菜单
Eliminate unreachable and unused code to reduce program size and improve performance.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Implement abstract machines for defining and executing operational semantics of programming languages.
Implements alias and points-to analysis for pointer programs. Use for: (1) Optimizing compilers, (2) Verifying memory safety, (3) Program understanding, (4) Parallelization.
Define program meaning through logical assertions and proof rules (Hoare logic).
Transforms closures to explicit environments. Use when: (1) Implementing functional languages, (2) Building compilers, (3) Understanding closures.
Implements common subexpression elimination (CSE). Use when: (1) Building compilers, (2) Optimizing code, (3) Program analysis.
Implements general dataflow analysis framework. Use when: (1) Building compilers, (2) Static analysis tools, (3) Program verification.
| name | dead-code-eliminator |
| description | Eliminate unreachable and unused code to reduce program size and improve performance. |
| version | 1.0.0 |
| tags | ["compilation","optimization","llvm","pldi"] |
| difficulty | intermediate |
| languages | ["c++","rust","python"] |
| dependencies | ["dataflow-analysis-framework","control-flow-analysis"] |
Dead code elimination (DCE) removes code that has no effect on program output, including unreachable code, unused computations, and redundant assignments. It is a fundamental optimization in compilers.
| Concept | Description |
|---|---|
| Dead Code | Code that has no effect on program output |
| Unreachable Code | Code that cannot be executed |
| Live Variable | Variable that may be used later |
| Critical Instruction | Instruction with side effects |
| Fixpoint | State where no more changes occur |
constant-propagation-pass - Often combined with DCEinline-expander - Creates dead code opportunitiesdataflow-analysis-framework - Foundation for livenessssa-constructor - Simplifies analysis| Reference | Why It Matters |
|---|---|
| Muchnick, "Advanced Compiler Design and Implementation", Ch. 18 (1997) | Comprehensive treatment |
| Click & Cooper, "Combining Analyses, Optimizing and Using Them" (1995) | Modern DCE approach |
| LLVM Dead Code Elimination Pass | Production implementation |
| Approach | Pros | Cons |
|---|---|---|
| Simple DCE | Fast, easy | Misses opportunities |
| Aggressive DCE | Removes more code | More complex |
| ADCE with interprocedural | Global optimization | Very expensive |
A high-quality implementation should have:
| Criterion | What to Look For |
|---|---|
| Soundness | Never removes code that affects output |
| Completeness | Removes all provably dead code |
| Speed | Fast enough for production use |
| Iteration | Reaches fixpoint |
✅ Good: Removes all dead code, preserves semantics, fast ⚠️ Warning: Removes some but not all dead code ❌ Bad: Removes code that affects program output
Real-world DCE implementations:
| Tool | Why It Matters |
|---|---|
| LLVM DCE pass | Production DCE in LLVM |
| GCC DCE | GCC's dead code elimination |
| Soot DCE | Java bytecode DCE |
| GraalVM DCE | Truffle-based DCE |
Current DCE research:
| Direction | Key Papers | Challenge |
|---|---|---|
| Interprocedural DCE | "Whole-program DCE" | Cross-module analysis |
| Type-based DCE | "Type-driven DCE" | Type information |
| Dynamically dead | "Dynamic Dead Code" | Runtime dead code |
Common DCE bugs:
| Pitfall | Real Example | Prevention |
|---|---|---|
| Side effects | Removing print statements | Track I/O precisely |
| Volatile | Removing volatile reads | Handle volatile |
| Finalizers | Removing finalize() | Don't remove finalizers |
| Self-modifying | Removing JIT code | Handle dynamic code |