for_loops_over_fallibles lint with macros give invalid suggestion

[theemathas]

Code

src/main.rs

fn main() {
    let _name = Some(1);
    dep::do_loop!(_name);
}

dep/src/lib.rs

#[macro_export]
macro_rules! identity {
    ($x:ident) => {
        $x
    };
}

#[macro_export]
macro_rules! do_loop {
    ($x:ident) => {
        for $crate::identity!($x) in $x {}
    };
}

Zip of all files needed for reproduction, for your convenience: repro.zip

Current output

warning: for loop over an `Option`. This is more readably written as an `if let` statement
 --> src\main.rs:3:5
  |
3 |     dep::do_loop!(_name);
  |     ^^^^^^^^^^^^^^^^^^^^
  |
  = note: `#[warn(for_loops_over_fallibles)]` on by default
help: to check pattern in a loop use `while let`
 --> C:\Users\theem\Downloads\repro\dep\src\lib.rs:4:9
  |
4 ~         while let Some(for $crate::identity!($x) in $x {}
5 ~         ) =
  |
help: consider using `if let` to clear intent
 --> C:\Users\theem\Downloads\repro\dep\src\lib.rs:4:9
  |
4 ~         if let Some(for $crate::identity!($x) in $x {}
5 ~         ) =
  |

warning: `repro` (bin "repro") generated 1 warning

Desired output

Rationale and extra context

No response

Other cases

Rust Version

rustc 1.90.0 (1159e78c4 2025-09-14)
binary: rustc
commit-hash: 1159e78c4747b02ef996e55082b704c09b970588
commit-date: 2025-09-14
host: x86_64-pc-windows-msvc
release: 1.90.0
LLVM version: 20.1.8

Anything else?

Discovered in #147973 as part of a crater run, in which rust gave the invalid suggestion in 1.90.0, but ICE'ed in beta.

[zachs18]

Bisecting with cargo bisect-rustc --preserve --start 1.87.0 (and a #[deny(for_loops_over_fallibles)]) gives:

searched toolchains nightly-2025-03-29 through nightly-2025-10-28


********************************************************************************
Regression in nightly-2025-05-10
********************************************************************************

fetching https://static.rust-lang.org/dist/2025-05-09/channel-rust-nightly-git-commit-hash.txt
nightly manifest 2025-05-09: 40 B / 40 B [=====================================================================================================================] 100.00 % 1.64 MB/s converted 2025-05-09 to 50aa04180709189a03dde5fd1c05751b2625ed37
fetching https://static.rust-lang.org/dist/2025-05-10/channel-rust-nightly-git-commit-hash.txt
nightly manifest 2025-05-10: 40 B / 40 B [=====================================================================================================================] 100.00 % 1.92 MB/s converted 2025-05-10 to dcecb99176edf2eec51613730937d21cdd5c8f6e
looking for regression commit between 2025-05-09 and 2025-05-10
fetching (via remote github) commits from max(50aa04180709189a03dde5fd1c05751b2625ed37, 2025-05-07) to dcecb99176edf2eec51613730937d21cdd5c8f6e
ending github query because we found starting sha: 50aa04180709189a03dde5fd1c05751b2625ed37
get_commits_between returning commits, len: 6
  commit[0] 2025-05-08: Auto merge of #140786 - Kobzol:try-builds-no-deny-warnings, r=jieyouxu
  commit[1] 2025-05-08: Auto merge of #140818 - matthiaskrgr:rollup-5eaotr2, r=matthiaskrgr
  commit[2] 2025-05-09: Auto merge of #140176 - dpaoliello:arm64ecdec, r=wesleywiser
  commit[3] 2025-05-09: Auto merge of #140252 - nnethercote:rm-Ident-empty, r=oli-obk
  commit[4] 2025-05-09: Auto merge of #140838 - Zalathar:rollup-13hybry, r=Zalathar
  commit[5] 2025-05-09: Auto merge of #140839 - pietroalbini:pa-version-bump, r=pietroalbini

Of those, #140838 includes #140801 "Use span before macro expansion in lint for-loops-over-falibles" which seems related cc @xizheyin .

Reverting 88c1796 seems to make the lint not fire here?

[xizheyin]

Thank you. I believe that to fix both #140747 and #148114, we probably just suppress the help information in this case, as it seems difficult to find a suitable span.🤔 @zachs18