Commit Graph

16 Commits

Author SHA1 Message Date
7f57a1bb05 feat(llvm): Phase 131-13/14 - MIR JSON順序修正 & 2パスsnapshot解決
## Phase 131-13: MIR JSON 命令順序修正
- copy 遅延ロジック削除(~80行)
- MIR の def→use 順序をそのまま出力(SSOT)
- PHI 先頭集約のみ維持

## Phase 131-14: jump-only block 2パス snapshot 解決
- Pass A: jump-only block はメタ記録のみ
- Pass B: resolve_jump_only_snapshots() で CFG ベース解決
- path compression で連鎖を効率的に解決
- サイクル検出で Fail-Fast

## 結果
-  STRICT モードでエラーなし
-  bb7 が bb5 の snapshot を正しく継承
-  ループが正しく動作(1, 2 出力確認)
- ⚠️ print/concat で segfault(別問題、次Phase)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2025-12-15 00:39:43 +09:00
eb70dfc5bb refactor(llvm): Phase 131-12-P2 - block_end_values SSOT 化(WIP)
## 実装内容
- get_end_values() API 追加
- _value_at_end_i64() を snapshot-only に変更
- def_blocks 即時更新
- PHI incoming を snapshot から取得

## 発見された問題
- 同一ブロック内の def→use が predecessor snapshot を見てしまう
- これは次フェーズで resolve_cur / resolve_incoming 分離で修正

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2025-12-14 21:55:44 +09:00
7dfd6ff1d9 feat(llvm): Phase 131-11-H/12 - ループキャリアPHI型修正 & vmap snapshot SSOT
## Phase 131-11-H: ループキャリアPHI型修正
- PHI生成時に初期値(entry block)の型のみ使用
- backedge の値を型推論に使わない(循環依存回避)
- NYASH_CARRIER_PHI_DEBUG=1 でトレース

## Phase 131-12-P0: def_blocks 登録 & STRICT エラー化
- safe_vmap_write() で PHI 上書き保護
- resolver miss を STRICT でエラー化(フォールバック 0 禁止)
- def_blocks 自動登録

## Phase 131-12-P1: vmap_cur スナップショット実装
- DeferredTerminator 構造体(block, term_ops, vmap_snapshot)
- Pass A で vmap_cur をスナップショット
- Pass C でスナップショット復元(try-finally)
- STRICT モード assert

## 結果
-  MIR PHI型: Integer(正しい)
-  VM: Result: 3
-  vmap snapshot 機構: 動作確認
- ⚠️ LLVM: Result: 0(別のバグ、次Phase で調査)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2025-12-14 21:28:41 +09:00
dda65b94b7 Phase 21.7 normalization: optimization pre-work + bench harness expansion
- Add opt-in optimizations (defaults OFF)
  - Ret purity verifier: NYASH_VERIFY_RET_PURITY=1
  - strlen FAST enhancement for const handles
  - FAST_INT gate for same-BB SSA optimization
  - length cache for string literals in llvmlite
- Expand bench harness (tools/perf/microbench.sh)
  - Add branch/call/stringchain/arraymap/chip8/kilo cases
  - Auto-calculate ratio vs C reference
  - Document in benchmarks/README.md
- Compiler health improvements
  - Unify PHI insertion to insert_phi_at_head()
  - Add NYASH_LLVM_SKIP_BUILD=1 for build reuse
- Runtime & safety enhancements
  - Clarify Rust/Hako ownership boundaries
  - Strengthen receiver localization (LocalSSA/pin/after-PHIs)
  - Stop excessive PluginInvoke→BoxCall rewrites
- Update CURRENT_TASK.md, docs, and canaries

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-13 16:40:58 +09:00
7b1f791395 feat(phase21.5): Loop FORCE direct assembly + PHI/compare fixes
## Loop FORCE Direct Assembly 
- Added: Direct MIR assembly bypass when HAKO_MIR_BUILDER_LOOP_FORCE_JSONFRAG=1
- Implementation: Extracts limit from Program(JSON), generates minimal while-form
- Structure: entry(0) → loop(1) → body(2) → exit(3)
- PHI: i = {i0, entry} | {i_next, body}
- Location: tools/hakorune_emit_mir.sh:70-126
- Tag: [selfhost-direct:ok] Direct MIR assembly (FORCE=1)

## PHI/Compare Fixes (ny-llvmc) 
- Fixed: vmap maintenance for PHI results across instructions
- Fixed: PHI placeholder name consistency (bytes vs str)
- Fixed: ensure_phi_alloca creates unique placeholders per block
- Fixed: resolve_i64_strict properly looks up PHI results
- Files:
  - src/llvm_py/phi_wiring/tagging.py
  - src/llvm_py/phi_wiring/wiring.py
  - src/llvm_py/instructions/compare.py
  - src/llvm_py/resolver.py

## Testing Results
- VM backend:  rc=10 (correct)
- Direct assembly MIR:  Structurally correct
- Crate backend: ⚠️ PHI/compare issues (being investigated)

## Implementation Principles
- 既定挙動不変 (FORCE=1 gated)
- Dev toggle controlled
- Minimal diff, surgical changes
- Bypasses using resolution when FORCE=1

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-11 17:04:33 +09:00
041cef875a json-native: token positions (line/column); escape utils BMP coverage + surrogate guard; add smokes for string escapes, nested, and error cases (AST/VM) 2025-09-26 00:42:55 +09:00
2d870664d5 current_task: refresh notes; cleanup unified; LLVM(harness): finalize_phis ownership, ret simplified, uses-predeclare; method-postfix cleanup return: PHI head grouping final TODOs 2025-09-19 02:43:33 +09:00
5e818eeb7e stage3: unify to cleanup; MIR return-defer; docs+smokes updated; LLVM(harness): finalize_phis ownership, ret.py simplified, uses-predeclare; cleanup return override green; method-postfix cleanup return WIP (PHI head) 2025-09-19 02:07:38 +09:00
a5054a271b llvm(py): introduce BuildCtx + trace hub; refactor if-merge prepass + PHI wiring into module; unify logs; ctx-enable compare/ret/call/boxcall/externcall/typeop/newbox/safepoint; curated smoke option for if-merge; README updates; keep behavior stable 2025-09-17 16:11:01 +09:00
d99b941218 docs: add MIR13 mode doc and set PHI-off as default; bridge lowering split (if/loop/try); llvmlite resolver stabilization; curated runner default PHI-off; refresh CURRENT_TASK.md 2025-09-17 10:58:12 +09:00
4c0e6726e3 🔧 refactor(llvm-py): Fix resolver PHI handling and add trace improvements
Changes to resolver.py:
- Improved PHI value tracking in _value_at_end_i64() (lines 268-285)
- Added trace logging for snap hits with PHI detection
- Fixed PHI placeholder reuse logic to preserve dominance
- PHI values now returned directly from snapshots when valid

Changes to llvm_builder.py:
- Fixed externcall instruction parsing (line 522: 'func' instead of 'name')
- Improved block snapshot tracing (line 439)
- Added PHI incoming metadata tracking (lines 316-376)
- Enhanced definition tracking for lifetime hints

This should help debug the string carry=0 issue in esc_dirname_smoke where
PHI values were being incorrectly coerced instead of preserved.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-09-14 16:25:21 +09:00
3e07763af8 docs: update CURRENT_TASK with Box Theory PHI plan (defer/finalize) and MIR v0.5 type meta; add parity tooling and PyVM scaffolding
impl(pyvm/llvmlite):
- add tools/parity.sh; tools/pyvm_runner.py; src/llvm_py/pyvm/*
- emit string const as handle type in MIR JSON; add dst_type hints
- unify '+' to concat_hh with from_i64/from_i8_string bridges; console print via to_i8p_h
- add runtime bridges: nyash.box.from_i64, nyash.string.to_i8p_h

tests:
- add apps/tests/min_str_cat_loop (minimal repro for string cat loop)
2025-09-14 04:51:33 +09:00
658a0d46da feat(llvm-py): Major breakthrough in Python LLVM backend! 🎉
 Print and FileBox paths now working correctly
 Resolver simplified by removing overly aggressive fast-path optimization
 Both OFF/ON in compare_harness_on_off.sh now use Python version
 String handle propagation issues resolved

Key changes:
- Removed instruction reordering in llvm_builder.py (respecting MIR order)
- Resolver now more conservative but reliable
- compare_harness_on_off.sh updated to use Python backend for both paths

This marks a major milestone towards Phase 15 self-hosting with Python/llvmlite!

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-09-14 00:44:28 +09:00
2a9aa5368d harness(llvm/py): fix PHI/dominance via Resolver-only; per-pred localization and constant GEPs; stabilize Main.esc_json/1, dirname/1, node_json/3; docs: add NYASH_LLVM_TRACE_FINAL and Resolver-only invariants
- Resolver-only reads across BBs; remove vmap fallbacks
- Create PHIs at block start; insert casts in preds before terminators
- Re-materialize int in preds to satisfy dominance (add/zext/trunc)
- Use constant GEP for method strings to avoid order dependency
- Order non-PHI lowering to preserve producer→consumer dominance
- Update docs: RESOLVER_API.md, LLVM_HARNESS.md
- compare_harness_on_off: ON/OFF exits match; linking green
2025-09-13 19:49:03 +09:00
1d6fab4eda 📚 Phase 15計画を詳細化・更新: Python/llvmlite正式採用とプラグイン全方向ビルド戦略
 主な更新内容:
- Python/llvmlite実装の正式採用を明記(開発速度10倍、~2400行)
- プラグイン全方向ビルド戦略(.so/.o/.a同時生成)で単一EXE生成可能に
- 各実装の予想コード量を具体化(パーサー800行、MIR Builder 2500行、VM 5000行)
- 循環依存問題の解決を明記(nyrtがC ABI経由で提供)
- 現実的なスケジュール調整(2025年9月~2026年3月)

🎉 最新進捗:
- dep_tree_min_string.nyashオブジェクト生成成功(10.4KB)
- LLVM verifier green - dominance違反解決
- Resolver patternでSSA安全性確保

🚀 次のマイルストーン:
- Python/llvmliteでEXE生成パイプライン完成
- nyash-llvm-compiler分離設計
- NyashパーサーMVP実装開始

Everything is Boxの究極形が、ついに実現へ!

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-09-13 15:37:58 +09:00
38aea59fc1 llvm: unify lowering via Resolver and Cursor; remove non-sealed PHI wiring; apply Resolver to extern/call/boxcall/arrays/maps/mem; add llvmlite harness docs; add LLVM layer overview; add LoopForm preheader 2025-09-12 20:40:48 +09:00