

Due to the complexity of the issue and the simplicity of the fix, I'm not being able to reproduce it in the tests (I'm not even being able to reproduce to issue to begin with), but testing it on upstream rust, the panic no longer occurs.

fixes #15338

changelog:[needless_return]: Do not run in proc macros.

[rustbot]

r? @Alexendoo

rustbot has assigned @Alexendoo.
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

[samueltardieu]

@blyxyas I can't reproduce the problem on the current rustc's master branch while running the full ./x clippy, even without your fix. Can you?

[rustbot]

☔ The latest upstream changes (possibly 1d0582a) made this pull request unmergeable. Please resolve the merge conflicts.

[y21]

In #13464 I did pretty much the same change as this PR, adding is_from_proc_macro to that function. It's a bit further down in a match arm, but as far as I can tell still on the same codepath to the emit_return_lint function where it supposedly ICEs.

So as far as I can tell, this change should have no effect, given that we now check from a proc macro twice? Or am I overlooking something?

[blyxyas]

Either the upstream rustc is fixed, or Clippy is fixed. If we don’t have a minimal repro and the suspect was fixed, I think that we can fix the upstream issue as well.