BE-266: HashQL: Implement PreInlining pass for MIR optimization

[indietyp]

🌟 What is the purpose of this PR?

Add a new PreInlining optimization pass that runs a sequence of transformations to prepare MIR bodies for inlining. This pass orchestrates multiple optimizations in a fixpoint loop to ensure code is maximally simplified before inlining occurs.

🔍 What does this change?

• Add new PreInlining pass that runs a carefully ordered sequence of optimizations:
  1. Administrative reduction - Removes structural clutter and normalizes shape
  2. Instruction simplification - Constant folding and algebraic simplification
  3. Value propagation (FS/CP alternating) - Propagates values through the code
  4. Dead store elimination - Removes stores made dead by propagation
  5. CFG simplification - Cleans up control flow after local changes
• Add GlobalTransformState to track per-body changes during global transformations
• Update AdministrativeReduction to use the new state tracking mechanism
• Add comprehensive test cases demonstrating the pass's effectiveness on various code patterns
• Add a new compiletest suite for the pre-inlining pass

🛡 What tests cover this?

• Added 8 new test cases covering various optimization patterns:
  • Constant folding and branch elimination
  • Cascading conditional simplification
  • Closure inlining with dead branch elimination
  • Dead code elimination after value propagation
  • Instruction simplification with value propagation
  • Nested conditional simplification
  • Nested let binding cleanup
  • Thunk inlining with dead code elimination

❓ How to test this?

1. Run the MIR tests: cargo test -p hashql_mir
2. Run the compiletest suite: cargo test -p hashql_compiletest -- mir/pass/transform/pre-inlining

Pre-Merge Checklist 🚀

🚢 Has this modified a publishable library?

This PR:

• does not modify any publishable blocks or libraries, or modifications do not need publishing

📜 Does this require a change to the docs?

The changes in this PR:

• are internal and do not require a docs change

🕸️ Does this require a change to the Turbo Graph?

The changes in this PR:

• do not affect the execution graph

[cursor]

PR Summary

Introduces a pre-inlining optimization driver and supporting infrastructure, and wires it into tests and tooling.

• Adds global pass PreInlining that runs a fixpoint of AdministrativeReduction → InstSimplify → ForwardSubstitution/CopyPropagation (alternating) → DeadStoreElimination → CfgSimplify, with an initial CP+CFG pre-pass
• Adds GlobalTransformState and OwnedGlobalTransformState to track per-body Changed during global transforms; updates GlobalTransformPass::run signature and AdministrativeReduction to use it
• Improves CfgSimplify dispatch to avoid unnecessary work and unreachable paths; minor helpers in AR (e.g., is_empty, state marking)
• Integrates new pass into compiletest (mir/pass/transform/pre-inlining), rendering (text/D2), and registers in suite/mod.rs
• Updates benches to run the pipeline via PreInlining; adjusts builders to assign DefId
• Adds comprehensive UI tests covering constant folding, cascading conditionals, closure/thunk inlining, propagation + DSE, nested lets/ifs

^{Written by Cursor Bugbot for commit 85aff24. This will update automatically on new commits. Configure here.}

[indietyp]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• BE-271: HashQL: Implement PostInline pass #8240
• BE-269: HashQL: Move PreInline out into Canonicalization pass #8239
• BE-268: HashQL: Rename PreInlining to PreInline for consistency #8238
• BE-270: HashQL: Optimize empty tuples to unit constants in MIR #8237
• BE-197: Implement MIR inline pass for HashQL #8236
• BE-267: Use bump scopes instead of full resets across transformation passes #8234
• BE-266: HashQL: Implement PreInlining pass for MIR optimization #8233 👈 (View in Graphite)
• BE-265: HashQL: Make CopyPropagation aware of local re-assignments #8231
• BE-264: HashQL: Rename SROA to ForwardSubstitution for better semantic clarity #8230
• BE-263: HashQL: add body! macro for declarative MIR construction in tests #8229
• BE-261: HashQL: Implement Administrative Reduction pass for MIR #8227
• BE-262: HashQL: Do not intern empty slices #8223
• BE-260: HashQL: Switch to bump allocator with scoping support #8218
• BE-259: HashQL: Add benchmarks for type system operations #8216
• BE-227: HashQL: Implement call graph analysis for MIR #8214
• BE-258: HashQL: Add Changed enum to track MIR transformation pass modifications #8210
• BE-255: HashQL: Implement Copy Propagation (CP) pass for MIR #8207
• BE-231: HashQL: Add InstSimplify pass for MIR instruction simplification and constant folding #8206
• BE-254: Add CodSpeed benchmarking for HashQL MIR transformation passes #8204
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[augmentcode]

🤖 Augment PR Summary

Summary: Introduces a new MIR optimization driver pass (PreInlining) to aggressively simplify MIR bodies before the inlining stage.

Changes:

• Add transform::PreInlining, which runs a fixpoint pipeline (AR → InstSimplify → FS/CP alternating → DSE → CfgSimplify) plus an initial CP+CFG pre-pass.
• Add GlobalTransformState (borrowed) and OwnedGlobalTransformState (owned) for per-body Changed tracking in global MIR transforms.
• Update GlobalTransformPass::run signature to accept &mut GlobalTransformState and adapt call sites/tests.
• Update AdministrativeReduction to mark per-body changes via the new state and return early when nothing is reducible.
• Minor CfgSimplify refactor to avoid unnecessary CFG recomputation for non-branching terminators.
• Add and register a new compiletest suite (mir/pass/transform/pre-inlining) plus multiple UI test cases exercising the new pipeline.

Technical notes: PreInlining maintains an internal “unstable body” set and iterates (bounded by MAX_ITERATIONS) until no further changes occur.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 1 suggestions posted.

Comment augment review to trigger a new review at any time.

[codecov]

Codecov Report

❌ Patch coverage is 92.52874% with 13 lines in your changes missing coverage. Please review.
✅ Project coverage is 59.87%. Comparing base (1c1b05d) to head (85aff24).

Files with missing lines	Patch %	Lines
libs/@local/hashql/mir/src/pass/mod.rs	61.90%	8 Missing ⚠️
...ocal/hashql/mir/src/pass/transform/pre_inlining.rs	96.92%	2 Missing and 2 partials ⚠️
.../hashql/mir/src/pass/transform/cfg_simplify/mod.rs	83.33%	1 Missing ⚠️

Additional details and impacted files

@@                                     Coverage Diff                                      @@
##           bm/be-265-hashql-track-locals-inside-of-copy-propagation    #8233      +/-   ##
============================================================================================
+ Coverage                                                     59.81%   59.87%   +0.06%     
============================================================================================
  Files                                                          1066     1067       +1     
  Lines                                                        106239   106408     +169     
  Branches                                                       4472     4479       +7     
============================================================================================
+ Hits                                                          63545    63710     +165     
- Misses                                                        41956    41961       +5     
+ Partials                                                        738      737       -1     

Flag	Coverage Δ
rust.hashql-compiletest	46.65% <ø> (ø)
rust.hashql-mir	87.58% <92.52%> (+0.19%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[codspeed-hq]

CodSpeed Performance Report

Merging #8233 will degrade performance by 47.86%

_{Comparing bm/be-266-hashql-implement-pre-inlining-fix-point-loop (85aff24) with bm/be-265-hashql-track-locals-inside-of-copy-propagation (1c1b05d)}

Summary

❌ 3 (👁 3) regressions
✅ 14 untouched

Benchmarks breakdown

	Benchmark	BASE	HEAD	Efficiency
👁	diamond	37.3 µs	71.6 µs	-47.86%
👁	linear	33 µs	40.5 µs	-18.65%
👁	complex	53.8 µs	103.2 µs	-47.86%