`rustfmt` wanted features & bugfixes

[ojeda]

Features that we would like to see

Required (we almost certainly want them)

• Stabilize the style rustfmt generates when using a trailing empty comment, but without having to use the comment.
  • Tracking Issue: Keep used items on individual lines when requested rust-lang/rustfmt#6829.
  • Docs: https://docs.kernel.org/rust/coding-guidelines.html#imports.
  • Lore: https://lore.kernel.org/rust-for-linux/20251010174351.948650-2-ojeda@kernel.org/.
  • Related issue (imports_layout): [unstable option] imports_layout rust-lang/rustfmt#3361.
  • Related issue (imports_granularity): [unstable option] imports_granularity rust-lang/rustfmt#4991.

Nice to have (not critical, we could workaround if needed, etc.)

• Stabilize wrap_comments.

  • Tracking Issue: [unstable option] wrap_comments rust-lang/rustfmt#3347.
  • Related: wrap_doc_comments (even just that would already be quite nice to have).
  • Related PR: Format markdown files and rust code blocks in markdown files rust-lang/rustfmt#5909.
  • Cc: @tgross35.

• Stabilize format_code_in_doc_comments.

  • Tracking Issue: [unstable option] format_code_in_doc_comments rust-lang/rustfmt#3348.
  • Related: doc_comment_code_block_width.

• Support RUSTC_BOOTSTRAP.

  • Issue (closed as WONTFIX): Support RUSTC_BOOTSTRAP rust-lang/rustfmt#4884.
  • Without it, we cannot easily test it nor ask kernel developers to run it from time to time to e.g. format the code in doc comments.
  • Old PR: RUSTFMT_BOOTSTRAP=1 allows the compiler's stage0 toolchain to be used upstream rust-lang/rustfmt#3900.

• Stabilize reorder_impl_items.

  • Tracking Issue: [unstable option] reorder_impl_items rust-lang/rustfmt#3363.

• Stabilize comment_width.

  • Tracking Issue: [unstable option] comment_width rust-lang/rustfmt#3349.

• Stabilize normalize_comments.

  • Tracking Issue: [unstable option] normalize_comments rust-lang/rustfmt#3350.

• Stabilize hex_literal_case.

  • Tracking Issue: [unstable option] hex_literal_case rust-lang/rustfmt#5081.
  • Context: https://lore.kernel.org/rust-for-linux/CALNs47u6+EFYkvpyHZD5zLcjQeb2CuZNTOjPuZ4MKewoKZYPMg@mail.gmail.com/
  • Cc: @tgross35.

• Support skip_macro_invocations, i.e. skipping the formatting of some macro invocations by the macro name.

  • Tracking Issue: [unstable option] skip_macro_invocations rust-lang/rustfmt#5346.
  • It could potentially useful in some cases (possibly e.g. module!) -- no concrete one yet though.
  • Lore: https://lore.kernel.org/rust-for-linux/CANiq72n=gFWrv9yFmzntaeR+RH5=qeD_uOs6NiW4dHkdaopKKw@mail.gmail.com/.

Low priority (we will likely not use them in the end)

Done (stabilized, fixed, not needed anymore, etc.)

---

Bugs that we would like to see fixed

Required (we almost certainly want them)

Nice to have (probably not critical, we could workaround if needed, etc.)

Low priority (we will likely not use them in the end)

Done (stabilized, fixed, or not needed anymore, etc.)

• #[no_mangle] functions should not have extern "Rust" removed.

  • Issue (to avoid removing it in all cases): Stripping extern "Rust" from functions rust-lang/rustfmt#5701.
  • PR (to avoid removing it in all cases): refactor ABI formatting rust-lang/rustfmt#5845 (1.73?).

• rustfmt ICEs (apparently on purpose) on certain expanded -Zunpretty=expanded (i.e. .rsi) constructs (e.g. offset_of!).

  • Status: rustfmt wants us to report the ICEs and they are also happy to use small snippets like the one we reported as test cases.
  • From: https://rust-for-linux.zulipchat.com/#narrow/stream/291565-Help/topic/Using.20the.20.2Ersi.20target/near/423973701.
  • Issue: rustfmt ICE with expanded code rust-lang/rustfmt#6105.
  • Issue (outside a const): ICE: unreachable src/tools/rustfmt/src/expr.rs:407:13 rust-lang/rustfmt#5885.
  • PR: Fix: ICE with expanded code #6105 rust-lang/rustfmt#6112 (1.81).

[bjorn3]

Which unstable rustfmt features are actually used that necessitate RUSTC_BOOTSTRAP?

[ojeda]

format_code_in_doc_comments is the one we are currently using.

[bjorn3]

That one sounds useful. https://github.com/rust-lang/rustfmt/issues?q=format_code_in_doc_comments+is%3Aopen lists a couple of issues with it. Once those are fixed it would make sense to push for stabilization IMHO.

[ojeda]

Indeed! We also have a few others in the .rustfmt.toml file that were not currently enabled. I have put them in the list above :-)

[tgross35]

hex_literal_case seems useful, and potentially pretty easy to get over the stabilization line. Thinking about this because I noticed a patch that had a mix of upper- and lowercase hex, grepping through the kernel it seems like lowercase is much more common (and used by all in-tree Rust code so far)

Context: https://lore.kernel.org/rust-for-linux/CALNs47u6+EFYkvpyHZD5zLcjQeb2CuZNTOjPuZ4MKewoKZYPMg@mail.gmail.com/

[ojeda]

Thanks Trevor! I agree, that one would be useful -- added above.