rust-build-times

Rust 构建时间

用途

指导 Agent 诊断并提升 Rust 编译速度：使用 cargo-timings 分析构建性能、使用 sccache 进行缓存、使用 Cranelift 代码生成后端加快开发构建、拆分工作区 crate、权衡 LTO 配置，以及使用快速链接器（mold/lld）。

触发条件

• “我的 Rust 项目编译时间太长”
• “如何分析哪些 crate 构建缓慢？”
• “如何为 Rust 配置 sccache？”
• “Cranelift 后端是什么？它有何帮助？”
• “我应该使用 thin LTO 还是 fat LTO？”
• “如何在 Rust 中使用 mold 链接器？”

工作流

1. 使用 cargo-timings 诊断

# Build with timing report
cargo build --timings

# Opens build/cargo-timings/cargo-timing.html
# Shows: crate compilation timeline, parallelism, bottlenecks

# For release builds
cargo build --release --timings

# Key things to look for in the timing report:
# - Long sequential chains (no parallelism)
# - Individual crates taking > 10s (candidates for optimization)
# - Proc-macro crates blocking everything downstream

# cargo-llvm-lines — count LLVM IR lines per function (monomorphization)
cargo install cargo-llvm-lines
cargo llvm-lines --release | head -20
# Shows functions generating the most LLVM IR (template explosion)

2. sccache——Rust 编译缓存

# Install
cargo install sccache
# or: brew install sccache

# Configure for Rust builds
export RUSTC_WRAPPER=sccache

# Add to .cargo/config.toml (project or global)
# ~/.cargo/config.toml
[build]
rustc-wrapper = "sccache"

# Check cache stats
sccache --show-stats

# S3 backend for CI teams
export SCCACHE_BUCKET=my-rust-cache
export SCCACHE_REGION=us-east-1
export AWS_ACCESS_KEY_ID=xxx
export AWS_SECRET_ACCESS_KEY=yyy
sccache --start-server

# GitHub Actions with sccache
# - uses: mozilla-actions/[email protected]

3. Cranelift 代码生成后端

Cranelift 是一种快速的代码生成后端（相较于 LLVM），生成的代码运行较慢，但编译速度快得多。适合开发构建：

# Install nightly (Cranelift requires nightly for now)
rustup toolchain install nightly
rustup component add rustc-codegen-cranelift-preview --toolchain nightly

# Use Cranelift for dev builds only
# .cargo/config.toml
[unstable]
codegen-backend = true

[profile.dev]
codegen-backend = "cranelift"

# Use per-build
CARGO_PROFILE_DEV_CODEGEN_BACKEND=cranelift \
RUSTFLAGS="-Zunstable-options" \
cargo +nightly build

Cranelift 与 LLVM 的权衡：

• 开发构建：使用 Cranelift 可将编译速度提升 20–40%
• 运行时性能：LLVM 编译的代码更快（Cranelift 会跳过许多优化）
• 发布构建：始终使用 LLVM

4. 拆分工作区以实现并行构建

单个大型 crate 会按顺序编译。将其拆分为较小的 crate，以启用 Cargo 并行处理：

# Before: one giant crate
[package]
name = "monolith"    # everything in one crate = sequential compile

# After: workspace with parallel crates
[workspace]
members = [
    "core",          # compiled in parallel
    "networking",    # no deps on ui → parallel with ui
    "ui",            # no deps on networking → parallel
    "server",        # depends on core + networking
    "cli",           # depends on core + ui
]

# Visualize dependency graph
cargo tree | head -30
cargo tree --graph | dot -Tsvg > deps.svg   # visual graph

# Check how many crates compile in parallel
cargo build -j$(nproc) --timings    # maximize parallelism

有效拆分工作区的规则：

• 首先消除循环依赖
• 将过程宏拆分到独立的 crate 中（它们会阻塞所有构建）
• 将经常变更的代码隔离开来（减少失效）

5. LTO 配置

LTO 可提升运行时性能，但会增加链接时间：

# Cargo.toml profile configuration
[profile.release]
lto = "thin"         # thin LTO: good performance, much faster than "fat"
codegen-units = 1    # needed for best optimization (but disables parallelism)

[profile.release-fast]
inherits = "release"
lto = "fat"          # full LTO: maximum performance, very slow link

[profile.dev]
lto = "off"          # never use LTO in dev (compilation speed)
codegen-units = 16   # maximize parallel codegen in dev

LTO 对比：

设置	链接时间	运行时性能	适用场景
lto = false	快	基准	开发构建
lto = "thin"	中等	+5–15%	大多数发布构建
lto = "fat"	慢	+15–30%	追求最高性能
codegen-units = 1	最慢	最佳	与 LTO 搭配用于发布构建

6. 快速链接器

链接器通常是大型 Rust 项目的性能瓶颈：

# mold — fastest general-purpose linker (Linux)
sudo apt-get install mold

# .cargo/config.toml
[target.x86_64-unknown-linux-gnu]
linker = "clang"
rustflags = ["-C", "link-arg=-fuse-ld=mold"]

# Or use cargo-zigbuild (uses zig cc as linker)
cargo install cargo-zigbuild
cargo zigbuild --release

# lld — LLVM's linker (faster than GNU ld, available everywhere)
# .cargo/config.toml
[target.x86_64-unknown-linux-gnu]
rustflags = ["-C", "link-arg=-fuse-ld=lld"]

# On macOS: zld or the default lld
[target.x86_64-apple-darwin]
rustflags = ["-C", "link-arg=-fuse-ld=/usr/local/bin/zld"]

链接器速度对比（大型项目，典型情况）：

• GNU ld：基准
• lld：快约 2 倍
• mold：快约 5–10 倍
• gold：快约 1.5 倍

7. 其他快速优化

# Reduce debug info level (faster but less debuggable)
# Cargo.toml
[profile.dev]
debug = 1           # 0=off, 1=line tables, 2=full (default)
# debug=1 saves 20-40% on debug build time

# Split debug info (reduces linker input)
[profile.dev]
split-debuginfo = "unpacked"   # macOS: equivalent of gsplit-dwarf

# Disable incremental compilation (sometimes faster for full rebuilds)
CARGO_INCREMENTAL=0 cargo build

# Reduce proc-macro compile time (pin heavy proc-macro deps)
# Heavy proc-macros: serde, tokio, axum — keep versions stable

相关技能

• 使用 skills/rust/cargo-workflows 配置 Cargo 工作区和构建配置
• 使用 skills/build-systems/build-acceleration 了解 C/C++ 对应的构建加速方法
• 使用 skills/debuggers/dwarf-debug-format 了解调试信息大小和 split-dwarf 的权衡
• 使用 skills/binaries/linkers-lto 了解 LTO 内部机制