a04b48416e
fix(joinir): Phase 287 P2 - Pattern6 nested loop latch overwrite fix
...
Fix infinite loop in Pattern6 (nested loop minimal) caused by main→loop_step
overwriting k_inner_exit→loop_step latch values.
Root cause: JoinIR main entry block was incorrectly treated as BackEdge,
causing it to overwrite the correct latch incoming values set by the true
back edge (k_inner_exit → loop_step).
Solution:
- Restrict latch recording to TailCallKind::BackEdge only
- Treat only MAIN's entry block as entry-like (not loop_step's entry block)
- Add debug_assert! to detect double latch set in future
Refactoring:
- Extract latch recording to latch_incoming_recorder module (SSOT)
- Add boundary.loop_header_func_name for explicit header identification
- Strengthen tail_call_classifier with is_source_entry_like parameter
Tests: apps/tests/phase1883_nested_minimal.hako → RC:9 (was infinite loop)
Smoke: 154/154 PASS, no regressions
🤖 Generated with [Claude Code](https://claude.com/claude-code )
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com >
2025-12-27 09:39:29 +09:00
bbfc3c1d94
refactor(joinir): Phase 287 P2 - Strengthen BackEdge/latch conditions (WIP)
...
**Problem**: Phase 188.3 Pattern6 (nested loop) encounters infinite loop
- inner_step → inner_step (self-recursion) incorrectly classified as BackEdge
- → redirects to outer header (loop_step) instead of inner_step entry
- is_recursive_call causes inner recursion to overwrite outer latch values
- → PHI receives wrong values → i doesn't increment → infinite loop
**Fix 1: BackEdge classification strictness** (tail_call_classifier.rs)
- Add `is_target_loop_entry` parameter to classify_tail_call()
- BackEdge ONLY when target==loop_step (entry_func), not inner_step
- Prevents inner_step → inner_step from redirecting to outer header
**Fix 2: latch_incoming guard** (instruction_rewriter.rs)
- Change condition from `is_recursive_call || is_target_loop_entry`
to `is_target_loop_entry` only
- Prevents inner_step self-recursion from overwriting outer loop's latch
- set_latch_incoming() now called with correct values (verified by debug)
**Status**: 🚧 WIP - Infinite loop still occurs
- set_latch_incoming('i', BasicBlockId(8), ValueId(21)) ✅ Called correctly
- But final PHI: `phi [%4, bb8]` instead of `phi [%21, bb8]` ❌
- Root cause likely in PHI generation (merge/mod.rs), not latch_incoming
- Next: Investigate why latch_incoming values aren't used in PHI
**Files**:
- src/mir/builder/control_flow/joinir/merge/tail_call_classifier.rs
- src/mir/builder/control_flow/joinir/merge/instruction_rewriter.rs
🤖 Generated with [Claude Code](https://claude.com/claude-code )
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com >
2025-12-27 08:32:14 +09:00
c6ffc06660
feat(joinir): Phase 188.3-P3.1,3.2 - AST extraction & validation helpers
...
Phase 3-1 完了: extract_inner_loop_ast() 実装
- 外側 loop body から内側 loop を抽出
- 正確に 1 つの inner loop を検証
- 0 個 or 2+ 個の場合は明示エラー (Fail-Fast)
Phase 3-2 完了: validate_strict_mode() プレースホルダー
- 現在は大半の検証を is_pattern6_lowerable() で実施
- 将来の strict mode チェック用の足場
ドキュメント追加:
- P1-INSTRUCTIONS.md: Phase 3-3 実装指示書 (4関数モデル詳細)
- README.md: max_loop_depth 定義の明確化
- 10-Now.md: Phase 188.3 方針の整合性維持
次タスク: Phase 3-3 (Continuation 生成) - P1-INSTRUCTIONS.md 参照
🤖 Generated with [Claude Code](https://claude.com/claude-code )
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com >
2025-12-27 06:22:59 +09:00
a2c5fd90fe
feat(joinir): Phase 188.3 - Pattern6 (NestedLoopMinimal) 選択ロジック実装
...
## Phase 188.3 進捗: Phase 2 完了 (6/13 tasks)
### 実装完了 ✅
**Phase 1: Fixture作成**
- apps/tests/phase1883_nested_minimal.hako 追加
- Add/Compare のみ（乗算なし）
- 期待 exit code: 9 (3×3 nested loops)
- 既存 lowering で fallback 動作確認
**Phase 2: 選択ロジック (SSOT)**
- LoopPatternContext に step_tree_max_loop_depth フィールド追加
- choose_pattern_kind() に Pattern6 選択ロジック実装:
1. Cheap check (has_inner_loop)
2. StepTree 構築 (max_loop_depth 取得)
3. AST validation (is_pattern6_lowerable)
- pattern6_nested_minimal.rs モジュール作成 (stub)
- LOOP_PATTERNS に Pattern6 entry 追加
- **検証**: Pattern6 が正しく選択される ✅
### 設計原則 (確認済み)
1. **Fail-Fast**: Pattern6 選択後は Ok(None) で逃げない
2. **outer 変数 write-back 検出 → validation false** (Phase 188.4+)
3. **最小実装**: inner local だけ、Pattern1 モデル二重化
4. **cfg! 依存なし**: production で動作
### 検証結果
```
[choose_pattern_kind] has_inner_loop=true
[choose_pattern_kind] max_loop_depth=2
[choose_pattern_kind] is_pattern6_lowerable=true
✅ Pattern6 SELECTED!
```
Stub からの期待エラー:
```
[ERROR] ❌ [Pattern6] Nested loop lowering not yet implemented
```
### 次: Phase 3 (Lowering 実装 - 推定4時間)
残りタスク:
- Phase 3-1: AST 抽出ヘルパー
- Phase 3-2: Validation ヘルパー
- Phase 3-3: Continuation 生成 (outer_step, inner_step, k_inner_exit)
- Phase 3-4: fixture が exit=9 を返すことを検証
### 変更ファイル
**新規**:
- apps/tests/phase1883_nested_minimal.hako
- src/mir/builder/control_flow/joinir/patterns/pattern6_nested_minimal.rs
- docs/development/current/main/phases/phase-188.{1,2,3}/README.md
**変更**:
- src/mir/builder/control_flow/joinir/routing.rs (Pattern6 選択)
- src/mir/builder/control_flow/joinir/patterns/router.rs (Context 拡張)
- src/mir/builder/control_flow/joinir/patterns/mod.rs (module 宣言)
🎯 Generated with [Claude Code](https://claude.com/claude-code )
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com >
2025-12-27 05:45:12 +09:00
