• Associated types are sometimes also called "output types". The key observation is that the implementer, not the caller, chooses this type.

• Many standard library traits have associated types, including arithmetic operators and Iterator.

派生功能是通过宏实现的，并且许多 crate 提供有用的派生宏，以添加实用功能。例如，serde 可以使用 #[derive(Serialize)] 为结构体派生序列化支持。

Exercise: Logger Trait

Let's design a simple logging utility, using a trait Logger with a log method. Code which might log its progress can then take an &impl Logger. In testing, this might put messages in the test logfile, while in a production build it would send messages to a log server.

However, the StderrLogger given below logs all messages, regardless of verbosity. Your task is to write a VerbosityFilter type that will ignore messages above a maximum verbosity.

This is a common pattern: a struct wrapping a trait implementation and implementing that same trait, adding behavior in the process. What other kinds of wrappers might be useful in a logging utility?

use std::fmt::Display;

pub trait Logger {
    /// Log a message at the given verbosity level.
    fn log(&self, verbosity: u8, message: impl Display);
}

struct StderrLogger;

impl Logger for StderrLogger {
    fn log(&self, verbosity: u8, message: impl Display) {
        eprintln!("verbosity={verbosity}: {message}");
    }
}

fn do_things(logger: &impl Logger) {
    logger.log(5, "FYI");
    logger.log(2, "Uhoh");
}

// TODO: Define and implement `VerbosityFilter`.

fn main() {
    let l = VerbosityFilter { max_verbosity: 3, inner: StderrLogger };
    do_things(&l);
}

解答

use std::fmt::Display;

pub trait Logger {
    /// Log a message at the given verbosity level.
    fn log(&self, verbosity: u8, message: impl Display);
}

struct StderrLogger;

impl Logger for StderrLogger {
    fn log(&self, verbosity: u8, message: impl Display) {
        eprintln!("verbosity={verbosity}: {message}");
    }
}

fn do_things(logger: &impl Logger) {
    logger.log(5, "FYI");
    logger.log(2, "Uhoh");
}

/// Only log messages up to the given verbosity level.
struct VerbosityFilter {
    max_verbosity: u8,
    inner: StderrLogger,
}

impl Logger for VerbosityFilter {
    fn log(&self, verbosity: u8, message: impl Display) {
        if verbosity <= self.max_verbosity {
            self.inner.log(verbosity, message);
        }
    }
}

fn main() {
    let l = VerbosityFilter { max_verbosity: 3, inner: StderrLogger };
    do_things(&l);
}

泛型

This segment should take about 40 minutes. It contains:

泛型函数

Rust supports generics, which lets you abstract algorithms or data structures (such as sorting or a binary tree) over the types used or stored.

/// Pick `even` or `odd` depending on the value of `n`.
fn pick<T>(n: i32, even: T, odd: T) -> T {
    if n % 2 == 0 {
        even
    } else {
        odd
    }
}

fn main() {
    println!("picked a number: {:?}", pick(97, 222, 333));
    println!("picked a tuple: {:?}", pick(28, ("dog", 1), ("cat", 2)));
}

泛型类型

您可以使用泛型对具体字段类型进行抽象化处理：

#[derive(Debug)]
struct Point<T> {
    x: T,
    y: T,
}

impl<T> Point<T> {
    fn coords(&self) -> (&T, &T) {
        (&self.x, &self.y)
    }

    fn set_x(&mut self, x: T) {
        self.x = x;
    }
}

fn main() {
    let integer = Point { x: 5, y: 10 };
    let float = Point { x: 1.0, y: 4.0 };
    println!("{integer:?} and {float:?}");
    println!("coords: {:?}", integer.coords());
}

泛型

Traits can also be generic, just like types and functions. A trait's parameters get concrete types when it is used.

#[derive(Debug)]
struct Foo(String);

impl From<u32> for Foo {
    fn from(from: u32) -> Foo {
        Foo(format!("Converted from integer: {from}"))
    }
}

impl From<bool> for Foo {
    fn from(from: bool) -> Foo {
        Foo(format!("Converted from bool: {from}"))
    }
}

fn main() {
    let from_int = Foo::from(123);
    let from_bool = Foo::from(true);
    println!("{from_int:?}, {from_bool:?}");
}

特征边界

使用泛型时，您通常会想要利用类型来实现某些特性， 这样才能调用此特征的方法。

您可以使用 T: Trait 或 impl Trait 执行此操作：

fn duplicate<T: Clone>(a: T) -> (T, T) {
    (a.clone(), a.clone())
}

// struct NotClonable;

fn main() {
    let foo = String::from("foo");
    let pair = duplicate(foo);
    println!("{pair:?}");
}

impl Trait

与特征边界类似，impl Trait 语法可以在函数形参 和返回值中使用：

// Syntactic sugar for:
//   fn add_42_millions<T: Into<i32>>(x: T) -> i32 {
fn add_42_millions(x: impl Into<i32>) -> i32 {
    x.into() + 42_000_000
}

fn pair_of(x: u32) -> impl std::fmt::Debug {
    (x + 1, x - 1)
}

fn main() {
    let many = add_42_millions(42_i8);
    println!("{many}");
    let many_more = add_42_millions(10_000_000);
    println!("{many_more}");
    let debuggable = pair_of(27);
    println!("debuggable: {debuggable:?}");
}

练习：通用 min 函数

In this short exercise, you will implement a generic min function that determines the minimum of two values, using the Ord trait.

use std::cmp::Ordering;

// TODO: implement the `min` function used in `main`.

fn main() {
    assert_eq!(min(0, 10), 0);
    assert_eq!(min(500, 123), 123);

    assert_eq!(min('a', 'z'), 'a');
    assert_eq!(min('7', '1'), '1');

    assert_eq!(min("hello", "goodbye"), "goodbye");
    assert_eq!(min("bat", "armadillo"), "armadillo");
}

解答

use std::cmp::Ordering;

fn min<T: Ord>(l: T, r: T) -> T {
    match l.cmp(&r) {
        Ordering::Less | Ordering::Equal => l,
        Ordering::Greater => r,
    }
}

fn main() {
    assert_eq!(min(0, 10), 0);
    assert_eq!(min(500, 123), 123);

    assert_eq!(min('a', 'z'), 'a');
    assert_eq!(min('7', '1'), '1');

    assert_eq!(min("hello", "goodbye"), "goodbye");
    assert_eq!(min("bat", "armadillo"), "armadillo");
}

Welcome Back

Including 10 minute breaks, this session should take about 3 hours and 10 minutes. It contains:

标准库类型

This segment should take about 1 hour and 20 minutes. It contains:

标准库

Rust comes with a standard library which helps establish a set of common types used by Rust libraries and programs. This way, two libraries can work together smoothly because they both use the same String type.

In fact, Rust contains several layers of the Standard Library: core, alloc and std.

• core includes the most basic types and functions that don't depend on libc, allocator or even the presence of an operating system.
• alloc 包括需要全局堆分配器的类型，例如 Vec、Box 和 Arc。
• 嵌入式 Rust 应用通常只使用 core，偶尔会使用 alloc。

文档

Rust comes with extensive documentation. For example:

• All of the details about loops.
• 'u8' 等基元类型。
• Standard library types like Option or BinaryHeap.

事实上，您可以为自己的代码编写文档：

/// Determine whether the first argument is divisible by the second argument.
///
/// If the second argument is zero, the result is false.
fn is_divisible_by(lhs: u32, rhs: u32) -> bool {
    if rhs == 0 {
        return false;
    }
    lhs % rhs == 0
}

The contents are treated as Markdown. All published Rust library crates are automatically documented at docs.rs using the rustdoc tool. It is idiomatic to document all public items in an API using this pattern.

如需从项内（例如在模块内）为项编写文档，请使用 //! 或 /*! .. */，这称为“内部文档注释”：

//! This module contains functionality relating to divisibility of integers.

Option

我们已经了解了 Option<T> 的一些用法。它可以存储“T”类型的值，或者不存储任何值。例如，'String::find' 会返回 Option<usize>。

fn main() {
    let name = "Löwe 老虎 Léopard Gepardi";
    let mut position: Option<usize> = name.find('é');
    println!("find returned {position:?}");
    assert_eq!(position.unwrap(), 14);
    position = name.find('Z');
    println!("find returned {position:?}");
    assert_eq!(position.expect("Character not found"), 0);
}

Result

Result 与 Option 相似，但表示操作成功或失败，且每个操作的类型不同。这类似于表达式练习中定义的 Res，但是一个泛型：Result<T, E>，其中 T 用于 Ok 变体，E 出现在 Err 变体中。

use std::fs::File;
use std::io::Read;

fn main() {
    let file: Result<File, std::io::Error> = File::open("diary.txt");
    match file {
        Ok(mut file) => {
            let mut contents = String::new();
            if let Ok(bytes) = file.read_to_string(&mut contents) {
                println!("Dear diary: {contents} ({bytes} bytes)");
            } else {
                println!("Could not read file content");
            }
        }
        Err(err) => {
            println!("The diary could not be opened: {err}");
        }
    }
}

String

String 是标准堆分配的可扩容 UTF-8 字符串缓冲区：

fn main() {
    let mut s1 = String::new();
    s1.push_str("Hello");
    println!("s1: len = {}, capacity = {}", s1.len(), s1.capacity());

    let mut s2 = String::with_capacity(s1.len() + 1);
    s2.push_str(&s1);
    s2.push('!');
    println!("s2: len = {}, capacity = {}", s2.len(), s2.capacity());

    let s3 = String::from("🇨🇭");
    println!("s3: len = {}, number of chars = {}", s3.len(), s3.chars().count());
}

String 会实现 Deref<Target = str>，这意味着您可以 对 String 调用所有 str 方法。

Vec

Vec 是标准的可调整大小堆分配缓冲区：

fn main() {
    let mut v1 = Vec::new();
    v1.push(42);
    println!("v1: len = {}, capacity = {}", v1.len(), v1.capacity());

    let mut v2 = Vec::with_capacity(v1.len() + 1);
    v2.extend(v1.iter());
    v2.push(9999);
    println!("v2: len = {}, capacity = {}", v2.len(), v2.capacity());

    // Canonical macro to initialize a vector with elements.
    let mut v3 = vec![0, 0, 1, 2, 3, 4];

    // Retain only the even elements.
    v3.retain(|x| x % 2 == 0);
    println!("{v3:?}");

    // Remove consecutive duplicates.
    v3.dedup();
    println!("{v3:?}");
}

Vec 会实现 Deref<Target = [T]>，这意味着您可以对 Vec 调用 slice 方法。

HashMap

标准的哈希映射，内含针对 HashDoS 攻击的保护措施：

use std::collections::HashMap;

fn main() {
    let mut page_counts = HashMap::new();
    page_counts.insert("Adventures of Huckleberry Finn", 207);
    page_counts.insert("Grimms' Fairy Tales", 751);
    page_counts.insert("Pride and Prejudice", 303);

    if !page_counts.contains_key("Les Misérables") {
        println!(
            "We know about {} books, but not Les Misérables.",
            page_counts.len()
        );
    }

    for book in ["Pride and Prejudice", "Alice's Adventure in Wonderland"] {
        match page_counts.get(book) {
            Some(count) => println!("{book}: {count} pages"),
            None => println!("{book} is unknown."),
        }
    }

    // Use the .entry() method to insert a value if nothing is found.
    for book in ["Pride and Prejudice", "Alice's Adventure in Wonderland"] {
        let page_count: &mut i32 = page_counts.entry(book).or_insert(0);
        *page_count += 1;
    }

    println!("{page_counts:#?}");
}

练习：计数器

在本练习中，您将学习一个非常简单的数据结构，并将其变成泛型的。该结构使用 std::collections::HashMap 来跟踪已经出现过的值以及每个值出现的次数。

Counter 的初始版本经过硬编码，仅适用于 u32 值。使结构体及其方法可用于所跟踪的值类型，以便 Counter 能够跟踪任何类型的值。

如果提前完成操作，请尝试使用 entry 方法将哈希查找次数减半，从而实现 count 方法。

use std::collections::HashMap;

/// Counter counts the number of times each value of type T has been seen.
struct Counter {
    values: HashMap<u32, u64>,
}

impl Counter {
    /// Create a new Counter.
    fn new() -> Self {
        Counter {
            values: HashMap::new(),
        }
    }

    /// Count an occurrence of the given value.
    fn count(&mut self, value: u32) {
        if self.values.contains_key(&value) {
            *self.values.get_mut(&value).unwrap() += 1;
        } else {
            self.values.insert(value, 1);
        }
    }

    /// Return the number of times the given value has been seen.
    fn times_seen(&self, value: u32) -> u64 {
        self.values.get(&value).copied().unwrap_or_default()
    }
}

fn main() {
    let mut ctr = Counter::new();
    ctr.count(13);
    ctr.count(14);
    ctr.count(16);
    ctr.count(14);
    ctr.count(14);
    ctr.count(11);

    for i in 10..20 {
        println!("saw {} values equal to {}", ctr.times_seen(i), i);
    }

    let mut strctr = Counter::new();
    strctr.count("apple");
    strctr.count("orange");
    strctr.count("apple");
    println!("got {} apples", strctr.times_seen("apple"));
}

解答

use std::collections::HashMap;
use std::hash::Hash;

/// Counter counts the number of times each value of type T has been seen.
struct Counter<T> {
    values: HashMap<T, u64>,
}

impl<T: Eq + Hash> Counter<T> {
    /// Create a new Counter.
    fn new() -> Self {
        Counter { values: HashMap::new() }
    }

    /// Count an occurrence of the given value.
    fn count(&mut self, value: T) {
        *self.values.entry(value).or_default() += 1;
    }

    /// Return the number of times the given value has been seen.
    fn times_seen(&self, value: T) -> u64 {
        self.values.get(&value).copied().unwrap_or_default()
    }
}

fn main() {
    let mut ctr = Counter::new();
    ctr.count(13);
    ctr.count(14);
    ctr.count(16);
    ctr.count(14);
    ctr.count(14);
    ctr.count(11);

    for i in 10..20 {
        println!("saw {} values equal to {}", ctr.times_seen(i), i);
    }

    let mut strctr = Counter::new();
    strctr.count("apple");
    strctr.count("orange");
    strctr.count("apple");
    println!("got {} apples", strctr.times_seen("apple"));
}

标准库特征

This segment should take about 1 hour and 40 minutes. It contains:

比较

这些 trait 支持在值之间进行比较。对于包含实现这些 trait 的字段，可以派生所有这些 trait。

PartialEq and Eq

PartialEq 指部分等价关系，其中包含必需的方法 eq 和提供的方法 ne。== 和 != 运算符会调用这些方法。

struct Key {
    id: u32,
    metadata: Option<String>,
}
impl PartialEq for Key {
    fn eq(&self, other: &Self) -> bool {
        self.id == other.id
    }
}

Eq is a full equivalence relation (reflexive, symmetric, and transitive) and implies PartialEq. Functions that require full equivalence will use Eq as a trait bound.

PartialOrd and Ord

PartialOrd 定义了使用 partial_cmp 方法的部分排序。它用于实现 <、<=、>= 和 > 运算符。

use std::cmp::Ordering;
#[derive(Eq, PartialEq)]
struct Citation {
    author: String,
    year: u32,
}
impl PartialOrd for Citation {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        match self.author.partial_cmp(&other.author) {
            Some(Ordering::Equal) => self.year.partial_cmp(&other.year),
            author_ord => author_ord,
        }
    }
}

Ord 是总排序，其中 cmp 返回 Ordering。

struct Key {
    id: u32,
    metadata: Option<String>,
}
impl PartialEq<u32> for Key {
    fn eq(&self, other: &u32) -> bool {
        self.id == *other
    }
}

运算符

运算符重载是通过 std::ops 中的特征实现的：

#[derive(Debug, Copy, Clone)]
struct Point {
    x: i32,
    y: i32,
}

impl std::ops::Add for Point {
    type Output = Self;

    fn add(self, other: Self) -> Self {
        Self { x: self.x + other.x, y: self.y + other.y }
    }
}

fn main() {
    let p1 = Point { x: 10, y: 20 };
    let p2 = Point { x: 100, y: 200 };
    println!("{:?} + {:?} = {:?}", p1, p2, p1 + p2);
}

From 和 Into

类型会实现 From 和 Into 以加快类型转换：

fn main() {
    let s = String::from("hello");
    let addr = std::net::Ipv4Addr::from([127, 0, 0, 1]);
    let one = i16::from(true);
    let bigger = i32::from(123_i16);
    println!("{s}, {addr}, {one}, {bigger}");
}

实现 From 后，系统会自动实现 Into：

fn main() {
    let s: String = "hello".into();
    let addr: std::net::Ipv4Addr = [127, 0, 0, 1].into();
    let one: i16 = true.into();
    let bigger: i32 = 123_i16.into();
    println!("{s}, {addr}, {one}, {bigger}");
}

类型转换

Rust 没有隐式类型转换，但支持使用 as 进行显式转换。转换写法通常和 C 语言的写法相一致。

fn main() {
    let value: i64 = 1000;
    println!("as u16: {}", value as u16);
    println!("as i16: {}", value as i16);
    println!("as u8: {}", value as u8);
}

使用 as 的结果在 Rust 中 始终 定义明确，并且在不同平台上保持一致。这可能和您对于更改符号或转换为更小类型的直观理解不一样。为清晰起见，请查看文档和注释。

使用 as 进行类型转换是一种快捷好用但容易出错的方法，也在未来进行代码更新时，比如改变了类型或类型范围时会导致潜伏的 Bug。 类型转换最好是用于明确是要进行无条件截取（比如截取低位的 u64 为 u32，忽略高位的数值）。

对于绝对不会出错的转换（比如 u32 转 u64），相比 as，更推荐用 From 或 Into 以肯定该转换是不会出错的。对于可能出错的转换，如果你想对这些不能成功转换的情况有不同处理方案时，可以考虑 TryFrom 和 TryInto。

Read 和 Write

您可以使用 Read 和 BufRead 对 u8 来源进行抽象化处理：

use std::io::{BufRead, BufReader, Read, Result};

fn count_lines<R: Read>(reader: R) -> usize {
    let buf_reader = BufReader::new(reader);
    buf_reader.lines().count()
}

fn main() -> Result<()> {
    let slice: &[u8] = b"foo\nbar\nbaz\n";
    println!("lines in slice: {}", count_lines(slice));

    let file = std::fs::File::open(std::env::current_exe()?)?;
    println!("lines in file: {}", count_lines(file));
    Ok(())
}

您同样可使用 Write 对 u8 接收器进行抽象化处理：

use std::io::{Result, Write};

fn log<W: Write>(writer: &mut W, msg: &str) -> Result<()> {
    writer.write_all(msg.as_bytes())?;
    writer.write_all("\n".as_bytes())
}

fn main() -> Result<()> {
    let mut buffer = Vec::new();
    log(&mut buffer, "Hello")?;
    log(&mut buffer, "World")?;
    println!("Logged: {:?}", buffer);
    Ok(())
}

Default 特征

Default 特征会为类型生成默认值。

#[derive(Debug, Default)]
struct Derived {
    x: u32,
    y: String,
    z: Implemented,
}

#[derive(Debug)]
struct Implemented(String);

impl Default for Implemented {
    fn default() -> Self {
        Self("John Smith".into())
    }
}

fn main() {
    let default_struct = Derived::default();
    println!("{default_struct:#?}");

    let almost_default_struct =
        Derived { y: "Y is set!".into(), ..Derived::default() };
    println!("{almost_default_struct:#?}");

    let nothing: Option<Derived> = None;
    println!("{:#?}", nothing.unwrap_or_default());
}

闭包

闭包或 lambda 表达式具有无法命名的类型。不过，它们会 实现特殊的 Fn， FnMut 和 FnOnce 特征：

fn apply_with_log(func: impl FnOnce(i32) -> i32, input: i32) -> i32 {
    println!("Calling function on {input}");
    func(input)
}

fn main() {
    let add_3 = |x| x + 3;
    println!("add_3: {}", apply_with_log(add_3, 10));
    println!("add_3: {}", apply_with_log(add_3, 20));

    let mut v = Vec::new();
    let mut accumulate = |x: i32| {
        v.push(x);
        v.iter().sum::<i32>()
    };
    println!("accumulate: {}", apply_with_log(&mut accumulate, 4));
    println!("accumulate: {}", apply_with_log(&mut accumulate, 5));

    let multiply_sum = |x| x * v.into_iter().sum::<i32>();
    println!("multiply_sum: {}", apply_with_log(multiply_sum, 3));
}

fn make_greeter(prefix: String) -> impl Fn(&str) {
    return move |name| println!("{} {}", prefix, name);
}

fn main() {
    let hi = make_greeter("Hi".to_string());
    hi("Greg");
}

练习：ROT13

在此示例中，您将实现经典的 “ROT13”加密。将此代码复制到 Playground，并实现缺失的位。请仅旋转 ASCII 字母字符，以确保结果仍为有效的 UTF-8 编码字符。

use std::io::Read;

struct RotDecoder<R: Read> {
    input: R,
    rot: u8,
}

// Implement the `Read` trait for `RotDecoder`.

fn main() {
    let mut rot =
        RotDecoder { input: "Gb trg gb gur bgure fvqr!".as_bytes(), rot: 13 };
    let mut result = String::new();
    rot.read_to_string(&mut result).unwrap();
    println!("{}", result);
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn joke() {
        let mut rot =
            RotDecoder { input: "Gb trg gb gur bgure fvqr!".as_bytes(), rot: 13 };
        let mut result = String::new();
        rot.read_to_string(&mut result).unwrap();
        assert_eq!(&result, "To get to the other side!");
    }

    #[test]
    fn binary() {
        let input: Vec<u8> = (0..=255u8).collect();
        let mut rot = RotDecoder::<&[u8]> { input: input.as_ref(), rot: 13 };
        let mut buf = [0u8; 256];
        assert_eq!(rot.read(&mut buf).unwrap(), 256);
        for i in 0..=255 {
            if input[i] != buf[i] {
                assert!(input[i].is_ascii_alphabetic());
                assert!(buf[i].is_ascii_alphabetic());
            }
        }
    }
}

如果将两个 RotDecoder 实例链接在一起，每个实例旋转 13 个字符，会发生什么情况？

解答

use std::io::Read;

struct RotDecoder<R: Read> {
    input: R,
    rot: u8,
}

impl<R: Read> Read for RotDecoder<R> {
    fn read(&mut self, buf: &mut [u8]) -> std::io::Result<usize> {
        let size = self.input.read(buf)?;
        for b in &mut buf[..size] {
            if b.is_ascii_alphabetic() {
                let base = if b.is_ascii_uppercase() { 'A' } else { 'a' } as u8;
                *b = (*b - base + self.rot) % 26 + base;
            }
        }
        Ok(size)
    }
}

fn main() {
    let mut rot =
        RotDecoder { input: "Gb trg gb gur bgure fvqr!".as_bytes(), rot: 13 };
    let mut result = String::new();
    rot.read_to_string(&mut result).unwrap();
    println!("{}", result);
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn joke() {
        let mut rot =
            RotDecoder { input: "Gb trg gb gur bgure fvqr!".as_bytes(), rot: 13 };
        let mut result = String::new();
        rot.read_to_string(&mut result).unwrap();
        assert_eq!(&result, "To get to the other side!");
    }

    #[test]
    fn binary() {
        let input: Vec<u8> = (0..=255u8).collect();
        let mut rot = RotDecoder::<&[u8]> { input: input.as_ref(), rot: 13 };
        let mut buf = [0u8; 256];
        assert_eq!(rot.read(&mut buf).unwrap(), 256);
        for i in 0..=255 {
            if input[i] != buf[i] {
                assert!(input[i].is_ascii_alphabetic());
                assert!(buf[i].is_ascii_alphabetic());
            }
        }
    }
}

欢迎参加第 3 天的课程

今日内容：

• 内存管理、生命周期和借用检查器：Rust 如何确保内存安全。
• 智能指针：标准库指针类型。

时间表

Including 10 minute breaks, this session should take about 2 hours and 20 minutes. It contains:

内存管理

This segment should take about 1 hour. It contains:

回顾：程序的内存分配

程序通过以下两种方式分配内存：

• 栈：局部变量的连续内存区域。

  • 值在编译时具有已知的固定大小。
  • 速度极快：只需移动一个栈指针。
  • 易于管理：遵循函数调用规则。
  • 优秀的内存局部性。

• 堆：函数调用之外的值的存储。

  • 值具有动态大小，具体大小需在运行时确定。
  • 比栈稍慢：需要向系统申请空间。
  • 不保证内存局部性。

示例

Creating a String puts fixed-sized metadata on the stack and dynamically sized data, the actual string, on the heap:

fn main() {
    let s1 = String::from("Hello");
}

fn main() {
    let mut s1 = String::from("Hello");
    s1.push(' ');
    s1.push_str("world");
    // DON'T DO THIS AT HOME! For educational purposes only.
    // String provides no guarantees about its layout, so this could lead to
    // undefined behavior.
    unsafe {
        let (capacity, ptr, len): (usize, usize, usize) = std::mem::transmute(s1);
        println!("capacity = {capacity}, ptr = {ptr:#x}, len = {len}");
    }
}

内存管理方法

传统上，语言分为两大类：

• 通过手动内存管理实现完全控制：C、C++、Pascal…
  • 程序员决定何时分配或释放堆内存。
  • 程序员必须确定指针是否仍指向有效内存。
  • 研究表明，程序员难免会犯错。
• 运行时通过自动内存管理实现完全安全：Java、Python、Go、Haskell…
  • 运行时系统可确保在内存无法被引用之前，不会释放该内存。
  • 通常通过引用计数、垃圾回收或 RAII 实现。

Rust 提供了一个全新的组合：

通过编译时强制执行正确的内存>管理来实现完全控制与安全。

它通过一个明确的所有权（ownership）概念来实现此目的。

所有权

所有变量绑定都有一个有效的“作用域”，使用 超出其作用域的变量是错误的：

struct Point(i32, i32);

fn main() {
    {
        let p = Point(3, 4);
        println!("x: {}", p.0);
    }
    println!("y: {}", p.1);
}

We say that the variable owns the value. Every Rust value has precisely one owner at all times.

At the end of the scope, the variable is dropped and the data is freed. A destructor can run here to free up resources.

移动语义

An assignment will transfer ownership between variables:

fn main() {
    let s1: String = String::from("Hello!");
    let s2: String = s1;
    println!("s2: {s2}");
    // println!("s1: {s1}");
}

• 将 s1 赋值给 s2，即转移了所有权。
• When s1 goes out of scope, nothing happens: it does not own anything.
• 当 s2 离开作用域时，字符串数据被释放。

移动到 s2 中之前：

移动到 s2 中之后：

你将值传递给函数时，该值会被赋给函数 参数。这就转移了所有权：

fn say_hello(name: String) {
    println!("Hello {name}")
}

fn main() {
    let name = String::from("Alice");
    say_hello(name);
    // say_hello(name);
}

std::string s1 = "Cpp";
std::string s2 = s1;  // Duplicate the data in s1.

Clone

有时，如需 复制某个值。Clone 特征 可以完成此操作。

#[derive(Default)]
struct Backends {
    hostnames: Vec<String>,
    weights: Vec<f64>,
}

impl Backends {
    fn set_hostnames(&mut self, hostnames: &Vec<String>) {
        self.hostnames = hostnames.clone();
        self.weights = hostnames.iter().map(|_| 1.0).collect();
    }
}

复合类型

虽然移动语义是默认的，但默认情况下会复制某些类型：

fn main() {
    let x = 42;
    let y = x;
    println!("x: {x}"); // would not be accessible if not Copy
    println!("y: {y}");
}

这些类型实现了 Copy trait。

你可以选择自己的类型来使用复制语义：

#[derive(Copy, Clone, Debug)]
struct Point(i32, i32);

fn main() {
    let p1 = Point(3, 4);
    let p2 = p1;
    println!("p1: {p1:?}");
    println!("p2: {p2:?}");
}

• 赋值之后，p1 和 p2 都拥有自己的数据。
• 我们还可以使用 p1.clone() 显式复制数据。

Drop 特征

用于实现 Drop 的值可以指定在超出范围时运行的代码：

struct Droppable {
    name: &'static str,
}

impl Drop for Droppable {
    fn drop(&mut self) {
        println!("Dropping {}", self.name);
    }
}

fn main() {
    let a = Droppable { name: "a" };
    {
        let b = Droppable { name: "b" };
        {
            let c = Droppable { name: "c" };
            let d = Droppable { name: "d" };
            println!("Exiting block B");
        }
        println!("Exiting block A");
    }
    drop(a);
    println!("Exiting main");
}

练习：构建器类型

在此示例中，我们将实现一个拥有全部数据所有权的复杂数据类型。我们将使用 “构建器模式”来支持逐步构建新值，通过便捷函数来实现。

填补缺失的内容。

#[derive(Debug)]
enum Language {
    Rust,
    Java,
    Perl,
}

#[derive(Clone, Debug)]
struct Dependency {
    name: String,
    version_expression: String,
}

/// A representation of a software package.
#[derive(Debug)]
struct Package {
    name: String,
    version: String,
    authors: Vec<String>,
    dependencies: Vec<Dependency>,
    language: Option<Language>,
}

impl Package {
    /// Return a representation of this package as a dependency, for use in
    /// building other packages.
    fn as_dependency(&self) -> Dependency {
        todo!("1")
    }
}

/// A builder for a Package. Use `build()` to create the `Package` itself.
struct PackageBuilder(Package);

impl PackageBuilder {
    fn new(name: impl Into<String>) -> Self {
        todo!("2")
    }

    /// Set the package version.
    fn version(mut self, version: impl Into<String>) -> Self {
        self.0.version = version.into();
        self
    }

    /// Set the package authors.
    fn authors(mut self, authors: Vec<String>) -> Self {
        todo!("3")
    }

    /// Add an additional dependency.
    fn dependency(mut self, dependency: Dependency) -> Self {
        todo!("4")
    }

    /// Set the language. If not set, language defaults to None.
    fn language(mut self, language: Language) -> Self {
        todo!("5")
    }

    fn build(self) -> Package {
        self.0
    }
}

fn main() {
    let base64 = PackageBuilder::new("base64").version("0.13").build();
    println!("base64: {base64:?}");
    let log =
        PackageBuilder::new("log").version("0.4").language(Language::Rust).build();
    println!("log: {log:?}");
    let serde = PackageBuilder::new("serde")
        .authors(vec!["djmitche".into()])
        .version(String::from("4.0"))
        .dependency(base64.as_dependency())
        .dependency(log.as_dependency())
        .build();
    println!("serde: {serde:?}");
}

解答

#[derive(Debug)]
enum Language {
    Rust,
    Java,
    Perl,
}

#[derive(Clone, Debug)]
struct Dependency {
    name: String,
    version_expression: String,
}

/// A representation of a software package.
#[derive(Debug)]
struct Package {
    name: String,
    version: String,
    authors: Vec<String>,
    dependencies: Vec<Dependency>,
    language: Option<Language>,
}

impl Package {
    /// Return a representation of this package as a dependency, for use in
    /// building other packages.
    fn as_dependency(&self) -> Dependency {
        Dependency {
            name: self.name.clone(),
            version_expression: self.version.clone(),
        }
    }
}

/// A builder for a Package. Use `build()` to create the `Package` itself.
struct PackageBuilder(Package);

impl PackageBuilder {
    fn new(name: impl Into<String>) -> Self {
        Self(Package {
            name: name.into(),
            version: "0.1".into(),
            authors: vec![],
            dependencies: vec![],
            language: None,
        })
    }

    /// Set the package version.
    fn version(mut self, version: impl Into<String>) -> Self {
        self.0.version = version.into();
        self
    }

    /// Set the package authors.
    fn authors(mut self, authors: Vec<String>) -> Self {
        self.0.authors = authors;
        self
    }

    /// Add an additional dependency.
    fn dependency(mut self, dependency: Dependency) -> Self {
        self.0.dependencies.push(dependency);
        self
    }

    /// Set the language. If not set, language defaults to None.
    fn language(mut self, language: Language) -> Self {
        self.0.language = Some(language);
        self
    }

    fn build(self) -> Package {
        self.0
    }
}

fn main() {
    let base64 = PackageBuilder::new("base64").version("0.13").build();
    println!("base64: {base64:?}");
    let log =
        PackageBuilder::new("log").version("0.4").language(Language::Rust).build();
    println!("log: {log:?}");
    let serde = PackageBuilder::new("serde")
        .authors(vec!["djmitche".into()])
        .version(String::from("4.0"))
        .dependency(base64.as_dependency())
        .dependency(log.as_dependency())
        .build();
    println!("serde: {serde:?}");
}

智能指针

This segment should take about 55 minutes. It contains:

Box<T>

Box 是指向堆上数据的自有指针：

fn main() {
    let five = Box::new(5);
    println!("five: {}", *five);
}

Box<T> 会实现 Deref<Target = T>，这意味着您可以直接在 Box<T> 上通过 T 调用相应方法。

递归数据类型或具有动态大小的数据类型需要使用 Box：

#[derive(Debug)]
enum List<T> {
    /// A non-empty list: first element and the rest of the list.
    Element(T, Box<List<T>>),
    /// An empty list.
    Nil,
}

fn main() {
    let list: List<i32> =
        List::Element(1, Box::new(List::Element(2, Box::new(List::Nil))));
    println!("{list:?}");
}

use std::mem::size_of_val;

struct Item(String);

fn main() {
    let just_box: Box<Item> = Box::new(Item("Just box".into()));
    let optional_box: Option<Box<Item>> =
        Some(Box::new(Item("Optional box".into())));
    let none: Option<Box<Item>> = None;

    assert_eq!(size_of_val(&just_box), size_of_val(&optional_box));
    assert_eq!(size_of_val(&just_box), size_of_val(&none));

    println!("Size of just_box: {}", size_of_val(&just_box));
    println!("Size of optional_box: {}", size_of_val(&optional_box));
    println!("Size of none: {}", size_of_val(&none));
}

Rc

Rc 是引用计数的共享指针。如果您需要从多个位置 引用相同的数据，请使用此指针：

use std::rc::Rc;

fn main() {
    let a = Rc::new(10);
    let b = Rc::clone(&a);

    println!("a: {a}");
    println!("b: {b}");
}

• See Arc and Mutex if you are in a multi-threaded context.
• 您可以将共享指针_降级_为 Weak 指针， 以便创建之后会被舍弃的循环引用。

特征对象

特征（Trait）对象可接受不同类型的值，举例来说，在集合中会是这样：

struct Dog {
    name: String,
    age: i8,
}
struct Cat {
    lives: i8,
}

trait Pet {
    fn talk(&self) -> String;
}

impl Pet for Dog {
    fn talk(&self) -> String {
        format!("Woof, my name is {}!", self.name)
    }
}

impl Pet for Cat {
    fn talk(&self) -> String {
        String::from("Miau!")
    }
}

fn main() {
    let pets: Vec<Box<dyn Pet>> = vec![
        Box::new(Cat { lives: 9 }),
        Box::new(Dog { name: String::from("Fido"), age: 5 }),
    ];
    for pet in pets {
        println!("Hello, who are you? {}", pet.talk());
    }
}

以下是分配 pets 后的内存布局：

练习：二叉树

二元树是一种树型数据结构，其中每个节点都有两个子节点（左侧和右侧）。我们将创建一个树状结构，其中每个节点存储一个值。对于给定的节点 N，N 的左侧子树中的所有节点都包含较小的值，而 N 的右侧子树中的所有节点都将包含较大的值。

实现以下类型，以便通过指定的测试。

额外提示：对按顺序返回值的二元树实现迭代器。

/// A node in the binary tree.
#[derive(Debug)]
struct Node<T: Ord> {
    value: T,
    left: Subtree<T>,
    right: Subtree<T>,
}

/// A possibly-empty subtree.
#[derive(Debug)]
struct Subtree<T: Ord>(Option<Box<Node<T>>>);

/// A container storing a set of values, using a binary tree.
///
/// If the same value is added multiple times, it is only stored once.
#[derive(Debug)]
pub struct BinaryTree<T: Ord> {
    root: Subtree<T>,
}

// Implement `new`, `insert`, `len`, and `has`.

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn len() {
        let mut tree = BinaryTree::new();
        assert_eq!(tree.len(), 0);
        tree.insert(2);
        assert_eq!(tree.len(), 1);
        tree.insert(1);
        assert_eq!(tree.len(), 2);
        tree.insert(2); // not a unique item
        assert_eq!(tree.len(), 2);
    }

    #[test]
    fn has() {
        let mut tree = BinaryTree::new();
        fn check_has(tree: &BinaryTree<i32>, exp: &[bool]) {
            let got: Vec<bool> =
                (0..exp.len()).map(|i| tree.has(&(i as i32))).collect();
            assert_eq!(&got, exp);
        }

        check_has(&tree, &[false, false, false, false, false]);
        tree.insert(0);
        check_has(&tree, &[true, false, false, false, false]);
        tree.insert(4);
        check_has(&tree, &[true, false, false, false, true]);
        tree.insert(4);
        check_has(&tree, &[true, false, false, false, true]);
        tree.insert(3);
        check_has(&tree, &[true, false, false, true, true]);
    }

    #[test]
    fn unbalanced() {
        let mut tree = BinaryTree::new();
        for i in 0..100 {
            tree.insert(i);
        }
        assert_eq!(tree.len(), 100);
        assert!(tree.has(&50));
    }
}

解答

use std::cmp::Ordering;

/// A node in the binary tree.
#[derive(Debug)]
struct Node<T: Ord> {
    value: T,
    left: Subtree<T>,
    right: Subtree<T>,
}

/// A possibly-empty subtree.
#[derive(Debug)]
struct Subtree<T: Ord>(Option<Box<Node<T>>>);

/// A container storing a set of values, using a binary tree.
///
/// If the same value is added multiple times, it is only stored once.
#[derive(Debug)]
pub struct BinaryTree<T: Ord> {
    root: Subtree<T>,
}

impl<T: Ord> BinaryTree<T> {
    fn new() -> Self {
        Self { root: Subtree::new() }
    }

    fn insert(&mut self, value: T) {
        self.root.insert(value);
    }

    fn has(&self, value: &T) -> bool {
        self.root.has(value)
    }

    fn len(&self) -> usize {
        self.root.len()
    }
}

impl<T: Ord> Subtree<T> {
    fn new() -> Self {
        Self(None)
    }

    fn insert(&mut self, value: T) {
        match &mut self.0 {
            None => self.0 = Some(Box::new(Node::new(value))),
            Some(n) => match value.cmp(&n.value) {
                Ordering::Less => n.left.insert(value),
                Ordering::Equal => {}
                Ordering::Greater => n.right.insert(value),
            },
        }
    }

    fn has(&self, value: &T) -> bool {
        match &self.0 {
            None => false,
            Some(n) => match value.cmp(&n.value) {
                Ordering::Less => n.left.has(value),
                Ordering::Equal => true,
                Ordering::Greater => n.right.has(value),
            },
        }
    }

    fn len(&self) -> usize {
        match &self.0 {
            None => 0,
            Some(n) => 1 + n.left.len() + n.right.len(),
        }
    }
}

impl<T: Ord> Node<T> {
    fn new(value: T) -> Self {
        Self { value, left: Subtree::new(), right: Subtree::new() }
    }
}

fn main() {
    let mut tree = BinaryTree::new();
    tree.insert("foo");
    assert_eq!(tree.len(), 1);
    tree.insert("bar");
    assert!(tree.has(&"foo"));
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn len() {
        let mut tree = BinaryTree::new();
        assert_eq!(tree.len(), 0);
        tree.insert(2);
        assert_eq!(tree.len(), 1);
        tree.insert(1);
        assert_eq!(tree.len(), 2);
        tree.insert(2); // not a unique item
        assert_eq!(tree.len(), 2);
    }

    #[test]
    fn has() {
        let mut tree = BinaryTree::new();
        fn check_has(tree: &BinaryTree<i32>, exp: &[bool]) {
            let got: Vec<bool> =
                (0..exp.len()).map(|i| tree.has(&(i as i32))).collect();
            assert_eq!(&got, exp);
        }

        check_has(&tree, &[false, false, false, false, false]);
        tree.insert(0);
        check_has(&tree, &[true, false, false, false, false]);
        tree.insert(4);
        check_has(&tree, &[true, false, false, false, true]);
        tree.insert(4);
        check_has(&tree, &[true, false, false, false, true]);
        tree.insert(3);
        check_has(&tree, &[true, false, false, true, true]);
    }

    #[test]
    fn unbalanced() {
        let mut tree = BinaryTree::new();
        for i in 0..100 {
            tree.insert(i);
        }
        assert_eq!(tree.len(), 100);
        assert!(tree.has(&50));
    }
}

Welcome Back

Including 10 minute breaks, this session should take about 1 hour and 50 minutes. It contains:

借用

This segment should take about 50 minutes. It contains:

借用值

As we saw before, instead of transferring ownership when calling a function, you can let a function borrow the value:

#[derive(Debug)]
struct Point(i32, i32);

fn add(p1: &Point, p2: &Point) -> Point {
    Point(p1.0 + p2.0, p1.1 + p2.1)
}

fn main() {
    let p1 = Point(3, 4);
    let p2 = Point(10, 20);
    let p3 = add(&p1, &p2);
    println!("{p1:?} + {p2:?} = {p3:?}");
}

• add 函数“借用”两个点并返回一个新点。
• 调用方会保留输入的所有权。

借用检查

Rust's borrow checker puts constraints on the ways you can borrow values. For a given value, at any time:

• You can have one or more shared references to the value, or
• You can have exactly one exclusive reference to the value.

fn main() {
    let mut a: i32 = 10;
    let b: &i32 = &a;

    {
        let c: &mut i32 = &mut a;
        *c = 20;
    }

    println!("a: {a}");
    println!("b: {b}");
}

内部可变性

In some situations, it's necessary to modify data behind a shared (read-only) reference. For example, a shared data structure might have an internal cache, and wish to update that cache from read-only methods.

The "interior mutability" pattern allows exclusive (mutable) access behind a shared reference. The standard library provides several ways to do this, all while still ensuring safety, typically by performing a runtime check.

RefCell

use std::cell::RefCell;
use std::rc::Rc;

#[derive(Debug, Default)]
struct Node {
    value: i64,
    children: Vec<Rc<RefCell<Node>>>,
}

impl Node {
    fn new(value: i64) -> Rc<RefCell<Node>> {
        Rc::new(RefCell::new(Node { value, ..Node::default() }))
    }

    fn sum(&self) -> i64 {
        self.value + self.children.iter().map(|c| c.borrow().sum()).sum::<i64>()
    }
}

fn main() {
    let root = Node::new(1);
    root.borrow_mut().children.push(Node::new(5));
    let subtree = Node::new(10);
    subtree.borrow_mut().children.push(Node::new(11));
    subtree.borrow_mut().children.push(Node::new(12));
    root.borrow_mut().children.push(subtree);

    println!("graph: {root:#?}");
    println!("graph sum: {}", root.borrow().sum());
}

Cell

Cell wraps a value and allows getting or setting the value, even with a shared reference to the Cell. However, it does not allow any references to the value. Since there are no references, borrowing rules cannot be broken.

练习：健康统计

你正在实现一个健康监控系统。作为其中的一部分，你需要对用户的健康统计数据进行追踪。

You'll start with a stubbed function in an impl block as well as a User struct definition. Your goal is to implement the stubbed out method on the User struct defined in the impl block.

Copy the code below to https://play.rust-lang.org/ and fill in the missing method:

// TODO: remove this when you're done with your implementation.
#![allow(unused_variables, dead_code)]


#![allow(dead_code)]
pub struct User {
    name: String,
    age: u32,
    height: f32,
    visit_count: usize,
    last_blood_pressure: Option<(u32, u32)>,
}

pub struct Measurements {
    height: f32,
    blood_pressure: (u32, u32),
}

pub struct HealthReport<'a> {
    patient_name: &'a str,
    visit_count: u32,
    height_change: f32,
    blood_pressure_change: Option<(i32, i32)>,
}

impl User {
    pub fn new(name: String, age: u32, height: f32) -> Self {
        Self { name, age, height, visit_count: 0, last_blood_pressure: None }
    }

    pub fn visit_doctor(&mut self, measurements: Measurements) -> HealthReport {
        todo!("Update a user's statistics based on measurements from a visit to the doctor")
    }
}

fn main() {
    let bob = User::new(String::from("Bob"), 32, 155.2);
    println!("I'm {} and my age is {}", bob.name, bob.age);
}

#[test]
fn test_visit() {
    let mut bob = User::new(String::from("Bob"), 32, 155.2);
    assert_eq!(bob.visit_count, 0);
    let report =
        bob.visit_doctor(Measurements { height: 156.1, blood_pressure: (120, 80) });
    assert_eq!(report.patient_name, "Bob");
    assert_eq!(report.visit_count, 1);
    assert_eq!(report.blood_pressure_change, None);

    let report =
        bob.visit_doctor(Measurements { height: 156.1, blood_pressure: (115, 76) });

    assert_eq!(report.visit_count, 2);
    assert_eq!(report.blood_pressure_change, Some((-5, -4)));
}

解答

#![allow(dead_code)]
pub struct User {
    name: String,
    age: u32,
    height: f32,
    visit_count: usize,
    last_blood_pressure: Option<(u32, u32)>,
}

pub struct Measurements {
    height: f32,
    blood_pressure: (u32, u32),
}

pub struct HealthReport<'a> {
    patient_name: &'a str,
    visit_count: u32,
    height_change: f32,
    blood_pressure_change: Option<(i32, i32)>,
}

impl User {
    pub fn new(name: String, age: u32, height: f32) -> Self {
        Self { name, age, height, visit_count: 0, last_blood_pressure: None }
    }

    pub fn visit_doctor(&mut self, measurements: Measurements) -> HealthReport {
        self.visit_count += 1;
        let bp = measurements.blood_pressure;
        let report = HealthReport {
            patient_name: &self.name,
            visit_count: self.visit_count as u32,
            height_change: measurements.height - self.height,
            blood_pressure_change: match self.last_blood_pressure {
                Some(lbp) => {
                    Some((bp.0 as i32 - lbp.0 as i32, bp.1 as i32 - lbp.1 as i32))
                }
                None => None,
            },
        };
        self.height = measurements.height;
        self.last_blood_pressure = Some(bp);
        report
    }
}

fn main() {
    let bob = User::new(String::from("Bob"), 32, 155.2);
    println!("I'm {} and my age is {}", bob.name, bob.age);
}

#[test]
fn test_visit() {
    let mut bob = User::new(String::from("Bob"), 32, 155.2);
    assert_eq!(bob.visit_count, 0);
    let report =
        bob.visit_doctor(Measurements { height: 156.1, blood_pressure: (120, 80) });
    assert_eq!(report.patient_name, "Bob");
    assert_eq!(report.visit_count, 1);
    assert_eq!(report.blood_pressure_change, None);

    let report =
        bob.visit_doctor(Measurements { height: 156.1, blood_pressure: (115, 76) });

    assert_eq!(report.visit_count, 2);
    assert_eq!(report.blood_pressure_change, Some((-5, -4)));
}

结构体生命周期

This segment should take about 50 minutes. It contains:

生命周期注解

A reference has a lifetime, which must not "outlive" the value it refers to. This is verified by the borrow checker.

The lifetime can be implicit - this is what we have seen so far. Lifetimes can also be explicit: &'a Point, &'document str. Lifetimes start with ' and 'a is a typical default name. Read &'a Point as "a borrowed Point which is valid for at least the lifetime a".

Lifetimes are always inferred by the compiler: you cannot assign a lifetime yourself. Explicit lifetime annotations create constraints where there is ambiguity; the compiler verifies that there is a valid solution.

当考虑向函数传递值和从函数返回值时，生命周期会变得更加复杂。

#[derive(Debug)]
struct Point(i32, i32);

fn left_most(p1: &Point, p2: &Point) -> &Point {
    if p1.0 < p2.0 {
        p1
    } else {
        p2
    }
}

fn main() {
    let p1: Point = Point(10, 10);
    let p2: Point = Point(20, 20);
    let p3 = left_most(&p1, &p2); // What is the lifetime of p3?
    println!("p3: {p3:?}");
}

fn left_most<'a>(p1: &'a Point, p2: &'a Point) -> &'a Point {

函数调用中的生命周期

Lifetimes for function arguments and return values must be fully specified, but Rust allows lifetimes to be elided in most cases with a few simple rules. This is not inference -- it is just a syntactic shorthand.

• 每个没有生命周期注解的参数都会添加一个生命周期注解。
• 如果只有一个参数生命周期，则将其赋予所有未加注解的返回值。
• 如果有多个参数生命周期，但第一个是用于 self 的，则将该生命周期赋予所有未加注解的返回值。

#[derive(Debug)]
struct Point(i32, i32);

fn cab_distance(p1: &Point, p2: &Point) -> i32 {
    (p1.0 - p2.0).abs() + (p1.1 - p2.1).abs()
}

fn nearest<'a>(points: &'a [Point], query: &Point) -> Option<&'a Point> {
    let mut nearest = None;
    for p in points {
        if let Some((_, nearest_dist)) = nearest {
            let dist = cab_distance(p, query);
            if dist < nearest_dist {
                nearest = Some((p, dist));
            }
        } else {
            nearest = Some((p, cab_distance(p, query)));
        };
    }
    nearest.map(|(p, _)| p)
}

fn main() {
    println!(
        "{:?}",
        nearest(
            &[Point(1, 0), Point(1, 0), Point(-1, 0), Point(0, -1),],
            &Point(0, 2)
        )
    );
}

fn nearest<'a, 'q>(points: &'a [Point], query: &'q Point) -> Option<&'q Point> {

数据结构中的生命周期

如果数据类型存储了借用的数据，则必须对其添加生命周期注释：

#[derive(Debug)]
struct Highlight<'doc>(&'doc str);

fn erase(text: String) {
    println!("Bye {text}!");
}

fn main() {
    let text = String::from("The quick brown fox jumps over the lazy dog.");
    let fox = Highlight(&text[4..19]);
    let dog = Highlight(&text[35..43]);
    // erase(text);
    println!("{fox:?}");
    println!("{dog:?}");
}

练习：Protobuf 解析

在本练习中，您将为 protobuf 二进制编码 构建一个解析器。别担心，其实非常简单！这展示了一种常见的解析模式，即传递数据 slice。底层数据本身永远不会被复制。

如要完整解析 protobuf 消息，需要知道字段的类型（按字段编号编入索引）。这通常会在 proto 文件中提供。在本练习中，我们将把这些信息编码成处理每个字段所调用的函数中的 match 语句。

我们将使用以下 proto：

message PhoneNumber {
  optional string number = 1;
  optional string type = 2;
}

message Person {
  optional string name = 1;
  optional int32 id = 2;
  repeated PhoneNumber phones = 3;
}

proto 消息被编码为连续的一系列字段。每个字段都通过 “标签”后面紧跟值的形式来实现。标签包含一个字段编号（例如Person 消息的 id 字段的值为 2）和线型（用于定义应如何从字节流确定载荷）。

整数（包括标签）使用名为 VARINT 的可变长度编码表示。幸运的是，下面为您提供了 parse_varint 的定义。该指定代码还定义了一些回调，用于处理 Person 和 PhoneNumber 字段，并将消息解析为对这些回调的一系列调用。

What remains for you is to implement the parse_field function and the ProtoMessage trait for Person and PhoneNumber.

use std::convert::TryFrom;
use thiserror::Error;

#[derive(Debug, Error)]
enum Error {
    #[error("Invalid varint")]
    InvalidVarint,
    #[error("Invalid wire-type")]
    InvalidWireType,
    #[error("Unexpected EOF")]
    UnexpectedEOF,
    #[error("Invalid length")]
    InvalidSize(#[from] std::num::TryFromIntError),
    #[error("Unexpected wire-type)")]
    UnexpectedWireType,
    #[error("Invalid string (not UTF-8)")]
    InvalidString,
}

/// A wire type as seen on the wire.
enum WireType {
    /// Varint WireType 表明该值为单个 VARINT。
    Varint,
    //I64,  -- not needed for this exercise
    /// The Len WireType indicates that the value is a length represented as a
    /// VARINT followed by exactly that number of bytes.
    Len,
    /// The I32 WireType indicates that the value is precisely 4 bytes in
    /// little-endian order containing a 32-bit signed integer.
    I32,
}

#[derive(Debug)]
/// A field's value, typed based on the wire type.
enum FieldValue<'a> {
    Varint(u64),
    //I64(i64),  -- not needed for this exercise
    Len(&'a [u8]),
    I32(i32),
}

#[derive(Debug)]
/// A field, containing the field number and its value.
struct Field<'a> {
    field_num: u64,
    value: FieldValue<'a>,
}

trait ProtoMessage<'a>: Default + 'a {
    fn add_field(&mut self, field: Field<'a>) -> Result<(), Error>;
}

impl TryFrom<u64> for WireType {
    type Error = Error;

    fn try_from(value: u64) -> Result<WireType, Error> {
        Ok(match value {
            0 => WireType::Varint,
            //1 => WireType::I64,  -- not needed for this exercise
            2 => WireType::Len,
            5 => WireType::I32,
            _ => return Err(Error::InvalidWireType),
        })
    }
}

impl<'a> FieldValue<'a> {
    fn as_string(&self) -> Result<&'a str, Error> {
        let FieldValue::Len(data) = self else {
            return Err(Error::UnexpectedWireType);
        };
        std::str::from_utf8(data).map_err(|_| Error::InvalidString)
    }

    fn as_bytes(&self) -> Result<&'a [u8], Error> {
        let FieldValue::Len(data) = self else {
            return Err(Error::UnexpectedWireType);
        };
        Ok(data)
    }

    fn as_u64(&self) -> Result<u64, Error> {
        let FieldValue::Varint(value) = self else {
            return Err(Error::UnexpectedWireType);
        };
        Ok(*value)
    }
}

/// Parse a VARINT, returning the parsed value and the remaining bytes.
fn parse_varint(data: &[u8]) -> Result<(u64, &[u8]), Error> {
    for i in 0..7 {
        let Some(b) = data.get(i) else {
            return Err(Error::InvalidVarint);
        };
        if b & 0x80 == 0 {
            // This is the last byte of the VARINT, so convert it to
            // a u64 and return it.
            let mut value = 0u64;
            for b in data[..=i].iter().rev() {
                value = (value << 7) | (b & 0x7f) as u64;
            }
            return Ok((value, &data[i + 1..]));
        }
    }

    // More than 7 bytes is invalid.
    Err(Error::InvalidVarint)
}

/// Convert a tag into a field number and a WireType.
fn unpack_tag(tag: u64) -> Result<(u64, WireType), Error> {
    let field_num = tag >> 3;
    let wire_type = WireType::try_from(tag & 0x7)?;
    Ok((field_num, wire_type))
}


/// Parse a field, returning the remaining bytes
fn parse_field(data: &[u8]) -> Result<(Field, &[u8]), Error> {
    let (tag, remainder) = parse_varint(data)?;
    let (field_num, wire_type) = unpack_tag(tag)?;
    let (fieldvalue, remainder) = match wire_type {
        _ => todo!("Based on the wire type, build a Field, consuming as many bytes as necessary.")
    };
    todo!("Return the field, and any un-consumed bytes.")
}

/// Parse a message in the given data, calling `T::add_field` for each field in
/// the message.
///
/// The entire input is consumed.
fn parse_message<'a, T: ProtoMessage<'a>>(mut data: &'a [u8]) -> Result<T, Error> {
    let mut result = T::default();
    while !data.is_empty() {
        let parsed = parse_field(data)?;
        result.add_field(parsed.0)?;
        data = parsed.1;
    }
    Ok(result)
}

#[derive(Debug, Default)]
struct PhoneNumber<'a> {
    number: &'a str,
    type_: &'a str,
}

#[derive(Debug, Default)]
struct Person<'a> {
    name: &'a str,
    id: u64,
    phone: Vec<PhoneNumber<'a>>,
}

// TODO: Implement ProtoMessage for Person and PhoneNumber.

fn main() {
    let person: Person = parse_message(&[
        0x0a, 0x07, 0x6d, 0x61, 0x78, 0x77, 0x65, 0x6c, 0x6c, 0x10, 0x2a, 0x1a,
        0x16, 0x0a, 0x0e, 0x2b, 0x31, 0x32, 0x30, 0x32, 0x2d, 0x35, 0x35, 0x35,
        0x2d, 0x31, 0x32, 0x31, 0x32, 0x12, 0x04, 0x68, 0x6f, 0x6d, 0x65, 0x1a,
        0x18, 0x0a, 0x0e, 0x2b, 0x31, 0x38, 0x30, 0x30, 0x2d, 0x38, 0x36, 0x37,
        0x2d, 0x35, 0x33, 0x30, 0x38, 0x12, 0x06, 0x6d, 0x6f, 0x62, 0x69, 0x6c,
        0x65,
    ])
    .unwrap();
    println!("{:#?}", person);
}

解答

use std::convert::TryFrom;
use thiserror::Error;

#[derive(Debug, Error)]
enum Error {
    #[error("Invalid varint")]
    InvalidVarint,
    #[error("Invalid wire-type")]
    InvalidWireType,
    #[error("Unexpected EOF")]
    UnexpectedEOF,
    #[error("Invalid length")]
    InvalidSize(#[from] std::num::TryFromIntError),
    #[error("Unexpected wire-type)")]
    UnexpectedWireType,
    #[error("Invalid string (not UTF-8)")]
    InvalidString,
}

/// A wire type as seen on the wire.
enum WireType {
    /// Varint WireType 表明该值为单个 VARINT。
    Varint,
    //I64,  -- not needed for this exercise
    /// The Len WireType indicates that the value is a length represented as a
    /// VARINT followed by exactly that number of bytes.
    Len,
    /// The I32 WireType indicates that the value is precisely 4 bytes in
    /// little-endian order containing a 32-bit signed integer.
    I32,
}

#[derive(Debug)]
/// A field's value, typed based on the wire type.
enum FieldValue<'a> {
    Varint(u64),
    //I64(i64),  -- not needed for this exercise
    Len(&'a [u8]),
    I32(i32),
}

#[derive(Debug)]
/// A field, containing the field number and its value.
struct Field<'a> {
    field_num: u64,
    value: FieldValue<'a>,
}

trait ProtoMessage<'a>: Default + 'a {
    fn add_field(&mut self, field: Field<'a>) -> Result<(), Error>;
}

impl TryFrom<u64> for WireType {
    type Error = Error;

    fn try_from(value: u64) -> Result<WireType, Error> {
        Ok(match value {
            0 => WireType::Varint,
            //1 => WireType::I64,  -- not needed for this exercise
            2 => WireType::Len,
            5 => WireType::I32,
            _ => return Err(Error::InvalidWireType),
        })
    }
}

impl<'a> FieldValue<'a> {
    fn as_string(&self) -> Result<&'a str, Error> {
        let FieldValue::Len(data) = self else {
            return Err(Error::UnexpectedWireType);
        };
        std::str::from_utf8(data).map_err(|_| Error::InvalidString)
    }

    fn as_bytes(&self) -> Result<&'a [u8], Error> {
        let FieldValue::Len(data) = self else {
            return Err(Error::UnexpectedWireType);
        };
        Ok(data)
    }

    fn as_u64(&self) -> Result<u64, Error> {
        let FieldValue::Varint(value) = self else {
            return Err(Error::UnexpectedWireType);
        };
        Ok(*value)
    }
}

/// Parse a VARINT, returning the parsed value and the remaining bytes.
fn parse_varint(data: &[u8]) -> Result<(u64, &[u8]), Error> {
    for i in 0..7 {
        let Some(b) = data.get(i) else {
            return Err(Error::InvalidVarint);
        };
        if b & 0x80 == 0 {
            // This is the last byte of the VARINT, so convert it to
            // a u64 and return it.
            let mut value = 0u64;
            for b in data[..=i].iter().rev() {
                value = (value << 7) | (b & 0x7f) as u64;
            }
            return Ok((value, &data[i + 1..]));
        }
    }

    // More than 7 bytes is invalid.
    Err(Error::InvalidVarint)
}

/// Convert a tag into a field number and a WireType.
fn unpack_tag(tag: u64) -> Result<(u64, WireType), Error> {
    let field_num = tag >> 3;
    let wire_type = WireType::try_from(tag & 0x7)?;
    Ok((field_num, wire_type))
}

/// Parse a field, returning the remaining bytes
fn parse_field(data: &[u8]) -> Result<(Field, &[u8]), Error> {
    let (tag, remainder) = parse_varint(data)?;
    let (field_num, wire_type) = unpack_tag(tag)?;
    let (fieldvalue, remainder) = match wire_type {
        WireType::Varint => {
            let (value, remainder) = parse_varint(remainder)?;
            (FieldValue::Varint(value), remainder)
        }
        WireType::Len => {
            let (len, remainder) = parse_varint(remainder)?;
            let len: usize = len.try_into()?;
            if remainder.len() < len {
                return Err(Error::UnexpectedEOF);
            }
            let (value, remainder) = remainder.split_at(len);
            (FieldValue::Len(value), remainder)
        }
        WireType::I32 => {
            if remainder.len() < 4 {
                return Err(Error::UnexpectedEOF);
            }
            let (value, remainder) = remainder.split_at(4);
            // Unwrap error because `value` is definitely 4 bytes long.
            let value = i32::from_le_bytes(value.try_into().unwrap());
            (FieldValue::I32(value), remainder)
        }
    };
    Ok((Field { field_num, value: fieldvalue }, remainder))
}

/// Parse a message in the given data, calling `T::add_field` for each field in
/// the message.
///
/// The entire input is consumed.
fn parse_message<'a, T: ProtoMessage<'a>>(mut data: &'a [u8]) -> Result<T, Error> {
    let mut result = T::default();
    while !data.is_empty() {
        let parsed = parse_field(data)?;
        result.add_field(parsed.0)?;
        data = parsed.1;
    }
    Ok(result)
}

#[derive(Debug, Default)]
struct PhoneNumber<'a> {
    number: &'a str,
    type_: &'a str,
}

#[derive(Debug, Default)]
struct Person<'a> {
    name: &'a str,
    id: u64,
    phone: Vec<PhoneNumber<'a>>,
}

impl<'a> ProtoMessage<'a> for Person<'a> {
    fn add_field(&mut self, field: Field<'a>) -> Result<(), Error> {
        match field.field_num {
            1 => self.name = field.value.as_string()?,
            2 => self.id = field.value.as_u64()?,
            3 => self.phone.push(parse_message(field.value.as_bytes()?)?),
            _ => {} // skip everything else
        }
        Ok(())
    }
}

impl<'a> ProtoMessage<'a> for PhoneNumber<'a> {
    fn add_field(&mut self, field: Field<'a>) -> Result<(), Error> {
        match field.field_num {
            1 => self.number = field.value.as_string()?,
            2 => self.type_ = field.value.as_string()?,
            _ => {} // skip everything else
        }
        Ok(())
    }
}

fn main() {
    let person: Person = parse_message(&[
        0x0a, 0x07, 0x6d, 0x61, 0x78, 0x77, 0x65, 0x6c, 0x6c, 0x10, 0x2a, 0x1a,
        0x16, 0x0a, 0x0e, 0x2b, 0x31, 0x32, 0x30, 0x32, 0x2d, 0x35, 0x35, 0x35,
        0x2d, 0x31, 0x32, 0x31, 0x32, 0x12, 0x04, 0x68, 0x6f, 0x6d, 0x65, 0x1a,
        0x18, 0x0a, 0x0e, 0x2b, 0x31, 0x38, 0x30, 0x30, 0x2d, 0x38, 0x36, 0x37,
        0x2d, 0x35, 0x33, 0x30, 0x38, 0x12, 0x06, 0x6d, 0x6f, 0x62, 0x69, 0x6c,
        0x65,
    ])
    .unwrap();
    println!("{:#?}", person);
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn as_string() {
        assert!(FieldValue::Varint(10).as_string().is_err());
        assert!(FieldValue::I32(10).as_string().is_err());
        assert_eq!(FieldValue::Len(b"hello").as_string().unwrap(), "hello");
    }

    #[test]
    fn as_bytes() {
        assert!(FieldValue::Varint(10).as_bytes().is_err());
        assert!(FieldValue::I32(10).as_bytes().is_err());
        assert_eq!(FieldValue::Len(b"hello").as_bytes().unwrap(), b"hello");
    }

    #[test]
    fn as_u64() {
        assert_eq!(FieldValue::Varint(10).as_u64().unwrap(), 10u64);
        assert!(FieldValue::I32(10).as_u64().is_err());
        assert!(FieldValue::Len(b"hello").as_u64().is_err());
    }
}

Welcome to Day 4

Today we will cover topics relating to building large-scale software in Rust:

• 迭代器：深入了解 Iterator 特征。
• 模块和可见性。
• Testing.
• 错误处理：panic、“Result”和 try 运算符“?”。
• 不安全 Rust：当无法用安全 Rust 表达您的意图时，则可将其作为应急方法。

时间表

Including 10 minute breaks, this session should take about 2 hours and 40 minutes. It contains:

迭代器

This segment should take about 45 minutes. It contains:

Iterator

'Iterator' trait 支持迭代集合中的值。它需要用到 next 方法，并提供很多方法。许多标准库类型均能实现 Iterator，您也可以自行实现：

struct Fibonacci {
    curr: u32,
    next: u32,
}

impl Iterator for Fibonacci {
    type Item = u32;

    fn next(&mut self) -> Option<Self::Item> {
        let new_next = self.curr + self.next;
        self.curr = self.next;
        self.next = new_next;
        Some(self.curr)
    }
}

fn main() {
    let fib = Fibonacci { curr: 0, next: 1 };
    for (i, n) in fib.enumerate().take(5) {
        println!("fib({i}): {n}");
    }
}

IntoIterator

The Iterator trait tells you how to iterate once you have created an iterator. The related trait IntoIterator defines how to create an iterator for a type. It is used automatically by the for loop.

struct Grid {
    x_coords: Vec<u32>,
    y_coords: Vec<u32>,
}

impl IntoIterator for Grid {
    type Item = (u32, u32);
    type IntoIter = GridIter;
    fn into_iter(self) -> GridIter {
        GridIter { grid: self, i: 0, j: 0 }
    }
}

struct GridIter {
    grid: Grid,
    i: usize,
    j: usize,
}

impl Iterator for GridIter {
    type Item = (u32, u32);

    fn next(&mut self) -> Option<(u32, u32)> {
        if self.i >= self.grid.x_coords.len() {
            self.i = 0;
            self.j += 1;
            if self.j >= self.grid.y_coords.len() {
                return None;
            }
        }
        let res = Some((self.grid.x_coords[self.i], self.grid.y_coords[self.j]));
        self.i += 1;
        res
    }
}

fn main() {
    let grid = Grid { x_coords: vec![3, 5, 7, 9], y_coords: vec![10, 20, 30, 40] };
    for (x, y) in grid {
        println!("point = {x}, {y}");
    }
}

FromIterator

FromIterator 让您可通过 Iterator 构建一个集合。

fn main() {
    let primes = vec![2, 3, 5, 7];
    let prime_squares = primes.into_iter().map(|p| p * p).collect::<Vec<_>>();
    println!("prime_squares: {prime_squares:?}");
}

fn collect<B>(self) -> B
where
    B: FromIterator<Self::Item>,
    Self: Sized

练习：迭代器方法链

In this exercise, you will need to find and use some of the provided methods in the Iterator trait to implement a complex calculation.

Copy the following code to https://play.rust-lang.org/ and make the tests pass. Use an iterator expression and collect the result to construct the return value.

#![allow(unused)]
fn main() {
/// Calculate the differences between elements of `values` offset by `offset`,
/// wrapping around from the end of `values` to the beginning.
///
/// Element `n` of the result is `values[(n+offset)%len] - values[n]`.
fn offset_differences<N>(offset: usize, values: Vec<N>) -> Vec<N>
where
    N: Copy + std::ops::Sub<Output = N>,
{
    unimplemented!()
}

#[test]
fn test_offset_one() {
    assert_eq!(offset_differences(1, vec![1, 3, 5, 7]), vec![2, 2, 2, -6]);
    assert_eq!(offset_differences(1, vec![1, 3, 5]), vec![2, 2, -4]);
    assert_eq!(offset_differences(1, vec![1, 3]), vec![2, -2]);
}

#[test]
fn test_larger_offsets() {
    assert_eq!(offset_differences(2, vec![1, 3, 5, 7]), vec![4, 4, -4, -4]);
    assert_eq!(offset_differences(3, vec![1, 3, 5, 7]), vec![6, -2, -2, -2]);
    assert_eq!(offset_differences(4, vec![1, 3, 5, 7]), vec![0, 0, 0, 0]);
    assert_eq!(offset_differences(5, vec![1, 3, 5, 7]), vec![2, 2, 2, -6]);
}

#[test]
fn test_custom_type() {
    assert_eq!(
        offset_differences(1, vec![1.0, 11.0, 5.0, 0.0]),
        vec![10.0, -6.0, -5.0, 1.0]
    );
}

#[test]
fn test_degenerate_cases() {
    assert_eq!(offset_differences(1, vec![0]), vec![0]);
    assert_eq!(offset_differences(1, vec![1]), vec![0]);
    let empty: Vec<i32> = vec![];
    assert_eq!(offset_differences(1, empty), vec![]);
}
}

解答

/// Calculate the differences between elements of `values` offset by `offset`,
/// wrapping around from the end of `values` to the beginning.
///
/// Element `n` of the result is `values[(n+offset)%len] - values[n]`.
fn offset_differences<N>(offset: usize, values: Vec<N>) -> Vec<N>
where
    N: Copy + std::ops::Sub<Output = N>,
{
    let a = (&values).into_iter();
    let b = (&values).into_iter().cycle().skip(offset);
    a.zip(b).map(|(a, b)| *b - *a).collect()
}

#[test]
fn test_offset_one() {
    assert_eq!(offset_differences(1, vec![1, 3, 5, 7]), vec![2, 2, 2, -6]);
    assert_eq!(offset_differences(1, vec![1, 3, 5]), vec![2, 2, -4]);
    assert_eq!(offset_differences(1, vec![1, 3]), vec![2, -2]);
}

#[test]
fn test_larger_offsets() {
    assert_eq!(offset_differences(2, vec![1, 3, 5, 7]), vec![4, 4, -4, -4]);
    assert_eq!(offset_differences(3, vec![1, 3, 5, 7]), vec![6, -2, -2, -2]);
    assert_eq!(offset_differences(4, vec![1, 3, 5, 7]), vec![0, 0, 0, 0]);
    assert_eq!(offset_differences(5, vec![1, 3, 5, 7]), vec![2, 2, 2, -6]);
}

#[test]
fn test_custom_type() {
    assert_eq!(
        offset_differences(1, vec![1.0, 11.0, 5.0, 0.0]),
        vec![10.0, -6.0, -5.0, 1.0]
    );
}

#[test]
fn test_degenerate_cases() {
    assert_eq!(offset_differences(1, vec![0]), vec![0]);
    assert_eq!(offset_differences(1, vec![1]), vec![0]);
    let empty: Vec<i32> = vec![];
    assert_eq!(offset_differences(1, empty), vec![]);
}

fn main() {}

模块

This segment should take about 40 minutes. It contains:

模块

我们已看了“impl”块如何让我们将函数的命名空间建为一种类型。

同样，“mod”让我们可为类型和函数建立命名空间：

mod foo {
    pub fn do_something() {
        println!("In the foo module");
    }
}

mod bar {
    pub fn do_something() {
        println!("In the bar module");
    }
}

fn main() {
    foo::do_something();
    bar::do_something();
}

文件系统层级结构

如果省略模块内容，则会指示 Rust 在另一个文件中查找：

mod garden;

这会告知 Rust 可以在“src/garden.rs”中找到“garden”模块内容。同样，您可以在“src/garden/vegetables.rs”中找到“garden::vegetables”模块。

“crate”根目录位于：

• “src/lib.rs”（对于库 crate）
• “src/main.rs”（对于二进制文件 crate）

也可以使用“内部文档注释”对文件中定义的模块进行记录。这些用于记录包含它们的项（在本例中为模块）。

//! This module implements the garden, including a highly performant germination
//! implementation.

// Re-export types from this module.
pub use garden::Garden;
pub use seeds::SeedPacket;

/// Sow the given seed packets.
pub fn sow(seeds: Vec<SeedPacket>) {
    todo!()
}

/// Harvest the produce in the garden that is ready.
pub fn harvest(garden: &mut Garden) {
    todo!()
}

可见性

模块是一种隐私边界：

• 默认情况下，模块项是私有的（隐藏实现详情）。
• 父项和同级子项始终可见。
• 换言之，如果某个项在模块“foo”中可见，那么该项在“foo”的所有后代中均可见。

mod outer {
    fn private() {
        println!("outer::private");
    }

    pub fn public() {
        println!("outer::public");
    }

    mod inner {
        fn private() {
            println!("outer::inner::private");
        }

        pub fn public() {
            println!("outer::inner::public");
            super::private();
        }
    }
}

fn main() {
    outer::public();
}

use、super、self

一个模块可以使用“use”将另一个模块的符号全部纳入。您通常在每个模块的顶部会看到如下内容：

use std::collections::HashSet;
use std::process::abort;

路径

路径解析如下：

1. 作为相对路径：

   • foo 或 self::foo 是指当前模块中的 foo，
   • “super::foo”是指父模块中的“foo”。

2. 作为绝对路径：

   • crate::foo 是指当前 crate 的根中的 foo，
   • “bar::foo”是指“bar”crate 中的“foo”。

练习：面向 GUI 库的模块

In this exercise, you will reorganize a small GUI Library implementation. This library defines a Widget trait and a few implementations of that trait, as well as a main function.

It is typical to put each type or set of closely-related types into its own module, so each widget type should get its own module.

Cargo Setup

Rust Playground 仅支持一个文件，因此您需要在本地文件系统上创建一个 Cargo 项目：

cargo init gui-modules
cd gui-modules
cargo run

Edit the resulting src/main.rs to add mod statements, and add additional files in the src directory.

Source

Here's the single-module implementation of the GUI library:

pub trait Widget {
    /// Natural width of `self`.
    fn width(&self) -> usize;

    /// Draw the widget into a buffer.
    fn draw_into(&self, buffer: &mut dyn std::fmt::Write);

    /// Draw the widget on standard output.
    fn draw(&self) {
        let mut buffer = String::new();
        self.draw_into(&mut buffer);
        println!("{buffer}");
    }
}

pub struct Label {
    label: String,
}

impl Label {
    fn new(label: &str) -> Label {
        Label { label: label.to_owned() }
    }
}

pub struct Button {
    label: Label,
}

impl Button {
    fn new(label: &str) -> Button {
        Button { label: Label::new(label) }
    }
}

pub struct Window {
    title: String,
    widgets: Vec<Box<dyn Widget>>,
}

impl Window {
    fn new(title: &str) -> Window {
        Window { title: title.to_owned(), widgets: Vec::new() }
    }

    fn add_widget(&mut self, widget: Box<dyn Widget>) {
        self.widgets.push(widget);
    }

    fn inner_width(&self) -> usize {
        std::cmp::max(
            self.title.chars().count(),
            self.widgets.iter().map(|w| w.width()).max().unwrap_or(0),
        )
    }
}

impl Widget for Window {
    fn width(&self) -> usize {
        // Add 4 paddings for borders
        self.inner_width() + 4
    }

    fn draw_into(&self, buffer: &mut dyn std::fmt::Write) {
        let mut inner = String::new();
        for widget in &self.widgets {
            widget.draw_into(&mut inner);
        }

        let inner_width = self.inner_width();

        // TODO: Change draw_into to return Result<(), std::fmt::Error>. Then use the
        // ?-operator here instead of .unwrap().
        writeln!(buffer, "+-{:-<inner_width$}-+", "").unwrap();
        writeln!(buffer, "| {:^inner_width$} |", &self.title).unwrap();
        writeln!(buffer, "+={:=<inner_width$}=+", "").unwrap();
        for line in inner.lines() {
            writeln!(buffer, "| {:inner_width$} |", line).unwrap();
        }
        writeln!(buffer, "+-{:-<inner_width$}-+", "").unwrap();
    }
}

impl Widget for Button {
    fn width(&self) -> usize {
        self.label.width() + 8 // add a bit of padding
    }

    fn draw_into(&self, buffer: &mut dyn std::fmt::Write) {
        let width = self.width();
        let mut label = String::new();
        self.label.draw_into(&mut label);

        writeln!(buffer, "+{:-<width$}+", "").unwrap();
        for line in label.lines() {
            writeln!(buffer, "|{:^width$}|", &line).unwrap();
        }
        writeln!(buffer, "+{:-<width$}+", "").unwrap();
    }
}

impl Widget for Label {
    fn width(&self) -> usize {
        self.label.lines().map(|line| line.chars().count()).max().unwrap_or(0)
    }

    fn draw_into(&self, buffer: &mut dyn std::fmt::Write) {
        writeln!(buffer, "{}", &self.label).unwrap();
    }
}

fn main() {
    let mut window = Window::new("Rust GUI Demo 1.23");
    window.add_widget(Box::new(Label::new("This is a small text GUI demo.")));
    window.add_widget(Box::new(Button::new("Click me!")));
    window.draw();
}

解答

src
├── main.rs
├── widgets
│   ├── button.rs
│   ├── label.rs
│   └── window.rs
└── widgets.rs

// ---- src/widgets.rs ----
mod button;
mod label;
mod window;

pub trait Widget {
    /// Natural width of `self`.
    fn width(&self) -> usize;

    /// Draw the widget into a buffer.
    fn draw_into(&self, buffer: &mut dyn std::fmt::Write);

    /// Draw the widget on standard output.
    fn draw(&self) {
        let mut buffer = String::new();
        self.draw_into(&mut buffer);
        println!("{buffer}");
    }
}

pub use button::Button;
pub use label::Label;
pub use window::Window;

// ---- src/widgets/label.rs ----
use super::Widget;

pub struct Label {
    label: String,
}

impl Label {
    pub fn new(label: &str) -> Label {
        Label { label: label.to_owned() }
    }
}

impl Widget for Label {
    fn width(&self) -> usize {
        // ANCHOR_END: Label-width
        self.label.lines().map(|line| line.chars().count()).max().unwrap_or(0)
    }

    // ANCHOR: Label-draw_into
    fn draw_into(&self, buffer: &mut dyn std::fmt::Write) {
        // ANCHOR_END: Label-draw_into
        writeln!(buffer, "{}", &self.label).unwrap();
    }
}

// ---- src/widgets/button.rs ----
use super::{Label, Widget};

pub struct Button {
    label: Label,
}

impl Button {
    pub fn new(label: &str) -> Button {
        Button { label: Label::new(label) }
    }
}

impl Widget for Button {
    fn width(&self) -> usize {
        // ANCHOR_END: Button-width
        self.label.width() + 8 // add a bit of padding
    }

    // ANCHOR: Button-draw_into
    fn draw_into(&self, buffer: &mut dyn std::fmt::Write) {
        // ANCHOR_END: Button-draw_into
        let width = self.width();
        let mut label = String::new();
        self.label.draw_into(&mut label);

        writeln!(buffer, "+{:-<width$}+", "").unwrap();
        for line in label.lines() {
            writeln!(buffer, "|{:^width$}|", &line).unwrap();
        }
        writeln!(buffer, "+{:-<width$}+", "").unwrap();
    }
}

// ---- src/widgets/window.rs ----
use super::Widget;

pub struct Window {
    title: String,
    widgets: Vec<Box<dyn Widget>>,
}

impl Window {
    pub fn new(title: &str) -> Window {
        Window { title: title.to_owned(), widgets: Vec::new() }
    }

    pub fn add_widget(&mut self, widget: Box<dyn Widget>) {
        self.widgets.push(widget);
    }

    fn inner_width(&self) -> usize {
        std::cmp::max(
            self.title.chars().count(),
            self.widgets.iter().map(|w| w.width()).max().unwrap_or(0),
        )
    }
}

impl Widget for Window {
    fn width(&self) -> usize {
        // ANCHOR_END: Window-width
        // Add 4 paddings for borders
        self.inner_width() + 4
    }

    // ANCHOR: Window-draw_into
    fn draw_into(&self, buffer: &mut dyn std::fmt::Write) {
        // ANCHOR_END: Window-draw_into
        let mut inner = String::new();
        for widget in &self.widgets {
            widget.draw_into(&mut inner);
        }

        let inner_width = self.inner_width();

        // TODO: after learning about error handling, you can change
        // draw_into to return Result<(), std::fmt::Error>. Then use
        // the ?-operator here instead of .unwrap().
        writeln!(buffer, "+-{:-<inner_width$}-+", "").unwrap();
        writeln!(buffer, "| {:^inner_width$} |", &self.title).unwrap();
        writeln!(buffer, "+={:=<inner_width$}=+", "").unwrap();
        for line in inner.lines() {
            writeln!(buffer, "| {:inner_width$} |", line).unwrap();
        }
        writeln!(buffer, "+-{:-<inner_width$}-+", "").unwrap();
    }
}

// ---- src/main.rs ----
mod widgets;

use widgets::Widget;

fn main() {
    let mut window = widgets::Window::new("Rust GUI Demo 1.23");
    window
        .add_widget(Box::new(widgets::Label::new("This is a small text GUI demo.")));
    window.add_widget(Box::new(widgets::Button::new("Click me!")));
    window.draw();
}

测试

This segment should take about 45 minutes. It contains:

单元测试

Rust 和 Cargo 随附了一个简单的单元测试框架：

• 单元测试在您的整个代码中都受支持。

• 您可以通过 tests/ 目录来支持集成测试。

Tests are marked with #[test]. Unit tests are often put in a nested tests module, using #[cfg(test)] to conditionally compile them only when building tests.

fn first_word(text: &str) -> &str {
    match text.find(' ') {
        Some(idx) => &text[..idx],
        None => &text,
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_empty() {
        assert_eq!(first_word(""), "");
    }

    #[test]
    fn test_single_word() {
        assert_eq!(first_word("Hello"), "Hello");
    }

    #[test]
    fn test_multiple_words() {
        assert_eq!(first_word("Hello World"), "Hello");
    }
}

• 这样一来，您可以对专用帮助程序进行单元测试。
• 仅当您运行 cargo test 时，#[cfg(test)] 属性才有效。

其他类型的测试

集成测试

如果您想要以客户的身份测试您的库，请使用集成测试。

在 tests/ 下方创建一个 .rs 文件：

// tests/my_library.rs
use my_library::init;

#[test]
fn test_init() {
    assert!(init().is_ok());
}

这些测试只能使用您的 crate 的公共 API。

文档测试

Rust 本身就支持文档测试：

#![allow(unused)]
fn main() {
/// Shortens a string to the given length.
///
/// ```
/// # use playground::shorten_string;
/// assert_eq!(shorten_string("Hello World", 5), "Hello");
/// assert_eq!(shorten_string("Hello World", 20), "Hello World");
/// ```
pub fn shorten_string(s: &str, length: usize) -> &str {
    &s[..std::cmp::min(length, s.len())]
}
}

• /// 注释中的代码块会自动被视为 Rust 代码。
• 代码会作为 cargo test 的一部分进行编译和执行。
• Adding # in the code will hide it from the docs, but will still compile/run it.
• 在 Rust Playground 上测试上述代码。

编译器 Lint 和 Clippy

Rust 编译器会生成出色的错误消息，并提供实用的内置 lint 功能。Clippy 提供了更多 lint 功能，采用按组分类方式，并可按项目灵活启用。

#[deny(clippy::cast_possible_truncation)]
fn main() {
    let x = 3;
    while (x < 70000) {
        x *= 2;
    }
    println!("X probably fits in a u16, right? {}", x as u16);
}

练习：卢恩算法

卢恩算法

卢恩算法用于验证信用卡号。该算法将字符串作为输入内容，并执行以下操作来验证信用卡号：

• Ignore all spaces. Reject number with fewer than two digits.

• 从右到左，将偶数位的数字乘二。对于数字“1234”，我们将“3”和“1”乘二；对于数字“98765”，将“6”和“8”乘二。

• 将一个数字乘二后，如果结果大于 9，则将每位数字相加。因此，将“7”乘二得“14”，然后“1 + 4 = 5”。

• 将所有未乘二和已乘二的数字相加。

• 如果总和以“0”结尾，则信用卡号有效。

提供的代码提供了一个有缺陷的 Luhn 算法实现，附带两个基本单元测试，用于验证大部分算法是否正确实现。

Copy the code below to https://play.rust-lang.org/ and write additional tests to uncover bugs in the provided implementation, fixing any bugs you find.

#![allow(unused)]
fn main() {
pub fn luhn(cc_number: &str) -> bool {
    let mut sum = 0;
    let mut double = false;

    for c in cc_number.chars().rev() {
        if let Some(digit) = c.to_digit(10) {
            if double {
                let double_digit = digit * 2;
                sum +=
                    if double_digit > 9 { double_digit - 9 } else { double_digit };
            } else {
                sum += digit;
            }
            double = !double;
        } else {
            continue;
        }
    }

    sum % 10 == 0
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn test_valid_cc_number() {
        assert!(luhn("4263 9826 4026 9299"));
        assert!(luhn("4539 3195 0343 6467"));
        assert!(luhn("7992 7398 713"));
    }

    #[test]
    fn test_invalid_cc_number() {
        assert!(!luhn("4223 9826 4026 9299"));
        assert!(!luhn("4539 3195 0343 6476"));
        assert!(!luhn("8273 1232 7352 0569"));
    }
}
}

解答

// This is the buggy version that appears in the problem.
#[cfg(never)]
pub fn luhn(cc_number: &str) -> bool {
    let mut sum = 0;
    let mut double = false;

    for c in cc_number.chars().rev() {
        if let Some(digit) = c.to_digit(10) {
            if double {
                let double_digit = digit * 2;
                sum +=
                    if double_digit > 9 { double_digit - 9 } else { double_digit };
            } else {
                sum += digit;
            }
            double = !double;
        } else {
            continue;
        }
    }

    sum % 10 == 0
}

// This is the solution and passes all of the tests below.
pub fn luhn(cc_number: &str) -> bool {
    let mut sum = 0;
    let mut double = false;
    let mut digits = 0;

    for c in cc_number.chars().rev() {
        if let Some(digit) = c.to_digit(10) {
            digits += 1;
            if double {
                let double_digit = digit * 2;
                sum +=
                    if double_digit > 9 { double_digit - 9 } else { double_digit };
            } else {
                sum += digit;
            }
            double = !double;
        } else if c.is_whitespace() {
            continue;
        } else {
            return false;
        }
    }

    digits >= 2 && sum % 10 == 0
}

fn main() {
    let cc_number = "1234 5678 1234 5670";
    println!(
        "Is {cc_number} a valid credit card number? {}",
        if luhn(cc_number) { "yes" } else { "no" }
    );
}

#[cfg(test)]
mod test {
    use super::*;

    #[test]
    fn test_valid_cc_number() {
        assert!(luhn("4263 9826 4026 9299"));
        assert!(luhn("4539 3195 0343 6467"));
        assert!(luhn("7992 7398 713"));
    }

    #[test]
    fn test_invalid_cc_number() {
        assert!(!luhn("4223 9826 4026 9299"));
        assert!(!luhn("4539 3195 0343 6476"));
        assert!(!luhn("8273 1232 7352 0569"));
    }

    #[test]
    fn test_non_digit_cc_number() {
        assert!(!luhn("foo"));
        assert!(!luhn("foo 0 0"));
    }

    #[test]
    fn test_empty_cc_number() {
        assert!(!luhn(""));
        assert!(!luhn(" "));
        assert!(!luhn("  "));
        assert!(!luhn("    "));
    }

    #[test]
    fn test_single_digit_cc_number() {
        assert!(!luhn("0"));
    }

    #[test]
    fn test_two_digit_cc_number() {
        assert!(luhn(" 0 0 "));
    }
}

Welcome Back

Including 10 minute breaks, this session should take about 2 hours and 10 minutes. It contains:

错误处理

This segment should take about 55 minutes. It contains:

Panics

Rust 通过 “panic”机制处理严重错误。

如果运行时发生严重错误，Rust 会触发 panic：

fn main() {
    let v = vec![10, 20, 30];
    println!("v[100]: {}", v[100]);
}

• Panic 用于指示不可恢复的意外错误。
  • Panic反映了程序中的 bug 问题。
  • 运行时失败（例如边界检查失败）可能会触发 panic
  • 断言（例如 assert!）在失败时会触发 panic
  • 针对特定用途的 panic 可以使用 panic! 宏。
• 使用 panic 会 “展开”堆栈，并丢弃对应的值，就像函数已经返回一样。
• 如果崩溃不可接受，请使用不会触发 panic 的 API（例如 Vec::get）。

use std::panic;

fn main() {
    let result = panic::catch_unwind(|| "No problem here!");
    println!("{result:?}");

    let result = panic::catch_unwind(|| {
        panic!("oh no!");
    });
    println!("{result:?}");
}

尝试运算符

Runtime errors like connection-refused or file-not-found are handled with the Result type, but matching this type on every call can be cumbersome. The try-operator ? is used to return errors to the caller. It lets you turn the common

match some_expression {
    Ok(value) => value,
    Err(err) => return Err(err),
}

转换成更简单的命令

some_expression?

We can use this to simplify our error handling code:

use std::io::Read;
use std::{fs, io};

fn read_username(path: &str) -> Result<String, io::Error> {
    let username_file_result = fs::File::open(path);
    let mut username_file = match username_file_result {
        Ok(file) => file,
        Err(err) => return Err(err),
    };

    let mut username = String::new();
    match username_file.read_to_string(&mut username) {
        Ok(_) => Ok(username),
        Err(err) => Err(err),
    }
}

fn main() {
    //fs::write("config.dat", "alice").unwrap();
    let username = read_username("config.dat");
    println!("username or error: {username:?}");
}

尝试转换

? 的有效展开比前面介绍的内容略微复杂一些：

expression?

效果等同于

match expression {
    Ok(value) => value,
    Err(err)  => return Err(From::from(err)),
}

The From::from call here means we attempt to convert the error type to the type returned by the function. This makes it easy to encapsulate errors into higher-level errors.

示例

use std::error::Error;
use std::fmt::{self, Display, Formatter};
use std::fs::File;
use std::io::{self, Read};

#[derive(Debug)]
enum ReadUsernameError {
    IoError(io::Error),
    EmptyUsername(String),
}

impl Error for ReadUsernameError {}

impl Display for ReadUsernameError {
    fn fmt(&self, f: &mut Formatter) -> fmt::Result {
        match self {
            Self::IoError(e) => write!(f, "IO error: {e}"),
            Self::EmptyUsername(path) => write!(f, "Found no username in {path}"),
        }
    }
}

impl From<io::Error> for ReadUsernameError {
    fn from(err: io::Error) -> Self {
        Self::IoError(err)
    }
}

fn read_username(path: &str) -> Result<String, ReadUsernameError> {
    let mut username = String::with_capacity(100);
    File::open(path)?.read_to_string(&mut username)?;
    if username.is_empty() {
        return Err(ReadUsernameError::EmptyUsername(String::from(path)));
    }
    Ok(username)
}

fn main() {
    //fs::write("config.dat", "").unwrap();
    let username = read_username("config.dat");
    println!("username or error: {username:?}");
}

动态错误类型

Sometimes we want to allow any type of error to be returned without writing our own enum covering all the different possibilities. The std::error::Error trait makes it easy to create a trait object that can contain any error.

use std::error::Error;
use std::fs;
use std::io::Read;

fn read_count(path: &str) -> Result<i32, Box<dyn Error>> {
    let mut count_str = String::new();
    fs::File::open(path)?.read_to_string(&mut count_str)?;
    let count: i32 = count_str.parse()?;
    Ok(count)
}

fn main() {
    fs::write("count.dat", "1i3").unwrap();
    match read_count("count.dat") {
        Ok(count) => println!("Count: {count}"),
        Err(err) => println!("Error: {err}"),
    }
}

thiserror 和 anyhow

The thiserror and anyhow crates are widely used to simplify error handling.

• thiserror is often used in libraries to create custom error types that implement From<T>.
• anyhow is often used by applications to help with error handling in functions, including adding contextual information to your errors.

use anyhow::{bail, Context, Result};
use std::fs;
use std::io::Read;
use thiserror::Error;

#[derive(Clone, Debug, Eq, Error, PartialEq)]
#[error("Found no username in {0}")]
struct EmptyUsernameError(String);

fn read_username(path: &str) -> Result<String> {
    let mut username = String::with_capacity(100);
    fs::File::open(path)
        .with_context(|| format!("Failed to open {path}"))?
        .read_to_string(&mut username)
        .context("Failed to read")?;
    if username.is_empty() {
        bail!(EmptyUsernameError(path.to_string()));
    }
    Ok(username)
}

fn main() {
    //fs::write("config.dat", "").unwrap();
    match read_username("config.dat") {
        Ok(username) => println!("Username: {username}"),
        Err(err) => println!("Error: {err:?}"),
    }
}

练习：使用 Result 进行重写

以下代码实现了一个非常简单的表达式语言解析器。不过，它通过 panic 机制来处理错误。请重写该代码，改用惯用的错误处理方式，并将错误传播到 main 函数的返回值。您可以随意使用 thiserror 和 anyhow。

提示：请先修复 parse 函数中的错误处理问题。该部分正常运行后，请更新 Tokenizer 以实现 Iterator<Item=Result<Token, TokenizerError>>，并在解析器中进行相应处理。

use std::iter::Peekable;
use std::str::Chars;

/// An arithmetic operator.
#[derive(Debug, PartialEq, Clone, Copy)]
enum Op {
    Add,
    Sub,
}

/// A token in the expression language.
#[derive(Debug, PartialEq)]
enum Token {
    Number(String),
    Identifier(String),
    Operator(Op),
}

/// An expression in the expression language.
#[derive(Debug, PartialEq)]
enum Expression {
    /// A reference to a variable.
    Var(String),
    /// A literal number.
    Number(u32),
    /// A binary operation.
    Operation(Box<Expression>, Op, Box<Expression>),
}

fn tokenize(input: &str) -> Tokenizer {
    return Tokenizer(input.chars().peekable());
}

struct Tokenizer<'a>(Peekable<Chars<'a>>);

impl<'a> Iterator for Tokenizer<'a> {
    type Item = Token;

    fn next(&mut self) -> Option<Token> {
        let c = self.0.next()?;
        match c {
            '0'..='9' => {
                let mut num = String::from(c);
                while let Some(c @ '0'..='9') = self.0.peek() {
                    num.push(*c);
                    self.0.next();
                }
                Some(Token::Number(num))
            }
            'a'..='z' => {
                let mut ident = String::from(c);
                while let Some(c @ ('a'..='z' | '_' | '0'..='9')) = self.0.peek() {
                    ident.push(*c);
                    self.0.next();
                }
                Some(Token::Identifier(ident))
            }
            '+' => Some(Token::Operator(Op::Add)),
            '-' => Some(Token::Operator(Op::Sub)),
            _ => panic!("Unexpected character {c}"),
        }
    }
}

fn parse(input: &str) -> Expression {
    let mut tokens = tokenize(input);

    fn parse_expr<'a>(tokens: &mut Tokenizer<'a>) -> Expression {
        let Some(tok) = tokens.next() else {
            panic!("Unexpected end of input");
        };
        let expr = match tok {
            Token::Number(num) => {
                let v = num.parse().expect("Invalid 32-bit integer'");
                Expression::Number(v)
            }
            Token::Identifier(ident) => Expression::Var(ident),
            Token::Operator(_) => panic!("Unexpected token {tok:?}"),
        };
        // Look ahead to parse a binary operation if present.
        match tokens.next() {
            None => expr,
            Some(Token::Operator(op)) => Expression::Operation(
                Box::new(expr),
                op,
                Box::new(parse_expr(tokens)),
            ),
            Some(tok) => panic!("Unexpected token {tok:?}"),
        }
    }

    parse_expr(&mut tokens)
}

fn main() {
    let expr = parse("10+foo+20-30");
    println!("{expr:?}");
}

解答

use thiserror::Error;
use std::iter::Peekable;
use std::str::Chars;

/// An arithmetic operator.
#[derive(Debug, PartialEq, Clone, Copy)]
enum Op {
    Add,
    Sub,
}

/// A token in the expression language.
#[derive(Debug, PartialEq)]
enum Token {
    Number(String),
    Identifier(String),
    Operator(Op),
}

/// An expression in the expression language.
#[derive(Debug, PartialEq)]
enum Expression {
    /// A reference to a variable.
    Var(String),
    /// A literal number.
    Number(u32),
    /// A binary operation.
    Operation(Box<Expression>, Op, Box<Expression>),
}

fn tokenize(input: &str) -> Tokenizer {
    return Tokenizer(input.chars().peekable());
}

#[derive(Debug, Error)]
enum TokenizerError {
    #[error("Unexpected character '{0}' in input")]
    UnexpectedCharacter(char),
}

struct Tokenizer<'a>(Peekable<Chars<'a>>);

impl<'a> Iterator for Tokenizer<'a> {
    type Item = Result<Token, TokenizerError>;

    fn next(&mut self) -> Option<Result<Token, TokenizerError>> {
        let c = self.0.next()?;
        match c {
            '0'..='9' => {
                let mut num = String::from(c);
                while let Some(c @ '0'..='9') = self.0.peek() {
                    num.push(*c);
                    self.0.next();
                }
                Some(Ok(Token::Number(num)))
            }
            'a'..='z' => {
                let mut ident = String::from(c);
                while let Some(c @ ('a'..='z' | '_' | '0'..='9')) = self.0.peek() {
                    ident.push(*c);
                    self.0.next();
                }
                Some(Ok(Token::Identifier(ident)))
            }
            '+' => Some(Ok(Token::Operator(Op::Add))),
            '-' => Some(Ok(Token::Operator(Op::Sub))),
            _ => Some(Err(TokenizerError::UnexpectedCharacter(c))),
        }
    }
}

#[derive(Debug, Error)]
enum ParserError {
    #[error("Tokenizer error: {0}")]
    TokenizerError(#[from] TokenizerError),
    #[error("Unexpected end of input")]
    UnexpectedEOF,
    #[error("Unexpected token {0:?}")]
    UnexpectedToken(Token),
    #[error("Invalid number")]
    InvalidNumber(#[from] std::num::ParseIntError),
}

fn parse(input: &str) -> Result<Expression, ParserError> {
    let mut tokens = tokenize(input);

    fn parse_expr<'a>(
        tokens: &mut Tokenizer<'a>,
    ) -> Result<Expression, ParserError> {
        let tok = tokens.next().ok_or(ParserError::UnexpectedEOF)??;
        let expr = match tok {
            Token::Number(num) => {
                let v = num.parse()?;
                Expression::Number(v)
            }
            Token::Identifier(ident) => Expression::Var(ident),
            Token::Operator(_) => return Err(ParserError::UnexpectedToken(tok)),
        };
        // Look ahead to parse a binary operation if present.
        Ok(match tokens.next() {
            None => expr,
            Some(Ok(Token::Operator(op))) => Expression::Operation(
                Box::new(expr),
                op,
                Box::new(parse_expr(tokens)?),
            ),
            Some(Err(e)) => return Err(e.into()),
            Some(Ok(tok)) => return Err(ParserError::UnexpectedToken(tok)),
        })
    }

    parse_expr(&mut tokens)
}

fn main() -> anyhow::Result<()> {
    let expr = parse("10+foo+20-30")?;
    println!("{expr:?}");
    Ok(())
}

不安全 Rust

This segment should take about 1 hour and 5 minutes. It contains:

不安全 Rust

Rust 语言包含两个部分：

• **安全 Rust：**内存安全，没有潜在的未定义行为。
• **不安全 Rust：**如果违反了前提条件，可能会触发未定义的行为。

We saw mostly safe Rust in this course, but it's important to know what Unsafe Rust is.

不安全的代码通常内容很少而且与其他代码隔离， 其正确性也应得到仔细记录。这类代码通常封装在安全的抽象层中。

不安全 Rust 提供了五种新功能：

• 解引用原始指针。
• 访问或修改可变的静态变量。
• 访问 union 字段。
• 调用 unsafe 函数，包括 extern 函数。
• 实现 unsafe trait。

下面，我们将简要介绍这些不安全功能。如需了解完整详情，请参阅 《Rust 手册》第 19.1 章 和 Rustonomicon。

解引用裸指针

创建指针是安全的操作，但解引用指针需要使用 unsafe 方法：

fn main() {
    let mut s = String::from("careful!");

    let r1 = &mut s as *mut String;
    let r2 = r1 as *const String;

    // SAFETY: r1 and r2 were obtained from references and so are guaranteed to
    // be non-null and properly aligned, the objects underlying the references
    // from which they were obtained are live throughout the whole unsafe
    // block, and they are not accessed either through the references or
    // concurrently through any other pointers.
    unsafe {
        println!("r1 is: {}", *r1);
        *r1 = String::from("uhoh");
        println!("r2 is: {}", *r2);
    }

    // NOT SAFE. DO NOT DO THIS.
    /*
    let r3: &String = unsafe { &*r1 };
    drop(s);
    println!("r3 is: {}", *r3);
    */
}

可变的静态变量

读取不可变的静态变量是安全的操作：

static HELLO_WORLD: &str = "Hello, world!";

fn main() {
    println!("HELLO_WORLD: {HELLO_WORLD}");
}

但是，读取和写入可变的静态变量是不安全的，因为这可能会 造成数据争用：

static mut COUNTER: u32 = 0;

fn add_to_counter(inc: u32) {
    // SAFETY: There are no other threads which could be accessing `COUNTER`.
    unsafe {
        COUNTER += inc;
    }
}

fn main() {
    add_to_counter(42);

    // SAFETY: There are no other threads which could be accessing `COUNTER`.
    unsafe {
        println!("COUNTER: {COUNTER}");
    }
}

联合体

联合体与枚举类似，但您需要自行跟踪活跃字段：

#[repr(C)]
union MyUnion {
    i: u8,
    b: bool,
}

fn main() {
    let u = MyUnion { i: 42 };
    println!("int: {}", unsafe { u.i });
    println!("bool: {}", unsafe { u.b }); // Undefined behavior!
}

Unsafe 函数

调用 Unsafe 函数

如果函数或方法具有额外的前提条件，您必须遵守这些前提条件来避免未定义的行为， 则可以将该函数或方法标记为 unsafe：

extern "C" {
    fn abs(input: i32) -> i32;
}

fn main() {
    let emojis = "🗻∈🌏";

    // SAFETY: The indices are in the correct order, within the bounds of the
    // string slice, and lie on UTF-8 sequence boundaries.
    unsafe {
        println!("emoji: {}", emojis.get_unchecked(0..4));
        println!("emoji: {}", emojis.get_unchecked(4..7));
        println!("emoji: {}", emojis.get_unchecked(7..11));
    }

    println!("char count: {}", count_chars(unsafe { emojis.get_unchecked(0..7) }));

    // SAFETY: `abs` doesn't deal with pointers and doesn't have any safety
    // requirements.
    unsafe {
        println!("Absolute value of -3 according to C: {}", abs(-3));
    }

    // Not upholding the UTF-8 encoding requirement breaks memory safety!
    // println!("emoji: {}", unsafe { emojis.get_unchecked(0..3) });
    // println!("char count: {}", count_chars(unsafe {
    // emojis.get_unchecked(0..3) }));
}

fn count_chars(s: &str) -> usize {
    s.chars().count()
}

编写 Unsafe 函数

如果您自己编写的函数需要满足特定条件以避免未定义的行为， 您可以将这些函数标记为 unsafe。

/// Swaps the values pointed to by the given pointers.
///
/// # Safety
///
/// The pointers must be valid and properly aligned.
unsafe fn swap(a: *mut u8, b: *mut u8) {
    let temp = *a;
    *a = *b;
    *b = temp;
}

fn main() {
    let mut a = 42;
    let mut b = 66;

    // SAFETY: ...
    unsafe {
        swap(&mut a, &mut b);
    }

    println!("a = {}, b = {}", a, b);
}

实现 Unsafe Trait

与函数一样，如果您在实现某个 trait 时必须保证特定条件来避免未定义的行为， 您也可以将该 trait 标记为 unsafe。

例如，zerocopy crate 包含一个不安全的 trait， 大致内容是这样的：

use std::mem::size_of_val;
use std::slice;

/// ...
/// # Safety
/// The type must have a defined representation and no padding.
pub unsafe trait AsBytes {
    fn as_bytes(&self) -> &[u8] {
        unsafe {
            slice::from_raw_parts(
                self as *const Self as *const u8,
                size_of_val(self),
            )
        }
    }
}

// SAFETY: `u32` has a defined representation and no padding.
unsafe impl AsBytes for u32 {}

安全 FFI 封装容器

Rust 为通过 外部函数接口 (FFI) 调用函数提供了出色的支持。我们将使用它为 libc 函数构建一个安全封装容器，用于从 C 代码中读取目录中的文件名称。

建议您参考以下手册页面：

• opendir(3)
• readdir(3)
• closedir(3)

您还需要浏览“std::ffi”模块。在下方，您会发现完成这个练习所需的多种字符串类型：

您将在以下所有类型之间进行转换：

• 将 &str 转换为 CString：您需要为尾随 \0 字符分配空格，
• 将 CString 转换为 \*const i8 ：您需要一个指针来调用 C 函数，
• 将 \*const i8 转换为 &CStr ：您需要一些能够找到尾随 \0 字符的内容，
• &CStr to &[u8]: a slice of bytes is the universal interface for "some unknown data",
• 将 &\[u8\] 转换为 &OsStr ：&OsStr 是向 OsString 迈进的一步，请使用OsStrExt来创建它，
• 将“&OsStr”转换为“OsString”：您需要克隆“&OsStr”中的数据，以便能够返回它并再次调用“readdir”。

秘典 中也有一个关于 FFI 的非常实用的章节。

将以下代码复制到 https://play.rust-lang.org/，并填入缺少的函数和方法：

// TODO: remove this when you're done with your implementation.
#![allow(unused_imports, unused_variables, dead_code)]

mod ffi {
    use std::os::raw::{c_char, c_int};
    #[cfg(not(target_os = "macos"))]
    use std::os::raw::{c_long, c_uchar, c_ulong, c_ushort};

    // Opaque type. See https://doc.rust-lang.org/nomicon/ffi.html.
    #[repr(C)]
    pub struct DIR {
        _data: [u8; 0],
        _marker: core::marker::PhantomData<(*mut u8, core::marker::PhantomPinned)>,
    }

    // Layout according to the Linux man page for readdir(3), where ino_t and
    // off_t are resolved according to the definitions in
    // /usr/include/x86_64-linux-gnu/{sys/types.h, bits/typesizes.h}.
    #[cfg(not(target_os = "macos"))]
    #[repr(C)]
    pub struct dirent {
        pub d_ino: c_ulong,
        pub d_off: c_long,
        pub d_reclen: c_ushort,
        pub d_type: c_uchar,
        pub d_name: [c_char; 256],
    }

    // Layout according to the macOS man page for dir(5).
    #[cfg(all(target_os = "macos"))]
    #[repr(C)]
    pub struct dirent {
        pub d_fileno: u64,
        pub d_seekoff: u64,
        pub d_reclen: u16,
        pub d_namlen: u16,
        pub d_type: u8,
        pub d_name: [c_char; 1024],
    }

    extern "C" {
        pub fn opendir(s: *const c_char) -> *mut DIR;

        #[cfg(not(all(target_os = "macos", target_arch = "x86_64")))]
        pub fn readdir(s: *mut DIR) -> *const dirent;

        // See https://github.com/rust-lang/libc/issues/414 and the section on
        // _DARWIN_FEATURE_64_BIT_INODE in the macOS man page for stat(2).
        //
        // "Platforms that existed before these updates were available" refers
        // to macOS (as opposed to iOS / wearOS / etc.) on Intel and PowerPC.
        #[cfg(all(target_os = "macos", target_arch = "x86_64"))]
        #[link_name = "readdir$INODE64"]
        pub fn readdir(s: *mut DIR) -> *const dirent;

        pub fn closedir(s: *mut DIR) -> c_int;
    }
}

use std::ffi::{CStr, CString, OsStr, OsString};
use std::os::unix::ffi::OsStrExt;

#[derive(Debug)]
struct DirectoryIterator {
    path: CString,
    dir: *mut ffi::DIR,
}

impl DirectoryIterator {
    fn new(path: &str) -> Result<DirectoryIterator, String> {
        // Call opendir and return a Ok value if that worked,
        // otherwise return Err with a message.
        unimplemented!()
    }
}

impl Iterator for DirectoryIterator {
    type Item = OsString;
    fn next(&mut self) -> Option<OsString> {
        // Keep calling readdir until we get a NULL pointer back.
        unimplemented!()
    }
}

impl Drop for DirectoryIterator {
    fn drop(&mut self) {
        // Call closedir as needed.
        unimplemented!()
    }
}

fn main() -> Result<(), String> {
    let iter = DirectoryIterator::new(".")?;
    println!("files: {:#?}", iter.collect::<Vec<_>>());
    Ok(())
}

解答

mod ffi {
    use std::os::raw::{c_char, c_int};
    #[cfg(not(target_os = "macos"))]
    use std::os::raw::{c_long, c_uchar, c_ulong, c_ushort};

    // Opaque type. See https://doc.rust-lang.org/nomicon/ffi.html.
    #[repr(C)]
    pub struct DIR {
        _data: [u8; 0],
        _marker: core::marker::PhantomData<(*mut u8, core::marker::PhantomPinned)>,
    }

    // Layout according to the Linux man page for readdir(3), where ino_t and
    // off_t are resolved according to the definitions in
    // /usr/include/x86_64-linux-gnu/{sys/types.h, bits/typesizes.h}.
    #[cfg(not(target_os = "macos"))]
    #[repr(C)]
    pub struct dirent {
        pub d_ino: c_ulong,
        pub d_off: c_long,
        pub d_reclen: c_ushort,
        pub d_type: c_uchar,
        pub d_name: [c_char; 256],
    }

    // Layout according to the macOS man page for dir(5).
    #[cfg(all(target_os = "macos"))]
    #[repr(C)]
    pub struct dirent {
        pub d_fileno: u64,
        pub d_seekoff: u64,
        pub d_reclen: u16,
        pub d_namlen: u16,
        pub d_type: u8,
        pub d_name: [c_char; 1024],
    }

    extern "C" {
        pub fn opendir(s: *const c_char) -> *mut DIR;

        #[cfg(not(all(target_os = "macos", target_arch = "x86_64")))]
        pub fn readdir(s: *mut DIR) -> *const dirent;

        // See https://github.com/rust-lang/libc/issues/414 and the section on
        // _DARWIN_FEATURE_64_BIT_INODE in the macOS man page for stat(2).
        //
        // "Platforms that existed before these updates were available" refers
        // to macOS (as opposed to iOS / wearOS / etc.) on Intel and PowerPC.
        #[cfg(all(target_os = "macos", target_arch = "x86_64"))]
        #[link_name = "readdir$INODE64"]
        pub fn readdir(s: *mut DIR) -> *const dirent;

        pub fn closedir(s: *mut DIR) -> c_int;
    }
}

use std::ffi::{CStr, CString, OsStr, OsString};
use std::os::unix::ffi::OsStrExt;

#[derive(Debug)]
struct DirectoryIterator {
    path: CString,
    dir: *mut ffi::DIR,
}

impl DirectoryIterator {
    fn new(path: &str) -> Result<DirectoryIterator, String> {
        // Call opendir and return a Ok value if that worked,
        // otherwise return Err with a message.
        let path =
            CString::new(path).map_err(|err| format!("Invalid path: {err}"))?;
        // SAFETY: path.as_ptr() cannot be NULL.
        let dir = unsafe { ffi::opendir(path.as_ptr()) };
        if dir.is_null() {
            Err(format!("Could not open {:?}", path))
        } else {
            Ok(DirectoryIterator { path, dir })
        }
    }
}

impl Iterator for DirectoryIterator {
    type Item = OsString;
    fn next(&mut self) -> Option<OsString> {
        // Keep calling readdir until we get a NULL pointer back.
        // SAFETY: self.dir is never NULL.
        let dirent = unsafe { ffi::readdir(self.dir) };
        if dirent.is_null() {
            // We have reached the end of the directory.
            return None;
        }
        // SAFETY: dirent is not NULL and dirent.d_name is NUL
        // terminated.
        let d_name = unsafe { CStr::from_ptr((*dirent).d_name.as_ptr()) };
        let os_str = OsStr::from_bytes(d_name.to_bytes());
        Some(os_str.to_owned())
    }
}

impl Drop for DirectoryIterator {
    fn drop(&mut self) {
        // Call closedir as needed.
        if !self.dir.is_null() {
            // SAFETY: self.dir is not NULL.
            if unsafe { ffi::closedir(self.dir) } != 0 {
                panic!("Could not close {:?}", self.path);
            }
        }
    }
}

fn main() -> Result<(), String> {
    let iter = DirectoryIterator::new(".")?;
    println!("files: {:#?}", iter.collect::<Vec<_>>());
    Ok(())
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::error::Error;

    #[test]
    fn test_nonexisting_directory() {
        let iter = DirectoryIterator::new("no-such-directory");
        assert!(iter.is_err());
    }

    #[test]
    fn test_empty_directory() -> Result<(), Box<dyn Error>> {
        let tmp = tempfile::TempDir::new()?;
        let iter = DirectoryIterator::new(
            tmp.path().to_str().ok_or("Non UTF-8 character in path")?,
        )?;
        let mut entries = iter.collect::<Vec<_>>();
        entries.sort();
        assert_eq!(entries, &[".", ".."]);
        Ok(())
    }

    #[test]
    fn test_nonempty_directory() -> Result<(), Box<dyn Error>> {
        let tmp = tempfile::TempDir::new()?;
        std::fs::write(tmp.path().join("foo.txt"), "The Foo Diaries\n")?;
        std::fs::write(tmp.path().join("bar.png"), "<PNG>\n")?;
        std::fs::write(tmp.path().join("crab.rs"), "//! Crab\n")?;
        let iter = DirectoryIterator::new(
            tmp.path().to_str().ok_or("Non UTF-8 character in path")?,
        )?;
        let mut entries = iter.collect::<Vec<_>>();
        entries.sort();
        assert_eq!(entries, &[".", "..", "bar.png", "crab.rs", "foo.txt"]);
        Ok(())
    }
}

欢迎来到Android 中的Rust

Rust is supported for system software on Android. This means that you can write new services, libraries, drivers or even firmware in Rust (or improve existing code as needed).

今天我们会尝试在你自己的项目中调用Rust。 所以试着在你的代码中找一小段来改成Rust。 代码中越少依赖(dependencies)，越少“独特”的类型，越好。比如 一段解析原始字符的代码就很理想。

设置

We will be using a Cuttlefish Android Virtual Device to test our code. Make sure you have access to one or create a new one with:

source build/envsetup.sh
lunch aosp_cf_x86_64_phone-trunk_staging-userdebug
acloud create

更多细节请参考 Android Developer Codelab.

构建规则

Android 构建系统（Soong）通过一系列模块来支持Rust：

下面我们来看看 rust_binary 和 rust_library。

Rust 二进制文件

让我们从一个简单的应用程序开始。在 AOSP 签出的根目录下，创建以下文件：

hello_rust/Android.bp:

rust_binary {
    name: "hello_rust",
    crate_name: "hello_rust",
    srcs: ["src/main.rs"],
}

hello_rust/src/main.rs:

//! Rust demo.

/// Prints a greeting to standard output.
fn main() {
    println!("Hello from Rust!");
}

你现在可以构建、推送和运行二进制文件：

m hello_rust
adb push "$ANDROID_PRODUCT_OUT/system/bin/hello_rust" /data/local/tmp
adb shell /data/local/tmp/hello_rust

Hello from Rust!

Rust 库

您可以使用 rust_library 为 Android 创建一个新的 Rust 库。

在这里，我们声明了对两个库的依赖：

• libgreeting, 我们在下面进行了定义，
• libtextwrap, 一个已经在 external/rust/crates/ 中提供的 crate。

hello_rust/Android.bp:

rust_binary {
    name: "hello_rust_with_dep",
    crate_name: "hello_rust_with_dep",
    srcs: ["src/main.rs"],
    rustlibs: [
        "libgreetings",
        "libtextwrap",
    ],
    prefer_rlib: true, // Need this to avoid dynamic link error.
}

rust_library {
    name: "libgreetings",
    crate_name: "greetings",
    srcs: ["src/lib.rs"],
}

hello_rust/src/main.rs:

//! Rust demo.

use greetings::greeting;
use textwrap::fill;

/// Prints a greeting to standard output.
fn main() {
    println!("{}", fill(&greeting("Bob"), 24));
}

hello_rust/src/lib.rs:

//! Greeting library.

/// Greet `name`.
pub fn greeting(name: &str) -> String {
    format!("Hello {name}, it is very nice to meet you!")
}

您可以像之前一样构建、推送和运行二进制文件：

m hello_rust_with_dep
adb push "$ANDROID_PRODUCT_OUT/system/bin/hello_rust_with_dep" /data/local/tmp
adb shell /data/local/tmp/hello_rust_with_dep

Hello Bob, it is very
nice to meet you!

AIDL

Rust 支持 Android 接口定义语言 (AIDL)：

• Rust 代码可以调用现有的 AIDL 服务器，
• 您可以在 Rust 中创建新的 AIDL 服务器。

/** Birthday service interface. */

To illustrate how to use Rust with Binder, we're going to walk through the process of creating a Binder interface. We're then going to both implement the described service and write client code that talks to that service.

AIDL 接口

您可以使用 AIDL 接口声明您的服务的 API：

birthday_service/aidl/com/example/birthdayservice/IBirthdayService.aidl:

/** Birthday service interface. */
interface IBirthdayService {
    /** Generate a Happy Birthday message. */
    String wishHappyBirthday(String name, int years);
}

birthday_service/aidl/Android.bp:

aidl_interface {
    name: "com.example.birthdayservice",
    srcs: ["com/example/birthdayservice/*.aidl"],
    unstable: true,
    backend: {
        rust: { // Rust is not enabled by default
            enabled: true,
        },
    },
}

Generated Service API

Binder generates a trait corresponding to the interface definition. trait to talk to the service.

birthday_service/aidl/com/example/birthdayservice/IBirthdayService.aidl:

/** Birthday service interface. */
interface IBirthdayService {
    /** Generate a Happy Birthday message. */
    String wishHappyBirthday(String name, int years);
}

Generated trait:

trait IBirthdayService {
    fn wishHappyBirthday(&self, name: &str, years: i32) -> binder::Result<String>;
}

Your service will need to implement this trait, and your client will use this trait to talk to the service.

服务实现

我们现在可以实现AIDL服务：

birthday_service/src/lib.rs:

use com_example_birthdayservice::aidl::com::example::birthdayservice::IBirthdayService::IBirthdayService;
use com_example_birthdayservice::binder;

/// The `IBirthdayService` implementation.
pub struct BirthdayService;

impl binder::Interface for BirthdayService {}

impl IBirthdayService for BirthdayService {
    fn wishHappyBirthday(&self, name: &str, years: i32) -> binder::Result<String> {
        Ok(format!("Happy Birthday {name}, congratulations with the {years} years!"))
    }
}

birthday_service/Android.bp:

rust_library {
    name: "libbirthdayservice",
    srcs: ["src/lib.rs"],
    crate_name: "birthdayservice",
    rustlibs: [
        "com.example.birthdayservice-rust",
        "libbinder_rs",
    ],
}

AIDL 服务器

最后，我们可以创建一个暴露服务的服务器：

birthday_service/src/server.rs:

//! Birthday service.
use birthdayservice::BirthdayService;
use com_example_birthdayservice::aidl::com::example::birthdayservice::IBirthdayService::BnBirthdayService;
use com_example_birthdayservice::binder;

const SERVICE_IDENTIFIER: &str = "birthdayservice";

/// Entry point for birthday service.
fn main() {
    let birthday_service = BirthdayService;
    let birthday_service_binder = BnBirthdayService::new_binder(
        birthday_service,
        binder::BinderFeatures::default(),
    );
    binder::add_service(SERVICE_IDENTIFIER, birthday_service_binder.as_binder())
        .expect("Failed to register service");
    binder::ProcessState::join_thread_pool()
}

birthday_service/Android.bp:

rust_binary {
    name: "birthday_server",
    crate_name: "birthday_server",
    srcs: ["src/server.rs"],
    rustlibs: [
        "com.example.birthdayservice-rust",
        "libbinder_rs",
        "libbirthdayservice",
    ],
    prefer_rlib: true, // To avoid dynamic link error.
}

部署

我们现在可以构建、推送和启动服务：

m birthday_server
adb push "$ANDROID_PRODUCT_OUT/system/bin/birthday_server" /data/local/tmp
adb root
adb shell /data/local/tmp/birthday_server

在另一个终端中，检查该服务是否正在运行：

adb shell service check birthdayservice

Service birthdayservice: found

您还可以使用 service call 命令调用该服务：

adb shell service call birthdayservice 1 s16 Bob i32 24

Result: Parcel(
  0x00000000: 00000000 00000036 00610048 00700070 '....6...H.a.p.p.'
  0x00000010: 00200079 00690042 00740072 00640068 'y. .B.i.r.t.h.d.'
  0x00000020: 00790061 00420020 0062006f 0020002c 'a.y. .B.o.b.,. .'
  0x00000030: 006f0063 0067006e 00610072 00750074 'c.o.n.g.r.a.t.u.'
  0x00000040: 0061006c 00690074 006e006f 00200073 'l.a.t.i.o.n.s. .'
  0x00000050: 00690077 00680074 00740020 00650068 'w.i.t.h. .t.h.e.'
  0x00000060: 00320020 00200034 00650079 00720061 ' .2.4. .y.e.a.r.'
  0x00000070: 00210073 00000000                   's.!.....        ')

AIDL 客户端

最后，我们可以为我们的新服务创建一个 Rust 客户端。

birthday_service/src/client.rs:

use com_example_birthdayservice::aidl::com::example::birthdayservice::IBirthdayService::IBirthdayService;
use com_example_birthdayservice::binder;

const SERVICE_IDENTIFIER: &str = "birthdayservice";

/// Call the birthday service.
fn main() -> Result<(), Box<dyn Error>> {
    let name = std::env::args().nth(1).unwrap_or_else(|| String::from("Bob"));
    let years = std::env::args()
        .nth(2)
        .and_then(|arg| arg.parse::<i32>().ok())
        .unwrap_or(42);

    binder::ProcessState::start_thread_pool();
    let service = binder::get_interface::<dyn IBirthdayService>(SERVICE_IDENTIFIER)
        .map_err(|_| "Failed to connect to BirthdayService")?;

    // Call the service.
    let msg = service.wishHappyBirthday(&name, years)?;
    println!("{msg}");
}

birthday_service/Android.bp:

rust_binary {
    name: "birthday_client",
    crate_name: "birthday_client",
    srcs: ["src/client.rs"],
    rustlibs: [
        "com.example.birthdayservice-rust",
        "libbinder_rs",
    ],
    prefer_rlib: true, // To avoid dynamic link error.
}

请注意，客户端不依赖于 libbirthdayservice。

在您的设备上构建、推送并运行客户端：

m birthday_client
adb push "$ANDROID_PRODUCT_OUT/system/bin/birthday_client" /data/local/tmp
adb shell /data/local/tmp/birthday_client Charlie 60

Happy Birthday Charlie, congratulations with the 60 years!

更改 API

让我们扩展API以提供更多功能：我们希望允许客户端指定生日贺卡的行列表：

package com.example.birthdayservice;

/** Birthday service interface. */
interface IBirthdayService {
    /** Generate a Happy Birthday message. */
    String wishHappyBirthday(String name, int years, in String[] text);
}

This results in an updated trait definition for IBirthdayService:

trait IBirthdayService {
    fn wishHappyBirthday(
        &self,
        name: &str,
        years: i32,
        text: &[String],
    ) -> binder::Result<String>;
}

Updating Client and Service

Update the client and server code to account for the new API.

birthday_service/src/lib.rs:

impl IBirthdayService for BirthdayService {
    fn wishHappyBirthday(
        &self,
        name: &str,
        years: i32,
        text: &[String],
    ) -> binder::Result<String> {
        let mut msg = format!(
            "Happy Birthday {name}, congratulations with the {years} years!",
        );

        for line in text {
            msg.push('\n');
            msg.push_str(line);
        }

        Ok(msg)
    }
}

birthday_service/src/client.rs:

let msg = service.wishHappyBirthday(
    &name,
    years,
    &[
        String::from("Habby birfday to yuuuuu"),
        String::from("And also: many more"),
    ],
)?;

Working With AIDL Types

AIDL types translate into the appropriate idiomatic Rust type:

• Primitive types map (mostly) to idiomatic Rust types.
• Collection types like slices, Vecs and string types are supported.
• References to AIDL objects and file handles can be sent between clients and services.
• File handles and parcelables are fully supported.

Primitive Types

Primitive types map (mostly) idiomatically:

数组（Arrays）

The array types (T[], byte[], and List<T>) get translated to the appropriate Rust array type depending on how they are used in the function signature:

特征对象

AIDL objects can be sent either as a concrete AIDL type or as the type-erased IBinder interface:

birthday_service/aidl/com/example/birthdayservice/IBirthdayInfoProvider.aidl:

package com.example.birthdayservice;

interface IBirthdayInfoProvider {
    String name();
    int years();
}

birthday_service/aidl/com/example/birthdayservice/IBirthdayService.aidl:

import com.example.birthdayservice.IBirthdayInfoProvider;

interface IBirthdayService {
    /** The same thing, but using a binder object. */
    String wishWithProvider(IBirthdayInfoProvider provider);

    /** The same thing, but using `IBinder`. */
    String wishWithErasedProvider(IBinder provider);
}

birthday_service/src/client.rs:

/// Rust struct implementing the `IBirthdayInfoProvider` interface.
struct InfoProvider {
    name: String,
    age: u8,
}

impl binder::Interface for InfoProvider {}

impl IBirthdayInfoProvider for InfoProvider {
    fn name(&self) -> binder::Result<String> {
        Ok(self.name.clone())
    }

    fn years(&self) -> binder::Result<i32> {
        Ok(self.age as i32)
    }
}

fn main() {
    binder::ProcessState::start_thread_pool();
    let service = connect().expect("Failed to connect to BirthdayService");

    // Create a binder object for the `IBirthdayInfoProvider` interface.
    let provider = BnBirthdayInfoProvider::new_binder(
        InfoProvider { name: name.clone(), age: years as u8 },
        BinderFeatures::default(),
    );

    // Send the binder object to the service.
    service.wishWithProvider(&provider)?;

    // Perform the same operation but passing the provider as an `SpIBinder`.
    service.wishWithErasedProvider(&provider.as_binder())?;
}

变量

Binder for Rust supports sending parcelables directly:

birthday_service/aidl/com/example/birthdayservice/BirthdayInfo.aidl:

package com.example.birthdayservice;

parcelable BirthdayInfo {
    String name;
    int years;
}

birthday_service/aidl/com/example/birthdayservice/IBirthdayService.aidl:

import com.example.birthdayservice.BirthdayInfo;

interface IBirthdayService {
    /** The same thing, but with a parcelable. */
    String wishWithInfo(in BirthdayInfo info);
}

birthday_service/src/client.rs:

fn main() {
    binder::ProcessState::start_thread_pool();
    let service = connect().expect("Failed to connect to BirthdayService");

    service.wishWithInfo(&BirthdayInfo { name: name.clone(), years })?;
}

Sending Files

Files can be sent between Binder clients/servers using the ParcelFileDescriptor type:

birthday_service/aidl/com/example/birthdayservice/IBirthdayService.aidl:

interface IBirthdayService {
    /** The same thing, but loads info from a file. */
    String wishFromFile(in ParcelFileDescriptor infoFile);
}

birthday_service/src/client.rs:

fn main() {
    binder::ProcessState::start_thread_pool();
    let service = connect().expect("Failed to connect to BirthdayService");

    // Open a file and put the birthday info in it.
    let mut file = File::create("/data/local/tmp/birthday.info").unwrap();
    writeln!(file, "{name}")?;
    writeln!(file, "{years}")?;

    // Create a `ParcelFileDescriptor` from the file and send it.
    let file = ParcelFileDescriptor::new(file);
    service.wishFromFile(&file)?;
}

birthday_service/src/lib.rs:

impl IBirthdayService for BirthdayService {
    fn wishFromFile(
        &self,
        info_file: &ParcelFileDescriptor,
    ) -> binder::Result<String> {
        // Convert the file descriptor to a `File`. `ParcelFileDescriptor` wraps
        // an `OwnedFd`, which can be cloned and then used to create a `File`
        // object.
        let mut info_file = info_file
            .as_ref()
            .try_clone()
            .map(File::from)
            .expect("Invalid file handle");

        let mut contents = String::new();
        info_file.read_to_string(&mut contents).unwrap();

        let mut lines = contents.lines();
        let name = lines.next().unwrap();
        let years: i32 = lines.next().unwrap().parse().unwrap();

        Ok(format!("Happy Birthday {name}, congratulations with the {years} years!"))
    }
}

Testing in Android

Building on Testing, we will now look at how unit tests work in AOSP. Use the rust_test module for your unit tests:

testing/Android.bp:

rust_library {
    name: "libleftpad",
    crate_name: "leftpad",
    srcs: ["src/lib.rs"],
}

rust_test {
    name: "libleftpad_test",
    crate_name: "leftpad_test",
    srcs: ["src/lib.rs"],
    host_supported: true,
    test_suites: ["general-tests"],
}

testing/src/lib.rs:

#![allow(unused)]
fn main() {
//! Left-padding library.

/// Left-pad `s` to `width`.
pub fn leftpad(s: &str, width: usize) -> String {
    format!("{s:>width$}")
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn short_string() {
        assert_eq!(leftpad("foo", 5), "  foo");
    }

    #[test]
    fn long_string() {
        assert_eq!(leftpad("foobar", 6), "foobar");
    }
}
}

You can now run the test with

atest --host libleftpad_test

The output looks like this:

INFO: Elapsed time: 2.666s, Critical Path: 2.40s
INFO: 3 processes: 2 internal, 1 linux-sandbox.
INFO: Build completed successfully, 3 total actions
//comprehensive-rust-android/testing:libleftpad_test_host            PASSED in 2.3s
    PASSED  libleftpad_test.tests::long_string (0.0s)
    PASSED  libleftpad_test.tests::short_string (0.0s)
Test cases: finished with 2 passing and 0 failing out of 2 test cases

Notice how you only mention the root of the library crate. Tests are found recursively in nested modules.

GoogleTest

The GoogleTest crate allows for flexible test assertions using matchers:

use googletest::prelude::*;

#[googletest::test]
fn test_elements_are() {
    let value = vec!["foo", "bar", "baz"];
    expect_that!(value, elements_are!(eq("foo"), lt("xyz"), starts_with("b")));
}

如果我们将最后一个元素更改为 "!"，测试将失败，并会提供详细的错误消息来指出错误的位置：

---- test_elements_are stdout ----
Value of: value
Expected: has elements:
  0. is equal to "foo"
  1. is less than "xyz"
  2. starts with prefix "!"
Actual: ["foo", "bar", "baz"],
  where element #2 is "baz", which does not start with "!"
  at src/testing/googletest.rs:6:5
Error: See failure output above

#[test]
fn test_multiline_string_diff() {
    let haiku = "Memory safety found,\n\
                 Rust's strong typing guides the way,\n\
                 Secure code you'll write.";
    assert_that!(
        haiku,
        eq("Memory safety found,\n\
            Rust's silly humor guides the way,\n\
            Secure code you'll write.")
    );
}

    Value of: haiku
Expected: is equal to "Memory safety found,\nRust's silly humor guides the way,\nSecure code you'll write."
Actual: "Memory safety found,\nRust's strong typing guides the way,\nSecure code you'll write.",
  which isn't equal to "Memory safety found,\nRust's silly humor guides the way,\nSecure code you'll write."
Difference(-actual / +expected):
 Memory safety found,
-Rust's strong typing guides the way,
+Rust's silly humor guides the way,
 Secure code you'll write.
  at src/testing/googletest.rs:17:5

模拟

对于模拟，Mockall 是一个广泛使用的库。您需要重构代码才能使用 trait，然后便可很快地对其进行模拟：

use std::time::Duration;

#[mockall::automock]
pub trait Pet {
    fn is_hungry(&self, since_last_meal: Duration) -> bool;
}

#[test]
fn test_robot_dog() {
    let mut mock_dog = MockPet::new();
    mock_dog.expect_is_hungry().return_const(true);
    assert_eq!(mock_dog.is_hungry(Duration::from_secs(10)), true);
}

#[test]
fn test_robot_cat() {
    let mut mock_cat = MockPet::new();
    mock_cat
        .expect_is_hungry()
        .with(mockall::predicate::gt(Duration::from_secs(3 * 3600)))
        .return_const(true);
    mock_cat.expect_is_hungry().return_const(false);
    assert_eq!(mock_cat.is_hungry(Duration::from_secs(1 * 3600)), false);
    assert_eq!(mock_cat.is_hungry(Duration::from_secs(5 * 3600)), true);
}

日志记录

你应该使用 log crate 来自动记录日志到 logcat （设备上）或 stdout（主机上）：

hello_rust_logs/Android.bp:

rust_binary {
    name: "hello_rust_logs",
    crate_name: "hello_rust_logs",
    srcs: ["src/main.rs"],
    rustlibs: [
        "liblog_rust",
        "liblogger",
    ],
    host_supported: true,
}

hello_rust_logs/src/main.rs:

//! Rust logging demo.

use log::{debug, error, info};

/// Logs a greeting.
fn main() {
    logger::init(
        logger::Config::default()
            .with_tag_on_device("rust")
            .with_min_level(log::Level::Trace),
    );
    debug!("Starting program.");
    info!("Things are going fine.");
    error!("Something went wrong!");
}

在你的设备上构建，推送，并运行二进制文件 ：

m hello_rust_logs
adb push "$ANDROID_PRODUCT_OUT/system/bin/hello_rust_logs" /data/local/tmp
adb shell /data/local/tmp/hello_rust_logs

日志将会在 adb logcat 中显示：

adb logcat -s rust

09-08 08:38:32.454  2420  2420 D rust: hello_rust_logs: Starting program.
09-08 08:38:32.454  2420  2420 I rust: hello_rust_logs: Things are going fine.
09-08 08:38:32.454  2420  2420 E rust: hello_rust_logs: Something went wrong!

互操作性

Rust 对于与其他编程语言的互操作性有着出色的支持。这意味着您可以：

• 从其他语言调用 Rust 函数。
• 从 Rust 调用用其他语言编写的函数。

当您从外部语言调用函数时，我们称之为使用 外部函数接口（Foreign Function Interface， FFI）。

与 C 的互操作性

Rust 对使用 C 调用约定链接目标文件提供了完整的支持。同样地，你可以导出 Rust 函数并从 C 中调用它们。

如果你愿意的话，你可以手工完成它：

extern "C" {
    fn abs(x: i32) -> i32;
}

fn main() {
    let x = -42;
    // SAFETY: `abs` doesn't have any safety requirements.
    let abs_x = unsafe { abs(x) };
    println!("{x}, {abs_x}");
}

我们已经在安全 FFI 封装容器练习中看到了这个例子。

这假设对目标平台拥有充分的了解，不建议用于生产环境。

接下来我们将探讨更好的选择。

使用 Bindgen

bindgen 工具可以自动生成 C 头文件的绑定代码。

首先创建一个小型的 C 语言库：

interoperability/bindgen/libbirthday.h:

typedef struct card {
  const char* name;
  int years;
} card;

void print_card(const card* card);

interoperability/bindgen/libbirthday.c:

#include <stdio.h>
#include "libbirthday.h"

void print_card(const card* card) {
  printf("+--------------\n");
  printf("| Happy Birthday %s!\n", card->name);
  printf("| Congratulations with the %i years!\n", card->years);
  printf("+--------------\n");
}

将该库添加到你的 Android.bp 文件中：

interoperability/bindgen/Android.bp:

cc_library {
    name: "libbirthday",
    srcs: ["libbirthday.c"],
}

为该库创建一个包装头文件（在此示例中不是必需的）：

interoperability/bindgen/libbirthday_wrapper.h:

#include "libbirthday.h"

您现在可以自动生成绑定代码：

interoperability/bindgen/Android.bp:

rust_bindgen {
    name: "libbirthday_bindgen",
    crate_name: "birthday_bindgen",
    wrapper_src: "libbirthday_wrapper.h",
    source_stem: "bindings",
    static_libs: ["libbirthday"],
}

最后，我们可以在 Rust 程序中使用这些绑定：

interoperability/bindgen/Android.bp:

rust_binary {
    name: "print_birthday_card",
    srcs: ["main.rs"],
    rustlibs: ["libbirthday_bindgen"],
}

interoperability/bindgen/main.rs:

//! Bindgen demo.

use birthday_bindgen::{card, print_card};

fn main() {
    let name = std::ffi::CString::new("Peter").unwrap();
    let card = card { name: name.as_ptr(), years: 42 };
    // SAFETY: The pointer we pass is valid because it came from a Rust
    // reference, and the `name` it contains refers to `name` above which also
    // remains valid. `print_card` doesn't store either pointer to use later
    // after it returns.
    unsafe {
        print_card(&card as *const card);
    }
}

在你的设备上构建，推送，并运行二进制文件 ：

m print_birthday_card
adb push "$ANDROID_PRODUCT_OUT/system/bin/print_birthday_card" /data/local/tmp
adb shell /data/local/tmp/print_birthday_card

最后，我们可以运行自动生成的测试来确保绑定代码正常工作：

interoperability/bindgen/Android.bp:

rust_test {
    name: "libbirthday_bindgen_test",
    srcs: [":libbirthday_bindgen"],
    crate_name: "libbirthday_bindgen_test",
    test_suites: ["general-tests"],
    auto_gen_config: true,
    clippy_lints: "none", // Generated file, skip linting
    lints: "none",
}

atest libbirthday_bindgen_test

调用 Rust

将 Rust 函数和类型导出到 C 很简单：

interoperability/rust/libanalyze/analyze.rs

//! Rust FFI demo.
#![deny(improper_ctypes_definitions)]

use std::os::raw::c_int;

/// Analyze the numbers.
#[no_mangle]
pub extern "C" fn analyze_numbers(x: c_int, y: c_int) {
    if x < y {
        println!("x ({x}) is smallest!");
    } else {
        println!("y ({y}) is probably larger than x ({x})");
    }
}

interoperability/rust/libanalyze/analyze.h

#ifndef ANALYSE_H
#define ANALYSE_H

extern "C" {
void analyze_numbers(int x, int y);
}

#endif

interoperability/rust/libanalyze/Android.bp

rust_ffi {
    name: "libanalyze_ffi",
    crate_name: "analyze_ffi",
    srcs: ["analyze.rs"],
    include_dirs: ["."],
}

我们现在可以从一个 C 二进制文件中调用它：

interoperability/rust/analyze/main.c

#include "analyze.h"

int main() {
  analyze_numbers(10, 20);
  analyze_numbers(123, 123);
  return 0;
}

interoperability/rust/analyze/Android.bp

cc_binary {
    name: "analyze_numbers",
    srcs: ["main.c"],
    static_libs: ["libanalyze_ffi"],
}

在你的设备上构建，推送，并运行二进制文件 ：

m analyze_numbers
adb push "$ANDROID_PRODUCT_OUT/system/bin/analyze_numbers" /data/local/tmp
adb shell /data/local/tmp/analyze_numbers

与 C++ 交互

CXX crate 使得在 Rust 和 C++ 之间进行安全的互操作成为可能。

整体的方法如下：

桥接模块

CXX依赖于提供的函数签名说明，这些签名会在不用语言之间进行交互使用。您可以在带有 #[cxx::bridge] 属性宏注解的 Rust 模块中使用 extern 代码块提供此说明。

#[allow(unsafe_op_in_unsafe_fn)]
#[cxx::bridge(namespace = "org::blobstore")]
mod ffi {
    // Shared structs with fields visible to both languages.
    struct BlobMetadata {
        size: usize,
        tags: Vec<String>,
    }

    // Rust types and signatures exposed to C++.
    extern "Rust" {
        type MultiBuf;

        fn next_chunk(buf: &mut MultiBuf) -> &[u8];
    }

    // C++ types and signatures exposed to Rust.
    unsafe extern "C++" {
        include!("include/blobstore.h");

        type BlobstoreClient;

        fn new_blobstore_client() -> UniquePtr<BlobstoreClient>;
        fn put(self: Pin<&mut BlobstoreClient>, parts: &mut MultiBuf) -> u64;
        fn tag(self: Pin<&mut BlobstoreClient>, blobid: u64, tag: &str);
        fn metadata(&self, blobid: u64) -> BlobMetadata;
    }
}

Rust Bridge Declarations

#[cxx::bridge]
mod ffi {
    extern "Rust" {
        type MyType; // Opaque type
        fn foo(&self); // Method on `MyType`
        fn bar() -> Box<MyType>; // Free function
    }
}

struct MyType(i32);

impl MyType {
    fn foo(&self) {
        println!("{}", self.0);
    }
}

fn bar() -> Box<MyType> {
    Box::new(MyType(123))
}

生成的 C++ 代码

#[cxx::bridge]
mod ffi {
    // Rust types and signatures exposed to C++.
    extern "Rust" {
        type MultiBuf;

        fn next_chunk(buf: &mut MultiBuf) -> &[u8];
    }
}

大致生成以下 C++：

struct MultiBuf final : public ::rust::Opaque {
  ~MultiBuf() = delete;

private:
  friend ::rust::layout;
  struct layout {
    static ::std::size_t size() noexcept;
    static ::std::size_t align() noexcept;
  };
};

::rust::Slice<::std::uint8_t const> next_chunk(::org::blobstore::MultiBuf &buf) noexcept;

C++ 桥接声明

#[cxx::bridge]
mod ffi {
    // C++ types and signatures exposed to Rust.
    unsafe extern "C++" {
        include!("include/blobstore.h");

        type BlobstoreClient;

        fn new_blobstore_client() -> UniquePtr<BlobstoreClient>;
        fn put(self: Pin<&mut BlobstoreClient>, parts: &mut MultiBuf) -> u64;
        fn tag(self: Pin<&mut BlobstoreClient>, blobid: u64, tag: &str);
        fn metadata(&self, blobid: u64) -> BlobMetadata;
    }
}

大致生成以下 Rust：

#[repr(C)]
pub struct BlobstoreClient {
    _private: ::cxx::private::Opaque,
}

pub fn new_blobstore_client() -> ::cxx::UniquePtr<BlobstoreClient> {
    extern "C" {
        #[link_name = "org$blobstore$cxxbridge1$new_blobstore_client"]
        fn __new_blobstore_client() -> *mut BlobstoreClient;
    }
    unsafe { ::cxx::UniquePtr::from_raw(__new_blobstore_client()) }
}

impl BlobstoreClient {
    pub fn put(&self, parts: &mut MultiBuf) -> u64 {
        extern "C" {
            #[link_name = "org$blobstore$cxxbridge1$BlobstoreClient$put"]
            fn __put(
                _: &BlobstoreClient,
                parts: *mut ::cxx::core::ffi::c_void,
            ) -> u64;
        }
        unsafe {
            __put(self, parts as *mut MultiBuf as *mut ::cxx::core::ffi::c_void)
        }
    }
}

// ...

共享类型

#[cxx::bridge]
mod ffi {
    #[derive(Clone, Debug, Hash)]
    struct PlayingCard {
        suit: Suit,
        value: u8,  // A=1, J=11, Q=12, K=13
    }

    enum Suit {
        Clubs,
        Diamonds,
        Hearts,
        Spades,
    }
}

共享枚举

#[cxx::bridge]
mod ffi {
    enum Suit {
        Clubs,
        Diamonds,
        Hearts,
        Spades,
    }
}

Generated Rust:

#![allow(unused)]
fn main() {
#[derive(Copy, Clone, PartialEq, Eq)]
#[repr(transparent)]
pub struct Suit {
    pub repr: u8,
}

#[allow(non_upper_case_globals)]
impl Suit {
    pub const Clubs: Self = Suit { repr: 0 };
    pub const Diamonds: Self = Suit { repr: 1 };
    pub const Hearts: Self = Suit { repr: 2 };
    pub const Spades: Self = Suit { repr: 3 };
}
}

Generated C++:

enum class Suit : uint8_t {
  Clubs = 0,
  Diamonds = 1,
  Hearts = 2,
  Spades = 3,
};

Rust 错误处理

#[cxx::bridge]
mod ffi {
    extern "Rust" {
        fn fallible(depth: usize) -> Result<String>;
    }
}

fn fallible(depth: usize) -> anyhow::Result<String> {
    if depth == 0 {
        return Err(anyhow::Error::msg("fallible1 requires depth > 0"));
    }

    Ok("Success!".into())
}

C++ 错误处理

#[cxx::bridge]
mod ffi {
    unsafe extern "C++" {
        include!("example/include/example.h");
        fn fallible(depth: usize) -> Result<String>;
    }
}

fn main() {
    if let Err(err) = ffi::fallible(99) {
        eprintln!("Error: {}", err);
        process::exit(1);
    }
}

其他类型

Building in Android

创建 cc_library_static 以构建 C++ 库，包括 CXX 生成的头文件和源文件。

cc_library_static {
    name: "libcxx_test_cpp",
    srcs: ["cxx_test.cpp"],
    generated_headers: [
        "cxx-bridge-header",
        "libcxx_test_bridge_header"
    ],
    generated_sources: ["libcxx_test_bridge_code"],
}

Building in Android

创建两个 genrule：一个用于生成 CXX 头文件，另一个用于生成 CXX 源文件。然后，这些内容会被用作 cc_library_static 的输入。

// Generate a C++ header containing the C++ bindings
// to the Rust exported functions in lib.rs.
genrule {
    name: "libcxx_test_bridge_header",
    tools: ["cxxbridge"],
    cmd: "$(location cxxbridge) $(in) --header > $(out)",
    srcs: ["lib.rs"],
    out: ["lib.rs.h"],
}

// Generate the C++ code that Rust calls into.
genrule {
    name: "libcxx_test_bridge_code",
    tools: ["cxxbridge"],
    cmd: "$(location cxxbridge) $(in) > $(out)",
    srcs: ["lib.rs"],
    out: ["lib.rs.cc"],
}

Building in Android

创建一个依赖于 libcxx 和 cc_library_static 的 rust_binary。

rust_binary {
    name: "cxx_test",
    srcs: ["lib.rs"],
    rustlibs: ["libcxx"],
    static_libs: ["libcxx_test_cpp"],
}

与 Java 的互操作性

Java可以通过 Java 本地接口 (JNI) 加载共享对象。jni crate 允许您创建一个兼容的库。

首先，我们创建一个可以导出到 Java 的 Rust 函数：

interoperability/java/src/lib.rs:

#![allow(unused)]
fn main() {
//! Rust <-> Java FFI demo.

use jni::objects::{JClass, JString};
use jni::sys::jstring;
use jni::JNIEnv;

/// HelloWorld::hello method implementation.
#[no_mangle]
pub extern "system" fn Java_HelloWorld_hello(
    env: JNIEnv,
    _class: JClass,
    name: JString,
) -> jstring {
    let input: String = env.get_string(name).unwrap().into();
    let greeting = format!("Hello, {input}!");
    let output = env.new_string(greeting).unwrap();
    output.into_inner()
}
}

interoperability/java/Android.bp:

rust_ffi_shared {
    name: "libhello_jni",
    crate_name: "hello_jni",
    srcs: ["src/lib.rs"],
    rustlibs: ["libjni"],
}

We then call this function from Java:

interoperability/java/HelloWorld.java:

class HelloWorld {
    private static native String hello(String name);

    static {
        System.loadLibrary("hello_jni");
    }

    public static void main(String[] args) {
        String output = HelloWorld.hello("Alice");
        System.out.println(output);
    }
}

interoperability/java/Android.bp:

java_binary {
    name: "helloworld_jni",
    srcs: ["HelloWorld.java"],
    main_class: "HelloWorld",
    required: ["libhello_jni"],
}

最后，您可以构建、同步和运行二进制文件：

m helloworld_jni
adb sync  # requires adb root && adb remount
adb shell /system/bin/helloworld_jni

习题

这是一个小组练习：我们将查看你们正在处理的项目之一，并尝试将一些 Rust 代码集成进去。以下是一些建议：

• 使用 Rust 编写的客户端调用你的 AIDL 服务。

• 将你项目中的某个函数迁移到 Rust 中并调用它。

Welcome to Rust in Chromium

Chromium 中的第三方库支持 Rust，并使用第一方粘合代码连接 Rust 和现有 Chromium C++ 代码。

今天，我们将调用 Rust 对字符串进行一些有趣的操作。如果您的代码中某个部分是用于向用户展示 UTF8 字符串，那么可以在代码库中的对应部分按照这个步骤来操作，而不一定要在我们所讨论的确切部分。

设置

请确保您可以构建并运行 Chromium。只要您的代码较新（提交位置始于 1223636，对应于 2023 年 11 月），则可在任何平台和任何一组 build 标志下运行。

gn gen out/Debug
autoninja -C out/Debug chrome
out/Debug/chrome # or on Mac, out/Debug/Chromium.app/Contents/MacOS/Chromium

（建议使用调试 build 组件，以缩短迭代时间。这是默认值！）

如果您不具备这点，请参阅 如何构建 Chromium。注意：设置 build Chromium 需要花些时间。

此外，我们还建议您安装 Visual Studio 代码。

About the exercises

本课程的这一部分包含一系列练习，它们之间是相辅相成的。我们将在整个课程中进行这些练习，而不仅仅是在最后阶段完成。如果您没有时间完成某个部分，也无需担心：可以在下一阶段赶上进度。

Chromium 和 Cargo 的生态对比

The Rust community typically uses cargo and libraries from crates.io. Chromium is built using gn and ninja and a curated set of dependencies.

使用 Rust 编写代码时，您可以选择：

• 借助 //build/rust/*.gni 模板（例如 rust_static_library，我们稍后会介绍）使用 gn 和 ninja。该操作会使用经过审核的 Chromium 工具链和 crate。
• 使用 cargo，但 仅限于经过审核的 Chromium 工具链和 crate
• 使用 cargo，信任 工具链 和/或 从互联网下载的 crate

From here on we'll be focusing on gn and ninja, because this is how Rust code can be built into the Chromium browser. At the same time, Cargo is an important part of the Rust ecosystem and you should keep it in your toolbox.

Mini exercise

分成各个小组开展以下活动：

• 思考 cargo 在哪些场景下具有优势，并评估这些场景的风险状况。
• 讨论在使用 gn 和 ninja 以及离线 cargo 等时，需要信任哪些工具、库和人群。

Chromium Rust 政策

Chromium 尚不支持使用第一方 Rust，除非是经过 区域技术主管 批准的特殊情况。

有关 Chromium 的第三方库政策，请参阅 此处。根据该这些政策，在很多情况下允许将 Rust 作为第三方库使用，包括无论是在性能还是安全方面，它们都是理想之选。

鲜少有 Rust 库会直接公开 C/C++ API，这意味着几乎所有此类库都需要使用少量的第一方粘合代码。

特定第三方 crate 的第一方 Rust 粘合代码通常应放在 third_party/rust/<crate>/<version>/wrapper 目录中。

因此，今天的课程将重点介绍以下内容：

• 引入第三方 Rust 库 ("crates")
• 编写粘合代码，以便能够从 Chromium C++ 中使用这些 crate。

如果此政策随时间而发生变化，则本课程也会随之更新。

Build rules

Rust 代码通常使用 cargo 构建而成。为提高效率，Chromium 使用 gn 和 ninja 进行构建，其静态规则支持实现最大程度的并行处理。Rust 也不例外。

Adding Rust code to Chromium

在一些现有的 Chromium BUILD.gn 文件中，声明 rust_static_library：

import("//build/rust/rust_static_library.gni")

rust_static_library("my_rust_lib") {
  crate_root = "lib.rs"
  sources = [ "lib.rs" ]
}

您还可以在其他 Rust 目标上添加 deps。稍后，我们通过该操作来使用第三方代码。

Including unsafe Rust Code

默认情况下，禁止在 rust_static_library 中使用不安全的 Rust 代码，因为此类代码无法编译。如需使用不安全的 Rust 代码，请将 allow_unsafe = true 添加到 gn 目标中。（在本课程的稍后部分，我们将介绍在哪些情形下必须这样做。）

import("//build/rust/rust_static_library.gni")

rust_static_library("my_rust_lib") {
  crate_root = "lib.rs"
  sources = [
    "lib.rs",
    "hippopotamus.rs"
  ]
  allow_unsafe = true
}

在 Chromium C++ 中导入 Rust 代码

只需将上述目标添加到某些 Chromium C++ 目标的 deps 中即可。

import("//build/rust/rust_static_library.gni")

rust_static_library("my_rust_lib") {
  crate_root = "lib.rs"
  sources = [ "lib.rs" ]
}

# or source_set, static_library etc.
component("preexisting_cpp") {
  deps = [ ":my_rust_lib" ]
}

Visual Studio Code

在 Rust 代码中，通常会省略类型，这使得拥有强大的 IDE 甚至比 C++ 更为有用。Visual Studio Code 非常适用于在 Chromium 中处理 Rust 代码。如需使用此功能，

• 请确保您的 VSCode 安装了 rust-analyzer 扩展程序，而不是较早版本的 Rust 支持插件。
• gn gen out/Debug --export-rust-project（或输出目录的等效项）
• In -s out/Debug/rust-project.json rust-project.json

Build rules exercise

在 Chromium build 中，向 //ui/base/BUILD.gn 添加新的 Rust 目标，其中包含：

#![allow(unused)]
fn main() {
#[no_mangle]
pub extern "C" fn hello_from_rust() {
    println!("Hello from Rust!")
}
}

Important: note that no_mangle here is considered a type of unsafety by the Rust compiler, so you'll need to allow unsafe code in your gn target.

将这个新的 Rust 目标添加为 //ui/base:base 的依赖项。在 ui/base/resource/resource_bundle.cc 顶部声明此函数（稍后，我们将介绍如何通过绑定生成工具来自动执行此操作）：

extern "C" void hello_from_rust();

从 ui/base/resource/resource_bundle.cc 中的某个位置调用此函数，我们建议在从 ResourceBundle::MaybeMangleLocalizedString 的顶部调用此函数。构建并运行 Chromium，并确保多次显示 "Hello from Rust!"。

如果您使用 VSCode，现在就请设置 Rust，以便其能在 VSCode 中正常运行。这对后续练习会很有帮助。如果操作成功，则可使用右键点击 println! 上的 "Go to definition"。

如何获取帮助

• 适用于 [rust_static_library gn 模板] 的选项 (https://source.chromium.org/chromium/chromium/src/+/main:build/rust/rust_static_library.gni;l=16)
• 关于 [#[no_mangle]] 的详细信息 (https://doc.rust-lang.org/beta/reference/abi.html#the-no_mangle-attribute)
• 关于 [extern "C"] 的详细信息 (https://doc.rust-lang.org/std/keyword.extern.html)
• 关于 gn 的 [--export-rust-project] 开关的详细信息 (https://gn.googlesource.com/gn/+/main/docs/reference.md#compilation-database)
• 如何在 VSCode 中安装 rust-analyzer

测试

Rust community typically authors unit tests in a module placed in the same source file as the code being tested. This was covered earlier in the course and looks like this:

#![allow(unused)]
fn main() {
#[cfg(test)]
mod tests {
    #[test]
    fn my_test() {
        todo!()
    }
}
}

In Chromium we place unit tests in a separate source file and we continue to follow this practice for Rust --- this makes tests consistently discoverable and helps to avoid rebuilding .rs files a second time (in the test configuration).

This results in the following options for testing Rust code in Chromium:

• Native Rust tests (i.e. #[test]). Discouraged outside of //third_party/rust.
• gtest tests authored in C++ and exercising Rust via FFI calls. Sufficient when Rust code is just a thin FFI layer and the existing unit tests provide sufficient coverage for the feature.
• gtest tests authored in Rust and using the crate under test through its public API (using pub mod for_testing { ... } if needed). This is the subject of the next few slides.

rust_gtest_interop 库

The rust_gtest_interop library provides a way to:

• Use a Rust function as a gtest testcase (using the #[gtest(...)] attribute)
• Use expect_eq! and similar macros (similar to assert_eq! but not panicking and not terminating the test when the assertion fails).

Example:

use rust_gtest_interop::prelude::*;

#[gtest(MyRustTestSuite, MyAdditionTest)]
fn test_addition() {
    expect_eq!(2 + 2, 4);
}

Rust 测试的 GN 规则

The simplest way to build Rust gtest tests is to add them to an existing test binary that already contains tests authored in C++. For example:

test("ui_base_unittests") {
  ...
  sources += [ "my_rust_lib_unittest.rs" ]
  deps += [ ":my_rust_lib" ]
}

Authoring Rust tests in a separate static_library also works, but requires manually declaring the dependency on the support libraries:

rust_static_library("my_rust_lib_unittests") {
  testonly = true
  is_gtest_unittests = true
  crate_root = "my_rust_lib_unittest.rs"
  sources = [ "my_rust_lib_unittest.rs" ]
  deps = [
    ":my_rust_lib",
    "//testing/rust_gtest_interop",
  ]
}

test("ui_base_unittests") {
  ...
  deps += [ ":my_rust_lib_unittests" ]
}

chromium::import! 宏

After adding :my_rust_lib to GN deps, we still need to learn how to import and use my_rust_lib from my_rust_lib_unittest.rs. We haven't provided an explicit crate_name for my_rust_lib so its crate name is computed based on the full target path and name. Fortunately we can avoid working with such an unwieldy name by using the chromium::import! macro from the automatically-imported chromium crate:

chromium::import! {
    "//ui/base:my_rust_lib";
}

use my_rust_lib::my_function_under_test;

Under the covers the macro expands to something similar to:

extern crate ui_sbase_cmy_urust_ulib as my_rust_lib;

use my_rust_lib::my_function_under_test;

More information can be found in the doc comment of the chromium::import macro.

Testing exercise

Time for another exercise!

In your Chromium build:

• Add a testable function next to hello_from_rust. Some suggestions: adding two integers received as arguments, computing the nth Fibonacci number, summing integers in a slice, etc.
• Add a separate ..._unittest.rs file with a test for the new function.
• Add the new tests to BUILD.gn.
• Build the tests, run them, and verify that the new test works.

与 C++ 的互操作性

Rust 社区提供了多种 C++/Rust 互操作选项，并且一直在不断开发新工具。目前，Chromium 使用一种名为 CXX 的工具。

您可以使用接口定义语言（与 Rust 极为相似）描述整个语言边界，然后 CXX 工具会据此生成 Rust 和 C++ 函数及类型的声明。

See the CXX tutorial for a full example of using this.

绑定示例

CXX requires that the whole C++/Rust boundary is declared in cxx::bridge modules inside .rs source code.

#[cxx::bridge]
mod ffi {
    extern "Rust" {
        type MultiBuf;

        fn next_chunk(buf: &mut MultiBuf) -> &[u8];
    }

    unsafe extern "C++" {
        include!("example/include/blobstore.h");

        type BlobstoreClient;

        fn new_blobstore_client() -> UniquePtr<BlobstoreClient>;
        fn put(self: &BlobstoreClient, buf: &mut MultiBuf) -> Result<u64>;
    }
}

// Definitions of Rust types and functions go here

CXX 的局限性

By far the most useful page when using CXX is the type reference.

CXX 基本适用于以下情况：

• 您的 Rust-C++ 接口非常简单，可以声明其中的全部内容。
• 目前，您只使用了 CXX 提供原生支持的类型，例如 std::unique_ptr、std::string、&[u8] 等。

这样具有很多局限性，例如缺少对 Rust 的 Option 类型的支持。

由于这些局限，我们在 Chromium 中只能将 Rust 用于隔离紧密的 “叶节点”，而无法用于任意的 Rust-C++ 互操作。当您打算在 Chromium 中探索 Rust 的应用场景时，推荐先从拟定针对语言边界的 CXX 绑定入手，以查看该语言边界是否足够简单明了。

CXX 错误处理

CXX's support for Result<T,E> relies on C++ exceptions, so we can't use that in Chromium. Alternatives:

• The T part of Result<T, E> can be:

  • Returned via out parameters (e.g. via &mut T). This requires that T can be passed across the FFI boundary - for example T has to be:
    • A primitive type (like u32 or usize)
    • A type natively supported by cxx (like UniquePtr<T>) that has a suitable default value to use in a failure case (unlike Box<T>).
  • Retained on the Rust side, and exposed via reference. This may be needed when T is a Rust type, which cannot be passed across the FFI boundary, and cannot be stored in UniquePtr<T>.

• The E part of Result<T, E> can be:

  • Returned as a boolean (e.g. true representing success, and false representing failure)
  • Preserving error details is in theory possible, but so far hasn't been needed in practice.

CXX Error Handling: QR Example

The QR code generator is an example where a boolean is used to communicate success vs failure, and where the successful result can be passed across the FFI boundary:

#[cxx::bridge(namespace = "qr_code_generator")]
mod ffi {
    extern "Rust" {
        fn generate_qr_code_using_rust(
            data: &[u8],
            min_version: i16,
            out_pixels: Pin<&mut CxxVector<u8>>,
            out_qr_size: &mut usize,
        ) -> bool;
    }
}

CXX Error Handling: PNG Example

A prototype of a PNG decoder illustrates what can be done when the successful result cannot be passed across the FFI boundary:

#[cxx::bridge(namespace = "gfx::rust_bindings")]
mod ffi {
    extern "Rust" {
        /// This returns an FFI-friendly equivalent of `Result<PngReader<'a>,
        /// ()>`.
        fn new_png_reader<'a>(input: &'a [u8]) -> Box<ResultOfPngReader<'a>>;

        /// C++ bindings for the `crate::png::ResultOfPngReader` type.
        type ResultOfPngReader<'a>;
        fn is_err(self: &ResultOfPngReader) -> bool;
        fn unwrap_as_mut<'a, 'b>(
            self: &'b mut ResultOfPngReader<'a>,
        ) -> &'b mut PngReader<'a>;

        /// C++ bindings for the `crate::png::PngReader` type.
        type PngReader<'a>;
        fn height(self: &PngReader) -> u32;
        fn width(self: &PngReader) -> u32;
        fn read_rgba8(self: &mut PngReader, output: &mut [u8]) -> bool;
    }
}

Using cxx in Chromium

在 Chromium 中，针对每个需要使用 Rust 的叶节点，我们定义独立的 #[cxx::bridge] mod。通常，每个 rust_static_library 都有对应的定义。只需将

cxx_bindings = [ "my_rust_file.rs" ]
   # list of files containing #[cxx::bridge], not all source files
allow_unsafe = true

添加到您现有的 rust_static_library 以及 crate_root 和 sources 的目标中。

C++ 头文件会在合理的位置生成，因此您只需

#include "ui/base/my_rust_file.rs.h"

您会发现，//base 中提供了一些实用函数，可将 Chromium C++ 类型与 CXX Rust 类型相互转换，例如 SpanToRustSlice。

Exercise: Interoperability with C++

第一部分

• 在您之前创建的 Rust 文件中，添加 #[cxx::bridge] 来指定一个将从 C++ 调用的函数（名为 hello_from_rust），该函数不接受任何参数也不返回任何值。
• 修改之前的 hello_from_rust 函数，移除 extern "C" 和 #[no_mangle]。现在，这只是一个标准的 Rust 函数。
• 请修改 gn 目标以构建这些绑定。
• 在 C++ 代码中，移除 hello_from_rust 的正向声明，然后添加生成的头文件。
• Build and run!

第二部分

建议尝试操作一下 CXX。这有助于您更好地理解 Rust 在Chromium 中的灵活性。

Some things to try:

• 从 Rust 回调到 C++。您需要执行以下操作：
  • 创建一个附加头文件，且您可以从 cxx::bridge 对其进行 include! 操作。您需要在这个新的头文件中声明要调用的 C++ 函数。
  • 创建一个 unsafe 代码块，用于调用此类函数，也可以在 #[cxx::bridge] 中指定 unsafe 关键字，如此处所述。
  • 您可能还需要添加 #include "third_party/rust/cxx/v1/crate/include/cxx.h"
• 将 C++ 字符串从 C++ 传递到 Rust。
• 将对 C++ 对象的引用传递到 Rust。
• 刻意让 Rust 函数签名与 #[cxx::bridge] 不匹配，并逐渐熟悉所看到的错误信息。
• 刻意让 C++ 函数签名与 #[cxx::bridge] 不匹配，并适应您看到的错误。
• 将某个类型的 std::unique_ptr 从 C++ 传递到 Rust，以便 Rust 拥有某个 C++ 对象的所有权。
• 创建一个 Rust 对象并将其传递到 C++，以便 C++ 拥有它的所有权。（提示：您需要使用 Box）。
• 声明调用某个 C++ 类型的方法。从 Rust 调用它们。
• 声明调用某个 Rust 类型的方法。从 C++ 调用它们。

第三部分

现在，您已经了解了 CXX 互操作性的优势和局限，请思考几个 Rust 在 Chromium 中的应用场景，其中接口要足够简单构思该如何定义该接口。

如何获取帮助

• The cxx binding reference
• rust_static_library gn 模板

添加第三方 Crate

Rust 库被称为 crate，可在 crates.io 中找到。Rust 的 crate 之间非常容易相互依赖。事实证明，他们确实如此！

对于 Chromium 工程师来说，这种依赖关系具有以下利弊：

• 所有 crate 都使用共同的构建系统，这样我们就可以自动将其收录到 Chromium 中...
• ...但是，crate 通常具有传递依赖项，因此可能需要引入多个库。

我们将讨论以下内容：

• 如何将 crate 添加到 Chromium 源代码树中
• 如何为其制定 gn 构建规则
• 如何审核其源代码以确保足够的安全性。

配置 Cargo.tom 文件以添加 crate

Chromium 具有一组集中管理的直接 crate 依赖项。这些依赖项通过单独的 Cargo.toml 文件进行管理：

[dependencies]
bitflags = "1"
cfg-if = "1"
cxx = "1"
# lots more...

与任何其他 Cargo.toml 一样，您可以指定 有关依赖项的更多详细信息。最常见的是，您需要指定要在 crate 中启用的 features。

向 Chromium 中添加 crate 时，通常需要在附加文件 gnrt_config.toml 中提供一些额外的信息，我们将在下文中加以介绍。

配置 gnrt_config.toml

与 Cargo.toml 一起使用的是 gnrt_config.toml。此文件包含 Chromium 专用扩展程序，可用于处理 crate。

如果添加新的 crate，至少要明确指定 group。可以为以下选项之一：

#   'safe': The library satisfies the rule-of-2 and can be used in any process.
#   'sandbox': The library does not satisfy the rule-of-2 and must be used in
#              a sandboxed process such as the renderer or a utility process.
#   'test': The library is only used in tests.

例如：

[crate.my-new-crate]
group = 'test' # only used in test code

根据 crate 源代码布局，您可能还需要使用此文件指定其 LICENSE 文件的所在位置。

稍后，我们将介绍需要在此文件中配置的其他内容，以便能够解决问题。

下载 Crate

有一款名为 gnrt 的工具，具有下载 crate 以及生成 BUILD.gn 规则的功能。

首先，按如下所示下载所需的 crate：

cd chromium/src
vpython3 tools/crates/run_gnrt.py -- vendor

虽然 gnrt 工具是 Chromium 源代码的一部分，但通过运行此命令，您可以从 crates.io 下载并运行其依赖项。有关该安全决策的讨论，请参阅 前面的部分。

运行此 vendor 命令可能会下载以下内容：

• Your crate
• 直接依赖项和传递依赖项
• cargo 要求的其他 crate 的新版本，用于解析 Chromium 所需的全部 crate。

Chromium 会修复一些 crate 的补丁，并将其保存在 //third_party/rust/chromium_crates_io/patches 中。系统会自动重新应用这些补丁，但如果补丁应用失败，您可能需要进行手动操作。

生成 gn 构建规则

下载 crate 后，按如下方式生成 BUILD.gn 文件：

vpython3 tools/crates/run_gnrt.py -- gen

现在，运行 git status。您应该会看到：

• third_party/rust/chromium_crates_io/vendor 中至少包含一个新的 crate 源代码
• third_party/rust/<crate name>/v<major semver version> 中至少包含一个新的 BUILD.gn
• 相应的 README.chromium

The "major semver version" is a Rust "semver" version number.

请仔细观察，尤其是 third_party/rust 中生成的内容。

浅谈下 semver，特别是在 Chromium 中，它支持使用多个不兼容的 crate 版本。虽然在 Cargo 生态系统中不鼓励这种方式，但在某些情况下却是必要的。

解决问题

如果构建失败，可能是 build.rs 文件所致：这些程序在构建过程中执行了任意操作。这与 gn 和 ninja 的设计完全不相符，它们旨在实现静态、确定性的构建规则，以最大限度地提高构建的并行性和可重复性。

系统支持自动进行某些 build.rs 操作；而有些需要进行额外的处理：

构建脚本效果	我们的 gn 模板均支持	您需要完成的工作
检查 rustc 版本以配置启用和停用功能	是	无
检查平台或 CPU 以配置启用和停用功能	是	无
Generating code	是	是的，在 gnrt_config.toml 中指定
构建 C/C++	否	进行补丁修复
Arbitrary other actions	否	进行补丁修复

幸运的是，大多数 crate 不包含构建脚本，而且大多数的构建脚本只执行前两项操作。

构建用于生成代码的脚本

如果 ninja 提示有文件缺失，请检查 build.rs，确认其是否写入了源代码文件。

如果是，请修改 gnrt_config.toml，将 build-script-outputs 添加到 crate。如果这是一个传递依赖项（即 Chromium 代码不应直接依赖的依赖项），还要添加 allow-first-party-usage=false。该文件中已经提供了若干示例：

[crate.unicode-linebreak]
allow-first-party-usage = false
build-script-outputs = ["tables.rs"]

现在，请重新运行 gnrt.py -- gen，重新生成 BUILD.gn 文件，以通知 ninja 此特定输出文件将被用作后续构建步骤的输入。

构建用于构建 C++ 或执行任意操作的脚本

有些 crate 使用 cc crate 来构建和关联 C/C++ 库。其他 crate 会在其构建脚本中使用 bindgen 解析 C/C++。Chromium 环境中不支持进行这些操作，因为我们的 gn、ninja 和 LLVM 构建系统在表达构建操作之间的关系方面具有非常严格具体的要求。

因此，您可以选择：

• 避开这些 crate
• 对 crate 应用补丁。

补丁应保存在 third_party/rust/chromium_crates_io/patches/<crate> 中，请参阅 面向 cxx crate 的补丁 中的示例。每当 gnrt 升级该 crate 时，将会自动应用补丁文件。

依赖于 Crate

添加第三方 crate 并生成构建规则后，就可轻松使用该 crate。请找到 rust_static_library 目标，并在 crate 中的 :lib 目标上添加 dep。

Specifically,

例如：

rust_static_library("my_rust_lib") {
  crate_root = "lib.rs"
  sources = [ "lib.rs" ]
  deps = [ "//third_party/rust/example_rust_crate/v1:lib" ]
}

Auditing Third Party Crates

添加新库须遵守 Chromium 的标准 政策，当然也要遵守安全审核标准。您可能不仅要引入单个 crate，而且还要引入传递依赖项，因此可能需要审核大量代码。另一方面，安全的 Rust 代码可能产生的副作用有限。应如何对其进行审核？

随着时间的推移，Chromium 计划逐步采用以 cargo vet 为核心的处理流程。

与此同时，每当添加新的 crate 时，我们都会检查以下内容：

• 了解每个 crate 的用途。crate 之间存在什么关系？如果每个 crate 的构建系统都包含 build.rs 或过程宏，请确定它们的用途。它们能否与 Chromium 的正常构建方式相兼容。
• 检查每个 crate 是否得到合理维护。
• 使用 cd third-party/rust/chromium_crates_io; cargo review 检查已知漏洞（首先需要运行 cargo install cargo-audit，令人意外的是，这个过程中需要从互联网下载大量的依赖项 2）
• 确保所有 unsafe 代码都符合 两大规则 的要求
• 检查是否使用了 fs 或 net API
• 尽可能地仔细阅读所有代码，查找任何可能属于恶意插入，稍显异常的地方。（但现实中，您不可能做到百无遗漏，因为代码量通常太庞大了。）

以下只是一些指导建议，请与 security@chromium.org 的审核者合租，共同找出能够确保 crate 安全的正确方法。

Checking Crates into Chromium Source Code

git status 应显示以下内容：

• //third_party/rust/chromium_crates_io 中的 crate 代码
• //third_party/rust/<crate>/<version> 中的元数据（BUILD.gn 和 README.chromium）

此外，请在后面的位置添加 OWNERS 文件。

您应将所有这些内容，以及对 Cargo.toml 和 gnrt_config.toml 的更改一起提交到 Chromium 仓库中。

重要提示：您需要使用 git add -f 命令，否则 .gitignore 文件可能会导致某些文件被跳过。

在此过程中，您可能会发现由于使用了非包容性语言，导致提交前检查失败。这是因为 Rust crate 数据往往包含 git 分支的名称，而许多项目仍然在使用非包容性术语。因此，您需要运行以下命令：

infra/update_inclusive_language_presubmit_exempt_dirs.sh > infra/inclusive_language_presubmit_exempt_dirs.txt
git add -p infra/inclusive_language_presubmit_exempt_dirs.txt # add whatever changes are yours

及时更新 Crate

作为任何第三方 Chromium 依赖项的所有者，您 应使用任何安全修复程序，确保该依赖项处于最新状态。我们希望能够尽快实现对 Rust crate 自动执行此操作，但目前仍由您负责执行，就像对待任何其他第三方依赖项一样。

练习

将 uwuify 添加到 Chromium，以停用 crate 的 默认功能。假设该 crate 会交付 Chromium 时被使用，但不会用于处理不可信的输入内容。

（在下一个练习中，我们将使用 Chromium 中的 uwuify；但您也可以跳过这一步，现在就开始此操作。或者，您可以创建一个使用 uwuify 的新 rust_executable 目标。

Students will need to download lots of transitive dependencies.

The total crates needed are:

• instant,
• lock_api,
• parking_lot,
• parking_lot_core,
• redox_syscall,
• scopeguard,
• smallvec, and
• uwuify.

If students are downloading even more than that, they probably forgot to turn off the default features.

Thanks to Daniel Liu for this crate!

Bringing It Together --- Exercise

在本练习中，您将运用之前所学的全部知识，添加一项全新的 Chromium 功能。

The Brief from Product Management

在偏僻的热带雨林中，发现住着一群小精灵。我们务必尽快创建一款精灵版 Chromium，交付给他们。

要求是将 Chromium 的所有界面字符串翻译为精灵语。

由于时间紧迫，无法等待准确的翻译。但幸运的是，精灵语与英语非常接近，然后我们发现有一个 Rust crate 可以执行此翻译任务。

事实上，您已经 在上一个练习中导入了该 crate。

（显然，对 Chrome 进行准确恰当的翻译，需要做到极度细致和全力以赴。请勿交付此产品！）

步骤

修改 ResourceBundle::MaybeMangleLocalizedString，在显示之前对所有字符串进行 uwu 处理。在这个特殊的 Chromium 版本中，无论 mangle_localized_strings_ 的设置如何，都应该始终执行此操作。

如果您正确完成了这些练习中的所有操作，那么恭喜您已经成功创建了一款精灵版 Chrome！

Students will likely need some hints here. Hints include:

• UTF16 与 UTF8。学生应注意，Rust 字符串始终采用 UTF8 编码，它们可能会决定在 C++ 端使用 base::UTF16ToUTF8 进行转换，然后再进行反向转换。
• 如果学生决定在 Rust 端进行转换，则需要考虑使用 String::from_utf16 方法，同时注意错误处理，并确定哪些 CXX 支持的类型可以传输大量 u16s。
• 学生可以通过多种方式设计 C++/Rust 边界，例如按值传递和返回字符串，或对字符串采取可变引用。如果使用可变引用，CXX 可能会提示学生需要使用 Pin 方法。您可能需要解释“Pin”的用途，以及为何 CXX 需要它来对 C++ 数据进行可变引用：原因是 C++ 数据无法像 Rust 数据那样自由移动，它可能包含自引用指针。
• 对于包含 ResourceBundle::MaybeMangleLocalizedString 的 C++ 目标，需要依赖于 rust_static_library 目标。学生可能已经执行了此操作。
• rust_static_library 目标需要依赖于 //third_party/rust/uwuify/v0_2:lib。

练习解答

Solutions to the Chromium exercises can be found in this series of CLs.

Welcome to Bare Metal Rust

这是单独为裸机 Rust 开设的课程，为期一天，主要面向熟悉 Rust 基础知识的人员（例如已学完 Comprehensive Rust），最好也有一些使用其他语言（例如 C）进行裸机编程的经验。

今天，我们将探讨 bare-metal Rust：即在没有操作系统支持的情况下运行 Rust 代码。该部分主要分为以下内容：

• 什么是 no_std Rust？
• 编写微控制器固件。
• 为应用处理器编写引导加载程序 / 内核代码。
• 有助于裸机 Rust 开发的一些实用 crate。

在本课程的微控制器部分，我们将使用 BBC micro:bit v2 作为示例。这是一款基于 Nordic nRF51822 微控制器的 开发板，配有一些 LED 和按钮、连接 I2C 的加速度计和罗盘，以及板载 SWD 调试程序。

首先，请安装我们稍后需要用到的一些工具。在 gLinux 或 Debian 上：

sudo apt install gcc-aarch64-linux-gnu gdb-multiarch libudev-dev picocom pkg-config qemu-system-arm
rustup update
rustup target add aarch64-unknown-none thumbv7em-none-eabihf
rustup component add llvm-tools-preview
cargo install cargo-binutils cargo-embed

然后，向 plugdev 组中的用户授予 micro:bit 编程器的访问权限：

echo 'SUBSYSTEM=="usb", ATTR{idVendor}=="0d28", MODE="0664", GROUP="plugdev"' |\
  sudo tee /etc/udev/rules.d/50-microbit.rules
sudo udevadm control --reload-rules

在 MacOS 上：

xcode-select --install
brew install gdb picocom qemu
brew install --cask gcc-aarch64-embedded
rustup update
rustup target add aarch64-unknown-none thumbv7em-none-eabihf
rustup component add llvm-tools-preview
cargo install cargo-binutils cargo-embed

no_std

core	alloc	std
Slice、&str、CStr NonZeroU8... Option、Result Display、Debug、write!... Iterator panic!、assert_eq!... NonNull 和所有常见的指针相关函数 Future 和 async/await fence、AtomicBool、AtomicPtr、AtomicU32... Duration	Box、Cow、Arc、Rc Vec、BinaryHeap、BtreeMap、LinkedList、VecDeque String、CString、format!	Error HashMap Mutex、Condvar、Barrier、Once、RwLock、mpsc File 和 fs 的其余部分 println!、Read、Write、Stdin、Stdout 以及 io 的其余部分 Path, OsString net Command、Child、ExitCode spawn、sleep 和 thread 的其余部分 SystemTime、Instant

• HashMap 依赖于 RNG。
• std 会重新导出 core 和 alloc 的内容。

极小的 no_std 程序

#![no_main]
#![no_std]

use core::panic::PanicInfo;

#[panic_handler]
fn panic(_panic: &PanicInfo) -> ! {
    loop {}
}

• 这将编译为空二进制文件。
• std 提供了一个 panic 处理程序；如果没有它，我们就必须自行提供。
• 其他 crate（例如 panic-halt）也可以提供该处理程序。
• 根据目标不同，可能需要使用 panic = "abort" 进行编译，以避免出现与 eh_personality 相关的错误。
• 请注意，未提供 main 函数或任何其他入口点；您可以自行定义入口点。通常需要使用链接器脚本和一些汇编代码进行设置工作，以便 Rust 代码能够顺利运行。

alloc

如需使用 alloc，您必须实现 全局（堆）分配器。

#![no_main]
#![no_std]

extern crate alloc;
extern crate panic_halt as _;

use alloc::string::ToString;
use alloc::vec::Vec;
use buddy_system_allocator::LockedHeap;

#[global_allocator]
static HEAP_ALLOCATOR: LockedHeap<32> = LockedHeap::<32>::new();

static mut HEAP: [u8; 65536] = [0; 65536];

pub fn entry() {
    // SAFETY: `HEAP` is only used here and `entry` is only called once.
    unsafe {
        // Give the allocator some memory to allocate.
        HEAP_ALLOCATOR.lock().init(HEAP.as_mut_ptr() as usize, HEAP.len());
    }

    // Now we can do things that require heap allocation.
    let mut v = Vec::new();
    v.push("A string".to_string());
}

• buddy_system_allocator 是第三方 crate，用于实现基本伙伴系统分配器。还可以使用其他 crate，或者自行编写 crate，或者接入现有分配器。
• LockedHeap 的常量参数是分配器的最大阶数；即在本例中，它可以最多分配 2**32 字节大小的区域。
• 如果依赖项树中的所有 crate 都依赖于 alloc，则您必须在二进制文件中明确定义一个全局分配器。通常，在顶级二进制 crate 中完成此操作。
• 为了确保能够成功关联 panic_halt crate，以便我们获取其 panic 处理程序，必须使用 extern crate panic_halt as _ 方法。
• 我们可以构建该示例，但由于没有入口点，无法运行。

微控制器

cortex_m_rt crate 提供针对 Cortex M 微控制器的重置处理程序（以及其他内容）。

#![no_main]
#![no_std]

extern crate panic_halt as _;

mod interrupts;

use cortex_m_rt::entry;

#[entry]
fn main() -> ! {
    loop {}
}

接下来，我们看看随着抽象层级的不断提升，该如何访问外围设备。

• cortex_m_rt::entry 宏要求函数的类型为 fn() -> !，因为返回重置处理程序会毫无意义。
• 使用 cargo embed --bin minimum 运行该示例

原始 MMIO

大多数微控制器通过内存映射 IO 访问外围设备。现在试着开启 micro:bit 上的 LED 指示灯：

#![no_main]
#![no_std]

extern crate panic_halt as _;

mod interrupts;

use core::mem::size_of;
use cortex_m_rt::entry;

/// GPIO port 0 peripheral address
const GPIO_P0: usize = 0x5000_0000;

// GPIO peripheral offsets
const PIN_CNF: usize = 0x700;
const OUTSET: usize = 0x508;
const OUTCLR: usize = 0x50c;

// PIN_CNF fields
const DIR_OUTPUT: u32 = 0x1;
const INPUT_DISCONNECT: u32 = 0x1 << 1;
const PULL_DISABLED: u32 = 0x0 << 2;
const DRIVE_S0S1: u32 = 0x0 << 8;
const SENSE_DISABLED: u32 = 0x0 << 16;

#[entry]
fn main() -> ! {
    // Configure GPIO 0 pins 21 and 28 as push-pull outputs.
    let pin_cnf_21 = (GPIO_P0 + PIN_CNF + 21 * size_of::<u32>()) as *mut u32;
    let pin_cnf_28 = (GPIO_P0 + PIN_CNF + 28 * size_of::<u32>()) as *mut u32;
    // SAFETY: The pointers are to valid peripheral control registers, and no
    // aliases exist.
    unsafe {
        pin_cnf_21.write_volatile(
            DIR_OUTPUT
                | INPUT_DISCONNECT
                | PULL_DISABLED
                | DRIVE_S0S1
                | SENSE_DISABLED,
        );
        pin_cnf_28.write_volatile(
            DIR_OUTPUT
                | INPUT_DISCONNECT
                | PULL_DISABLED
                | DRIVE_S0S1
                | SENSE_DISABLED,
        );
    }

    // Set pin 28 low and pin 21 high to turn the LED on.
    let gpio0_outset = (GPIO_P0 + OUTSET) as *mut u32;
    let gpio0_outclr = (GPIO_P0 + OUTCLR) as *mut u32;
    // SAFETY: The pointers are to valid peripheral control registers, and no
    // aliases exist.
    unsafe {
        gpio0_outclr.write_volatile(1 << 28);
        gpio0_outset.write_volatile(1 << 21);
    }

    loop {}
}

• 将 GPIO 0 的引脚 21 连接到 LED 矩阵的第一列，将引脚 28 连接到第一行。

使用以下命令运行该示例：

cargo embed --bin mmio

外围设备访问 crate

svd2rust 使用 CMSIS-SVD 文件为内存映射外围设备生成了大部分安全的 Rust 封装容器。

#![no_main]
#![no_std]

extern crate panic_halt as _;

use cortex_m_rt::entry;
use nrf52833_pac::Peripherals;

#[entry]
fn main() -> ! {
    let p = Peripherals::take().unwrap();
    let gpio0 = p.P0;

    // Configure GPIO 0 pins 21 and 28 as push-pull outputs.
    gpio0.pin_cnf[21].write(|w| {
        w.dir().output();
        w.input().disconnect();
        w.pull().disabled();
        w.drive().s0s1();
        w.sense().disabled();
        w
    });
    gpio0.pin_cnf[28].write(|w| {
        w.dir().output();
        w.input().disconnect();
        w.pull().disabled();
        w.drive().s0s1();
        w.sense().disabled();
        w
    });

    // Set pin 28 low and pin 21 high to turn the LED on.
    gpio0.outclr.write(|w| w.pin28().clear());
    gpio0.outset.write(|w| w.pin21().set());

    loop {}
}

• SVD（系统视图描述）文件通常是由芯片供应商提供的 XML 文件，用于描述设备的内存映射。
  • 这些文件按照外围设备、寄存器、字段和值进行组织分类，其中包含名称、内容描述、地址等信息。
  • 由于 SVD 文件常常存在错误和不完整的情况，因此有许多项目会修复这些错误，补充缺失的相关信息，并发布生成的 crate。
• cortex-m-rt 提供矢量表以及其他功能。
• 如果您使用 cargo install cargo-binutils，则可以运行 cargo objdump --bin pac -- -d --no-show-raw-insn，查看生成的二进制文件。

使用以下命令运行该示例：

cargo embed --bin pac

HAL crates

许多微控制器的 HAL crate 为各种外围设备提供了封装容器。通常，这些封装容器可以实现 embedded-hal 中的各种 trait。

#![no_main]
#![no_std]

extern crate panic_halt as _;

use cortex_m_rt::entry;
use nrf52833_hal::gpio::{p0, Level};
use nrf52833_hal::pac::Peripherals;
use nrf52833_hal::prelude::*;

#[entry]
fn main() -> ! {
    let p = Peripherals::take().unwrap();

    // Create HAL wrapper for GPIO port 0.
    let gpio0 = p0::Parts::new(p.P0);

    // Configure GPIO 0 pins 21 and 28 as push-pull outputs.
    let mut col1 = gpio0.p0_28.into_push_pull_output(Level::High);
    let mut row1 = gpio0.p0_21.into_push_pull_output(Level::Low);

    // Set pin 28 low and pin 21 high to turn the LED on.
    col1.set_low().unwrap();
    row1.set_high().unwrap();

    loop {}
}

• set_low 和 set_high 是 embedded_hal OutputPin trait 上的方法。
• HAL crate 被广泛用于许多 Cortex-M 和 RISC-V 设备，包括各种 STM32、GD32、nRF、NXP、MSP430、AVR 和 PIC 微控制器。

使用以下命令运行该示例：

cargo embed --bin hal

Board support crates

为了方便使用，板级支持 crate 为特定开发板提供了更高级别的封装功能。

#![no_main]
#![no_std]

extern crate panic_halt as _;

use cortex_m_rt::entry;
use microbit::hal::prelude::*;
use microbit::Board;

#[entry]
fn main() -> ! {
    let mut board = Board::take().unwrap();

    board.display_pins.col1.set_low().unwrap();
    board.display_pins.row1.set_high().unwrap();

    loop {}
}

• 在本例中，该板级支持 crate 仅提供了一些实用的名称和初始化功能。
• 除微控制器以外，该 crate 还包含一些可用于板载设备的驱动程序。
  • microbit-v2 包含一个可用于 LED 矩阵的简单驱动程序。

使用以下命令运行该示例：

cargo embed --bin board_support

类型状态模式

#[entry]
fn main() -> ! {
    let p = Peripherals::take().unwrap();
    let gpio0 = p0::Parts::new(p.P0);

    let pin: P0_01<Disconnected> = gpio0.p0_01;

    // let gpio0_01_again = gpio0.p0_01; // Error, moved.
    let pin_input: P0_01<Input<Floating>> = pin.into_floating_input();
    if pin_input.is_high().unwrap() {
        // ...
    }
    let mut pin_output: P0_01<Output<OpenDrain>> = pin_input
        .into_open_drain_output(OpenDrainConfig::Disconnect0Standard1, Level::Low);
    pin_output.set_high().unwrap();
    // pin_input.is_high(); // Error, moved.

    let _pin2: P0_02<Output<OpenDrain>> = gpio0
        .p0_02
        .into_open_drain_output(OpenDrainConfig::Disconnect0Standard1, Level::Low);
    let _pin3: P0_03<Output<PushPull>> =
        gpio0.p0_03.into_push_pull_output(Level::Low);

    loop {}
}

• 引脚无法实现 Copy 或 Clone，因此，每个引脚只能存在一个实例。一旦某个引脚从端口结构体移出，任何人都无法再使用它。
• 更改引脚的配置会使旧的引脚实例失效，因此您无法再继续使用旧实例。
• 值的类型表示其所处状态：例如，在本例中，表示 GPIO 引脚的配置状态。通过这种方式，可将状态机编码成类型系统，并确保在未正确完成引脚配置之前，不会尝试以某种方式使用引脚。编译时捕获了非法状态转换。
• 您可以在输入引脚上调用 is_high，在输出引脚上调用 set_high，反之则不行。
• 许多 HAL crate 都遵循此模式。

embedded-hal

embedded-hal crate 提供许多适用于常见微控制器外围设备的 trait。

• GPIO
• ADC
• I2C、SPI、UART、CAN
• RNG
• 定时器
• 监控定时器

然后，其他 crate 可以根据这些 trait 实现 驱动程序，例如加速度计驱动程序可能需要通过 I2C 或 SPI 总线实现。

• 还有适用于许多微控制器以及其他平台（例如 Raspberry Pi 上的 Linux）的实现。
• 我们正在开发 async 版本的 embedded-hal，但目前还欠缺稳定性。

probe-rs 和 cargo-embed

probe-rs 是一个方便的嵌入式调试工具集，跟 OpenOCD 较为相似，但集成度更高。

• 通过 CMSIS-DAP、ST-Link 和 J-Link 探针实现 SWD（串行调试）和 JTAG
• GDB 桩和 Microsoft DAP（调试适配器协议）服务器
• Cargo integration

cargo-embed 是一个 cargo 子命令，用于构建和刷写二进制文件、记录 RTT（实时传输）输出以及连接 GDB。您可通过项目目录中的 Embed.toml 文件对其进行配置。

• CMSIS-DAP 是一项基于 USB 的 Arm 标准协议，使得电路内调试程序能够接入各种 Arm Cortex 处理器的 CoreSight 调试访问端口。BBC micro:bit 的板载调试程序所使用的便是此协议。
• ST-Link 属于 ST Microelectronics 产品系列，J-Link 是 SEGGER 系列。
• 调试访问端口通常为 5 针 JTAG 接口或 2 针串行线调试接口。
• probe-rs 是一个库，如有需要，可以将其集成到您的工具中。
• 借助 Microsoft 调试适配器协议，VSCode 和其他 IDE 可以调试任何受支持的微控制器上运行的代码。
• cargo-embed 是使用 probe-rs 库构建的二进制文件。
• RTT（实时传输）是一种通过多个环形缓冲区，在调试主机和目标之间进行数据传输的机制。

调试

Embed.toml:

[default.general]
chip = "nrf52833_xxAA"

[debug.gdb]
enabled = true

在 src/bare-metal/microcontrollers/examples/ 目录下某个终端中：

cargo embed --bin board_support debug

在同一目录下的另一个终端中：

在 gLinux 或 Debian 上：

gdb-multiarch target/thumbv7em-none-eabihf/debug/board_support --eval-command="target remote :1337"

在 MacOS 上：

arm-none-eabi-gdb target/thumbv7em-none-eabihf/debug/board_support --eval-command="target remote :1337"

在 GDB 中，请尝试运行以下命令：

b src/bin/board_support.rs:29
b src/bin/board_support.rs:30
b src/bin/board_support.rs:32
c
c
c

Other projects

• RTIC
  • "Real-Time Interrupt-driven Concurrency"
  • 共享资源管理、消息传递、任务调度、定时器队列
• Embassy
  • 具有优先级、计时器、网络和 USB 功能的 async 执行器
• TockOS
  • 以安全为中心的 RTOS，具有抢占式调度功能和提供内存保护单元支持
• Hubris
  • Oxide Computer Company 开发的微内核 RTOS，提供内存保护、非特权驱动程序和 IPC 功能
• FreeRTOS 的绑定
• 有些平台可以实现 std，例如 esp-idf。

• RTIC 可被视为 RTOS 或并发框架，
  • 但不包含任何 HAL。
  • 它使用 Cortex-M NVIC（嵌套虚拟中断控制器）进行调度，而不是选用适合的内核。
  • 仅限 Cortex-M。
• Google 在 Haven 微控制器上使用 TockOS 作为 Titan 安全密钥的操作系统。
• FreeRTOS 主要使用 C 语言编写，但也提供了专用于编写应用的 Rust 绑定。

习题

我们将从 I2C 罗盘读取方向，并将读数记录到串行端口。

After looking at the exercises, you can look at the solutions provided.

罗盘

我们将从 I2C 罗盘读取方向，并将读数记录到串行端口。如有时间，请尝试通过 LED 灯亮起的方式，或者使用按钮来显示方向。

提示：

• 请参阅 lsm303agr 和 microbit-v2 crate，以及 micro:bit 硬件 相关文档。
• LSM303AGR 惯性测量装置与内部 I2C 总线相连接。
• TWI 是 I2C 的别称，因此 I2C 主外围设备称为 TWIM。
• LSM303AGR 驱动程序需要使用某种方法来实现 embedded_hal::blocking::i2c::WriteRead trait。microbit::hal::Twim 结构体便可以做到这一点。
• 您拥有一个 microbit::Board 结构体，其中包含各种引脚和外围设备的字段。
• 如有需要，您还可以查看 nRF52833 数据表，但对本练习来说这不是必需的。

下载 练习模板 并在 compass 目录中查找以下文件。

src/main.rs:

#![no_main]
#![no_std]

extern crate panic_halt as _;

use core::fmt::Write;
use cortex_m_rt::entry;
use microbit::{hal::uarte::{Baudrate, Parity, Uarte}, Board};

#[entry]
fn main() -> ! {
    let board = Board::take().unwrap();

    // Configure serial port.
    let mut serial = Uarte::new(
        board.UARTE0,
        board.uart.into(),
        Parity::EXCLUDED,
        Baudrate::BAUD115200,
    );

    // Use the system timer as a delay provider.
    let mut delay = Delay::new(board.SYST);

    // Set up the I2C controller and Inertial Measurement Unit.
    // TODO

    writeln!(serial, "Ready.").unwrap();

    loop {
        // Read compass data and log it to the serial port.
        // TODO
    }
}

Cargo.toml（无需对此进行更改）：

[workspace]

[package]
name = "compass"
version = "0.1.0"
edition = "2021"
publish = false

[dependencies]
cortex-m-rt = "0.7.3"
embedded-hal = "1.0.0"
lsm303agr = "0.3.0"
microbit-v2 = "0.13.0"
panic-halt = "0.2.0"

Embed.toml（无需对此进行更改）：

[default.general]
chip = "nrf52833_xxAA"

[debug.gdb]
enabled = true

[debug.reset]
halt_afterwards = true

.cargo/config.toml（无需对此进行更改）：

[build]
target = "thumbv7em-none-eabihf" # Cortex-M4F

[target.'cfg(all(target_arch = "arm", target_os = "none"))']
rustflags = ["-C", "link-arg=-Tlink.x"]

运行以下命令查看 Linux 上的串行输出：

picocom --baud 115200 --imap lfcrlf /dev/ttyACM0

或者在 Mac OS 上，如下所示（设备名称可能略有不同）：

picocom --baud 115200 --imap lfcrlf /dev/tty.usbmodem14502

使用 Ctrl+A Ctrl+Q 退出 picocom。

裸机 Rust 上午练习

罗盘

(返回练习)

#![no_main]
#![no_std]

extern crate panic_halt as _;

use core::fmt::Write;
use cortex_m_rt::entry;
use core::cmp::{max, min};
use lsm303agr::{
    AccelMode, AccelOutputDataRate, Lsm303agr, MagMode, MagOutputDataRate,
};
use microbit::display::blocking::Display;
use microbit::hal::prelude::*;
use microbit::hal::twim::Twim;
use microbit::hal::uarte::{Baudrate, Parity, Uarte};
use microbit::hal::{Delay, Timer};
use microbit::pac::twim0::frequency::FREQUENCY_A;
use microbit::Board;

const COMPASS_SCALE: i32 = 30000;
const ACCELEROMETER_SCALE: i32 = 700;

#[entry]
fn main() -> ! {
    let board = Board::take().unwrap();

    // Configure serial port.
    let mut serial = Uarte::new(
        board.UARTE0,
        board.uart.into(),
        Parity::EXCLUDED,
        Baudrate::BAUD115200,
    );

    // Use the system timer as a delay provider.
    let mut delay = Delay::new(board.SYST);

    // Set up the I2C controller and Inertial Measurement Unit.
    writeln!(serial, "Setting up IMU...").unwrap();
    let i2c = Twim::new(board.TWIM0, board.i2c_internal.into(), FREQUENCY_A::K100);
    let mut imu = Lsm303agr::new_with_i2c(i2c);
    imu.init().unwrap();
    imu.set_mag_mode_and_odr(
        &mut delay,
        MagMode::HighResolution,
        MagOutputDataRate::Hz50,
    )
    .unwrap();
    imu.set_accel_mode_and_odr(
        &mut delay,
        AccelMode::Normal,
        AccelOutputDataRate::Hz50,
    )
    .unwrap();
    let mut imu = imu.into_mag_continuous().ok().unwrap();

    // Set up display and timer.
    let mut timer = Timer::new(board.TIMER0);
    let mut display = Display::new(board.display_pins);

    let mut mode = Mode::Compass;
    let mut button_pressed = false;

    writeln!(serial, "Ready.").unwrap();

    loop {
        // Read compass data and log it to the serial port.
        while !(imu.mag_status().unwrap().xyz_new_data()
            && imu.accel_status().unwrap().xyz_new_data())
        {}
        let compass_reading = imu.magnetic_field().unwrap();
        let accelerometer_reading = imu.acceleration().unwrap();
        writeln!(
            serial,
            "{},{},{}\t{},{},{}",
            compass_reading.x_nt(),
            compass_reading.y_nt(),
            compass_reading.z_nt(),
            accelerometer_reading.x_mg(),
            accelerometer_reading.y_mg(),
            accelerometer_reading.z_mg(),
        )
        .unwrap();

        let mut image = [[0; 5]; 5];
        let (x, y) = match mode {
            Mode::Compass => (
                scale(-compass_reading.x_nt(), -COMPASS_SCALE, COMPASS_SCALE, 0, 4)
                    as usize,
                scale(compass_reading.y_nt(), -COMPASS_SCALE, COMPASS_SCALE, 0, 4)
                    as usize,
            ),
            Mode::Accelerometer => (
                scale(
                    accelerometer_reading.x_mg(),
                    -ACCELEROMETER_SCALE,
                    ACCELEROMETER_SCALE,
                    0,
                    4,
                ) as usize,
                scale(
                    -accelerometer_reading.y_mg(),
                    -ACCELEROMETER_SCALE,
                    ACCELEROMETER_SCALE,
                    0,
                    4,
                ) as usize,
            ),
        };
        image[y][x] = 255;
        display.show(&mut timer, image, 100);

        // If button A is pressed, switch to the next mode and briefly blink all LEDs
        // on.
        if board.buttons.button_a.is_low().unwrap() {
            if !button_pressed {
                mode = mode.next();
                display.show(&mut timer, [[255; 5]; 5], 200);
            }
            button_pressed = true;
        } else {
            button_pressed = false;
        }
    }
}

#[derive(Copy, Clone, Debug, Eq, PartialEq)]
enum Mode {
    Compass,
    Accelerometer,
}

impl Mode {
    fn next(self) -> Self {
        match self {
            Self::Compass => Self::Accelerometer,
            Self::Accelerometer => Self::Compass,
        }
    }
}

fn scale(value: i32, min_in: i32, max_in: i32, min_out: i32, max_out: i32) -> i32 {
    let range_in = max_in - min_in;
    let range_out = max_out - min_out;
    cap(min_out + range_out * (value - min_in) / range_in, min_out, max_out)
}

fn cap(value: i32, min_value: i32, max_value: i32) -> i32 {
    max(min_value, min(value, max_value))
}

应用处理器

到目前为止，我们已经讨论了微控制器，例如 Arm Cortex-M 系列。现在，尝试为 Cortex-A 编写一些代码。为简单起见，我们只使用 QEMU 的 aarch64 virt 开发板 进行编写。

• 一般来说，微控制器不具备 MMU 或多级特权（例如，Arm CPU 中的异常级别，x86 中的铃声级别）的功能，而应用处理器则具备这些功能。
• QEMU 支持针对每种架构模拟不同的机器或板级模型。“虚拟”开发板并不适用于任何特定的真实硬件，而是专为虚拟机设计的。

准备使用 Rust

在开始运行 Rust 代码之前，我们需要进行一些初始化设置。

.section .init.entry, "ax"
.global entry
entry:
    /*
     * Load and apply the memory management configuration, ready to enable MMU and
     * caches.
     */
    adrp x30, idmap
    msr ttbr0_el1, x30

    mov_i x30, .Lmairval
    msr mair_el1, x30

    mov_i x30, .Ltcrval
    /* Copy the supported PA range into TCR_EL1.IPS.*/
    mrs x29, id_aa64mmfr0_el1
    bfi x30, x29, #32, #4

    msr tcr_el1, x30

    mov_i x30, .Lsctlrval

    /*
     * Ensure everything before this point has completed, then invalidate any
     * potentially stale local TLB entries before they start being used.
     */
    isb
    tlbi vmalle1
    ic iallu
    dsb nsh
    isb

    /*
     * Configure sctlr_el1 to enable MMU and cache and don't proceed until this
     * has completed.
     */
    msr sctlr_el1, x30
    isb

    /* Disable trapping floating point access in EL1.*/
    mrs x30, cpacr_el1
    orr x30, x30, #(0x3 << 20)
    msr cpacr_el1, x30
    isb

    /* Zero out the bss section.*/
    adr_l x29, bss_begin
    adr_l x30, bss_end
0:  cmp x29, x30
    b.hs 1f
    stp xzr, xzr, [x29], #16
    b 0b

1:  /* Prepare the stack.*/
    adr_l x30, boot_stack_end
    mov sp, x30

    /* Set up exception vector.*/
    adr x30, vector_table_el1
    msr vbar_el1, x30

    /* Call into Rust code.*/
    bl main

    /* Loop forever waiting for interrupts.*/
2:  wfi
    b 2b

• 这与 C 语言的情况相同：初始化处理器状态，将 BSS 清零，然后设置堆栈指针。
  • BSS（由于历史原因，称为代码块起始符）属于对象文件的一部分，其中包含静态分配的变量，这些变量被初始化为零。图像中省略了这些符号，以避免因存储零值而占用过多空间。编译器假定加载器会负责将它们清零。
• BSS 可能已经被清零，具体取决于内存的初始化方式以及图像的加载方式，但为了确保起见，我们会将其手动清零。
• 我们需要先启用 MMU 和缓存功能，然后才能读取或写入任何内存。否则：
  • 非对齐访问将会出错。我们为 aarch64-unknown-none 目标构建 Rust 代码，该目标会设置 +Strict-align 以防止编译器生成非对齐访问，因此在本例中应该没有问题，但一般情况下并不一定如此。
  • 如果是在虚拟机中运行该命令，可能会导致缓存一致性问题。问题在于，虚拟机是在禁用缓存的情况下直接访问内存，而主机具有同一内存的缓存别名。即使主机并没有明确访问该内存，推测性访问仍然会导致缓存被填充，然后在清除缓存或虚拟机启用缓存时，任何一方对于该内存进行的更改就会丢失。（使用物理地址来键控缓存，而 VA 或 IPA。）
• 为简单起见，我们只使用硬编码的分页表（请参阅 dmap.S），其通过身份映射将前一个 1 GiB 的地址空间用于设备，紧接着的 1 GiB 用于 DRAM，然后在更高位置预留了 1 GiB 给其他设备。这与 QEMU 使用的内存布局一致。
• 我们还设置了异常矢量 (vbar_el1)，稍后会对此进行详细介绍。
• 今天下午的所有示例都假定我们将在异常级别 1 (EL1) 下运行。如果您需要在其他异常级别下运行，则需要修改相应的 entry.S。

内嵌汇编

有时，如果无法通过 Rust 代码实现某些操作，我们就需要使用汇编来解决。例如，如需发出 HVC（Hypervisor 调用）来指示固件关闭系统，请使用以下命令：

#![no_main]
#![no_std]

use core::arch::asm;
use core::panic::PanicInfo;

mod exceptions;

const PSCI_SYSTEM_OFF: u32 = 0x84000008;

#[no_mangle]
extern "C" fn main(_x0: u64, _x1: u64, _x2: u64, _x3: u64) {
    // SAFETY: this only uses the declared registers and doesn't do anything
    // with memory.
    unsafe {
        asm!("hvc #0",
            inout("w0") PSCI_SYSTEM_OFF => _,
            inout("w1") 0 => _,
            inout("w2") 0 => _,
            inout("w3") 0 => _,
            inout("w4") 0 => _,
            inout("w5") 0 => _,
            inout("w6") 0 => _,
            inout("w7") 0 => _,
            options(nomem, nostack)
        );
    }

    loop {}
}

（如果确实想要这样做，请使用 smccc crate，其中包含适用于所有这些函数的封装容器。）

• PSCI 是 Arm 电源状态协调接口，为一组标准函数，用于管理系统和 CPU 电源状态等。在许多系统中，通过 EL3 固件和 Hypervisor 来实现该函数。
• 0 => _ 语法表示在运行内嵌汇编代码之前将寄存器初始化为 0，并在之后忽略寄存器中的内容。我们需要使用 inout 而非 in，因为该调用操作可能会破坏寄存器中的内容。
• 所用 main 函数必须是 #[no_mangle] 和 extern "C"，因为是从 entry.S 中的入口点调用该函数。
• _x0–_x3 表示寄存器 x0-x3 的值，引导加载程序通常使用这些值来传递各种内容（例如将指针传递到设备树）。根据标准的 aarch64 调用规范（extern "C"指定使用此规范），需要使用寄存器 x0-x7 将前 8 个参数传递给函数，因此 entry.S 无需执行任何特殊操作，只要确保不会更改这些寄存器。
• 在 QEMU 中，使用 src/bare-metal/aps/examples 目录下的 make qemu_psci 运行该示例。

MMIO 的易失性内存访问

• 使用 pointer::read_volatile 和 pointer::write_volatile。
• 切勿提及引用。
• 借助 addr_of!，您无需创建中间引用即可获取结构体字段。

• 易失性访问：执行读取或写入操作可能会产生副作用，因此应阻止编译器或硬件对这些操作进行重新排序、复制或省略。
  • 通常情况下，如果您先写入操作，紧接着进行读取操作（例如通过可变引用），则编译器可能会认为读取的值是最新写入的值，就不再执行实际的内存读取过程。
• 虽然在对硬件进行易失性访问时，一些 crate 确实会提及引用，但这很不安全。只要存在引用，编译器就会选择对其进行解引用操作。
• 使用 addr_of! 宏可以从结构体指针中获取结构体字段的指针。

编写 UART 驱动程序

QEMU “虚拟机”具有 PL011 UART，现在为其编写驱动程序。

const FLAG_REGISTER_OFFSET: usize = 0x18;
const FR_BUSY: u8 = 1 << 3;
const FR_TXFF: u8 = 1 << 5;

/// Minimal driver for a PL011 UART.
#[derive(Debug)]
pub struct Uart {
    base_address: *mut u8,
}

impl Uart {
    /// Constructs a new instance of the UART driver for a PL011 device at the
    /// given base address.
    ///
    /// # Safety
    ///
    /// The given base address must point to the 8 MMIO control registers of a
    /// PL011 device, which must be mapped into the address space of the process
    /// as device memory and not have any other aliases.
    pub unsafe fn new(base_address: *mut u8) -> Self {
        Self { base_address }
    }

    /// Writes a single byte to the UART.
    pub fn write_byte(&self, byte: u8) {
        // Wait until there is room in the TX buffer.
        while self.read_flag_register() & FR_TXFF != 0 {}

        // SAFETY: We know that the base address points to the control
        // registers of a PL011 device which is appropriately mapped.
        unsafe {
            // Write to the TX buffer.
            self.base_address.write_volatile(byte);
        }

        // Wait until the UART is no longer busy.
        while self.read_flag_register() & FR_BUSY != 0 {}
    }

    fn read_flag_register(&self) -> u8 {
        // SAFETY: We know that the base address points to the control
        // registers of a PL011 device which is appropriately mapped.
        unsafe { self.base_address.add(FLAG_REGISTER_OFFSET).read_volatile() }
    }
}

• 请注意，使用 Uart::new 方法不安全，而其他方法则安全。原因在于，只要 Uart::new 的调用方保证满足其安全要求（即所指定的 UART 只有一个驱动程序实例，且没有其他内容与其地址空间存在重叠），那么后续调用 write_byte 始终是安全的，因为我们假定需要满足的前提条件。
• 我们也可以采用相反的方式（即确保 new 安全，但 write_byte 不安全），不过这样会很不方便，因为每当调用 write_byte 时都需要推断是否安全。
• 这是安全地封装不安全代码时常见的策略：即在少数调用代码的地方进行安全验证，而不是在很多地方进行。

更多 trait

已经派生了 Debug trait。如果再实现更多 trait，会大有帮助。

use core::fmt::{self, Write};

impl Write for Uart {
    fn write_str(&mut self, s: &str) -> fmt::Result {
        for c in s.as_bytes() {
            self.write_byte(*c);
        }
        Ok(())
    }
}

// SAFETY: `Uart` just contains a pointer to device memory, which can be
// accessed from any context.
unsafe impl Send for Uart {}

• 通过实现 Write，我们可以将 write! 和 writeln! 宏与 Uart 类型搭配使用。
• 在 QEMU 中，使用 src/bare-metal/aps/examples 目录下的 make qemu_minimal 运行该示例。

更好的 UART 驱动程序

事实上，PL011 具有 很多额外的寄存器，但通过添加偏移量构建指针的方式来访问这些寄存器，既容易出错又难以读取。此外，其中有些寄存器是位字段，非常便于以结构化方式访问。

偏移	寄存器名称	宽度
0x00	DR	12
0x04	RSR	4
0x18	FR	9
0x20	ILPR	8
0x24	IBRD	16
0x28	FBRD	6
0x2c	LCR_H	8
0x30	CR	16
0x34	IFLS	6
0x38	IMSC	11
0x3c	RIS	11
0x40	MIS	11
0x44	ICR	11
0x48	DMACR	3

• 为简洁起见，我们省略了一些 ID 寄存器。

Bitflags

bitflags crate 非常适用于处理 bitflag。

use bitflags::bitflags;

bitflags! {
    /// Flags from the UART flag register.
    #[repr(transparent)]
    #[derive(Copy, Clone, Debug, Eq, PartialEq)]
    struct Flags: u16 {
        /// Clear to send.
        const CTS = 1 << 0;
        /// Data set ready.
        const DSR = 1 << 1;
        /// Data carrier detect.
        const DCD = 1 << 2;
        /// UART busy transmitting data.
        const BUSY = 1 << 3;
        /// Receive FIFO is empty.
        const RXFE = 1 << 4;
        /// Transmit FIFO is full.
        const TXFF = 1 << 5;
        /// Receive FIFO is full.
        const RXFF = 1 << 6;
        /// Transmit FIFO is empty.
        const TXFE = 1 << 7;
        /// Ring indicator.
        const RI = 1 << 8;
    }
}

• bitflags! 宏会创建类似于 Flags(u16) 的新类型，以及一系列用于获取和设置标记的方法实现。

多个寄存器

我们可以使用结构体来表示 UART 寄存器的内存布局。

#[repr(C, align(4))]
struct Registers {
    dr: u16,
    _reserved0: [u8; 2],
    rsr: ReceiveStatus,
    _reserved1: [u8; 19],
    fr: Flags,
    _reserved2: [u8; 6],
    ilpr: u8,
    _reserved3: [u8; 3],
    ibrd: u16,
    _reserved4: [u8; 2],
    fbrd: u8,
    _reserved5: [u8; 3],
    lcr_h: u8,
    _reserved6: [u8; 3],
    cr: u16,
    _reserved7: [u8; 3],
    ifls: u8,
    _reserved8: [u8; 3],
    imsc: u16,
    _reserved9: [u8; 2],
    ris: u16,
    _reserved10: [u8; 2],
    mis: u16,
    _reserved11: [u8; 2],
    icr: u16,
    _reserved12: [u8; 2],
    dmacr: u8,
    _reserved13: [u8; 3],
}

• 通过运行 #[repr(C)] 命令，指示编译器按顺序布置结构体字段，遵循与 C 语言相同的规则。这是确保结构体具有可预测布局的必要条件，因为默认的 Rust 表示法允许编译器（以及其他内容）按照其认为合适的方式重新排列字段。

驱动程序

现在将新的 Registers 结构体用于我们的驱动程序。

/// Driver for a PL011 UART.
#[derive(Debug)]
pub struct Uart {
    registers: *mut Registers,
}

impl Uart {
    /// Constructs a new instance of the UART driver for a PL011 device at the
    /// given base address.
    ///
    /// # Safety
    ///
    /// The given base address must point to the 8 MMIO control registers of a
    /// PL011 device, which must be mapped into the address space of the process
    /// as device memory and not have any other aliases.
    pub unsafe fn new(base_address: *mut u32) -> Self {
        Self { registers: base_address as *mut Registers }
    }

    /// Writes a single byte to the UART.
    pub fn write_byte(&self, byte: u8) {
        // Wait until there is room in the TX buffer.
        while self.read_flag_register().contains(Flags::TXFF) {}

        // SAFETY: We know that self.registers points to the control registers
        // of a PL011 device which is appropriately mapped.
        unsafe {
            // Write to the TX buffer.
            addr_of_mut!((*self.registers).dr).write_volatile(byte.into());
        }

        // Wait until the UART is no longer busy.
        while self.read_flag_register().contains(Flags::BUSY) {}
    }

    /// Reads and returns a pending byte, or `None` if nothing has been
    /// received.
    pub fn read_byte(&self) -> Option<u8> {
        if self.read_flag_register().contains(Flags::RXFE) {
            None
        } else {
            // SAFETY: We know that self.registers points to the control
            // registers of a PL011 device which is appropriately mapped.
            let data = unsafe { addr_of!((*self.registers).dr).read_volatile() };
            // TODO: Check for error conditions in bits 8-11.
            Some(data as u8)
        }
    }

    fn read_flag_register(&self) -> Flags {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL011 device which is appropriately mapped.
        unsafe { addr_of!((*self.registers).fr).read_volatile() }
    }
}

• 请注意，只使用 addr_of! / addr_of_mut! 获取指向各个字段的指针，而不创建中间引用，这样很不安全。

开始使用

使用驱动程序编写一个小程序，将数据写入串行控制台，并回显传入的字节。

#![no_main]
#![no_std]

mod exceptions;
mod pl011;

use crate::pl011::Uart;
use core::fmt::Write;
use core::panic::PanicInfo;
use log::error;
use smccc::psci::system_off;
use smccc::Hvc;

/// Base address of the primary PL011 UART.
const PL011_BASE_ADDRESS: *mut u32 = 0x900_0000 as _;

#[no_mangle]
extern "C" fn main(x0: u64, x1: u64, x2: u64, x3: u64) {
    // SAFETY: `PL011_BASE_ADDRESS` is the base address of a PL011 device, and
    // nothing else accesses that address range.
    let mut uart = unsafe { Uart::new(PL011_BASE_ADDRESS) };

    writeln!(uart, "main({x0:#x}, {x1:#x}, {x2:#x}, {x3:#x})").unwrap();

    loop {
        if let Some(byte) = uart.read_byte() {
            uart.write_byte(byte);
            match byte {
                b'\r' => {
                    uart.write_byte(b'\n');
                }
                b'q' => break,
                _ => {}
            }
        }
    }

    writeln!(uart, "Bye!").unwrap();
    system_off::<Hvc>().unwrap();
}

• 与 内嵌汇编 示例一样，从 entry.S 中的入口点代码调用此 main 函数。如需了解详情，请参阅演讲者备注。
• 在 QEMU 中，使用 src/bare-metal/aps/examples 目录下的 make qemu 运行该示例。

日志记录

最好能够使用 log crate 中的日志记录宏。可以通过实现“Log”trait 来做到这一点。

use crate::pl011::Uart;
use core::fmt::Write;
use log::{LevelFilter, Log, Metadata, Record, SetLoggerError};
use spin::mutex::SpinMutex;

static LOGGER: Logger = Logger { uart: SpinMutex::new(None) };

struct Logger {
    uart: SpinMutex<Option<Uart>>,
}

impl Log for Logger {
    fn enabled(&self, _metadata: &Metadata) -> bool {
        true
    }

    fn log(&self, record: &Record) {
        writeln!(
            self.uart.lock().as_mut().unwrap(),
            "[{}] {}",
            record.level(),
            record.args()
        )
        .unwrap();
    }

    fn flush(&self) {}
}

/// Initialises UART logger.
pub fn init(uart: Uart, max_level: LevelFilter) -> Result<(), SetLoggerError> {
    LOGGER.uart.lock().replace(uart);

    log::set_logger(&LOGGER)?;
    log::set_max_level(max_level);
    Ok(())
}

• 使用 log 进行解封装是安全的，因为我们会在调用 set_logger 之前初始化 LOGGER。

开始使用

需要先初始化日志记录器，然后才能使用它。

#![no_main]
#![no_std]

mod exceptions;
mod logger;
mod pl011;

use crate::pl011::Uart;
use core::panic::PanicInfo;
use log::{error, info, LevelFilter};
use smccc::psci::system_off;
use smccc::Hvc;

/// Base address of the primary PL011 UART.
const PL011_BASE_ADDRESS: *mut u32 = 0x900_0000 as _;

#[no_mangle]
extern "C" fn main(x0: u64, x1: u64, x2: u64, x3: u64) {
    // SAFETY: `PL011_BASE_ADDRESS` is the base address of a PL011 device, and
    // nothing else accesses that address range.
    let uart = unsafe { Uart::new(PL011_BASE_ADDRESS) };
    logger::init(uart, LevelFilter::Trace).unwrap();

    info!("main({x0:#x}, {x1:#x}, {x2:#x}, {x3:#x})");

    assert_eq!(x1, 42);

    system_off::<Hvc>().unwrap();
}

#[panic_handler]
fn panic(info: &PanicInfo) -> ! {
    error!("{info}");
    system_off::<Hvc>().unwrap();
    loop {}
}

• 请注意，panic 紧急处理程序现在可以记录各类 panic 详细信息。
• 在 QEMU 中，使用 src/bare-metal/aps/examples 目录下的 make qemu_logger 运行该示例。

异常

AArch64 定义了一个包含 16 个条目的异常矢量表，适用于处理 4 种状态（当前 EL 使用 SP0，当前 EL 使用 SPx，较低 EL 使用 AArch64，较低 EL 使用 AArch32）下的 4 种异常（同步、IRQ、FIQ、SError）。可以通过汇编方式实现这一操作，以便在调用 Rust 代码之前将易失性寄存器保存到堆栈：

use log::error;
use smccc::psci::system_off;
use smccc::Hvc;

#[no_mangle]
extern "C" fn sync_exception_current(_elr: u64, _spsr: u64) {
    error!("sync_exception_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn irq_current(_elr: u64, _spsr: u64) {
    error!("irq_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn fiq_current(_elr: u64, _spsr: u64) {
    error!("fiq_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn serr_current(_elr: u64, _spsr: u64) {
    error!("serr_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn sync_lower(_elr: u64, _spsr: u64) {
    error!("sync_lower");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn irq_lower(_elr: u64, _spsr: u64) {
    error!("irq_lower");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn fiq_lower(_elr: u64, _spsr: u64) {
    error!("fiq_lower");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn serr_lower(_elr: u64, _spsr: u64) {
    error!("serr_lower");
    system_off::<Hvc>().unwrap();
}

• EL 指异常级别；我们今天下午的所有示例都在 EL1 级别下运行。
• 为简单起见，我们没有区分当前 EL 异常中的 SP0 和 SPx，也没有区分较低 EL 异常中的 AArch32 和 AArch64。
• 在本示例中，只需记录异常并进行关机操作，因为预计不会发生任何此类异常。
• 我们可以将异常处理程序和主执行上下文视为不同的线程。通过 Send 和 Sync 控制它们之间可以共享的内容，就像使用线程进行共享一样。例如，如果想在异常处理程序和程序的其余部分之间共享某个值，并且使用 Send 而非 Sync，则需要将该值封装在诸如 Mutex 之类的内容中，并放入静态变量。

Other projects

• oreboot
  • "coreboot without the C"
  • 支持 x86、aarch64 和 RISC-V。
  • 依赖于 LinuxBoot，而不是许多驱动程序本身。
• Rust RaspberryPi 操作系统教程
  • 初始化、UART 驱动程序、简单引导加载程序、JTAG、异常级别、异常处理、分页表
  • 在 Rust 中，有些用于处理缓存维护和初始化的方法并不安全，不适宜照搬到正式版代码中。
• cargo-call-stack
  • 使用静态分析来确定堆栈用量上限。

• 在 RaspberryPi 操作系统教程中，先运行 Rust 代码然后启用 MMU 和缓存。此操作会读取和写入内存（例如堆栈）。不过：
  • 如果不启用 MMU 和缓存，非对齐访问将会出错。它使用 aarch64-unknown-none 进行构建，后者会设置 +strict-align 以防止编译器生成非对齐访问，因此应该没有问题，但一般情况下并不一定如此。
  • 如果是在虚拟机中运行该命令，可能会导致缓存一致性问题。问题在于，虚拟机是在禁用缓存的情况下直接访问内存，而主机具有同一内存的缓存别名。即使主机并没有明确访问该内存，推测性访问仍然会导致缓存被填充，然后任何一方对于该内存进行的更改就会丢失。再次强调，尽管在此特定示例中（即在硬件上直接运行且无 Hypervisor）这种做法是可行的，但总的来说，这并非一种良好策略。

Useful crates

接下来介绍几个 crate，用于解决裸机编程中的一些常见问题。

zerocopy

zerocopy crate（源自 Fuchsia）提供了 trait 和宏，用于确保在字节序列和其他类型之间进行安全转换。

use zerocopy::AsBytes;

#[repr(u32)]
#[derive(AsBytes, Debug, Default)]
enum RequestType {
    #[default]
    In = 0,
    Out = 1,
    Flush = 4,
}

#[repr(C)]
#[derive(AsBytes, Debug, Default)]
struct VirtioBlockRequest {
    request_type: RequestType,
    reserved: u32,
    sector: u64,
}

fn main() {
    let request = VirtioBlockRequest {
        request_type: RequestType::Flush,
        sector: 42,
        ..Default::default()
    };

    assert_eq!(
        request.as_bytes(),
        &[4, 0, 0, 0, 0, 0, 0, 0, 42, 0, 0, 0, 0, 0, 0, 0]
    );
}

这不适用于 MMIO（因为它不使用易失性读取和写入），但在与硬件共享的结构（例如通过 DMA 传输或发送到外部接口）中进行操作时会很有用。

• 对于可以接受任何字节模式的类型，都可以实现 FromBytes方法，因此可以对不受信任的字节序列进行安全转换。
• 如果尝试为这些类型派生 FromBytes，都将会失败，因为 RequestType 不会将所有可能的 u32 值用作判别标识，所以并非所有的字节模式都有效。
• zerocopy::byteorder 提供了适用于字节顺序感知的数字基元类型。
• 使用 src/bare-metal/useful-crates/zerocopy-example/ 目录下的 cargo run 运行该示例。（由于存在 crate 依赖项，无法在 Playground 中运行该示例。）

aarch64-paging

借助 aarch64-paging crate，您可根据 AArch64 虚拟内存系统架构创建分页表。

use aarch64_paging::{
    idmap::IdMap,
    paging::{Attributes, MemoryRegion},
};

const ASID: usize = 1;
const ROOT_LEVEL: usize = 1;

// Create a new page table with identity mapping.
let mut idmap = IdMap::new(ASID, ROOT_LEVEL);
// Map a 2 MiB region of memory as read-only.
idmap.map_range(
    &MemoryRegion::new(0x80200000, 0x80400000),
    Attributes::NORMAL | Attributes::NON_GLOBAL | Attributes::READ_ONLY,
).unwrap();
// Set `TTBR0_EL1` to activate the page table.
idmap.activate();

• 目前，该方法仅支持 EL1 级别，但也可以直接添加对其他异常级别的支持。
• 在 Android 中，该方法适用于 受保护的虚拟机固件。
• 由于此示例需要在真实硬件上或在 QEMU 中运行，因此没有简单的运行方法可用。

buddy_system_allocator

buddy_system_allocator 是第三方 crate，用于实现基本伙伴系统分配器。同时，也可将其用于 LockedHeap 以实现 GlobalAlloc，这样便可以使用标准的 alloc crate（正如在 之前部分 中所示），或者用于分配其他地址空间。例如，我们可能需要为 PCI BAR 分配 MMIO 空间：

use buddy_system_allocator::FrameAllocator;
use core::alloc::Layout;

fn main() {
    let mut allocator = FrameAllocator::<32>::new();
    allocator.add_frame(0x200_0000, 0x400_0000);

    let layout = Layout::from_size_align(0x100, 0x100).unwrap();
    let bar = allocator
        .alloc_aligned(layout)
        .expect("Failed to allocate 0x100 byte MMIO region");
    println!("Allocated 0x100 byte MMIO region at {:#x}", bar);
}

• PCI BAR 的对齐方式始终与其大小相等。
• 使用 src/bare-metal/useful-crates/allocator-example/ 目录下的 cargo run 运行该示例。（由于存在 crate 依赖项，无法在 Playground 中运行该示例。）

tinyvec

有时，需要一些像 Vec 一样能够调整大小的特性，但无需进行堆分配。tinyvec 提供了以下特性：由数组或 slice 支持的矢量，该矢量支持进行静态分配或堆分配；用于跟踪使用的元素数量，如果元素使用量超过了分配额度，则会出现 panic。

use tinyvec::{array_vec, ArrayVec};

fn main() {
    let mut numbers: ArrayVec<[u32; 5]> = array_vec!(42, 66);
    println!("{numbers:?}");
    numbers.push(7);
    println!("{numbers:?}");
    numbers.remove(1);
    println!("{numbers:?}");
}

• 根据 tinyvec要求，元素类型需实现初始化 Default。
• Rust Playground 中包含 tinyvec，因此本示例将以内嵌方式正常运行。

spin

在 core 或 alloc 中无法使用 std::sync::Mutex 和 std::sync 中的其他同步基元。那么该如何管理同步或内部可变性，例如在不同 CPU 之间共享状态？

spin crate 为许多基元提供了基于自旋锁的等效方法。

use spin::mutex::SpinMutex;

static counter: SpinMutex<u32> = SpinMutex::new(0);

fn main() {
    println!("count: {}", counter.lock());
    *counter.lock() += 2;
    println!("count: {}", counter.lock());
}

• 在中断处理程序中进行锁定操作时，请注意避免出现死锁的情况。
• spin also has a ticket lock mutex implementation; equivalents of RwLock, Barrier and Once from std::sync; and Lazy for lazy initialisation.
• once_cell crate 也提供了一些适用于延迟初始化的实用类型，它们与 spin::once::Once 所用方法略有不同。
• Rust Playground 中包含 spin，因此本示例将以内嵌方式正常运行。

Android

如需在 AOSP 中构建裸机 Rust 二进制文件，应使用 rust_ffi_static Soong 规则来构建 Rust 代码，然后通过 cc_binary 和链接器脚本生成二进制文件本身，再使用 raw_binary 将 ELF 转换为可以正常运行的原始二进制文件。

rust_ffi_static {
    name: "libvmbase_example",
    defaults: ["vmbase_ffi_defaults"],
    crate_name: "vmbase_example",
    srcs: ["src/main.rs"],
    rustlibs: [
        "libvmbase",
    ],
}

cc_binary {
    name: "vmbase_example",
    defaults: ["vmbase_elf_defaults"],
    srcs: [
        "idmap.S",
    ],
    static_libs: [
        "libvmbase_example",
    ],
    linker_scripts: [
        "image.ld",
        ":vmbase_sections",
    ],
}

raw_binary {
    name: "vmbase_example_bin",
    stem: "vmbase_example.bin",
    src: ":vmbase_example",
    enabled: false,
    target: {
        android_arm64: {
            enabled: true,
        },
    },
}

vmbase

对于在 aarch64 上使用 crosvm 运行的虚拟机，vmbase 库提供了链接器脚本和实用的默认构建规则，以及入口点、UART 控制台日志记录等功能。

#![no_main]
#![no_std]

use vmbase::{main, println};

main!(main);

pub fn main(arg0: u64, arg1: u64, arg2: u64, arg3: u64) {
    println!("Hello world");
}

• main! 宏用于标记需从 vmbase 入口点调用的 main 函数。
• vmbase 入口点用于处理控制台初始化，并在 main 函数返回时发送 PSCI_SYSTEM_OFF 命令以关闭虚拟机。

习题

我们将为 PL031 实时时钟设备编写驱动程序。

After looking at the exercises, you can look at the solutions provided.

RTC 驱动程序

QEMU aarch64 虚拟机在 0x9010000 地址处 配备了 PL031 实时时钟。对于本练习，应该为其编写驱动程序。

1. 使用该时钟可将当前时间输出到串行控制台。您可以使用 chrono crate 设置日期/时间格式。
2. 通过匹配寄存器和原始中断状态，使得系统在某段指定的时间内一直进行繁忙等待（例如 3 秒后）。（在循环操作中调用 core::hint::spin_loop。）
3. _进行扩展（如有时间）：_启用并处理由 RTC 匹配产生的中断。可以使用 arm-gic crate 中提供的驱动程序来配置 Arm 通用中断控制器。
   • 请使用 RTC 中断，将其作为 IntId::spi(2) 连接到 GIC。
   • 启用中断后，可以通过 arm_gic::wfi() 让核心进入休眠状态，直到它收到中断信号。

下载 练习模板 并在 rtc 目录中查找以下文件。

src/main.rs:

#![no_main]
#![no_std]

mod exceptions;
mod logger;
mod pl011;

use crate::pl011::Uart;
use arm_gic::gicv3::GicV3;
use core::panic::PanicInfo;
use log::{error, info, trace, LevelFilter};
use smccc::psci::system_off;
use smccc::Hvc;

/// Base addresses of the GICv3.
const GICD_BASE_ADDRESS: *mut u64 = 0x800_0000 as _;
const GICR_BASE_ADDRESS: *mut u64 = 0x80A_0000 as _;

/// Base address of the primary PL011 UART.
const PL011_BASE_ADDRESS: *mut u32 = 0x900_0000 as _;

#[no_mangle]
extern "C" fn main(x0: u64, x1: u64, x2: u64, x3: u64) {
    // SAFETY: `PL011_BASE_ADDRESS` is the base address of a PL011 device, and
    // nothing else accesses that address range.
    let uart = unsafe { Uart::new(PL011_BASE_ADDRESS) };
    logger::init(uart, LevelFilter::Trace).unwrap();

    info!("main({:#x}, {:#x}, {:#x}, {:#x})", x0, x1, x2, x3);

    // SAFETY: `GICD_BASE_ADDRESS` and `GICR_BASE_ADDRESS` are the base
    // addresses of a GICv3 distributor and redistributor respectively, and
    // nothing else accesses those address ranges.
    let mut gic = unsafe { GicV3::new(GICD_BASE_ADDRESS, GICR_BASE_ADDRESS) };
    gic.setup();

    // TODO: Create instance of RTC driver and print current time.

    // TODO: Wait for 3 seconds.

    system_off::<Hvc>().unwrap();
}

#[panic_handler]
fn panic(info: &PanicInfo) -> ! {
    error!("{info}");
    system_off::<Hvc>().unwrap();
    loop {}
}

src/exceptions.rs（只需在本练习的第 3 部分更改此项）：

#![allow(unused)]
fn main() {
// Copyright 2023 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//      http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

use arm_gic::gicv3::GicV3;
use log::{error, info, trace};
use smccc::psci::system_off;
use smccc::Hvc;

#[no_mangle]
extern "C" fn sync_exception_current(_elr: u64, _spsr: u64) {
    error!("sync_exception_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn irq_current(_elr: u64, _spsr: u64) {
    trace!("irq_current");
    let intid =
        GicV3::get_and_acknowledge_interrupt().expect("No pending interrupt");
    info!("IRQ {intid:?}");
}

#[no_mangle]
extern "C" fn fiq_current(_elr: u64, _spsr: u64) {
    error!("fiq_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn serr_current(_elr: u64, _spsr: u64) {
    error!("serr_current");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn sync_lower(_elr: u64, _spsr: u64) {
    error!("sync_lower");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn irq_lower(_elr: u64, _spsr: u64) {
    error!("irq_lower");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn fiq_lower(_elr: u64, _spsr: u64) {
    error!("fiq_lower");
    system_off::<Hvc>().unwrap();
}

#[no_mangle]
extern "C" fn serr_lower(_elr: u64, _spsr: u64) {
    error!("serr_lower");
    system_off::<Hvc>().unwrap();
}
}

src/logger.rs（无需对此进行更改）：

#![allow(unused)]
fn main() {
// Copyright 2023 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//      http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

// ANCHOR: main
use crate::pl011::Uart;
use core::fmt::Write;
use log::{LevelFilter, Log, Metadata, Record, SetLoggerError};
use spin::mutex::SpinMutex;

static LOGGER: Logger = Logger { uart: SpinMutex::new(None) };

struct Logger {
    uart: SpinMutex<Option<Uart>>,
}

impl Log for Logger {
    fn enabled(&self, _metadata: &Metadata) -> bool {
        true
    }

    fn log(&self, record: &Record) {
        writeln!(
            self.uart.lock().as_mut().unwrap(),
            "[{}] {}",
            record.level(),
            record.args()
        )
        .unwrap();
    }

    fn flush(&self) {}
}

/// Initialises UART logger.
pub fn init(uart: Uart, max_level: LevelFilter) -> Result<(), SetLoggerError> {
    LOGGER.uart.lock().replace(uart);

    log::set_logger(&LOGGER)?;
    log::set_max_level(max_level);
    Ok(())
}
}

src/pl011.rs（无需对此进行更改）：

#![allow(unused)]
fn main() {
// Copyright 2023 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//      http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

#![allow(unused)]

use core::fmt::{self, Write};
use core::ptr::{addr_of, addr_of_mut};

// ANCHOR: Flags
use bitflags::bitflags;

bitflags! {
    /// Flags from the UART flag register.
    #[repr(transparent)]
    #[derive(Copy, Clone, Debug, Eq, PartialEq)]
    struct Flags: u16 {
        /// Clear to send.
        const CTS = 1 << 0;
        /// Data set ready.
        const DSR = 1 << 1;
        /// Data carrier detect.
        const DCD = 1 << 2;
        /// UART busy transmitting data.
        const BUSY = 1 << 3;
        /// Receive FIFO is empty.
        const RXFE = 1 << 4;
        /// Transmit FIFO is full.
        const TXFF = 1 << 5;
        /// Receive FIFO is full.
        const RXFF = 1 << 6;
        /// Transmit FIFO is empty.
        const TXFE = 1 << 7;
        /// Ring indicator.
        const RI = 1 << 8;
    }
}
// ANCHOR_END: Flags

bitflags! {
    /// Flags from the UART Receive Status Register / Error Clear Register.
    #[repr(transparent)]
    #[derive(Copy, Clone, Debug, Eq, PartialEq)]
    struct ReceiveStatus: u16 {
        /// Framing error.
        const FE = 1 << 0;
        /// Parity error.
        const PE = 1 << 1;
        /// Break error.
        const BE = 1 << 2;
        /// Overrun error.
        const OE = 1 << 3;
    }
}

// ANCHOR: Registers
#[repr(C, align(4))]
struct Registers {
    dr: u16,
    _reserved0: [u8; 2],
    rsr: ReceiveStatus,
    _reserved1: [u8; 19],
    fr: Flags,
    _reserved2: [u8; 6],
    ilpr: u8,
    _reserved3: [u8; 3],
    ibrd: u16,
    _reserved4: [u8; 2],
    fbrd: u8,
    _reserved5: [u8; 3],
    lcr_h: u8,
    _reserved6: [u8; 3],
    cr: u16,
    _reserved7: [u8; 3],
    ifls: u8,
    _reserved8: [u8; 3],
    imsc: u16,
    _reserved9: [u8; 2],
    ris: u16,
    _reserved10: [u8; 2],
    mis: u16,
    _reserved11: [u8; 2],
    icr: u16,
    _reserved12: [u8; 2],
    dmacr: u8,
    _reserved13: [u8; 3],
}
// ANCHOR_END: Registers

// ANCHOR: Uart
/// Driver for a PL011 UART.
#[derive(Debug)]
pub struct Uart {
    registers: *mut Registers,
}

impl Uart {
    /// Constructs a new instance of the UART driver for a PL011 device at the
    /// given base address.
    ///
    /// # Safety
    ///
    /// The given base address must point to the MMIO control registers of a
    /// PL011 device, which must be mapped into the address space of the process
    /// as device memory and not have any other aliases.
    pub unsafe fn new(base_address: *mut u32) -> Self {
        Self { registers: base_address as *mut Registers }
    }

    /// Writes a single byte to the UART.
    pub fn write_byte(&self, byte: u8) {
        // Wait until there is room in the TX buffer.
        while self.read_flag_register().contains(Flags::TXFF) {}

        // SAFETY: We know that self.registers points to the control registers
        // of a PL011 device which is appropriately mapped.
        unsafe {
            // Write to the TX buffer.
            addr_of_mut!((*self.registers).dr).write_volatile(byte.into());
        }

        // Wait until the UART is no longer busy.
        while self.read_flag_register().contains(Flags::BUSY) {}
    }

    /// Reads and returns a pending byte, or `None` if nothing has been
    /// received.
    pub fn read_byte(&self) -> Option<u8> {
        if self.read_flag_register().contains(Flags::RXFE) {
            None
        } else {
            // SAFETY: We know that self.registers points to the control
            // registers of a PL011 device which is appropriately mapped.
            let data = unsafe { addr_of!((*self.registers).dr).read_volatile() };
            // TODO: Check for error conditions in bits 8-11.
            Some(data as u8)
        }
    }

    fn read_flag_register(&self) -> Flags {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL011 device which is appropriately mapped.
        unsafe { addr_of!((*self.registers).fr).read_volatile() }
    }
}
// ANCHOR_END: Uart

impl Write for Uart {
    fn write_str(&mut self, s: &str) -> fmt::Result {
        for c in s.as_bytes() {
            self.write_byte(*c);
        }
        Ok(())
    }
}

// Safe because it just contains a pointer to device memory, which can be
// accessed from any context.
unsafe impl Send for Uart {}
}

Cargo.toml（无需对此进行更改）：

[workspace]

[package]
name = "rtc"
version = "0.1.0"
edition = "2021"
publish = false

[dependencies]
arm-gic = "0.1.0"
bitflags = "2.5.0"
chrono = { version = "0.4.37", default-features = false }
log = "0.4.21"
smccc = "0.1.1"
spin = "0.9.8"

[build-dependencies]
cc = "1.0.94"

build.rs（无需对此进行更改）：

// Copyright 2023 Google LLC
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//      http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

use cc::Build;
use std::env;

fn main() {
    #[cfg(target_os = "linux")]
    env::set_var("CROSS_COMPILE", "aarch64-linux-gnu");
    #[cfg(not(target_os = "linux"))]
    env::set_var("CROSS_COMPILE", "aarch64-none-elf");

    Build::new()
        .file("entry.S")
        .file("exceptions.S")
        .file("idmap.S")
        .compile("empty")
}

entry.S（无需对此进行更改）：

/*
 * Copyright 2023 Google LLC
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     https://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

.macro adr_l, reg:req, sym:req
	adrp \reg, \sym
	add \reg, \reg, :lo12:\sym
.endm

.macro mov_i, reg:req, imm:req
	movz \reg, :abs_g3:\imm
	movk \reg, :abs_g2_nc:\imm
	movk \reg, :abs_g1_nc:\imm
	movk \reg, :abs_g0_nc:\imm
.endm

.set .L_MAIR_DEV_nGnRE,	0x04
.set .L_MAIR_MEM_WBWA,	0xff
.set .Lmairval, .L_MAIR_DEV_nGnRE | (.L_MAIR_MEM_WBWA << 8)

/* 4 KiB granule size for TTBR0_EL1. */
.set .L_TCR_TG0_4KB, 0x0 << 14
/* 4 KiB granule size for TTBR1_EL1. */
.set .L_TCR_TG1_4KB, 0x2 << 30
/* Disable translation table walk for TTBR1_EL1, generating a translation fault instead. */
.set .L_TCR_EPD1, 0x1 << 23
/* Translation table walks for TTBR0_EL1 are inner sharable. */
.set .L_TCR_SH_INNER, 0x3 << 12
/*
 * Translation table walks for TTBR0_EL1 are outer write-back read-allocate write-allocate
 * cacheable.
 */
.set .L_TCR_RGN_OWB, 0x1 << 10
/*
 * Translation table walks for TTBR0_EL1 are inner write-back read-allocate write-allocate
 * cacheable.
 */
.set .L_TCR_RGN_IWB, 0x1 << 8
/* Size offset for TTBR0_EL1 is 2**39 bytes (512 GiB). */
.set .L_TCR_T0SZ_512, 64 - 39
.set .Ltcrval, .L_TCR_TG0_4KB | .L_TCR_TG1_4KB | .L_TCR_EPD1 | .L_TCR_RGN_OWB
.set .Ltcrval, .Ltcrval | .L_TCR_RGN_IWB | .L_TCR_SH_INNER | .L_TCR_T0SZ_512

/* Stage 1 instruction access cacheability is unaffected. */
.set .L_SCTLR_ELx_I, 0x1 << 12
/* SP alignment fault if SP is not aligned to a 16 byte boundary. */
.set .L_SCTLR_ELx_SA, 0x1 << 3
/* Stage 1 data access cacheability is unaffected. */
.set .L_SCTLR_ELx_C, 0x1 << 2
/* EL0 and EL1 stage 1 MMU enabled. */
.set .L_SCTLR_ELx_M, 0x1 << 0
/* Privileged Access Never is unchanged on taking an exception to EL1. */
.set .L_SCTLR_EL1_SPAN, 0x1 << 23
/* SETEND instruction disabled at EL0 in aarch32 mode. */
.set .L_SCTLR_EL1_SED, 0x1 << 8
/* Various IT instructions are disabled at EL0 in aarch32 mode. */
.set .L_SCTLR_EL1_ITD, 0x1 << 7
.set .L_SCTLR_EL1_RES1, (0x1 << 11) | (0x1 << 20) | (0x1 << 22) | (0x1 << 28) | (0x1 << 29)
.set .Lsctlrval, .L_SCTLR_ELx_M | .L_SCTLR_ELx_C | .L_SCTLR_ELx_SA | .L_SCTLR_EL1_ITD | .L_SCTLR_EL1_SED
.set .Lsctlrval, .Lsctlrval | .L_SCTLR_ELx_I | .L_SCTLR_EL1_SPAN | .L_SCTLR_EL1_RES1

/**
 * This is a generic entry point for an image. It carries out the operations required to prepare the
 * loaded image to be run. Specifically, it zeroes the bss section using registers x25 and above,
 * prepares the stack, enables floating point, and sets up the exception vector. It preserves x0-x3
 * for the Rust entry point, as these may contain boot parameters.
 */
.section .init.entry, "ax"
.global entry
entry:
	/* Load and apply the memory management configuration, ready to enable MMU and caches. */
	adrp x30, idmap
	msr ttbr0_el1, x30

	mov_i x30, .Lmairval
	msr mair_el1, x30

	mov_i x30, .Ltcrval
	/* Copy the supported PA range into TCR_EL1.IPS. */
	mrs x29, id_aa64mmfr0_el1
	bfi x30, x29, #32, #4

	msr tcr_el1, x30

	mov_i x30, .Lsctlrval

	/*
	 * Ensure everything before this point has completed, then invalidate any potentially stale
	 * local TLB entries before they start being used.
	 */
	isb
	tlbi vmalle1
	ic iallu
	dsb nsh
	isb

	/*
	 * Configure sctlr_el1 to enable MMU and cache and don't proceed until this has completed.
	 */
	msr sctlr_el1, x30
	isb

	/* Disable trapping floating point access in EL1. */
	mrs x30, cpacr_el1
	orr x30, x30, #(0x3 << 20)
	msr cpacr_el1, x30
	isb

	/* Zero out the bss section. */
	adr_l x29, bss_begin
	adr_l x30, bss_end
0:	cmp x29, x30
	b.hs 1f
	stp xzr, xzr, [x29], #16
	b 0b

1:	/* Prepare the stack. */
	adr_l x30, boot_stack_end
	mov sp, x30

	/* Set up exception vector. */
	adr x30, vector_table_el1
	msr vbar_el1, x30

	/* Call into Rust code. */
	bl main

	/* Loop forever waiting for interrupts. */
2:	wfi
	b 2b

exceptions.S（无需对此进行更改）：

/*
 * Copyright 2023 Google LLC
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     https://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

/**
 * Saves the volatile registers onto the stack. This currently takes 14
 * instructions, so it can be used in exception handlers with 18 instructions
 * left.
 *
 * On return, x0 and x1 are initialised to elr_el2 and spsr_el2 respectively,
 * which can be used as the first and second arguments of a subsequent call.
 */
.macro save_volatile_to_stack
	/* Reserve stack space and save registers x0-x18, x29 & x30. */
	stp x0, x1, [sp, #-(8 * 24)]!
	stp x2, x3, [sp, #8 * 2]
	stp x4, x5, [sp, #8 * 4]
	stp x6, x7, [sp, #8 * 6]
	stp x8, x9, [sp, #8 * 8]
	stp x10, x11, [sp, #8 * 10]
	stp x12, x13, [sp, #8 * 12]
	stp x14, x15, [sp, #8 * 14]
	stp x16, x17, [sp, #8 * 16]
	str x18, [sp, #8 * 18]
	stp x29, x30, [sp, #8 * 20]

	/*
	 * Save elr_el1 & spsr_el1. This such that we can take nested exception
	 * and still be able to unwind.
	 */
	mrs x0, elr_el1
	mrs x1, spsr_el1
	stp x0, x1, [sp, #8 * 22]
.endm

/**
 * Restores the volatile registers from the stack. This currently takes 14
 * instructions, so it can be used in exception handlers while still leaving 18
 * instructions left; if paired with save_volatile_to_stack, there are 4
 * instructions to spare.
 */
.macro restore_volatile_from_stack
	/* Restore registers x2-x18, x29 & x30. */
	ldp x2, x3, [sp, #8 * 2]
	ldp x4, x5, [sp, #8 * 4]
	ldp x6, x7, [sp, #8 * 6]
	ldp x8, x9, [sp, #8 * 8]
	ldp x10, x11, [sp, #8 * 10]
	ldp x12, x13, [sp, #8 * 12]
	ldp x14, x15, [sp, #8 * 14]
	ldp x16, x17, [sp, #8 * 16]
	ldr x18, [sp, #8 * 18]
	ldp x29, x30, [sp, #8 * 20]

	/* Restore registers elr_el1 & spsr_el1, using x0 & x1 as scratch. */
	ldp x0, x1, [sp, #8 * 22]
	msr elr_el1, x0
	msr spsr_el1, x1

	/* Restore x0 & x1, and release stack space. */
	ldp x0, x1, [sp], #8 * 24
.endm

/**
 * This is a generic handler for exceptions taken at the current EL while using
 * SP0. It behaves similarly to the SPx case by first switching to SPx, doing
 * the work, then switching back to SP0 before returning.
 *
 * Switching to SPx and calling the Rust handler takes 16 instructions. To
 * restore and return we need an additional 16 instructions, so we can implement
 * the whole handler within the allotted 32 instructions.
 */
.macro current_exception_sp0 handler:req
	msr spsel, #1
	save_volatile_to_stack
	bl \handler
	restore_volatile_from_stack
	msr spsel, #0
	eret
.endm

/**
 * This is a generic handler for exceptions taken at the current EL while using
 * SPx. It saves volatile registers, calls the Rust handler, restores volatile
 * registers, then returns.
 *
 * This also works for exceptions taken from EL0, if we don't care about
 * non-volatile registers.
 *
 * Saving state and jumping to the Rust handler takes 15 instructions, and
 * restoring and returning also takes 15 instructions, so we can fit the whole
 * handler in 30 instructions, under the limit of 32.
 */
.macro current_exception_spx handler:req
	save_volatile_to_stack
	bl \handler
	restore_volatile_from_stack
	eret
.endm

.section .text.vector_table_el1, "ax"
.global vector_table_el1
.balign 0x800
vector_table_el1:
sync_cur_sp0:
	current_exception_sp0 sync_exception_current

.balign 0x80
irq_cur_sp0:
	current_exception_sp0 irq_current

.balign 0x80
fiq_cur_sp0:
	current_exception_sp0 fiq_current

.balign 0x80
serr_cur_sp0:
	current_exception_sp0 serr_current

.balign 0x80
sync_cur_spx:
	current_exception_spx sync_exception_current

.balign 0x80
irq_cur_spx:
	current_exception_spx irq_current

.balign 0x80
fiq_cur_spx:
	current_exception_spx fiq_current

.balign 0x80
serr_cur_spx:
	current_exception_spx serr_current

.balign 0x80
sync_lower_64:
	current_exception_spx sync_lower

.balign 0x80
irq_lower_64:
	current_exception_spx irq_lower

.balign 0x80
fiq_lower_64:
	current_exception_spx fiq_lower

.balign 0x80
serr_lower_64:
	current_exception_spx serr_lower

.balign 0x80
sync_lower_32:
	current_exception_spx sync_lower

.balign 0x80
irq_lower_32:
	current_exception_spx irq_lower

.balign 0x80
fiq_lower_32:
	current_exception_spx fiq_lower

.balign 0x80
serr_lower_32:
	current_exception_spx serr_lower

idmap.S（无需对此进行更改）

/*
 * Copyright 2023 Google LLC
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     https://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

.set .L_TT_TYPE_BLOCK, 0x1
.set .L_TT_TYPE_PAGE,  0x3
.set .L_TT_TYPE_TABLE, 0x3

/* Access flag. */
.set .L_TT_AF, 0x1 << 10
/* Not global. */
.set .L_TT_NG, 0x1 << 11
.set .L_TT_XN, 0x3 << 53

.set .L_TT_MT_DEV, 0x0 << 2			// MAIR #0 (DEV_nGnRE)
.set .L_TT_MT_MEM, (0x1 << 2) | (0x3 << 8)	// MAIR #1 (MEM_WBWA), inner shareable

.set .L_BLOCK_DEV, .L_TT_TYPE_BLOCK | .L_TT_MT_DEV | .L_TT_AF | .L_TT_XN
.set .L_BLOCK_MEM, .L_TT_TYPE_BLOCK | .L_TT_MT_MEM | .L_TT_AF | .L_TT_NG

.section ".rodata.idmap", "a", %progbits
.global idmap
.align 12
idmap:
	/* level 1 */
	.quad		.L_BLOCK_DEV | 0x0		    // 1 GiB of device mappings
	.quad		.L_BLOCK_MEM | 0x40000000	// 1 GiB of DRAM
	.fill		254, 8, 0x0			// 254 GiB of unmapped VA space
	.quad		.L_BLOCK_DEV | 0x4000000000 // 1 GiB of device mappings
	.fill		255, 8, 0x0			// 255 GiB of remaining VA space

image.ld（无需对此进行更改）：

/*
 * Copyright 2023 Google LLC
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     https://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

/*
 * Code will start running at this symbol which is placed at the start of the
 * image.
 */
ENTRY(entry)

MEMORY
{
	image : ORIGIN = 0x40080000, LENGTH = 2M
}

SECTIONS
{
	/*
	 * Collect together the code.
	 */
	.init : ALIGN(4096) {
		text_begin = .;
		*(.init.entry)
		*(.init.*)
	} >image
	.text : {
		*(.text.*)
	} >image
	text_end = .;

	/*
	 * Collect together read-only data.
	 */
	.rodata : ALIGN(4096) {
		rodata_begin = .;
		*(.rodata.*)
	} >image
	.got : {
		*(.got)
	} >image
	rodata_end = .;

	/*
	 * Collect together the read-write data including .bss at the end which
	 * will be zero'd by the entry code.
	 */
	.data : ALIGN(4096) {
		data_begin = .;
		*(.data.*)
		/*
		 * The entry point code assumes that .data is a multiple of 32
		 * bytes long.
		 */
		. = ALIGN(32);
		data_end = .;
	} >image

	/* Everything beyond this point will not be included in the binary. */
	bin_end = .;

	/* The entry point code assumes that .bss is 16-byte aligned. */
	.bss : ALIGN(16)  {
		bss_begin = .;
		*(.bss.*)
		*(COMMON)
		. = ALIGN(16);
		bss_end = .;
	} >image

	.stack (NOLOAD) : ALIGN(4096) {
		boot_stack_begin = .;
		. += 40 * 4096;
		. = ALIGN(4096);
		boot_stack_end = .;
	} >image

	. = ALIGN(4K);
	PROVIDE(dma_region = .);

	/*
	 * Remove unused sections from the image.
	 */
	/DISCARD/ : {
		/* The image loads itself so doesn't need these sections. */
		*(.gnu.hash)
		*(.hash)
		*(.interp)
		*(.eh_frame_hdr)
		*(.eh_frame)
		*(.note.gnu.build-id)
	}
}

Makefile（无需对此进行更改）：

# Copyright 2023 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

UNAME := $(shell uname -s)
ifeq ($(UNAME),Linux)
	TARGET = aarch64-linux-gnu
else
	TARGET = aarch64-none-elf
endif
OBJCOPY = $(TARGET)-objcopy

.PHONY: build qemu_minimal qemu qemu_logger

all: rtc.bin

build:
	cargo build

rtc.bin: build
	$(OBJCOPY) -O binary target/aarch64-unknown-none/debug/rtc $@

qemu: rtc.bin
	qemu-system-aarch64 -machine virt,gic-version=3 -cpu max -serial mon:stdio -display none -kernel $< -s

clean:
	cargo clean
	rm -f *.bin

.cargo/config.toml（无需对此进行更改）：

[build]
target = "aarch64-unknown-none"
rustflags = ["-C", "link-arg=-Timage.ld"]

使用 make qemu 在 QEMU 中运行代码。

嵌入式 Rust：进阶篇

RTC 驱动程序

(返回练习)

main.rs:

#![no_main]
#![no_std]

mod exceptions;
mod logger;
mod pl011;
mod pl031;

use crate::pl031::Rtc;
use arm_gic::gicv3::{IntId, Trigger};
use arm_gic::{irq_enable, wfi};
use chrono::{TimeZone, Utc};
use core::hint::spin_loop;
use crate::pl011::Uart;
use arm_gic::gicv3::GicV3;
use core::panic::PanicInfo;
use log::{error, info, trace, LevelFilter};
use smccc::psci::system_off;
use smccc::Hvc;

/// Base addresses of the GICv3.
const GICD_BASE_ADDRESS: *mut u64 = 0x800_0000 as _;
const GICR_BASE_ADDRESS: *mut u64 = 0x80A_0000 as _;

/// Base address of the primary PL011 UART.
const PL011_BASE_ADDRESS: *mut u32 = 0x900_0000 as _;

/// Base address of the PL031 RTC.
const PL031_BASE_ADDRESS: *mut u32 = 0x901_0000 as _;
/// The IRQ used by the PL031 RTC.
const PL031_IRQ: IntId = IntId::spi(2);

#[no_mangle]
extern "C" fn main(x0: u64, x1: u64, x2: u64, x3: u64) {
    // SAFETY: `PL011_BASE_ADDRESS` is the base address of a PL011 device, and
    // nothing else accesses that address range.
    let uart = unsafe { Uart::new(PL011_BASE_ADDRESS) };
    logger::init(uart, LevelFilter::Trace).unwrap();

    info!("main({:#x}, {:#x}, {:#x}, {:#x})", x0, x1, x2, x3);

    // SAFETY: `GICD_BASE_ADDRESS` and `GICR_BASE_ADDRESS` are the base
    // addresses of a GICv3 distributor and redistributor respectively, and
    // nothing else accesses those address ranges.
    let mut gic = unsafe { GicV3::new(GICD_BASE_ADDRESS, GICR_BASE_ADDRESS) };
    gic.setup();

    // SAFETY: `PL031_BASE_ADDRESS` is the base address of a PL031 device, and
    // nothing else accesses that address range.
    let mut rtc = unsafe { Rtc::new(PL031_BASE_ADDRESS) };
    let timestamp = rtc.read();
    let time = Utc.timestamp_opt(timestamp.into(), 0).unwrap();
    info!("RTC: {time}");

    GicV3::set_priority_mask(0xff);
    gic.set_interrupt_priority(PL031_IRQ, 0x80);
    gic.set_trigger(PL031_IRQ, Trigger::Level);
    irq_enable();
    gic.enable_interrupt(PL031_IRQ, true);

    // Wait for 3 seconds, without interrupts.
    let target = timestamp + 3;
    rtc.set_match(target);
    info!("Waiting for {}", Utc.timestamp_opt(target.into(), 0).unwrap());
    trace!(
        "matched={}, interrupt_pending={}",
        rtc.matched(),
        rtc.interrupt_pending()
    );
    while !rtc.matched() {
        spin_loop();
    }
    trace!(
        "matched={}, interrupt_pending={}",
        rtc.matched(),
        rtc.interrupt_pending()
    );
    info!("Finished waiting");

    // Wait another 3 seconds for an interrupt.
    let target = timestamp + 6;
    info!("Waiting for {}", Utc.timestamp_opt(target.into(), 0).unwrap());
    rtc.set_match(target);
    rtc.clear_interrupt();
    rtc.enable_interrupt(true);
    trace!(
        "matched={}, interrupt_pending={}",
        rtc.matched(),
        rtc.interrupt_pending()
    );
    while !rtc.interrupt_pending() {
        wfi();
    }
    trace!(
        "matched={}, interrupt_pending={}",
        rtc.matched(),
        rtc.interrupt_pending()
    );
    info!("Finished waiting");

    system_off::<Hvc>().unwrap();
}

#[panic_handler]
fn panic(info: &PanicInfo) -> ! {
    error!("{info}");
    system_off::<Hvc>().unwrap();
    loop {}
}

pl031.rs：

#![allow(unused)]
fn main() {
use core::ptr::{addr_of, addr_of_mut};

#[repr(C, align(4))]
struct Registers {
    /// Data register
    dr: u32,
    /// Match register
    mr: u32,
    /// Load register
    lr: u32,
    /// Control register
    cr: u8,
    _reserved0: [u8; 3],
    /// Interrupt Mask Set or Clear register
    imsc: u8,
    _reserved1: [u8; 3],
    /// Raw Interrupt Status
    ris: u8,
    _reserved2: [u8; 3],
    /// Masked Interrupt Status
    mis: u8,
    _reserved3: [u8; 3],
    /// Interrupt Clear Register
    icr: u8,
    _reserved4: [u8; 3],
}

/// Driver for a PL031 real-time clock.
#[derive(Debug)]
pub struct Rtc {
    registers: *mut Registers,
}

impl Rtc {
    /// Constructs a new instance of the RTC driver for a PL031 device at the
    /// given base address.
    ///
    /// # Safety
    ///
    /// The given base address must point to the MMIO control registers of a
    /// PL031 device, which must be mapped into the address space of the process
    /// as device memory and not have any other aliases.
    pub unsafe fn new(base_address: *mut u32) -> Self {
        Self { registers: base_address as *mut Registers }
    }

    /// Reads the current RTC value.
    pub fn read(&self) -> u32 {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL031 device which is appropriately mapped.
        unsafe { addr_of!((*self.registers).dr).read_volatile() }
    }

    /// Writes a match value. When the RTC value matches this then an interrupt
    /// will be generated (if it is enabled).
    pub fn set_match(&mut self, value: u32) {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL031 device which is appropriately mapped.
        unsafe { addr_of_mut!((*self.registers).mr).write_volatile(value) }
    }

    /// Returns whether the match register matches the RTC value, whether or not
    /// the interrupt is enabled.
    pub fn matched(&self) -> bool {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL031 device which is appropriately mapped.
        let ris = unsafe { addr_of!((*self.registers).ris).read_volatile() };
        (ris & 0x01) != 0
    }

    /// Returns whether there is currently an interrupt pending.
    ///
    /// This should be true if and only if `matched` returns true and the
    /// interrupt is masked.
    pub fn interrupt_pending(&self) -> bool {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL031 device which is appropriately mapped.
        let ris = unsafe { addr_of!((*self.registers).mis).read_volatile() };
        (ris & 0x01) != 0
    }

    /// Sets or clears the interrupt mask.
    ///
    /// When the mask is true the interrupt is enabled; when it is false the
    /// interrupt is disabled.
    pub fn enable_interrupt(&mut self, mask: bool) {
        let imsc = if mask { 0x01 } else { 0x00 };
        // SAFETY: We know that self.registers points to the control registers
        // of a PL031 device which is appropriately mapped.
        unsafe { addr_of_mut!((*self.registers).imsc).write_volatile(imsc) }
    }

    /// Clears a pending interrupt, if any.
    pub fn clear_interrupt(&mut self) {
        // SAFETY: We know that self.registers points to the control registers
        // of a PL031 device which is appropriately mapped.
        unsafe { addr_of_mut!((*self.registers).icr).write_volatile(0x01) }
    }
}

// SAFETY: `Rtc` just contains a pointer to device memory, which can be
// accessed from any context.
unsafe impl Send for Rtc {}
}

欢迎了解 Rust 中的并发

Rust 完全支持使用带有互斥锁和通道的操作系统线程进行并发。

Rust 类型系统能帮助我们把许多并发bug转换为编译期bug 发挥着重要作用。这通常称为“无畏并发”，因为你可以依靠编译器来确保 运行时的正确性。

• Rust lets us access OS concurrency toolkit: threads, sync. primitives, etc.
• The type system gives us safety for concurrency without any special features.
• The same tools that help with "concurrent" access in a single thread (e.g., a called function that might mutate an argument or save references to it to read later) save us from multi-threading issues.

线程

Rust 线程的运作方式与其他语言中的线程类似：

use std::thread;
use std::time::Duration;

fn main() {
    thread::spawn(|| {
        for i in 1..10 {
            println!("Count in thread: {i}!");
            thread::sleep(Duration::from_millis(5));
        }
    });

    for i in 1..5 {
        println!("Main thread: {i}");
        thread::sleep(Duration::from_millis(5));
    }
}

• 线程均为守护程序线程，主线程不会等待这些线程。
• 线程紧急警报 (panic) 是彼此独立的。
  • 紧急警报可以携带载荷，并可以使用 downcast_ref 对载荷进行解压缩。

• Rust thread APIs look not too different from e.g. C++ ones.

• Run the example.

  • 5ms timing is loose enough that main and spawned threads stay mostly in lockstep.
  • Notice that the program ends before the spawned thread reaches 10!
  • This is because main ends the program and spawned threads do not make it persist.
    • Compare to pthreads/C++ std::thread/boost::thread if desired.

• How do we wait around for the spawned thread to complete?

• thread::spawn returns a JoinHandle. Look at the docs.

  • JoinHandle has a .join() method that blocks.

• Use let handle = thread::spawn(...) and later handle.join() to wait for the thread to finish and have the program count all the way to 10.

• Now what if we want to return a value?

• Look at docs again:

  • thread::spawn's closure returns T
  • JoinHandle .join() returns thread::Result<T>

• Use the Result return value from handle.join() to get access to the returned value.

• Ok, what about the other case?

  • Trigger a panic in the thread. Note that this doesn't panic main.
  • Access the panic payload. This is a good time to talk about Any.

• Now we can return values from threads! What about taking inputs?

  • Capture something by reference in the thread closure.
  • An error message indicates we must move it.
  • Move it in, see we can compute and then return a derived value.

• If we want to borrow?

  • Main kills child threads when it returns, but another function would just return and leave them running.
  • That would be stack use-after-return, which violates memory safety!
  • How do we avoid this? see next slide.

范围线程

常规线程不能从它们所处的环境中借用：

use std::thread;

fn foo() {
    let s = String::from("Hello");
    thread::spawn(|| {
        println!("Length: {}", s.len());
    });
}

fn main() {
    foo();
}

不过，你可以使用范围线程来实现此目的：

use std::thread;

fn main() {
    let s = String::from("Hello");

    thread::scope(|scope| {
        scope.spawn(|| {
            println!("Length: {}", s.len());
        });
    });
}

• 其原因在于，在 thread::scope 函数完成后，可保证所有线程都已联结在一起，使得线程能够返回借用的数据。
• 此时须遵守常规 Rust 借用规则：你可以通过一个线程以可变的方式借用，也可以通过任意数量的线程以不可变的方式借用。

通道

Rust 通道（Channel）包含两个部分：Sender<T> 和 Receiver<T>。这两个部分 通过通道进行连接，但你只能看到端点。

use std::sync::mpsc;

fn main() {
    let (tx, rx) = mpsc::channel();

    tx.send(10).unwrap();
    tx.send(20).unwrap();

    println!("Received: {:?}", rx.recv());
    println!("Received: {:?}", rx.recv());

    let tx2 = tx.clone();
    tx2.send(30).unwrap();
    println!("Received: {:?}", rx.recv());
}

• mpsc 代表多个生产方，单个使用方。Sender 和 SyncSender 会实现 Clone（因此， 你可以设置多个生产方），但 Receiver 不会实现。
• send() 和 recv() 会返回 Result。如果它们返回 Err，则表示对应的 Sender 或 Receiver 已被丢弃，且通道已关闭。

无界通道

你可以使用 mpsc::channel() 获得无边界的异步通道：

use std::sync::mpsc;
use std::thread;
use std::time::Duration;

fn main() {
    let (tx, rx) = mpsc::channel();

    thread::spawn(move || {
        let thread_id = thread::current().id();
        for i in 1..10 {
            tx.send(format!("Message {i}")).unwrap();
            println!("{thread_id:?}: sent Message {i}");
        }
        println!("{thread_id:?}: done");
    });
    thread::sleep(Duration::from_millis(100));

    for msg in rx.iter() {
        println!("Main: got {msg}");
    }
}

有界通道

With bounded (synchronous) channels, send can block the current thread:

use std::sync::mpsc;
use std::thread;
use std::time::Duration;

fn main() {
    let (tx, rx) = mpsc::sync_channel(3);

    thread::spawn(move || {
        let thread_id = thread::current().id();
        for i in 1..10 {
            tx.send(format!("Message {i}")).unwrap();
            println!("{thread_id:?}: sent Message {i}");
        }
        println!("{thread_id:?}: done");
    });
    thread::sleep(Duration::from_millis(100));

    for msg in rx.iter() {
        println!("Main: got {msg}");
    }
}

• 调用 send 将阻塞当前线程，直到通道中有足够的空间放置新消息。如果无人从通道读取数据，线程会被无限期地阻塞。
• 如果通道关闭，调用 send 将中止并返回错误（这就是它会返回 Result 的原因）。当接收器被丢弃时，通道将关闭。
• A bounded channel with a size of zero is called a "rendezvous channel". Every send will block the current thread until another thread calls recv.

Send 和 Sync

How does Rust know to forbid shared access across threads? The answer is in two traits:

• Send：如果跨线程边界移动 T 是安全的，则类型 T 为 Send。
• Sync：如果跨线程边界移动 &T 是安全的，则类型 T 为 Sync。

Send 和 Sync 均为不安全特征。只要类型仅包含 Send 和 Sync 类型，编译器就会自动为类型派生 这两种特征。你也可以手动实现它们（如果你确定这样 有效的话）。

• 不妨将这些特征视为类型包含某些线程安全属性的标记。
• 它们可以在泛型约束中作为常规特征使用。

Send

如果将 T 值移动到另一个线程是安全的，则类型 T 为 Send。

将所有权转移到另一个线程的影响是，“析构函数”将在相应线程中 运行。因此，问题在于你何时可以在一个线程中分配某个值，然后在 另一个线程中取消分配该值。

例如，与 SQLite 库的连接只能通过 单个线程访问。

Sync

如果同时从多个线程访问 T 值是安全的，则类型 T 为 Sync。

更准确地说，定义是：

当且仅当 &T 为 Send 时，T 为 Sync

该语句实质上是一种简写形式，表示如果某个类型对于共享使用是线程安全的，那么跨线程传递对该类型的引用也是线程安全的。

这是因为如果某个类型为 Sync，则意味着它可以在多个线程之间共享，而不存在数据争用或其他同步问题的风险，因此将其移动到另一个线程是安全的。对该类型的引用同样可以安全地移动到另一个线程，因为它引用的数据可以从任何线程安全地访问。

示例

Send + Sync

你遇到的类型大都属于 Send + Sync：

• i8、f32、bool、char、&str…
• (T1, T2)、[T; N]、&[T]、struct { x: T }…
• String、Option<T>、Vec<T>、Box<T>…
• Arc<T>：明确通过原子引用计数实现线程安全。
• Mutex<T>：明确通过内部锁定实现线程安全。
• mpsc::Sender<T>: As of 1.72.0.
• AtomicBool、AtomicU8…：使用特殊的原子指令。

当类型参数为 Send + Sync 时，泛型类型通常 为 Send + Sync。

Send + !Sync

这些类型可以移动到其他线程，但它们不是线程安全的。 这通常是由内部可变性造成的：

• mpsc::Receiver<T>
• Cell<T>
• RefCell<T>

!Send + Sync

这些类型是线程安全的，但它们不能移动到另一个线程：

• MutexGuard<T: Sync>: Uses OS level primitives which must be deallocated on the thread which created them.

!Send + !Sync

这些类型不是线程安全的，不能移动到其他线程：

• Rc<T>：每个 Rc<T> 都具有对 RcBox<T> 的引用，其中包含 非原子引用计数。
• *const T、*mut T：Rust 会假定原始指针可能 在并发方面有特殊的注意事项。

共享状态

Rust 使用类型系统来强制同步共享数据。这主要 通过两种类型实现：

• Arc<T>，对 T 进行原子计数：用于处理线程之间的共享，并负责 在最后一个引用被丢弃时取消分配 T。
• Mutex<T>：确保对 T 值的互斥访问。

Arc

Arc<T> 允许通过 Arc::clone 实现共享只读权限：

use std::sync::Arc;
use std::thread;

fn main() {
    let v = Arc::new(vec![10, 20, 30]);
    let mut handles = Vec::new();
    for _ in 1..5 {
        let v = Arc::clone(&v);
        handles.push(thread::spawn(move || {
            let thread_id = thread::current().id();
            println!("{thread_id:?}: {v:?}");
        }));
    }

    handles.into_iter().for_each(|h| h.join().unwrap());
    println!("v: {v:?}");
}

• Arc 代表“原子引用计数”，它是使用原子操作的 Rc 的 线程安全版本。
• Arc<T> implements Clone whether or not T does. It implements Send and Sync if and only if T implements them both.
• Arc::clone() 在执行原子操作方面有开销，但在此之后，T 便可 随意使用，而没有任何开销。
• 请警惕引用循环，Arc 不会使用垃圾回收器检测引用循环。
  • std::sync::Weak 对此有所帮助。

互斥器（Mutex）

Mutex<T> ensures mutual exclusion and allows mutable access to T behind a read-only interface (another form of interior mutability):

use std::sync::Mutex;

fn main() {
    let v = Mutex::new(vec![10, 20, 30]);
    println!("v: {:?}", v.lock().unwrap());

    {
        let mut guard = v.lock().unwrap();
        guard.push(40);
    }

    println!("v: {:?}", v.lock().unwrap());
}

请注意我们如何设置 impl<T: Send> Sync for Mutex<T> 通用 实现。

• Mutex in Rust looks like a collection with just one element --- the protected data.
  • 在访问受保护的数据之前不可能忘记获取互斥量。
• 你可以通过获取锁，从 &Mutex<T> 中获取 &mut T。MutexGuard 能够确保 &mut T 存在的时间不会比持有锁的时间更长。
• Mutex<T> implements both Send and Sync iff (if and only if) T implements Send.
• A read-write lock counterpart: RwLock.
• Why does lock() return a Result?
  • 如果持有 Mutex 的线程发生panic，Mutex 便会“中毒”并发出信号， 表明其所保护的数据可能处于不一致状态。对中毒的互斥量调用 lock() 将会失败， 并将显示 PoisonError。无论如何，你可以对该错误调用 into_inner() 来 恢复数据。

示例

让我们看看 Arc 和 Mutex 的实际效果：

use std::thread;
// use std::sync::{Arc, Mutex};

fn main() {
    let v = vec![10, 20, 30];
    let handle = thread::spawn(|| {
        v.push(10);
    });
    v.push(1000);

    handle.join().unwrap();
    println!("v: {v:?}");
}

可能有用的解决方案：

use std::sync::{Arc, Mutex};
use std::thread;

fn main() {
    let v = Arc::new(Mutex::new(vec![10, 20, 30]));

    let v2 = Arc::clone(&v);
    let handle = thread::spawn(move || {
        let mut v2 = v2.lock().unwrap();
        v2.push(10);
    });

    {
        let mut v = v.lock().unwrap();
        v.push(1000);
    }

    handle.join().unwrap();

    println!("v: {v:?}");
}

值得注意的部分：

• Arc 和 Mutex 中都封装了 v，因为它们的关注点是正交的。
  • 将 Mutex 封装在 Arc 中是一种在线程之间共享可变状态的常见模式。
• v: Arc<_> 必须先克隆为 v2，然后才能移动到另一个线程中。请注意，lambda 签名中添加了 move。
• 我们引入了块，以尽可能缩小 LockGuard 的作用域。

习题

现在通过

• 哲学家用餐示例来练习我们新学习到的并发技巧：该示例是典型的并发问题。

• 多线程链接检查器：对于大型项目，需要使用 Cargo 下载依赖项，然后并行检查链接。

After looking at the exercises, you can look at the solutions provided.

哲学家就餐问题

哲学家用餐示例是一个典型的并发问题：

五位哲学家在同一桌子上用餐。每位哲学家在桌前都有自己的座位。每个盘子之间都有一把叉子。上的菜品是一种意大利面，需要用两把叉子才能吃。每位哲学家只能交替进行思考和用餐。此外，只有当哲学家们同时拿到左边和右边的叉子才能吃这个意大利面。因此，只有当两旁坐着的人在思考，而非在吃面时，他们才能使用两把叉子。每位哲学家吃完饭后，就会放下手中的两把叉子。

在本练习中，需要使用本地 Cargo 安装。将以下代码复制到名为 src/main.rs 的文件中，并填写空白的地方，然后测试 cargo run 不会死锁：

use std::sync::{mpsc, Arc, Mutex};
use std::thread;
use std::time::Duration;

struct Fork;

struct Philosopher {
    name: String,
    // left_fork: ...
    // right_fork: ...
    // thoughts: ...
}

impl Philosopher {
    fn think(&self) {
        self.thoughts
            .send(format!("Eureka! {} has a new idea!", &self.name))
            .unwrap();
    }

    fn eat(&self) {
        // Pick up forks...
        println!("{} is eating...", &self.name);
        thread::sleep(Duration::from_millis(10));
    }
}

static PHILOSOPHERS: &[&str] =
    &["Socrates", "Hypatia", "Plato", "Aristotle", "Pythagoras"];

fn main() {
    // Create forks

    // Create philosophers

    // Make each of them think and eat 100 times

    // Output their thoughts
}

您可以使用以下 Cargo.toml：

[package]
name = "dining-philosophers"
version = "0.1.0"
edition = "2021"

多线程链接检查器

运用掌握的新知识创建一个多线程链接检查工具。应先从网页入手，并检查网页上的链接是否有效。该工具应以递归方式检查同一网域中的其他网页，并且一直执行此操作，直到所有网页都通过验证。

For this, you will need an HTTP client such as reqwest. You will also need a way to find links, we can use scraper. Finally, we'll need some way of handling errors, we will use thiserror.

Create a new Cargo project and reqwest it as a dependency with:

cargo new link-checker
cd link-checker
cargo add --features blocking,rustls-tls reqwest
cargo add scraper
cargo add thiserror

如果 cargo add 操作失败并显示 error: no such subcommand，请手动修改 Cargo.toml 文件。添加下面列出的依赖项。

cargo add 调用会将 Cargo.toml 文件更新为如下所示：

[package]
name = "link-checker"
version = "0.1.0"
edition = "2021"
publish = false

[dependencies]
reqwest = { version = "0.11.12", features = ["blocking", "rustls-tls"] }
scraper = "0.13.0"
thiserror = "1.0.37"

您现在可以下载初始页了。请尝试使用一个小网站，例如 https://www.google.org/。

您的 src/main.rs 文件应如下所示：

use reqwest::blocking::Client;
use reqwest::Url;
use scraper::{Html, Selector};
use thiserror::Error;

#[derive(Error, Debug)]
enum Error {
    #[error("request error: {0}")]
    ReqwestError(#[from] reqwest::Error),
    #[error("bad http response: {0}")]
    BadResponse(String),
}

#[derive(Debug)]
struct CrawlCommand {
    url: Url,
    extract_links: bool,
}

fn visit_page(client: &Client, command: &CrawlCommand) -> Result<Vec<Url>, Error> {
    println!("Checking {:#}", command.url);
    let response = client.get(command.url.clone()).send()?;
    if !response.status().is_success() {
        return Err(Error::BadResponse(response.status().to_string()));
    }

    let mut link_urls = Vec::new();
    if !command.extract_links {
        return Ok(link_urls);
    }

    let base_url = response.url().to_owned();
    let body_text = response.text()?;
    let document = Html::parse_document(&body_text);

    let selector = Selector::parse("a").unwrap();
    let href_values = document
        .select(&selector)
        .filter_map(|element| element.value().attr("href"));
    for href in href_values {
        match base_url.join(href) {
            Ok(link_url) => {
                link_urls.push(link_url);
            }
            Err(err) => {
                println!("On {base_url:#}: ignored unparsable {href:?}: {err}");
            }
        }
    }
    Ok(link_urls)
}

fn main() {
    let client = Client::new();
    let start_url = Url::parse("https://www.google.org").unwrap();
    let crawl_command = CrawlCommand{ url: start_url, extract_links: true };
    match visit_page(&client, &crawl_command) {
        Ok(links) => println!("Links: {links:#?}"),
        Err(err) => println!("Could not extract links: {err:#}"),
    }
}

使用以下命令运行 src/main.rs 中的代码

cargo run

任务

• 通过线程并行检查链接：将要检查的网址发送到某个通道，然后使用多个线程并行检查这些网址。
• 您可以对此进行扩展，以递归方式从 www.google.org 网域的所有网页中提取链接。设置网页上限（例如 100 个），以免被网站屏蔽。

并发编程：上午练习

哲学家就餐问题

(返回练习)

use std::sync::{mpsc, Arc, Mutex};
use std::thread;
use std::time::Duration;

struct Fork;

struct Philosopher {
    name: String,
    left_fork: Arc<Mutex<Fork>>,
    right_fork: Arc<Mutex<Fork>>,
    thoughts: mpsc::SyncSender<String>,
}

impl Philosopher {
    fn think(&self) {
        self.thoughts
            .send(format!("Eureka! {} has a new idea!", &self.name))
            .unwrap();
    }

    fn eat(&self) {
        println!("{} is trying to eat", &self.name);
        let _left = self.left_fork.lock().unwrap();
        let _right = self.right_fork.lock().unwrap();

        println!("{} is eating...", &self.name);
        thread::sleep(Duration::from_millis(10));
    }
}

static PHILOSOPHERS: &[&str] =
    &["Socrates", "Hypatia", "Plato", "Aristotle", "Pythagoras"];

fn main() {
    let (tx, rx) = mpsc::sync_channel(10);

    let forks = (0..PHILOSOPHERS.len())
        .map(|_| Arc::new(Mutex::new(Fork)))
        .collect::<Vec<_>>();

    for i in 0..forks.len() {
        let tx = tx.clone();
        let mut left_fork = Arc::clone(&forks[i]);
        let mut right_fork = Arc::clone(&forks[(i + 1) % forks.len()]);

        // To avoid a deadlock, we have to break the symmetry
        // somewhere. This will swap the forks without deinitializing
        // either of them.
        if i == forks.len() - 1 {
            std::mem::swap(&mut left_fork, &mut right_fork);
        }

        let philosopher = Philosopher {
            name: PHILOSOPHERS[i].to_string(),
            thoughts: tx,
            left_fork,
            right_fork,
        };

        thread::spawn(move || {
            for _ in 0..100 {
                philosopher.eat();
                philosopher.think();
            }
        });
    }

    drop(tx);
    for thought in rx {
        println!("{thought}");
    }
}

Link Checker

(back to exercise)

use std::sync::{mpsc, Arc, Mutex};
use std::thread;

use reqwest::blocking::Client;
use reqwest::Url;
use scraper::{Html, Selector};
use thiserror::Error;

#[derive(Error, Debug)]
enum Error {
    #[error("request error: {0}")]
    ReqwestError(#[from] reqwest::Error),
    #[error("bad http response: {0}")]
    BadResponse(String),
}

#[derive(Debug)]
struct CrawlCommand {
    url: Url,
    extract_links: bool,
}

fn visit_page(client: &Client, command: &CrawlCommand) -> Result<Vec<Url>, Error> {
    println!("Checking {:#}", command.url);
    let response = client.get(command.url.clone()).send()?;
    if !response.status().is_success() {
        return Err(Error::BadResponse(response.status().to_string()));
    }

    let mut link_urls = Vec::new();
    if !command.extract_links {
        return Ok(link_urls);
    }

    let base_url = response.url().to_owned();
    let body_text = response.text()?;
    let document = Html::parse_document(&body_text);

    let selector = Selector::parse("a").unwrap();
    let href_values = document
        .select(&selector)
        .filter_map(|element| element.value().attr("href"));
    for href in href_values {
        match base_url.join(href) {
            Ok(link_url) => {
                link_urls.push(link_url);
            }
            Err(err) => {
                println!("On {base_url:#}: ignored unparsable {href:?}: {err}");
            }
        }
    }
    Ok(link_urls)
}

struct CrawlState {
    domain: String,
    visited_pages: std::collections::HashSet<String>,
}

impl CrawlState {
    fn new(start_url: &Url) -> CrawlState {
        let mut visited_pages = std::collections::HashSet::new();
        visited_pages.insert(start_url.as_str().to_string());
        CrawlState { domain: start_url.domain().unwrap().to_string(), visited_pages }
    }

    /// Determine whether links within the given page should be extracted.
    fn should_extract_links(&self, url: &Url) -> bool {
        let Some(url_domain) = url.domain() else {
            return false;
        };
        url_domain == self.domain
    }

    /// Mark the given page as visited, returning false if it had already
    /// been visited.
    fn mark_visited(&mut self, url: &Url) -> bool {
        self.visited_pages.insert(url.as_str().to_string())
    }
}

type CrawlResult = Result<Vec<Url>, (Url, Error)>;
fn spawn_crawler_threads(
    command_receiver: mpsc::Receiver<CrawlCommand>,
    result_sender: mpsc::Sender<CrawlResult>,
    thread_count: u32,
) {
    let command_receiver = Arc::new(Mutex::new(command_receiver));

    for _ in 0..thread_count {
        let result_sender = result_sender.clone();
        let command_receiver = command_receiver.clone();
        thread::spawn(move || {
            let client = Client::new();
            loop {
                let command_result = {
                    let receiver_guard = command_receiver.lock().unwrap();
                    receiver_guard.recv()
                };
                let Ok(crawl_command) = command_result else {
                    // The sender got dropped. No more commands coming in.
                    break;
                };
                let crawl_result = match visit_page(&client, &crawl_command) {
                    Ok(link_urls) => Ok(link_urls),
                    Err(error) => Err((crawl_command.url, error)),
                };
                result_sender.send(crawl_result).unwrap();
            }
        });
    }
}

fn control_crawl(
    start_url: Url,
    command_sender: mpsc::Sender<CrawlCommand>,
    result_receiver: mpsc::Receiver<CrawlResult>,
) -> Vec<Url> {
    let mut crawl_state = CrawlState::new(&start_url);
    let start_command = CrawlCommand { url: start_url, extract_links: true };
    command_sender.send(start_command).unwrap();
    let mut pending_urls = 1;

    let mut bad_urls = Vec::new();
    while pending_urls > 0 {
        let crawl_result = result_receiver.recv().unwrap();
        pending_urls -= 1;

        match crawl_result {
            Ok(link_urls) => {
                for url in link_urls {
                    if crawl_state.mark_visited(&url) {
                        let extract_links = crawl_state.should_extract_links(&url);
                        let crawl_command = CrawlCommand { url, extract_links };
                        command_sender.send(crawl_command).unwrap();
                        pending_urls += 1;
                    }
                }
            }
            Err((url, error)) => {
                bad_urls.push(url);
                println!("Got crawling error: {:#}", error);
                continue;
            }
        }
    }
    bad_urls
}

fn check_links(start_url: Url) -> Vec<Url> {
    let (result_sender, result_receiver) = mpsc::channel::<CrawlResult>();
    let (command_sender, command_receiver) = mpsc::channel::<CrawlCommand>();
    spawn_crawler_threads(command_receiver, result_sender, 16);
    control_crawl(start_url, command_sender, result_receiver)
}

fn main() {
    let start_url = reqwest::Url::parse("https://www.google.org").unwrap();
    let bad_urls = check_links(start_url);
    println!("Bad URLs: {:#?}", bad_urls);
}

异步 Rust

“异步”是一种并发模型，可以同时执行多个任务。具体做法是逐个执行任务直至阻塞，然后切换到另一项可以继续进行的任务。该模型支持在有限数量的线程上运行更多任务。原因在于，每个任务的开销通常很低，并且操作系统提供了基元来高效识别能够执行的 I/O 任务。

Rust 的异步操作基于 "Futures" 来实现，即表示未来可能完成的工作。系统会对这些 Future 进行 “轮询”，直到显示全部已完成。

由异步运行时对这些 Future 进行轮询，并且有多种不同的运行时可供选择。

比较

• Python 的 asyncio 中也有类似的模型。不过，其 Future 类型基于回调的实现方式，而非通过轮询。使用异步 Python 程序需要类似于 Rust 中运行时的 “循环” 进行管理。

• JavaScript 的 Promise 与之类似，但同样基于回调的实现方式。语言运行时实现了事件循环，因此许多与 Promise 解析相关的细节被隐藏起来。

async/await

从高层次上看，异步 Rust 代码与“正常”的顺序代码非常类似：

use futures::executor::block_on;

async fn count_to(count: i32) {
    for i in 1..=count {
        println!("Count is: {i}!");
    }
}

async fn async_main(count: i32) {
    count_to(count).await;
}

fn main() {
    block_on(async_main(10));
}

关键点：

• 请注意，这只是一个简单的示例，用于展示语法。其中没有长时间运行的操作或任何真正的并发！

• 异步调用的返回类型是什么？

  • 在 main 中使用 let future: () = async_main(10); 来查看类型。

• The "async" keyword is syntactic sugar. The compiler replaces the return type with a future.

• 你不能将 main 声明为异步函数，除非在编译器中加入额外的指令来告诉它如何使用返回的 future。

• You need an executor to run async code. block_on blocks the current thread until the provided future has run to completion.

• .await 会异步地等待另一个操作的完成。与 block_on 不同，.await 不会阻塞当前线程。

• .await can only be used inside an async function (or block; these are introduced later).

Futures

Future 是一种 trait，由表示尚未完成的操作的对象所实现。可以轮询 Future，并且 poll 会返回 一个 Poll 对象。

#![allow(unused)]
fn main() {
use std::pin::Pin;
use std::task::Context;

pub trait Future {
    type Output;
    fn poll(self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Self::Output>;
}

pub enum Poll<T> {
    Ready(T),
    Pending,
}
}

异步函数会返回 impl Future。对于自定义的类型，也可以实现 Future（但不常见）。例如，从 tokio::spawn 返回的 JoinHandle 会实现 Future，以允许加入该任务。

在 Future 中使用 .await 关键字会导致当前异步函数暂停，直到该 Future 准备就绪，然后计算其输出。

• Future 和 Poll 类型的实现完全如下所示：请点击链接查看文档中的实现。

• 我们的重点在于编写异步代码，而不是构建新的异步基元，因此不会涉及 Pin 和 Context。简言之：

  • 通过 Context，Future 在事件发生时可自行安排重新进行轮询。

  • Pin 确保 Future 不会移到内存中，以便指向该 Future 的指针仍然有效。为了确保使用 .await 之后引用依然有效，必须执行此操作。

Runtimes

运行时 支持异步执行操作（即 _反应器），并负责执行 Future （即 _执行器）。Rust 没有 “内置”运行时，但有以下几个选项可供选择：

• Tokio：性能出色，拥有成熟的功能生态系统，例如适用于 HTTP 的 Hyper 或 适用于 gRPC 的 Tonic。
• async-std：旨在成为 “异步编程的标准库”，并在 async::task 中包含基本运行时。
• smol：简单且轻量

有些大型应用具有自己的运行时。例如，Fuchsia 已有一个运行时。

• 请注意，在列出的运行时中，Rust Playground 仅支持 Tokio。该 Playground 也不支持任何 I/O 操作，因此大多数有趣的异步操作无法在该平台上运行。

• Future 是 “惯性”的，除非有执行程序对其进行轮询，否则它们不会执行任何操作（甚至不会启动 I/O 操作）。这与 JS promise 不同，例如，后者即使从未使用也会完成运行过程。

Tokio

Tokio provides:

• 用于执行异步代码的多线程运行时。
• An asynchronous version of the standard library.
• 庞大的库生态系统。

use tokio::time;

async fn count_to(count: i32) {
    for i in 1..=count {
        println!("Count in task: {i}!");
        time::sleep(time::Duration::from_millis(5)).await;
    }
}

#[tokio::main]
async fn main() {
    tokio::spawn(count_to(10));

    for i in 1..5 {
        println!("Main task: {i}");
        time::sleep(time::Duration::from_millis(5)).await;
    }
}

• 借助 tokio::main 宏，我们现在可以将 main 设为异步函数。

• spawn 函数会创建新的并发 “任务”。

• 注意：spawn 使用 Future 方法实现，而不是对 count_to 调用 .await。

深入探索：

• 为何 count_to 通常无法计数到 10？这是一个异步取消的示例。tokio::spawn 会返回一个句柄，可以等待该句柄直至其代表的任务执行完毕。

• 尝试使用 count_to(10).await，而不是派生方法。

• 尝试等待 tokio::spawn 返回的任务执行完毕。

任务

Rust 有一个任务系统，这是一种轻量级线程处理形式。

每个任务只有一个顶级 Future，执行器会对此进行轮询来推进任务进度。该 Future 可能包含一个或多个嵌套的 Future，可以通过其 poll 方法对它们进行轮询，类似于调用堆栈。可以通过轮询多个子 Future（例如争用定时器和 I/O 操作）在任务内部实现并发操作。

use tokio::io::{self, AsyncReadExt, AsyncWriteExt};
use tokio::net::TcpListener;

#[tokio::main]
async fn main() -> io::Result<()> {
    let listener = TcpListener::bind("127.0.0.1:0").await?;
    println!("listening on port {}", listener.local_addr()?.port());

    loop {
        let (mut socket, addr) = listener.accept().await?;

        println!("connection from {addr:?}");

        tokio::spawn(async move {
            socket.write_all(b"Who are you?\n").await.expect("socket error");

            let mut buf = vec![0; 1024];
            let name_size = socket.read(&mut buf).await.expect("socket error");
            let name = std::str::from_utf8(&buf[..name_size]).unwrap().trim();
            let reply = format!("Thanks for dialing in, {name}!\n");
            socket.write_all(reply.as_bytes()).await.expect("socket error");
        });
    }
}

将此示例复制到准备好的 src/main.rs 文件中，并从该文件运行它。

请尝试使用像 nc 或 telnet 这样的 TCP 连接工具进行连接。

• 让学生想象一下，当连接多个客户端时，示例服务器会达到怎样的状态。存在哪些任务？具有哪些 Future？

• This is the first time we've seen an async block. This is similar to a closure, but does not take any arguments. Its return value is a Future, similar to an async fn.

• Refactor the async block into a function, and improve the error handling using ?.

异步通道

有些 crate 支持使用异步通道。例如 tokio：

use tokio::sync::mpsc::{self, Receiver};

async fn ping_handler(mut input: Receiver<()>) {
    let mut count: usize = 0;

    while let Some(_) = input.recv().await {
        count += 1;
        println!("Received {count} pings so far.");
    }

    println!("ping_handler complete");
}

#[tokio::main]
async fn main() {
    let (sender, receiver) = mpsc::channel(32);
    let ping_handler_task = tokio::spawn(ping_handler(receiver));
    for i in 0..10 {
        sender.send(()).await.expect("Failed to send ping.");
        println!("Sent {} pings so far.", i + 1);
    }

    drop(sender);
    ping_handler_task.await.expect("Something went wrong in ping handler task.");
}

• 将通道大小更改为 3，然后看看对操作执行会有什么影响。

• 总体而言，该接口类似于 上午课程 中所讲的 sync 通道。

• 尝试移除 std::mem::drop 调用。会出现什么情况？这是为什么？

• Flume crate 包含可以同时实现 sync、async send 和 recv 的渠道，为涉及 IO 和大量 CPU 处理任务的复杂应用提供了极大便利。

• 使用 async 通道的优势在于，我们能够将它们与其他 future 结合起来，从而创建复杂的控制流。

Futures Control Flow

可以将很多 Future 组合在一起生成并发计算流图。我们已经介绍过用作独立线程的任务类型。

• 联接
• 选择

加入

联接操作会等待一组 Future 全部就绪，然后返回它们的结果集合。这类似于 JavaScript 中的 Promise.al 或 Python 中的 asyncio.gather。

use anyhow::Result;
use futures::future;
use reqwest;
use std::collections::HashMap;

async fn size_of_page(url: &str) -> Result<usize> {
    let resp = reqwest::get(url).await?;
    Ok(resp.text().await?.len())
}

#[tokio::main]
async fn main() {
    let urls: [&str; 4] = [
        "https://google.com",
        "https://httpbin.org/ip",
        "https://play.rust-lang.org/",
        "BAD_URL",
    ];
    let futures_iter = urls.into_iter().map(size_of_page);
    let results = future::join_all(futures_iter).await;
    let page_sizes_dict: HashMap<&str, Result<usize>> =
        urls.into_iter().zip(results.into_iter()).collect();
    println!("{:?}", page_sizes_dict);
}

将此示例复制到准备好的 src/main.rs 文件中，并从该文件运行它。

• 对于多个类型不相交的 Future，可以使用 std::future::join! 进行处理，但必须要确定在编译时 Future 的数量。目前，可在 futures crate 中使用该功能，但很快也会在 std::future 中正式发布。

• The risk of join is that one of the futures may never resolve, this would cause your program to stall.

• 还可以将 join_all 与 join! 结合使用，并行处理所有对 http 服务的请求和数据库查询。尝试使用 futures::join! 将 tokio::time::sleep 添加到 Future 中。这不是一个超时操作（其需要使用 select!，下一章会详细介绍），而是展示了 join! 的使用方式。

选择

选择操作会等待一组 Future 中的任意一个就绪，并对 Future 产生的结果进行响应。在 JavaScript 中，该操作类似于 Promise.race。在 Python 中，它相当于 asyncio.wait(task_set, return_when=asyncio.FIRST_COMPLETED)。

Similar to a match statement, the body of select! has a number of arms, each of the form pattern = future => statement. When a future is ready, its return value is destructured by the pattern. The statement is then run with the resulting variables. The statement result becomes the result of the select! macro.

use tokio::sync::mpsc::{self, Receiver};
use tokio::time::{sleep, Duration};

#[derive(Debug, PartialEq)]
enum Animal {
    Cat { name: String },
    Dog { name: String },
}

async fn first_animal_to_finish_race(
    mut cat_rcv: Receiver<String>,
    mut dog_rcv: Receiver<String>,
) -> Option<Animal> {
    tokio::select! {
        cat_name = cat_rcv.recv() => Some(Animal::Cat { name: cat_name? }),
        dog_name = dog_rcv.recv() => Some(Animal::Dog { name: dog_name? })
    }
}

#[tokio::main]
async fn main() {
    let (cat_sender, cat_receiver) = mpsc::channel(32);
    let (dog_sender, dog_receiver) = mpsc::channel(32);
    tokio::spawn(async move {
        sleep(Duration::from_millis(500)).await;
        cat_sender.send(String::from("Felix")).await.expect("Failed to send cat.");
    });
    tokio::spawn(async move {
        sleep(Duration::from_millis(50)).await;
        dog_sender.send(String::from("Rex")).await.expect("Failed to send dog.");
    });

    let winner = first_animal_to_finish_race(cat_receiver, dog_receiver)
        .await
        .expect("Failed to receive winner");

    println!("Winner is {winner:?}");
}

• 在本示例中，猫和狗之间进行了一场比赛。first_animal_to_finish_race 会同时监听这两个通道，并选择最先到达终点的作为胜者。由于狗用时 50 毫秒，而猫用时 500 毫秒，前者在此比赛中大获全胜。

• 在本示例中，可以使用 oneshot 通道，因为这些通道只能接收一次 send 信号。

• 尝试为比赛添加截至时间，演示如何选择不同类型的 Future。

• 请注意，select! 会丢弃不匹配的分支，相对应的 Future 也会随之取消。最简单的方法是，每次执行 select! 时创建新的 Future。

  • 另一种方法是传递 &mut future 而不是 future 本身，但这可能会导致问题，在本幻灯片的 “固定” 部分进行了详细介绍。

关于 async/await 的误区

async / await 为处理并发异步编程提供了一种方便高效的抽象方法。然而，Rust 中的 async/await 模型也存在一些误区和隐患。本章将展示其中的部分内容：

• 阻塞执行器
• 固定
• 异步 trait
• 取消

阻塞执行器

大多数异步运行时支持并发运行 IO 任务。这意味着 CPU 的阻塞性任务会阻塞执行器，并阻止执行其他任务。最简单的方法是，尽可能使用异步等效方法。

use futures::future::join_all;
use std::time::Instant;

async fn sleep_ms(start: &Instant, id: u64, duration_ms: u64) {
    std::thread::sleep(std::time::Duration::from_millis(duration_ms));
    println!(
        "future {id} slept for {duration_ms}ms, finished after {}ms",
        start.elapsed().as_millis()
    );
}

#[tokio::main(flavor = "current_thread")]
async fn main() {
    let start = Instant::now();
    let sleep_futures = (1..=10).map(|t| sleep_ms(&start, t, t * 10));
    join_all(sleep_futures).await;
}

• 运行该代码，您会发现休眠操作是连续发生的，而不是并发进行的。

• "current_thread" 变种将所有任务放在单个线程上。这样做效果会更明显，但 bug 仍然存在于多线程变种中。

• 将 std::thread::sleep 切换为 tokio::time::sleep，并等待结果。

• 另一个修复方案是 tokio::task::spawn_blocking，其会生成实际线程并将句柄转换为 Future，且不会阻塞执行器。

• 不应将任务视为操作系统线程。它们之间并非一对一的映射关系，并且大多数执行器都支持在单个操作系统线程上运行多个任务。尤其是通过 FFI 与其他库交互时，会更容易出现问题，因为在 FFI 中，因为该库可能依赖于线程本地存储或映射到特定的操作系统线程（例如，CUDA）。在这些情况下，首选 tokio::task::spawn_blocking。

• 请谨慎使用同步互斥操作。对 .await 一直执行互斥操作能会导致另一个任务阻塞，并且该任务可能与其在同一线程上运行。

Pin

Async blocks and functions return types implementing the Future trait. The type returned is the result of a compiler transformation which turns local variables into data stored inside the future.

Some of those variables can hold pointers to other local variables. Because of that, the future should never be moved to a different memory location, as it would invalidate those pointers.

To prevent moving the future type in memory, it can only be polled through a pinned pointer. Pin is a wrapper around a reference that disallows all operations that would move the instance it points to into a different memory location.

use tokio::sync::{mpsc, oneshot};
use tokio::task::spawn;
use tokio::time::{sleep, Duration};

// A work item. In this case, just sleep for the given time and respond
// with a message on the `respond_on` channel.
#[derive(Debug)]
struct Work {
    input: u32,
    respond_on: oneshot::Sender<u32>,
}

// A worker which listens for work on a queue and performs it.
async fn worker(mut work_queue: mpsc::Receiver<Work>) {
    let mut iterations = 0;
    loop {
        tokio::select! {
            Some(work) = work_queue.recv() => {
                sleep(Duration::from_millis(10)).await; // Pretend to work.
                work.respond_on
                    .send(work.input * 1000)
                    .expect("failed to send response");
                iterations += 1;
            }
            // TODO: report number of iterations every 100ms
        }
    }
}

// A requester which requests work and waits for it to complete.
async fn do_work(work_queue: &mpsc::Sender<Work>, input: u32) -> u32 {
    let (tx, rx) = oneshot::channel();
    work_queue
        .send(Work { input, respond_on: tx })
        .await
        .expect("failed to send on work queue");
    rx.await.expect("failed waiting for response")
}

#[tokio::main]
async fn main() {
    let (tx, rx) = mpsc::channel(10);
    spawn(worker(rx));
    for i in 0..100 {
        let resp = do_work(&tx, i).await;
        println!("work result for iteration {i}: {resp}");
    }
}

• 您可能认为这是执行器模式的一个示例。执行器通常会循环调用 select!。

• 本部分是对前面几节课的总结，因此请多花时间用心学习。

  • 只是单纯地在 select! 中添加 _ = sleep(Duration::from_millis(100)) => { println!(..) }，该行代码将不会执行任何操作。这是为什么？

  • 请改为在 loop 外部添加包含该 Future 的 timeout_fut：

    #![allow(unused)]
    fn main() {
    let mut timeout_fut = sleep(Duration::from_millis(100));
    loop {
        select! {
            ..,
            _ = timeout_fut => { println!(..); },
        }
    }
    }

  • 这仍然不起作用。根据编译器提示的错误，通过向 select! 中的 timeout_fut 添加 &mut 解决移动问题，然后使用 Box::pin：

    #![allow(unused)]
    fn main() {
    let mut timeout_fut = Box::pin(sleep(Duration::from_millis(100)));
    loop {
        select! {
            ..,
            _ = &mut timeout_fut => { println!(..); },
        }
    }
    }

  • 可以编译这段代码了，但超时过期后，每次迭代都会变为 Poll::Ready（使用混合 Future 有助于解决此问题）。每次超时过期后，通过更新重置 timeout_fut。

• Box 在堆上进行分配。在某些情况下，也可以选择使用 std::pin::pin!（最近才正式发布，较旧的代码通常使用 tokio::pin!），但对于重新分配的 Future，使用此功能较为困难。

• 另一种替代方案是完全不使用 pin，而是生成另一个任务，该任务每隔 100 毫秒就会发送到 oneshot 通道。

• Data that contains pointers to itself is called self-referential. Normally, the Rust borrow checker would prevent self-referential data from being moved, as the references cannot outlive the data they point to. However, the code transformation for async blocks and functions is not verified by the borrow checker.

• Pin is a wrapper around a reference. An object cannot be moved from its place using a pinned pointer. However, it can still be moved through an unpinned pointer.

• The poll method of the Future trait uses Pin<&mut Self> instead of &mut Self to refer to the instance. That's why it can only be called on a pinned pointer.

异步特征

Async methods in traits are were stabilized only recently, in the 1.75 release. This required support for using return-position impl Trait (RPIT) in traits, as the desugaring for async fn includes -> impl Future<Output = ...>.

However, even with the native support today there are some pitfalls around async fn and RPIT in traits:

• Return-position impl Trait captures all in-scope lifetimes (so some patterns of borrowing cannot be expressed)

• Traits whose methods use return-position impl trait or async are not dyn compatible.

If we do need dyn support, the crate async_trait provides a workaround through a macro, with some caveats:

use async_trait::async_trait;
use std::time::Instant;
use tokio::time::{sleep, Duration};

#[async_trait]
trait Sleeper {
    async fn sleep(&self);
}

struct FixedSleeper {
    sleep_ms: u64,
}

#[async_trait]
impl Sleeper for FixedSleeper {
    async fn sleep(&self) {
        sleep(Duration::from_millis(self.sleep_ms)).await;
    }
}

async fn run_all_sleepers_multiple_times(
    sleepers: Vec<Box<dyn Sleeper>>,
    n_times: usize,
) {
    for _ in 0..n_times {
        println!("running all sleepers..");
        for sleeper in &sleepers {
            let start = Instant::now();
            sleeper.sleep().await;
            println!("slept for {}ms", start.elapsed().as_millis());
        }
    }
}

#[tokio::main]
async fn main() {
    let sleepers: Vec<Box<dyn Sleeper>> = vec![
        Box::new(FixedSleeper { sleep_ms: 50 }),
        Box::new(FixedSleeper { sleep_ms: 100 }),
    ];
    run_all_sleepers_multiple_times(sleepers, 5).await;
}

• async_trait 易于使用，但请注意，它通过堆分配来实现这一点。这种堆分配会产生性能开销。

• 对于 async trait 的语言支持中的挑战是深入 Rust的，并且可能不值得深入描述。如果您对深入了解感兴趣，Niko Matsakis 在这篇文章中对它们做了很好的解释。

• 尝试创建一个新的 sleeper 结构，使其随机休眠一段时间，并将其添加到 Vec 中。

消除

丢弃 Future 意味着无法再对其进行轮询。这称为 取消，在任何 await 点都可能发生。请务必小心谨慎，确保即使 Future 任务被取消，系统也能正常运行。例如，系统不应死锁或丢失数据。

use std::io::{self, ErrorKind};
use std::time::Duration;
use tokio::io::{AsyncReadExt, AsyncWriteExt, DuplexStream};

struct LinesReader {
    stream: DuplexStream,
}

impl LinesReader {
    fn new(stream: DuplexStream) -> Self {
        Self { stream }
    }

    async fn next(&mut self) -> io::Result<Option<String>> {
        let mut bytes = Vec::new();
        let mut buf = [0];
        while self.stream.read(&mut buf[..]).await? != 0 {
            bytes.push(buf[0]);
            if buf[0] == b'\n' {
                break;
            }
        }
        if bytes.is_empty() {
            return Ok(None);
        }
        let s = String::from_utf8(bytes)
            .map_err(|_| io::Error::new(ErrorKind::InvalidData, "not UTF-8"))?;
        Ok(Some(s))
    }
}

async fn slow_copy(source: String, mut dest: DuplexStream) -> std::io::Result<()> {
    for b in source.bytes() {
        dest.write_u8(b).await?;
        tokio::time::sleep(Duration::from_millis(10)).await
    }
    Ok(())
}

#[tokio::main]
async fn main() -> std::io::Result<()> {
    let (client, server) = tokio::io::duplex(5);
    let handle = tokio::spawn(slow_copy("hi\nthere\n".to_owned(), client));

    let mut lines = LinesReader::new(server);
    let mut interval = tokio::time::interval(Duration::from_millis(60));
    loop {
        tokio::select! {
            _ = interval.tick() => println!("tick!"),
            line = lines.next() => if let Some(l) = line? {
                print!("{}", l)
            } else {
                break
            },
        }
    }
    handle.await.unwrap()?;
    Ok(())
}

• 编译器无法确保取消操作的安全性。您需要阅读 API 文档，并考虑 async fn 所持状态。

• 与 panic 和 ? 不同，取消属于正常控制流的一部分（而非错误处理）。

• 该示例丢失了字符串的某些部分。

  • 每当 tick() 分支先完成操作时，next() 及其 buf` 均会被丢弃。

  • 通过将 buf 整合到结构体中，`LinesReader 可以确保取消操作的安全性：

    #![allow(unused)]
    fn main() {
    struct LinesReader {
        stream: DuplexStream,
        bytes: Vec<u8>,
        buf: [u8; 1],
    }
    
    impl LinesReader {
        fn new(stream: DuplexStream) -> Self {
            Self { stream, bytes: Vec::new(), buf: [0] }
        }
        async fn next(&mut self) -> io::Result<Option<String>> {
            // prefix buf and bytes with self.
            // ...
            let raw = std::mem::take(&mut self.bytes);
            let s = String::from_utf8(raw)
            // ...
        }
    }
    }

• Interval::tick is cancellation-safe because it keeps track of whether a tick has been 'delivered'.

• AsyncReadExt::read is cancellation-safe because it either returns or doesn't read data.

• AsyncBufReadExt::read_line is similar to the example and isn't cancellation-safe. See its documentation for details and alternatives.

习题

为了练习您的异步 Rust 技能，我们再次为您提供了两个练习：

• 哲学家进餐：我们已经在上午看到了这个问题。这次你将使用异步 Rust 来实现它。

• 广播聊天应用：这是一个更大的项目，允许您尝试更高级的异步Rust功能。

After looking at the exercises, you can look at the solutions provided.

Dining Philosophers --- Async

查看哲学家进餐以获取问题的描述。

As before, you will need a local Cargo installation for this exercise. Copy the code below to a file called src/main.rs, fill out the blanks, and test that cargo run does not deadlock:

use std::sync::Arc;
use tokio::sync::mpsc::{self, Sender};
use tokio::sync::Mutex;
use tokio::time;

struct Fork;

struct Philosopher {
    name: String,
    // left_fork: ...
    // right_fork: ...
    // thoughts: ...
}

impl Philosopher {
    async fn think(&self) {
        self.thoughts
            .send(format!("Eureka! {} has a new idea!", &self.name))
            .await
            .unwrap();
    }

    async fn eat(&self) {
        // Keep trying until we have both forks
        println!("{} is eating...", &self.name);
        time::sleep(time::Duration::from_millis(5)).await;
    }
}

static PHILOSOPHERS: &[&str] =
    &["Socrates", "Hypatia", "Plato", "Aristotle", "Pythagoras"];

#[tokio::main]
async fn main() {
    // Create forks

    // Create philosophers

    // Make them think and eat

    // Output their thoughts
}

因为这次您正在使用异步Rust，您将需要一个 tokio 依赖。您可以使用以下的 Cargo.toml：

[package]
name = "dining-philosophers-async-dine"
version = "0.1.0"
edition = "2021"

[dependencies]
tokio = { version = "1.26.0", features = ["sync", "time", "macros", "rt-multi-thread"] }

另外，请注意，这次您必须使用来自 tokio 包的 Mutex 和 mpsc 模块。

• Can you make your implementation single-threaded?

广播聊天应用

在本练习中，我们想要使用我们的新知识来实现一个广播聊天应用。我们有一个聊天服务器，客户端连接到该服务器并发布他们的消息。客户端从标准输入读取用户消息，并将其发送到服务器。聊天服务器将收到的每条消息广播给所有客户端。

For this, we use a broadcast channel on the server, and tokio_websockets for the communication between the client and the server.

创建一个新的 Cargo 项目并添加以下依赖：

Cargo.toml:

[package]
name = "chat-async"
version = "0.1.0"
edition = "2021"

[dependencies]
futures-util = { version = "0.3.30", features = ["sink"] }
http = "1.1.0"
tokio = { version = "1.37.0", features = ["full"] }
tokio-websockets = { version = "0.7.0", features = ["client", "fastrand", "server", "sha1_smol"] }

所需的API

You are going to need the following functions from tokio and tokio_websockets. Spend a few minutes to familiarize yourself with the API.

• StreamExt::next() implemented by WebSocketStream: for asynchronously reading messages from a Websocket Stream.
• SinkExt::send() implemented by WebSocketStream: for asynchronously sending messages on a Websocket Stream.
• Lines::next_line(): for asynchronously reading user messages from the standard input.
• Sender::subscribe()：用于订阅广播频道。

两个可执行文件

Normally in a Cargo project, you can have only one binary, and one src/main.rs file. In this project, we need two binaries. One for the client, and one for the server. You could potentially make them two separate Cargo projects, but we are going to put them in a single Cargo project with two binaries. For this to work, the client and the server code should go under src/bin (see the documentation).

Copy the following server and client code into src/bin/server.rs and src/bin/client.rs, respectively. Your task is to complete these files as described below.

src/bin/server.rs:

use futures_util::sink::SinkExt;
use futures_util::stream::StreamExt;
use std::error::Error;
use std::net::SocketAddr;
use tokio::net::{TcpListener, TcpStream};
use tokio::sync::broadcast::{channel, Sender};
use tokio_websockets::{Message, ServerBuilder, WebSocketStream};

async fn handle_connection(
    addr: SocketAddr,
    mut ws_stream: WebSocketStream<TcpStream>,
    bcast_tx: Sender<String>,
) -> Result<(), Box<dyn Error + Send + Sync>> {

    // TODO: For a hint, see the description of the task below.

}

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error + Send + Sync>> {
    let (bcast_tx, _) = channel(16);

    let listener = TcpListener::bind("127.0.0.1:2000").await?;
    println!("listening on port 2000");

    loop {
        let (socket, addr) = listener.accept().await?;
        println!("New connection from {addr:?}");
        let bcast_tx = bcast_tx.clone();
        tokio::spawn(async move {
            // Wrap the raw TCP stream into a websocket.
            let ws_stream = ServerBuilder::new().accept(socket).await?;

            handle_connection(addr, ws_stream, bcast_tx).await
        });
    }
}

src/bin/client.rs:

use futures_util::stream::StreamExt;
use futures_util::SinkExt;
use http::Uri;
use tokio::io::{AsyncBufReadExt, BufReader};
use tokio_websockets::{ClientBuilder, Message};

#[tokio::main]
async fn main() -> Result<(), tokio_websockets::Error> {
    let (mut ws_stream, _) =
        ClientBuilder::from_uri(Uri::from_static("ws://127.0.0.1:2000"))
            .connect()
            .await?;

    let stdin = tokio::io::stdin();
    let mut stdin = BufReader::new(stdin).lines();


    // TODO: For a hint, see the description of the task below.

}

运行可执行文件

Run the server with:

cargo run --bin server

and the client with:

cargo run --bin client

任务

• 在 src/bin/server.rs 中实现 handle_connection 函数。
  • 提示：使用 tokio::select! 在一个连续的循环中并发执行两个任务。一个任务从客户端接收消息并广播它们。另一个任务将服务器接收到的消息发送给客户端。
• 完成 src/bin/client.rs 中的 main 函数。
  • Hint: As before, use tokio::select! in a continuous loop for concurrently performing two tasks: (1) reading user messages from standard input and sending them to the server, and (2) receiving messages from the server, and displaying them for the user.
• Optional: Once you are done, change the code to broadcast messages to all clients, but the sender of the message.

并发编程：下午练习

Dining Philosophers --- Async

(返回练习)

use std::sync::Arc;
use tokio::sync::mpsc::{self, Sender};
use tokio::sync::Mutex;
use tokio::time;

struct Fork;

struct Philosopher {
    name: String,
    left_fork: Arc<Mutex<Fork>>,
    right_fork: Arc<Mutex<Fork>>,
    thoughts: Sender<String>,
}

impl Philosopher {
    async fn think(&self) {
        self.thoughts
            .send(format!("Eureka! {} has a new idea!", &self.name))
            .await
            .unwrap();
    }

    async fn eat(&self) {
        // Keep trying until we have both forks
        let (_left_fork, _right_fork) = loop {
            // Pick up forks...
            let left_fork = self.left_fork.try_lock();
            let right_fork = self.right_fork.try_lock();
            let Ok(left_fork) = left_fork else {
                // If we didn't get the left fork, drop the right fork if we
                // have it and let other tasks make progress.
                drop(right_fork);
                time::sleep(time::Duration::from_millis(1)).await;
                continue;
            };
            let Ok(right_fork) = right_fork else {
                // If we didn't get the right fork, drop the left fork and let
                // other tasks make progress.
                drop(left_fork);
                time::sleep(time::Duration::from_millis(1)).await;
                continue;
            };
            break (left_fork, right_fork);
        };

        println!("{} is eating...", &self.name);
        time::sleep(time::Duration::from_millis(5)).await;

        // The locks are dropped here
    }
}

static PHILOSOPHERS: &[&str] =
    &["Socrates", "Hypatia", "Plato", "Aristotle", "Pythagoras"];

#[tokio::main]
async fn main() {
    // Create forks
    let mut forks = vec![];
    (0..PHILOSOPHERS.len()).for_each(|_| forks.push(Arc::new(Mutex::new(Fork))));

    // Create philosophers
    let (philosophers, mut rx) = {
        let mut philosophers = vec![];
        let (tx, rx) = mpsc::channel(10);
        for (i, name) in PHILOSOPHERS.iter().enumerate() {
            let left_fork = Arc::clone(&forks[i]);
            let right_fork = Arc::clone(&forks[(i + 1) % PHILOSOPHERS.len()]);
            philosophers.push(Philosopher {
                name: name.to_string(),
                left_fork,
                right_fork,
                thoughts: tx.clone(),
            });
        }
        (philosophers, rx)
        // tx is dropped here, so we don't need to explicitly drop it later
    };

    // Make them think and eat
    for phil in philosophers {
        tokio::spawn(async move {
            for _ in 0..100 {
                phil.think().await;
                phil.eat().await;
            }
        });
    }

    // Output their thoughts
    while let Some(thought) = rx.recv().await {
        println!("Here is a thought: {thought}");
    }
}

广播聊天应用

(返回练习)

src/bin/server.rs:

use futures_util::sink::SinkExt;
use futures_util::stream::StreamExt;
use std::error::Error;
use std::net::SocketAddr;
use tokio::net::{TcpListener, TcpStream};
use tokio::sync::broadcast::{channel, Sender};
use tokio_websockets::{Message, ServerBuilder, WebSocketStream};

async fn handle_connection(
    addr: SocketAddr,
    mut ws_stream: WebSocketStream<TcpStream>,
    bcast_tx: Sender<String>,
) -> Result<(), Box<dyn Error + Send + Sync>> {

    ws_stream
        .send(Message::text("Welcome to chat! Type a message".to_string()))
        .await?;
    let mut bcast_rx = bcast_tx.subscribe();

    // A continuous loop for concurrently performing two tasks: (1) receiving
    // messages from `ws_stream` and broadcasting them, and (2) receiving
    // messages on `bcast_rx` and sending them to the client.
    loop {
        tokio::select! {
            incoming = ws_stream.next() => {
                match incoming {
                    Some(Ok(msg)) => {
                        if let Some(text) = msg.as_text() {
                            println!("From client {addr:?} {text:?}");
                            bcast_tx.send(text.into())?;
                        }
                    }
                    Some(Err(err)) => return Err(err.into()),
                    None => return Ok(()),
                }
            }
            msg = bcast_rx.recv() => {
                ws_stream.send(Message::text(msg?)).await?;
            }
        }
    }
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error + Send + Sync>> {
    let (bcast_tx, _) = channel(16);

    let listener = TcpListener::bind("127.0.0.1:2000").await?;
    println!("listening on port 2000");

    loop {
        let (socket, addr) = listener.accept().await?;
        println!("New connection from {addr:?}");
        let bcast_tx = bcast_tx.clone();
        tokio::spawn(async move {
            // Wrap the raw TCP stream into a websocket.
            let ws_stream = ServerBuilder::new().accept(socket).await?;

            handle_connection(addr, ws_stream, bcast_tx).await
        });
    }
}

src/bin/client.rs:

use futures_util::stream::StreamExt;
use futures_util::SinkExt;
use http::Uri;
use tokio::io::{AsyncBufReadExt, BufReader};
use tokio_websockets::{ClientBuilder, Message};

#[tokio::main]
async fn main() -> Result<(), tokio_websockets::Error> {
    let (mut ws_stream, _) =
        ClientBuilder::from_uri(Uri::from_static("ws://127.0.0.1:2000"))
            .connect()
            .await?;

    let stdin = tokio::io::stdin();
    let mut stdin = BufReader::new(stdin).lines();

    // Continuous loop for concurrently sending and receiving messages.
    loop {
        tokio::select! {
            incoming = ws_stream.next() => {
                match incoming {
                    Some(Ok(msg)) => {
                        if let Some(text) = msg.as_text() {
                            println!("From server: {}", text);
                        }
                    },
                    Some(Err(err)) => return Err(err.into()),
                    None => return Ok(()),
                }
            }
            res = stdin.next_line() => {
                match res {
                    Ok(None) => return Ok(()),
                    Ok(Some(line)) => ws_stream.send(Message::text(line.to_string())).await?,
                    Err(err) => return Err(err.into()),
                }
            }

        }
    }
}

谢谢！

Thank you for taking Comprehensive Rust 🦀! We hope you enjoyed it and that it was useful.

组织这门课程让我们收获了很多乐趣。本课程并非完美无缺，因此，如果您发现任何错误或有任何改进建议，请在 GitHub 上与我们联系。我们期待收到您的宝贵意见。

词汇表

本页面的词汇表提供了许多 Rust 术语的简要定义。同时提供翻译版本和英语原文的对应。

• 分配（allocate）：
  在 堆 上进行动态内存分配。
• 参数（argument）：
  传入某个函数或方法中的信息。
• 裸机 Rust（Bare-metal Rust）：
  底层 Rust 开发方式，通常部署于没有操作系统的系统。请参阅 裸机 Rust。
• 代码块（block）：
  请参阅 代码块 和 作用域。
• 借用（borrow）：
  请参阅 借用。
• 借用检查器（borrow checker）：
  Rust 编译器的一部分，用于检查所有借用操作是否有效。
• 大括号（brace）：
  { 和 }。也称为 花括号，用于分隔 代码块。
• 构建（build）：
  将源代码转换为可执行代码或可用程序的过程。
• 调用（call）：
  调用或执行某个函数或方法。
• 通道（channel）：
  用于安全地 在线程之间 传递消息。
• Comprehensive Rust 🦀：
  本课程统称为 Comprehensive Rust 🦀。
• 并发（concurrency）：
  同时执行多个任务或进程。
• Rust 中的并发（Concurrency in Rust）：
  请参阅 Rust 中的并发.
• 常量（constant）：
  在程序执行期间不会改变的值。
• 控制流（control flow）：
  程序中各个语句或指令的执行顺序。
• 崩溃（crash）：
  程序出现意外的、未处理的故障或终止。
• 枚举（enumeration）：
  一种用于保存多个已命名常量中的一个的数据类型，可能还有一个相关的元组或结构体。
• 错误（error）：
  与预期行为存在偏差的意外情况或结果。
• 错误处理（error handling）：
  对程序执行期间发生的错误进行管理和响应的过程。
• 练习（exercise）：
  专为练习和测试编程技能而设计的任务或问题。
• 函数（function）：
  用于执行特定任务且可重复使用的代码块。
• 垃圾回收器（garbage collector）：
  一种自动释放不再使用的对象所占内存的机制。
• 泛型（generics）：
  这项功能支持使用类型占位符编写代码，支持对不同数据类型的代码进行重复使用。
• 不可变（immutable）：
  创建后无法再进行更改。
• 集成测试（integration test）：
  一种验证系统不同部分或组件之间交互的测试类型。
• 关键字（keyword）：
  编程语言中的保留字，具有特定含义且不能用作标识符。
• 库（library）：
  程序可以使用的一组预编译例程或代码。
• 宏（macro）：
  Rust 宏可通过名称中的 ! 符号识别。当普通函数无法满足需求时，可以使用宏。一个典型示例是 format!，其接受可变数量的参数，但 Rust 函数不支持这种类型。
• main 函数（main function）：
  Rust 程序从 main 函数开始执行。
• 匹配（match）：
  Rust 中的控制流结构，允许对表达式的值进行模式匹配。
• 内存泄漏（memory leak）：
  程序无法释放不再不要的内存的情况，会导致内存用量不断增加。
• 方法（method）：
  与 Rust 中的某个对象或类型相关联的函数。
• 模块（module）：
  Rust 中用于归纳整理代码的命名空间，其中包含函数、类型或特性等定义。
• 移动（move）：
  在 Rust 中，将值的所有权从一个变量转移到另一个变量。
• 可变（mutable）：
  Rust 中的一个属性，支持在声明变量后对其进行修改。
• 所有权（ownership）：
  Rust 中的概念，用于定义代码中的哪一部分负责管理与值关联的内存。
• panic：
  Rust 中导致程序终止且不可恢复的错误情况。
• 参数（parameter）：
  在调用函数或方法时传入函数或方法的值。
• 模式（pattern）：
  Rust 中可与表达式匹配的值、字面量或结构的组合。
• 载荷（payload）：
  消息、事件或数据结构所携带的数据或信息。
• 程序（program）：
  计算机为执行特定任务或解决特定问题而执行的一组指令。
• 编程语言（programming language）：
  用于向计算机传递指令的正式系统，例如 Rust。
• 接收器（receiver）：
  Rust 方法中的首个参数，表示正在调用该方法的实例。
• 引用计数（reference counting）：
  一种内存管理方法，可以跟踪某个对象的引用数量，并在计数为零时释放该对象。
• 返回（return）：
  Rust 中的一个关键字，用于表示从函数返回的值。
• Rust：
  一种系统编程语言，专注于安全性、性能和并发性。
• Rust 基础（Rust Fundamentals）：
  本课程第 1 天到第 4 天的内容。
• Android 中的 Rust（Rust in Android）：
  请参阅 Android 中的 Rust.
• Chromium 中的 Rust（Rust in Chromium）:
  请参阅 Chromium 中的 Rust.
• 安全（safe）：
  指代码遵循 Rust 的所有权和借用规则，以防止出现与内存相关的错误。
• 作用域（scope）：
  程序中变量有效且可使用的区域。
• 标准库（standard library）：
  Rust 中提供基本功能的一系列模块。
• 静态（static）：
  Rust 中的关键字，用于定义具有 'static 生命周期的静态变量或项。
• 字符串（string）：
  一种存储文本数据的数据类型。如需了解详情，请参阅 String 与 str。
• 结构体（struct）：
  Rust 中的复合数据类型，可将不同类型的变量归到同一名称下。
• 测试（test）：
  Rust 中的模块，其中包含用于测试其他函数是否正确的函数。
• 线程（thread）：
  程序中的单独执行顺序，支持并发执行。
• 线程安全（thread safety）：
  一种程序属性，用于确保多线程环境中的行为正确无误。
• 特征（trait）：
  用于定义未知类型的一系列方法，为在 Rust 中实现多态性提供了方法。
• 特征约束（trait bound）:
  一种可以要求类型实现一些感兴趣的特性的抽象。
• 元组（tuple）：
  包含不同类型变量的复合数据类型。元组的字段没有名称，需要通过序号访问。
• 类型（type）：
  一种分类方式，用于指定可以对 Rust 中特定类型的值执行哪些操作。
• 类型推导（type inference）：
  Rust 编译器能够推断变量或表达式的类型。
• 未定义行为（undefined behavior）：
  Rust 中未指定结果的操作或条件，通常会导致不可预测的程序行为。
• 联合体（union）：
  一种数据类型，可以存储不同类型的值，但一次只能保存一个值。
• 单元测试（unit test）：
  Rust 内置了运行小型单元测试和大型集成测试的支持功能。请参阅 单元测试。
• 单元类型（unit type）:
  不保存数据的类型，写为没有成员的元组。
• 不安全（unsafe）：
  Rust 的子集，允许触发 未定义行为。请参阅 不安全 Rust。
• 变量（variable）：
  用于存储数据的内存位置。变量在 作用域 内有效。

其他 Rust 资源

Rust 社区已经创造了丰富的高质量免费资源在线提供。

官方文档

Rust 项目提供了许多资源。这些资源涵盖了 Rust 的一般内容：

• Rust 程序设计语言：一部有关 Rust 的免费权威图书。书中详细介绍了该语言，并包含一些可供读者构建的项目。
• 通过例子学 Rust：通过一系列展示不同结构的示例介绍 Rust 语法。有时会包括一些小练习，会要求您充分地阐述示例中的代码。
• Rust 标准库：Rust 标准库的完整文档。
• Rust 参考手册：一本未完成的书，介绍了 Rust 语法和内存模型。

Rust 官方网站上有更多专业指南：

• Rust 秘典：介绍了不安全 Rust，包括使用原始指针以及与其他语言 (FFI) 交互。
• Rust 中的异步编程：介绍了在《Rust 程序设计语言》成书后引入的新异步编程模型。
• 嵌入式 Rust 之书：介绍如何在没有操作系统的嵌入式设备上使用 Rust。

非官方学习资料

其他 Rust 指南和教程的小选集：

• Learn Rust the Dangerous Way（以危险的方式学 Rust）：从低级 C 语言程序员的角度介绍 Rust。
• 面向嵌入式 C 程序员的 Rust：从使用 C 语言编写固件的开发者的角度介绍 Rust。
• Rust for professionals（面向专业人士的 Rust）：通过与其他语言（例如 C、C++、Java、JavaScript 和 Python）进行并排比较，介绍 Rust 的语法。
• Rust on Exercism（在 Exercism 上学 Rust）：100 多项练习助您学习 Rust。
• Ferrous Teaching Material：一系列小演示文稿，涵盖 Rust 语言的基础知识和高级部分。还涵盖了 WebAssembly 和 async/await 等其他主题。
• 面向 Rust 的初学者系列和使用 Rust 迈出第一步：两个面向新手开发者的 Rust 指南。第一个指南包含 35 个视频，第二个指南包含 11 个模块，内容涵盖 Rust 语法和基本结构。
• 通过大量的链表学习Rust：通过实现几种不同类型的列表结构，深入探索 Rust 的内存管理规则。

如需更多 Rust 图书，请查看 Rust 小册。

鸣谢

本课中的资料以众多优秀的 Rust 文档资源为基础。 如需查看实用资源的完整列表， 请参阅关于其他资源的页面。

The material of Comprehensive Rust is licensed under the terms of the Apache 2.0 license, please see LICENSE for details.

Rust 示例

部分示例和练习复制并 改编自Rust by Example。如需了解详情（包括许可 条款），请参阅 third_party/rust-by-example/ 目录。

Rust on Exercism

部分练习复制并 改编自 Rust on Exercism。如需了解详情（包括许可 条款），请参阅 third_party/rust-on-exercism/ 目录。

CXX

“与 C++ 的互操作性”部分引用了一张 来自 CXX 的图片。如需了解详情（包括许可条款）， 请参阅 third_party/cxx/ 目录。