Fuzz fix 2 by g-r-a-n-t · Pull Request #1311 · argotorg/fe

g-r-a-n-t · 2026-03-02T00:51:54Z

#	Commit	Area	Why It Exists	What It Changes	Main Coverage
1	`eff91898e` `hir: avoid self-assumptions in trait headers`	HIR trait headers	Trait headers could recurse through invalid self-assumptions involving associated projections.	Stops trait-header self-assumptions and adds direct regression coverage.	`trait_header_assoc_projection_recursive_bound.fe`
2	`accc364f6` `hir: recover trait-ref and assoc-bound cycles as diagnostics`	HIR trait lowering and constraint expansion	Several recursive trait and assoc-bound cases could hang or panic instead of surfacing a normal error.	Adds cycle recovery in `lower_trait_ref`, makes recursive bound expansion safer, keeps unnamed impl methods from poisoning trait collection, and covers the affected crash-regression paths.	`499d4ead`, `1670ec70`, `acb16143`, `a8d7d78c`, `assoc_ty_cycle`
3	`11316942b` `hir: harden associated-type resolution and normalization`	HIR assoc-type resolution	Assoc-type lookup and normalization could reuse bad inference state or hit panic paths, especially around qualified traits.	Hardens canonicalization, normalization, unify remapping, and qualified-trait / assoc-type resolution.	`b02aa0c9`, `c2482ead`, `7856b3f9`
4	`f2250a523` `hir: harden malformed impl and pattern analysis`	HIR pattern analysis and ops selection	Malformed impls and malformed patterns could still reach panic or `unreachable!()` paths during analysis.	Turns ops-selection failures into diagnostics and makes malformed pattern cases recoverable instead of fatal.	`a64bec5e`, `76e16ce9`, `25d24f34`, `b8fc85b7`
5	`e81f3c594` `mir/index: avoid scalar address-space lookup on by-value indexing`	MIR indexing / repr queries	MIR could query scalar address-space information for by-value indexing cases where that lookup is not meaningful.	Narrows the MIR-side fix to the address-space/indexing issue only, without keeping the older invalid-HIR fallback machinery.	by-value indexing / repr-address-space lowering path
6	`d51fafedf` `mir: short-circuit diagnostics and lowering on invalid HIR`	Driver, CLI, LSP, MIR entry points	MIR work was still being attempted after HIR had already failed, and the behavior could diverge by entry point.	Centralizes MIR diagnostics flow and stops MIR lowering/diagnostics from running on already-invalid HIR. This commit also owns the invalid-HIR crash fixtures that used to motivate the old MIR fallback.	`mir_borrow_operand_not_place.fe`, `mir_match_lowering_non_ref_projection_d58afcf7.fe`, `mir_value_address_space_fallback_143863d4.fe`, `monomorphize_failed_instantiate_from_parse_error.fe`, `monomorphize_unlowered_hir_expr_survives.fe`, `monomorphize_unresolved_trait_method.fe`
7	`6195de534` `tests: minimize coverage-only crash fixtures`	Regression test corpus	A few leftover crash fixtures were still coverage-only cleanup rather than owning a distinct semantic fix.	Keeps the cleanup isolated to the residual tests-only reductions that did not clearly belong to one owning fix commit.	`hir_pretty_print_missing_param_name_5ecc2069.fe` and related coverage-only fixture cleanup
8	`8cadca252` `tests: refresh inline Fe snippets for current parser syntax`	Test maintenance only	Some inline Fe snippets and parser-facing fixtures were stale after the history cleanup and caused otherwise unrelated tests to fail.	Updates codegen/MIR/parser-facing test inputs to current Fe syntax. This is a tests-only maintenance commit, not a fuzz-fix semantic change.	`crates/codegen/src/sonatina/tests.rs`, `crates/mir/src/transform.rs`, parser-facing fixtures

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c2c6b91388

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

crates/fe/tests/crash_regressions.rs

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: aeec92d371

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

crates/resolver/Cargo.toml

...es/fe/tests/fixtures/crash_regressions/trait_constraints_missing_core_effect_ref_c5d7f979.fe

crates/mir/src/ir.rs

crates/mir/src/lower/match_lowering.rs

crates/hir/src/analysis/ty/unify.rs

crates/hir/src/analysis/ty/binder.rs

sbillig · 2026-03-03T02:22:25Z

Some good stuff in here, but also some cases that are papering over real bugs by working around panics. These panics let the fuzzer find bugs; if we just remove them all with fallbacks the fuzzing becomes useless.

crates/hir/src/analysis/ty/trait_resolution/mod.rs

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5892bcc750

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

crates/fe/tests/crash_regressions.rs

crates/hir/src/analysis/ty/trait_resolution/mod.rs

sbillig · 2026-03-05T22:55:14Z

@g-r-a-n-t Do you mind if I push a few commits to this branch?

g-r-a-n-t · 2026-03-05T22:56:14Z

@g-r-a-n-t Do you mind if I push a few commits to this branch?

Please do.

Co-authored-by: Sean Billig <sean.billig@gmail.com>

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8cadca2522

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

< BEA4 ul dir="auto">

Open a pull request for review

Mark a draft as ready

Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-03-20T03:29:48Z

crates/hir/src/analysis/name_resolution/path_resolver.rs

+                let folded_inst_keys = inference_keys(db, &folded_inst);
+                let folded_ty_keys = inference_keys(db, &folded_ty);
+                if folded_inst_keys.is_subset(&lhs_keys) && folded_ty_keys.is_subset(&lhs_keys) {


Keep bound-carried generics when resolving associated types

In find_associated_type, this new lhs_keys filter drops any candidate whose projection still mentions a generic that comes from the trait bound rather than from Self. That breaks valid cases like where U: Trait<T> followed by U::Assoc: after unification, folded_ty still contains T, so the candidate is discarded and the path resolves as “not found” even though the bound uniquely defines the associated type. The same pattern affects nested projections that depend on extra bound arguments.

Useful? React with 👍 / 👎.

chatgpt-codex-connector · 2026-03-20T03:29:48Z

crates/hir/src/analysis/ty/simplified_pattern.rs

+                    let (expected_base, _) = expected_ty.decompose_ty_app(db);
+                    let (resolved_base, _) = ty_id.decompose_ty_app(db);
+                    if expected_base == resolved_base {


Preserve explicit type arguments in constructor patterns

This base-type check is broad enough to fire for explicitly written generic patterns like Foo<bool> against a Foo<i32> scrutinee. Once it does, the branch below rewrites the constructor to expected_ty, so exhaustiveness/reachability analysis treats the invalid arm as if it were Foo<i32>. That can produce bogus unreachable pattern or missing-arm diagnostics whenever a struct/tuple pattern names the right constructor with the wrong type arguments.

Useful? React with 👍 / 👎.

g-r-a-n-t marked this pull request as draft March 2, 2026 00:52

chatgpt-codex-connector bot reviewed Mar 2, 2026

View reviewed changes

crates/fe/tests/crash_regressions.rs Show resolved Hide resolved

g-r-a-n-t force-pushed the fuzz-fix2 branch 5 times, most recently from f8a25e5 to aeec92d Compare March 3, 2026 00:17

g-r-a-n-t marked this pull request as ready for review March 3, 2026 00:25

chatgpt-codex-connector bot reviewed Mar 3, 2026

View reviewed changes

crates/resolver/Cargo.toml Outdated Show resolved Hide resolved

sbillig requested changes Mar 3, 2026

View reviewed changes

...es/fe/tests/fixtures/crash_regressions/trait_constraints_missing_core_effect_ref_c5d7f979.fe Show resolved Hide resolved

crates/mir/src/ir.rs Show resolved Hide resolved

sbillig reviewed Mar 3, 2026

View reviewed changes

crates/mir/src/lower/match_lowering.rs Outdated Show resolved Hide resolved

sbillig reviewed Mar 3, 2026

View reviewed changes

crates/hir/src/analysis/ty/unify.rs Show resolved Hide resolved

sbillig reviewed Mar 3, 2026

View reviewed changes

crates/hir/src/analysis/ty/binder.rs Outdated Show resolved Hide resolved

sbillig reviewed Mar 3, 2026

View reviewed changes

crates/hir/src/analysis/ty/trait_resolution/mod.rs Show resolved Hide resolved

g-r-a-n-t force-pushed the fuzz-fix2 branch 6 times, most recently from a50f8fd to 3992bb3 Compare March 4, 2026 22:21

g-r-a-n-t marked this pull request as draft March 4, 2026 22:41

g-r-a-n-t force-pushed the fuzz-fix2 branch 4 times, most recently from e51641c to 5892bcc Compare March 5, 2026 15:21

g-r-a-n-t marked this pull request as ready for review March 5, 2026 15:31

g-r-a-n-t requested a review from sbillig March 5, 2026 15:31

chatgpt-codex-connector bot reviewed Mar 5, 2026

View reviewed changes

crates/fe/tests/crash_regressions.rs Show resolved Hide resolved

crates/hir/src/analysis/ty/trait_resolution/mod.rs Show resolved Hide resolved

g-r-a-n-t force-pushed the fuzz-fix2 branch 4 times, most recently from 57eb711 to 20b6efd Compare March 10, 2026 00:05

g-r-a-n-t marked this pull request as draft March 10, 2026 01:28

g-r-a-n-t force-pushed the fuzz-fix2 branch 2 times, most recently from 7fe04e1 to ed9aca8 Compare March 19, 2026 17:04

hir: avoid self-assumptions in trait headers

eff9189

g-r-a-n-t force-pushed the fuzz-fix2 branch from ed9aca8 to bdff42a Compare March 20, 2026 00:21

g-r-a-n-t and others added 7 commits March 19, 2026 18:27

hir: recover trait-ref and assoc-bound cycles as diagnostics

accc364

Co-authored-by: Sean Billig <sean.billig@gmail.com>

hir: harden associated-type resolution and normalization

1131694

hir: harden malformed impl and pattern analysis

f2250a5

mir/index: avoid scalar address-space lookup on by-value indexing

e81f3c5

mir: short-circuit diagnostics and lowering on invalid HIR

d51fafe

tests: minimize coverage-only crash fixtures

4D24

6195de5

tests: refresh inline Fe snippets for current parser syntax

8cadca2

g-r-a-n-t force-pushed the fuzz-fix2 branch from bdff42a to 8cadca2 Compare March 20, 2026 00:32

g-r-a-n-t marked this pull request as ready for review March 20, 2026 03:20

chatgpt-codex-connector bot reviewed Mar 20, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fuzz fix 2#1311

Fuzz fix 2#1311
g-r-a-n-t wants to merge 8 commits intoargotorg:masterfrom
g-r-a-n-t:fuzz-fix2

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

Uh oh!

Uh oh!

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants