[youknowone]

Problem

origin/main does not compile. pyre-jit-trace fails with 4× E0609:

error[E0609]: no field `last_caught_exception_value` on type `majit_metainterp::MIFrame`
  --> pyre/pyre-jit-trace/src/jitcode_dispatch/residual_call.rs:240
error[E0609]: no field `last_caught_exception_value` on type `InlineFrame`
  --> pyre/pyre-jit-trace/src/jitcode_dispatch/residual_call.rs:241
  (same at :273 for MIFrame and &InlineFrame)

Because PR CI builds merge(PR, main), this reddens every open PR, not just the one that introduced it.

Root cause

Stale-branch merge:

• jit: instruction-keyed, per-frame, and loop-close exception traceback recording #756 (dc3559c94aa, "instruction-keyed, per-frame, and loop-close exception traceback recording") deliberately removed last_caught_exception_value from MIFrame, BlackholeInterpreter, InlineFrame, and WalkSession. Its commit body: "no traceback decision keys on object identity." Traceback recording is now keyed on the raising instruction.
• jit(fbw): multi-frame blackhole-resume build path + input-arg _resref seed (adoption gated) #763 (80d590b0f6d, "multi-frame blackhole-resume build path") was developed on a pre-jit: instruction-keyed, per-frame, and loop-close exception traceback recording #756 base and still reads/writes that field in build_multi_frame_miframe. It merged 4 commits after jit: instruction-keyed, per-frame, and loop-close exception traceback recording #756 (#756 → #753 → #748 → #735 → #763), so the field references no longer resolve.

Fix

Drop #763's vestigial per-frame last_caught_exception_value propagation to match #756's instruction-keyed model — not restore the field (that would undo #756's design). The removed lines only copied the field between reconstructed frames; the multi-frame build path is PYRE_FBW_MULTIFRAME-gated (default off), so there is no behavior change.

Verification

python3 ./pyre/check.py on the fix:

ALL PASSED: dynasm 303/303
ALL PASSED: cranelift 303/303
ALL PASSED: wasm 300/300

— commented by Claude

Summary by CodeRabbit

• Bug Fixes
  • Corrected exception state handling when resuming execution after a residual call.
  • Prevented exception-catch information from being incorrectly carried over between nested execution frames.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 84d49125-d2ec-458e-b2de-4af02eb9ede8

📥 Commits

Reviewing files that changed from the base of the PR and between 80d590b and 5fb7a11.

📒 Files selected for processing (1)

• pyre/pyre-jit-trace/src/jitcode_dispatch/residual_call.rs

---

Walkthrough

Multi-frame blackhole construction in the residual-call path no longer copies or overrides last_caught_exception_value while rebuilding the MIFrameStack.

Changes

Residual-call exception state

Layer / File(s)	Summary
Preserve per-frame exception state pyre/pyre-jit-trace/src/jitcode_dispatch/residual_call.rs	Intermediate and innermost MIFrame values are used without propagating or mutating last_caught_exception_value.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• youknowone/pyre#563: Adds related last_caught_exception_value bookkeeping and JIT traceback handling.
• youknowone/pyre#593: Modifies residual-call exception handling and catch-resume paths.
• youknowone/pyre#743: Expands related exception and traceback recording behavior.

Poem

I’m a small rabbit hopping through the frame,
No caught exception copied just the same.
Each blackhole keeps its state in view,
The innermost frame stays honest too.
Hop, hop—clean bookkeeping through!

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly matches the change: removing stale last_caught_exception_value propagation in the fbw residual-call path.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix-763-last-caught-build

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

🤖 Codex parity review

Static analysis of this diff vs the local RPython/PyPy sources (commit 5fb7a11).
Updated: 2026-07-24T06:29:31.733Z

Files in the reviewed diff

pyre/pyre-jit-trace/src/jitcode_dispatch/residual_call.rs

1. Regressions to PyPy parity introduced by this patch

None.

2. Other mismatches introduced by this patch

None.

3. Pre-existing mismatches (already present before this patch)

None.

4. Structural adaptations

None.