cargo fmt crashes in Rust 1.81.0

[ajewellamz]

lib.rs.gz

Un-gzip the attached lib.rs
make it the only file in a crate
run cargo fmt
See

thread 'main' panicked at library/core/src/slice/sort/shared/smallsort.rs:862:5:
user-provided comparison function does not correctly implement a total order
stack backtrace:
   0:        0x103e69aa3 - <std::sys::backtrace::BacktraceLock::print::DisplayBacktrace as core::fmt::Display>::fmt::haf41a54e7ed5fe07
   1:        0x103eb45cb - core::fmt::write::h526161fad96c5ad5
   2:        0x103e5f69e - std::io::Write::write_fmt::h4de0398536c8b0e4

As an aside - with previous versions of Rust, cargo fmt also failed on this file.
It didn't crash, but it left trailing spaces on some lines, and then reported failure.

[ytmimi]

Hey @ajewellamz what exact version of cargo fmt / rustfmt are you using. You can run cargo fmt --version to get that info.

Also, any chance you can get this down to a reproducible code snippet that you can add inline? That would be best. rustfmt can run just fine on code that doesn't compile (as long as it parses) so my suggestion would be to slowly remove function, struct, enum, macro, and trait definitions / impls from the file until you're able to narrow down what's causing the issue. It would be a great help 🙏🏼

[ajewellamz]

rustfmt 1.7.1-stable (eeb90cda 2024-09-04)

I won't be able to do any work on this until next week at the earliest.
I was assuming you could just run it in the debugger or something, if you had a fail case.

[ajewellamz]

Ok, here's a minimal fail case. I've tried deleting various parts, and it seems to always succeed.

pub use ::dafny_runtime::Sequence;
pub use ::std::rc::Rc;
pub use crate::r#_StructuredEncryptionUtil_Compile::CanonCryptoItem;
pub use ::dafny_runtime::itertools::Itertools;
pub use crate::software::amazon::cryptography::dbencryptionsdk::structuredencryption::internaldafny::types::CryptoAction;
pub use crate::software::amazon::cryptography::materialproviders::internaldafny::types::EncryptedDataKey;
pub use ::dafny_runtime::Object;
pub use crate::software::amazon::cryptography::primitives::internaldafny::types::IAwsCryptographicPrimitivesClient;
pub use crate::software::amazon::cryptography::materialproviders::internaldafny::types::AlgorithmSuiteInfo;
pub use ::dafny_runtime::DafnyCharUTF16;
pub use crate::r#_StructuredEncryptionUtil_Compile::MessageID;
pub use crate::software::amazon::cryptography::materialproviders::internaldafny::types::EncryptionMaterials;
pub use crate::r#_Wrappers_Compile::Outcome;
pub use ::dafny_runtime::string_utf16_of;
pub use ::dafny_runtime::int;
pub use ::dafny_runtime::MapBuilder;
pub use ::dafny_runtime::_System::nat;
pub use crate::software::amazon::cryptography::primitives::internaldafny::types::HMacInput;
pub use ::dafny_runtime::rd;
pub use ::dafny_runtime::truncate;
pub use ::dafny_runtime::DafnyTypeEq;
pub use ::dafny_runtime::DafnyType;

[ytmimi]

Thanks! confirming that with rustfmt +1.81 (rustfmt 1.7.1-stable (eeb90cda 2024-09-04)) I'm able to reproduce the panic with the above example:

Backtrace

thread 'main' panicked at library/core/src/slice/sort/shared/smallsort.rs:862:5:
user-provided comparison function does not correctly implement a total order
stack backtrace:
   0:        0x10316373c - <std::sys::backtrace::BacktraceLock::print::DisplayBacktrace as core::fmt::Display>::fmt::h243268f17d714c7f
   1:        0x1031a6688 - core::fmt::write::hb3cfb8a30e72d7ff
   2:        0x103159720 - std::io::Write::write_fmt::hfb2314975de9ecf1
   3:        0x103165c4c - std::panicking::default_hook::{{closure}}::h14c7718ccf39d316
   4:        0x103165870 - std::panicking::default_hook::hc62e60da3be2f352
   5:        0x10cd5c5b8 - <alloc[47bc6d386d7ae45f]::boxed::Box<rustc_driver_impl[54c40c94c6cfc0b2]::install_ice_hook::{closure#0}> as core[f827f14b5e761a5d]::ops::function::Fn<(&dyn for<'a, 'b> core[f827f14b5e761a5d]::ops::function::Fn<(&'a std[4f7d7c3ef984657a]::panic::PanicHookInfo<'b>,), Output = ()> + core[f827f14b5e761a5d]::marker::Sync + core[f827f14b5e761a5d]::marker::Send, &std[4f7d7c3ef984657a]::panic::PanicHookInfo)>>::call
   6:        0x103166868 - std::panicking::rust_panic_with_hook::h09e8a656f11e82b2
   7:        0x103166150 - std::panicking::begin_panic_handler::{{closure}}::h1230eb3cc91b241c
   8:        0x103163bc8 - std::sys::backtrace::__rust_end_short_backtrace::hc3491307aceda2c2
   9:        0x103165e40 - _rust_begin_unwind
  10:        0x1031c0ad8 - core::panicking::panic_fmt::ha4b80a05b9fff47a
  11:        0x1031a8c84 - core::slice::sort::shared::smallsort::panic_on_ord_violation::h9a9b9a4ea28b52bf
  12:        0x1020fed34 - core[f827f14b5e761a5d]::slice::sort::shared::smallsort::small_sort_general_with_scratch::<rustfmt_nightly[2afeb25615a25c73]::imports::UseTree, <rustfmt_nightly[2afeb25615a25c73]::imports::UseTree as core[f827f14b5e761a5d]::cmp::PartialOrd>::lt>
  13:        0x1021469a4 - core[f827f14b5e761a5d]::slice::sort::stable::quicksort::quicksort::<rustfmt_nightly[2afeb25615a25c73]::imports::UseTree, <rustfmt_nightly[2afeb25615a25c73]::imports::UseTree as core[f827f14b5e761a5d]::cmp::PartialOrd>::lt>
  14:        0x1021a1108 - core[f827f14b5e761a5d]::slice::sort::stable::drift::sort::<rustfmt_nightly[2afeb25615a25c73]::imports::UseTree, <rustfmt_nightly[2afeb25615a25c73]::imports::UseTree as core[f827f14b5e761a5d]::cmp::PartialOrd>::lt>
  15:        0x1020bdedc - core[f827f14b5e761a5d]::slice::sort::stable::driftsort_main::<rustfmt_nightly[2afeb25615a25c73]::imports::UseTree, <rustfmt_nightly[2afeb25615a25c73]::imports::UseTree as core[f827f14b5e761a5d]::cmp::PartialOrd>::lt, alloc[47bc6d386d7ae45f]::vec::Vec<rustfmt_nightly[2afeb25615a25c73]::imports::UseTree>>
  16:        0x102192af8 - rustfmt_nightly[2afeb25615a25c73]::reorder::rewrite_reorderable_or_regroupable_items
  17:        0x1021ae510 - <rustfmt_nightly[2afeb25615a25c73]::visitor::FmtVisitor>::visit_items_with_reordering
  18:        0x1021b9dd4 - <rustfmt_nightly[2afeb25615a25c73]::visitor::FmtVisitor>::format_separate_mod
  19:        0x102047714 - rustfmt_nightly[2afeb25615a25c73]::formatting::format_project::<rustfmt_nightly[2afeb25615a25c73]::Session<std[4f7d7c3ef984657a]::io::stdio::Stdout>>
  20:        0x102045adc - <scoped_tls[df49f867320abf2e]::ScopedKey<rustc_span[ab16d476329f5d04]::SessionGlobals>>::with::<<rustfmt_nightly[2afeb25615a25c73]::Session<std[4f7d7c3ef984657a]::io::stdio::Stdout>>::format_input_inner::{closure#0}, core[f827f14b5e761a5d]::result::Result<rustfmt_nightly[2afeb25615a25c73]::FormatReport, rustfmt_nightly[2afeb25615a25c73]::ErrorKind>>
  21:        0x102043bd4 - <scoped_tls[df49f867320abf2e]::ScopedKey<rustc_span[ab16d476329f5d04]::SessionGlobals>>::set::<rustc_span[ab16d476329f5d04]::create_session_if_not_set_then<core[f827f14b5e761a5d]::result::Result<rustfmt_nightly[2afeb25615a25c73]::FormatReport, rustfmt_nightly[2afeb25615a25c73]::ErrorKind>, <rustfmt_nightly[2afeb25615a25c73]::Session<std[4f7d7c3ef984657a]::io::stdio::Stdout>>::format_input_inner::{closure#0}>::{closure#0}, core[f827f14b5e761a5d]::result::Result<rustfmt_nightly[2afeb25615a25c73]::FormatReport, rustfmt_nightly[2afeb25615a25c73]::ErrorKind>>
  22:        0x102047060 - rustc_span[ab16d476329f5d04]::create_session_if_not_set_then::<core[f827f14b5e761a5d]::result::Result<rustfmt_nightly[2afeb25615a25c73]::FormatReport, rustfmt_nightly[2afeb25615a25c73]::ErrorKind>, <rustfmt_nightly[2afeb25615a25c73]::Session<std[4f7d7c3ef984657a]::io::stdio::Stdout>>::format_input_inner::{closure#0}>
  23:        0x1020325a8 - rustfmt[4ddeb7d779b6c40f]::format_and_emit_report::<std[4f7d7c3ef984657a]::io::stdio::Stdout>
  24:        0x102031c40 - rustfmt[4ddeb7d779b6c40f]::execute
  25:        0x10202ffac - rustfmt[4ddeb7d779b6c40f]::main
  26:        0x10204a668 - std[4f7d7c3ef984657a]::sys::backtrace::__rust_begin_short_backtrace::<fn(), ()>
  27:        0x102042734 - std[4f7d7c3ef984657a]::rt::lang_start::<()>::{closure#0}
  28:        0x10314c0b8 - std::rt::lang_start_internal::hdd117cb81a316264
  29:        0x102033080 - _main

error: the compiler unexpectedly panicked. this is a bug.

note: we would appreciate a bug report: https://github.com/rust-lang/rustfmt/issues/new?labels=bug

[mitsuhiko]

I just ran into this trying to rustfmt some pretty terrifying expanded macro code. I found the reduced example above more useful to debug than the mess I had. I was not particularly successful in finding the issue but the issue appears to be in Ord for UseSegment somewhere. I'm guessing the issue might be that the different kind pairs are not given a total order?

rustfmt/src/imports.rs

Lines 922 to 933 in 6157568

	(Slf(ref a), Slf(ref b))
	| (Super(ref a), Super(ref b))
	| (Crate(ref a), Crate(ref b)) => match (a, b) {
	(Some(sa), Some(sb)) => {
	if self.style_edition >= StyleEdition::Edition2024 {
	version_sort(sa.trim_start_matches("r#"), sb.trim_start_matches("r#"))
	} else {
	a.cmp(b)
	}
	}
	(_, _) => a.cmp(b),
	},

[ajewellamz]

Still broken in Rust 1.82, not surprising since rustfmt is still 1.7.1-stable

[ajewellamz]

I was trying to fix this myself, but I'm having trouble building it.

I assumed that all I had to do was

git clone git@github.com:rust-lang/rust.git
cd rust/src/tools/rustfmt
git checkout 1.82.0 # same failure with or without this step
cargo build

but that fails to build, with 40 compile time errors in various file within src/tools/rustfmt/src

Am I missing something?

[ytmimi]

You want to clone the rustfmt repo, not the rust-lang/rust repo.

git clone https://github.com/rust-lang/rustfmt.git
cd rustfmt
cargo build # should automatically install the relevant toolchain / components

[ajewellamz]

Thanks, that works.

I've found the problem, but I'm not sure of the proper fix
Ord for UseSegment is not transitive.

In imports.rs there exists impl Ord for UseSegment {
within that function is

match (&self.kind, &other.kind) {
   ...
   (Ident(ref pia, ref aa), Ident(ref pib, ref ab)) => {
   ...
          1)        if ia.starts_with(char::is_uppercase) && ib.starts_with(char::is_lowercase) {
                        return Ordering::Greater;
                    }
          2)        if ia.starts_with(char::is_lowercase) && ib.starts_with(char::is_uppercase) {
                        return Ordering::Less;
                    }
          3)      if is_upper_snake_case(ia) && !is_upper_snake_case(ib) {
                        return Ordering::Greater;
                    }
          4)      if !is_upper_snake_case(ia) && is_upper_snake_case(ib) {
                        return Ordering::Less;
                    }
          5)         ia.cmp(ib)

let A = DafnyType
let B = _System
let C = truncate

A < B by 5)
B < C by 5)
A > C by 1)

hence user-provided comparison function does not correctly implement a total order

[ajewellamz]

Fixed in #6375