mega mir redo by sbillig · Pull Request #1404 · argotorg/fe

sbillig · 2026-04-22T09:42:24Z

Note: I haven't fully vetted this on the external test cases since rebasing, will do so in the morning.

The big change here is to split MIR into phases. SMIR -> NSMIR -> MIR -> Sonatina/Yul

SMIR ("semantic mir") - this is basically hir in cfg form. Used for const time function evaluation (and thus needs to live in the hir crate to avoid a dependency cycle between hir and mir). This allows us to cleanly expand const fn capabilities, and replaces the previous ugly hir-based implementation.
NSMIR (normalized) - includes concrete borrowing and place/origin information. Used for borrow checking and other diagnostics.
MIR - low-level, but backend neutral representation that's lowered into sonatina/yul. Includes concrete layouts, const/code regions, etc

(Aside: SMIR and NSMIR could maybe be collapsed together. NSMIR just derives a bunch of information that isn't relevant for ctfe)

This fixes a lot of tricky bugs we had previously due to sidecar info alongside the insufficient representation, and information being deduced/inferred in late stages because it wasn't propagated forward by earlier stages.

Unfortunately basically every snapshot has been updated. Name mangling that was previously done in the old MIR is now deferred to codegen, and yul currently uses a simple incremented suffix for conflicting names, which should be changed.

(Aside: I think we should consider replacing the big codegen snapshots with small targeted snapshots.)

TODO, here or in quick follow-up prs

remove diagnostic blocks preventing const fn functionality that's already supported (mut, while, ..?)
add uninitialized var analysis. most of the machinery for this is in place.
port over previous naming scheme, or make a new one
newsfragment

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 17ac59eacb

ℹ️ About Codex in GitHub

Your team has set up Codex to 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 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

sbillig · 2026-04-22T10:28:44Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f3afd41a74

ℹ️ About Codex in GitHub

Your team has set up Codex to 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 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

sbillig · 2026-04-22T19:31:28Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 33f491b0a1

ℹ️ About Codex in GitHub

Your team has set up Codex to 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 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

cburgdorf · 2026-04-22T20:20:41Z

Not a proper review but looks like the bountiful tests crash against this branch:

$ ~/Documents/hacking/fe/fe/target/debug/fe test
running `fe test` for 3 inputs (jobs=8)

COMPILING        games    /home/cburgdorf/Documents/hacking/fe/bountiful/contracts/ingots/games

thread '<unnamed>' (2999015) panicked at crates/mir/src/runtime/lower/classify.rs:1036:17:
runtime call resolution failed during return-class inference for SemanticInstanceKey(Id(33ef3), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>): runtime trait-call resolution failed to resolve a concrete impl body: caller=SemanticInstanceKey(Id(33ef3), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>) decl=SemanticInstanceKey(Id(33f04), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>) method=encode_payload concrete_inst=Encode<Sol> original_inst=Encode<Sol>
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
                 games    Sonatina backend panicked while emitting test module: runtime call resolution failed during return-class inference for SemanticInstanceKey(Id(33ef3), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>): runtime trait-call resolution failed to resolve a concrete impl body: caller=SemanticInstanceKey(Id(33ef3), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>) decl=SemanticInstanceKey(Id(33f04), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>) method=encode_payload concrete_inst=Encode<Sol> original_inst=Encode<Sol>
ERROR [66.845s]  games    Sonatina backend panicked while emitting test module: runtime call resolution failed during return-class inference for SemanticInstanceKey(Id(33ef3), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>): runtime trait-call resolution failed to resolve a concrete impl body: caller=SemanticInstanceKey(Id(33ef3), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>) decl=SemanticInstanceKey(Id(33f04), PhantomData<&salsa::interned::Value<fe_hir::analysis::semantic::instance::semantic::SemanticInstanceKey<'_>>>) method=encode_payload concrete_inst=Encode<Sol> original_inst=Encode<Sol>
COMPILING        registry /home/cburgdorf/Documents/hacking/fe/bountiful/contracts/ingots/registry

This is against latest bountiful master

sbillig · 2026-04-22T20:42:42Z

Not a proper review but looks like the bountiful tests crash against this branch:

shiiiiit

sbillig · 2026-04-22T20:52:43Z

The core/std abi changes were meant to be a separate pr that this one depends on. I'll see if I can split that out cleanly. (Though I guess I'm not sure if it's worthwhile, given that the old defunct mir code would need to be updated)

sbillig · 2026-04-22T22:15:38Z

@cburgdorf
Initial fix for the bountiful panic is in, which is not lowering to MIR when there are errors in dependencies. The errors are due to the core abi trait changes that I unintentionally snuck in.

The deeper hardening to not panic when MIR lowering runs into this issue, but I'm going to defer that to a followup.

sbillig · 2026-04-22T22:47:36Z

@cburgdorf Initial fix for the bountiful panic is in, which is not lowering to MIR when there are errors in dependencies. The errors are due to the core abi trait changes that I unintentionally snuck in.

The deeper hardening to not panic when MIR lowering runs into this issue, but I'm going to defer that to a followup.

I'll push up a bountiful pr with the required changes.

Edit: fe-lang/bountiful#8

micahscopes · 2026-04-23T08:39:43Z

Ran rosetta-fe against this branch vs v26.0.1.

Two compile failures

EffectRef<Evm> on capability passthrough. Any fn that declares uses (evm: mut Evm) and calls another fn with the same declaration fails with `evm` doesn't implement `EffectRef<Evm>`. Smallest case:

use std::evm::Evm
pub struct Buf { pub ptr: u256, pub len: u256 }
impl Copy for Buf {}
impl Buf {
    pub fn put(self, v: u256) -> Buf uses (evm: mut Evm) {
        evm.mstore(addr: self.ptr, value: v); self
    }
    pub fn wrapper(self) -> Buf uses (evm: mut Evm) { self.put(v: 1) }
}

Every call shape hits it, not just chaining:

// fails on the single call
pub fn one(self, a: u256) -> Buf uses (evm: mut Evm) { self.put(v: a) }

// fails at both links
pub fn chained(self, a: u256, b: u256) -> Buf uses (evm: mut Evm) {
    self.put(v: a).put(v: b)
}

// also fails when the intermediate goes through a let
pub fn seq(self, a: u256, b: u256) -> Buf uses (evm: mut Evm) {
    let x = self.put(v: a); x.put(v: b)
}

Trait-method dispatch is affected too: rosetta-fe's Transcript::write<T: Writable> calls val.write_to(ptr: ...) (a trait method declared uses (evm: mut Evm)) and hits the same error.

Suspected spot: effect_provider_target_resolution in ty_check/expr.rs around L2332 — the as_capability fast path doesn't seem to fire for forwarded evm, so resolution falls through to the EffectRef goal where mut Evm isn't covered by the reflexive impl. In rosetta-fe the verifier example hits this 182 times (plonk.fe, halo2.fe, transcript.fe).

"runtime package has no root objects" on packages whose contract lives in a submodule re-exported from lib.fe. Smallest case:

# src/lib.fe
pub use sub::{self, *}

// src/sub.fe
use std::abi::sol
msg M { #[selector = sol("ping()")] Ping -> u256 }
pub contract C {
    init() {}
    recv M { Ping -> u256 { 42 } }
}

Moving contract C (and its msg) inline into lib.fe works around it, so the shape that breaks is specifically "contract reachable only via re-export from the root module." Guard is at crates/codegen/src/runtime_package.rs:9; looks like RuntimePackage::root_objects isn't following the re-export. Hits math, merkle, and shared/ec in rosetta-fe (amm / erc20 / escrow / governance all declare the contract in lib.fe directly and build fine).

Gas deltas on what still compiles

testGas_fe_* only; fuzz trimmed to 32 runs; percentages relative to v26.0.1.

amm        addLiquidity -1.9   getK +0.7   swapAForB -2.1
erc20      approve      +0.2   transfer +0.5
escrow     deposit      -0.1   getBalance -0.5   release +2.6
governance balanceOf    -0.5   delegate +1.2   execute +0.9   mint -0.3
           propose      +1.3   transfer +1.3   vote +1.7

amm's hot paths down ~2%, governance uniformly up ~1-1.7%, erc20/escrow flat.

Caveats: v26.0.1 baseline (master since is only docs work); had to exclude verifier/math/merkle/ec from the workspace to make fe build ../.. succeed in setUp.

chatgpt-codex-connector Bot reviewed Apr 22, 2026

View reviewed changes

Comment thread crates/fe/src/test/mod.rs Outdated

chatgpt-codex-connector Bot reviewed Apr 22, 2026

View reviewed changes

Comment thread crates/driver/src/db.rs Outdated

sbillig force-pushed the quagmir branch from 062cfee to 33f491b Compare April 22, 2026 19:30

chatgpt-codex-connector Bot reviewed Apr 22, 2026

View reviewed changes

Comment thread crates/driver/src/db.rs Outdated

sbillig force-pushed the quagmir branch from 33f491b to ba591a3 Compare April 22, 2026 20:32

sbillig added 12 commits April 22, 2026 16:13

Clean up core/std abi traits and helpers

4b14e23

Implement semantic MIR runtime pipeline

3328c86

Rename mir crate and preserve revert semantics

d194b7e

Add borrow diagnostics AnalysisPass

79cd8df

codegen: conflict-resistant name mangling

acfe5e0

Add semantic instance normalization helpers

6ecf427

Add fe_test fixtures for applicative and custom effects

b4b5a87

Use test package lowering for fe_test roundtrip

84672d0

Cache root provider lookups

e72eb18

Remove MIR diagnostics mode

b30f625

Gate tests on dependency diagnostics

613050f

sonatina a774c70

417e835

sbillig force-pushed the quagmir branch from 986346f to 417e835 Compare April 23, 2026 04:07

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

mega mir redo#1404

mega mir redo#1404
sbillig wants to merge 12 commits intoargotorg:masterfrom
sbillig:quagmir

sbillig commented Apr 22, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

cburgdorf commented Apr 22, 2026

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

sbillig commented Apr 22, 2026 •

edited

Loading

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

sbillig commented Apr 22, 2026 •

edited

Loading

Uh oh!

micahscopes commented Apr 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

sbillig commented Apr 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

cburgdorf commented Apr 22, 2026

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

sbillig commented Apr 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sbillig commented Apr 22, 2026

Uh oh!

sbillig commented Apr 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

micahscopes commented Apr 23, 2026

Two compile failures

Gas deltas on what still compiles

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

sbillig commented Apr 22, 2026 •

edited

Loading

sbillig commented Apr 22, 2026 •

edited

Loading

sbillig commented Apr 22, 2026 •

edited

Loading