mohitmishra786/low-level-dev-skills

cpp-coroutines

C++20 coroutines skill for understanding coroutine mechanics and debugging.

Voir la source

Mode d’emploi

  1. Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
  2. Consultez le document original pour les outils, identifiants et environnements requis.
  3. Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/mohitmishra786/low-level-dev-skills --skill cpp-coroutines

Avant de commencer

  • L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
  • Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Document Skill original

Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.

C++20 Coroutines

Purpose

Guide agents through C++20 coroutine mechanics: co_await, co_yield, co_return, implementing the required promise_type, understanding coroutine frame memory layout, debugging suspended coroutines in GDB, and reducing frame allocation overhead.

Triggers

  • "How do coawait, coyield, and co_return work?"
  • "How do I implement promise_type for a coroutine?"
  • "How does a coroutine suspend and resume?"
  • "How do I debug a suspended coroutine in GDB?"
  • "How much memory does a coroutine frame use?"
  • "How do I write a generator with co_yield?"

Workflow

1. The three coroutine keywords

cpp
// co_return — return a value and end the coroutine
co_return value;

// co_yield — produce a value, suspend, resume later
co_yield value;

// co_await — suspend until an awaitable completes
auto result = co_await some_awaitable;

A function is a coroutine if it contains any of these three keywords. Its return type must be a coroutine type with a promise_type.

2. Minimal coroutine type — Task

cpp
#include <coroutine>
#include <stdexcept>
#include <optional>

template <typename T>
struct Task {
    struct promise_type {
        std::optional<T> value;
        std::exception_ptr exception;

        Task get_return_object() {
            return Task{std::coroutine_handle<promise_type>::from_promise(*this)};
        }

        std::suspend_always initial_suspend() { return {}; }  // lazy start
        std::suspend_always final_suspend() noexcept { return {}; }

        void return_value(T v) { value = std::move(v); }

        void unhandled_exception() { exception = std::current_exception(); }
    };

    std::coroutine_handle<promise_type> handle;

    explicit Task(std::coroutine_handle<promise_type> h) : handle(h) {}

    Task(Task&&) = default;
    Task& operator=(Task&&) = default;

    ~Task() { if (handle) handle.destroy(); }

    T get() {
        handle.resume();                      // resume to completion
        if (handle.promise().exception)
            std::rethrow_exception(handle.promise().exception);
        return std::move(*handle.promise().value);
    }
};

// Usage
Task<int> compute() {
    co_return 42;
}

int main() {
    auto task = compute();
    int result = task.get();   // 42
}

3. Generator with co_yield

cpp
template <typename T>
struct Generator {
    struct promise_type {
        T current_value;

        Generator get_return_object() {
            return Generator{std::coroutine_handle<promise_type>::from_promise(*this)};
        }

        std::suspend_always initial_suspend() { return {}; }
        std::suspend_always final_suspend() noexcept { return {}; }
        void return_void() {}
        void unhandled_exception() { throw; }

        std::suspend_always yield_value(T value) {
            current_value = value;
            return {};                     // suspend after yielding
        }
    };

    std::coroutine_handle<promise_type> handle;

    explicit Generator(std::coroutine_handle<promise_type> h) : handle(h) {}
    ~Generator() { if (handle) handle.destroy(); }

    struct iterator {
        std::coroutine_handle<promise_type> handle;
        bool done;

        iterator& operator++() {
            handle.resume();
            done = handle.done();
            return *this;
        }
        T operator*() const { return handle.promise().current_value; }
        bool operator!=(std::default_sentinel_t) const { return !done; }
    };

    iterator begin() {
        handle.resume();                   // advance to first yield
        return {handle, handle.done()};
    }
    std::default_sentinel_t end() { return {}; }
};

// Usage
Generator<int> iota(int start, int end) {
    for (int i = start; i < end; ++i)
        co_yield i;
}

for (int x : iota(0, 5)) {
    std::cout << x << ' ';   // 0 1 2 3 4
}

4. Awaitable — custom co_await target

cpp
// An awaitable has three methods:
// await_ready() — true means don't suspend
// await_suspend(handle) — suspend: store handle, schedule resume
// await_resume() — return value of co_await expression

struct TimerAwaitable {
    int delay_ms;

    bool await_ready() const noexcept { return delay_ms <= 0; }

    void await_suspend(std::coroutine_handle<> h) {
        // Schedule h.resume() to be called after delay
        std::thread([h, this]() {
            std::this_thread::sleep_for(std::chrono::milliseconds(delay_ms));
            h.resume();
        }).detach();
    }

    void await_resume() const noexcept {}  // no return value
};

// suspend_always and suspend_never are built-in awaitables
std::suspend_always{};   // always suspends
std::suspend_never{};    // never suspends (no-op)

5. Coroutine frame layout and memory

The compiler allocates a coroutine frame (heap object) containing:

  • Local variables that live across suspension points
  • The promise object
  • The current suspension state (where to resume)
  • A pointer to the resumption/destruction functions
cpp
// Inspect frame size with Compiler Explorer (godbolt.org)
// Compile with: g++ -std=c++20 -O2 -S
// Look for: operator new call size in the generated asm
// Or: clang -std=c++20 -O2 -emit-llvm -S | grep "coro.size"

// Reduce frame size:
// 1. Don't keep large objects alive across co_await
struct Bad {
    std::vector<char> large_buf;   // whole vector lives in frame
    co_return large_buf.size();    // large_buf crosses suspension
};

// 2. Move data out before suspending
std::vector<char> buf = get_data();
size_t sz = buf.size();            // capture only what's needed
buf.clear();                       // release before suspension
co_await next_event;
// sz still valid; buf released

6. Debugging suspended coroutines in GDB

bash
# Coroutines appear as regular stack frames after resume()
# To inspect a suspended coroutine:

(gdb) info locals
# Look for coroutine_handle variables

# Print the promise object
(gdb) p *(promise_type*)(handle.__handle_)
# GDB 14+ has coroutine-specific support
(gdb) info coroutines        # GCC coroutine support (experimental)

# Step through coroutine execution
(gdb) step     # enters co_await implementation
(gdb) finish   # returns from coroutine frame function
(gdb) next     # step over suspension point

# View all threads (coroutines running on thread pool)
(gdb) info threads
(gdb) thread 2
(gdb) bt

7. Boost.Asio co_spawn and co_await

cpp
#include <boost/asio.hpp>
#include <boost/asio/co_spawn.hpp>
#include <boost/asio/awaitable.hpp>

namespace net = boost::asio;

net::awaitable<void> echo_session(net::ip::tcp::socket socket) {
    char buf[1024];
    for (;;) {
        std::size_t n = co_await socket.async_read_some(net::buffer(buf));
        co_await net::async_write(socket, net::buffer(buf, n));
    }
}

int main() {
    net::io_context io;
    net::co_spawn(io, listen_accept(io), net::detached);
    io.run();
}

co_spawn launches coroutines on an executor; co_await chains completion tokens without callback nesting.

8. std::generator (C++23)

cpp
#include <generator>
#include <ranges>

std::generator<int> fibonacci() {
    int a = 0, b = 1;
    while (true) {
        co_yield a;
        auto next = a + b;
        a = b;
        b = next;
    }
}

// Usage
for (int v : fibonacci() | std::views::take(10))
    printf("%d\n", v);

Lazy sequences without manual coroutine handle management — compiler provides std::generator promise type.

9. Coroutine frame layout in GDB

bash
# Compile with debug info
g++ -std=c++20 -g -O0 -o app app.cpp
gdb ./app
gdb
(gdb) break my_coro
(gdb) run
(gdb) info frame                    # current stack frame
(gdb) info locals                   # promise, handle in scope

# Inspect coroutine frame pointer (compiler-specific mangling)
(gdb) p *(MyPromise*)h.address()    # h = coroutine_handle

# GCC coroutine support (GCC 14+)
(gdb) info coroutines

# Pretty-print promise state
(gdb) set print pretty on
(gdb) p promise

Suspended coroutines may not appear on stack until resumed — trace via stored coroutine_handle.

10. Compilation time impact

Coroutines increase template instantiation and header parsing cost:

MitigationEffect
-O2 HALOReduces generated frame glue
Out-of-line co_await in .cppCuts recompilation cascade
Pimpl for coroutine return typesHides awaitable templates from headers
ccache / modulesSee skills/rust/rust-build-times patterns for C++

Measure with g++ -ftime-report or clang -ftime-trace. Coroutine-heavy headers (Asio) benefit from unity builds sparingly — balance with RAM use.

11. Common pitfalls

IssueCauseFix
co_await in a non-coroutineFunction missing coroutine return typeChange return type to a coroutine type
Dangling handle after co_returnUsing handle after coroutine finishesCheck handle.done() before resume
Double-resumeResuming an already-resumed coroutineTrack state; only resume when suspended
Coroutine frame never freedForgot handle.destroy()Use RAII wrapper (Task, Generator)
Heap allocation overheadNew frame per coroutine callEnable HALO (Heap Allocation eLision Optimization) with -O2
Recursive co_await depthStack overflow from deep chainsUse std::coroutine_handle<> tail-call pattern

Related skills

  • Use skills/compilers/cpp-templates for other advanced C++20 features
  • Use skills/rust/rust-async-internals for Rust's equivalent Future/Poll model
  • Use skills/debuggers/gdb for GDB session management

Tâches connexes

du même dépôt

Autres Skills

Tous les Skills
mohitmishra786
Communauté

core-dumps

Core dump analysis skill for production crash triage. Use when loading core files in GDB or LLDB, enabling core dump generation on Linux/macOS, mapping symbols with debuginfo or debuginfod, or extracting backtraces from crashes without re-running the program. Activates on queries about core files, ulimit, coredumpctl, debuginfod, crash triage, or analyzing segfaults from production binaries.

installations
2
GitHub Stars
267
Mis à jour
27 juin
mohitmishra786
Communauté

custom-allocators

Custom allocator skill for memory allocation strategies. Use when implementing pool/slab/arena allocators, tuning jemalloc/mimalloc, writing Rust GlobalAlloc, or benchmarking allocator performance. Activates on queries about jemalloc, mimalloc, tcmalloc, arena allocator, GlobalAlloc, or fragmentation.

installations
2
GitHub Stars
267
Mis à jour
27 juin
mohitmishra786
Communauté

ebpf

eBPF skill for Linux observability and networking. Use when writing eBPF programs with libbpf or bpftrace, attaching kprobes/tracepoints/XDP hooks, debugging verifier errors, working with eBPF maps, or achieving CO-RE portability across kernel versions. Activates on queries about eBPF, bpftool, bpftrace, XDP programs, libbpf, verifier errors, eBPF maps, or kernel tracing with BPF.

installations
2
GitHub Stars
267
Mis à jour
27 juin
mohitmishra786
Communauté

flamegraphs

Flamegraph generation and interpretation skill. Use when converting perf, Valgrind Callgrind, or other profiler output into SVG flamegraphs using Brendan Gregg's FlameGraph tools, or when reading flamegraphs to identify performance bottlenecks. Activates on queries about flamegraphs, stackcollapse, flamegraph.svg, identifying hot frames, wide vs tall frames, or performance visualisation.

installations
2
GitHub Stars
267
Mis à jour
27 juin