More tweaking

case-study-borrow-checker
Bradlee Speice 2019-01-04 00:08:36 -05:00
parent ab09489d32
commit 5efdcad0d2
1 changed files with 46 additions and 28 deletions

View File

@ -8,17 +8,17 @@ tags: [rust]
There's an alchemy of distilling complex technical topics into articles and videos
that change the way programmers see the tools they interact with on a regular basis.
I knew what a linker was, but there's a staggering complexity to get from
[from `main()` to an executable](https://www.youtube.com/watch?v=dOfucXtyEsU).
I knew what a linker was, but there's a staggering amount of complexity in between
[`main()` and your executable](https://www.youtube.com/watch?v=dOfucXtyEsU).
Rust programmers use the [`Box`](https://doc.rust-lang.org/stable/std/boxed/struct.Box.html)
type all the time, but there's a rich history of the Rust language itself wrapped up in
[how special it is](https://manishearth.github.io/blog/2017/01/10/rust-tidbits-box-is-special/).
In a similar vein, I want you to look at code and understand memory;
the complex choreography of processor, operating system, and program that frees you
In a similar vein, I want you to look at code and understand how memory is used;
the complex choreography of operating system, compiler, and program that frees you
to focus on functionality far-flung from frivolous book-keeping. The Rust compiler relieves
a great deal of the cognitive burden associated with memory management, but let's make time
to explore what's going on under the hood.
a great deal of the cognitive burden associated with memory management, but we're going
to step into its world for a while.
Let's learn a bit about memory in Rust.
@ -30,16 +30,15 @@ section for easy citation in the future. To that end, a table of contents is pro
to assist in easy navigation:
- [Foreword](#foreword)
- [Non-Heap Memory Types](#non-heap-memory-types)
- [Piling On - Rust and the Heap](#piling-on-rust-and-the-heap)
- [Stacking Up: Non-Heap Memory Types](#non-heap-memory-types)
- [Piling On: Rust and the Heap](#piling-on-rust-and-the-heap)
- [Compiler Optimizations Make Everything Complicated](#compiler-optimizations-make-everything-complicated)
- Summary: When Does Rust Allocate?
- [Appendix and Further Reading](#appendix-and-further-reading)
# Foreword
There's a simple way to guarantee you never need to know the content
of this article:
There's a simple checklist to see if you can skip over reading this article. You must:
1. Only write `#![no_std]` crates
2. Never use `unsafe`
@ -47,35 +46,54 @@ of this article:
For some uses of Rust, typically embedded devices, these constraints make sense.
They're working with very limited memory, and the program binary size itself may
affect the memory available! There's no operating system able to manage the heap,
but that's not an issue because your program is likely the only one running.
The [embedonomicon] is ever in mind, and you just might interact with extra
peripherals by reading and writing to exact memory addresses.
significantly affect what's available! There's no operating system able to manage
this "virtual memory" junk, but that's not an issue because there's only one
running application. The [embedonomicon] is ever in mind, and interacting with the
"real world" through extra peripherals is accomplished by reading and writing to
exact memory addresses.
Most Rust programs find these requirements overly burdensome though. C++ developers
would struggle without access to [`std::vector`](https://en.cppreference.com/w/cpp/container/vector),
and Rust developers would struggle without [`std::vec`](https://doc.rust-lang.org/std/vec/struct.Vec.html).
But in this scenario, `std::vec` is actually part of the
[`alloc` crate](https://doc.rust-lang.org/alloc/vec/struct.Vec.html), and thus off-limits.
Or how would you use trait objects? Rust's monomorphization still works, but there's no
would struggle without access to [`std::vector`](https://en.cppreference.com/w/cpp/container/vector)
(except those hardcore no-STL guys), and Rust developers would struggle without
[`std::vec`](https://doc.rust-lang.org/std/vec/struct.Vec.html). But in this scenario,
`std::vec` is actually part of the [`alloc` crate](https://doc.rust-lang.org/alloc/vec/struct.Vec.html),
and thus off-limits (because the `alloc` crate requires `#![feature(alloc)]`).
Or how would you use trait objects? There's no
[`Box<dyn Trait>`](https://doc.rust-lang.org/alloc/boxed/struct.Box.html)
available to use for dynamic dispatch.
Given a target audience of "basically every Rust developer," let's talk about
some of the details you don't normally have to worry about. This article will focus
on "safe" Rust only; `unsafe` mode allows you to make use of platform-specific
allocation APIs (think [libc] and [winapi] implementations of [malloc]) that
we'll ignore for the time being. We'll also assume a "debug" build of libraries
and applications (what you get with `cargo run` and `cargo test`) and address
"release" mode at the end (`cargo run --release` and `cargo test --release`).
Whether writing code for embedded devices or not, the important thing in both situations
is how much you know *before your application starts* about what your memory usage looks like.
In the embedded device example, there's a small, fixed amount of memory you can possibly use.
In a browser, however, you have no idea how large [google.com's home page] is until you start
trying to download it. The compiler uses this information (or lack thereof) to optimize
how memory is used; put simply, your code runs faster when the compiler can guarantee exactly
how much memory your program needs while it's running. This post is all about understanding
the optimization tricks the compiler uses, and how you can help the compiler and make
your programs more efficient.
Finally, a caveat: while the details are unlikely to change, the Rust docs
Now let's address some conditions and caveats before going much further.
This article will focus on "safe" Rust only; `unsafe` mode allows you
to make use of platform-specific allocation API's (think the [libc] and [winapi]
implementations of [malloc]) that we'll ignore. We'll also assume a "debug"
build of libraries and applications (what you get with `cargo run` and `cargo test`)
and address (hehe) "release" mode at the end (`cargo run --release` and `cargo test --release`).
Finally, while the details are unlikely to change, the Rust docs
include a warning worth repeating here:
> Rust does not currently have a rigorously and formally defined memory model.
> - the [Rust docs](https://doc.rust-lang.org/std/ptr/fn.read_volatile.html)
# Non-Heap Memory Types
# Stacking Up: Non-Heap Memory Types
Languages like Java and Python do an amazing job of simplifying the memory model
needed for programmers. You can essentially treat
Most of the reason this post was written is because I
Everyone's agreed that [compilers](https://www.youtube.com/watch?v=bSkpMdDe4g4) are
[smart](https://www.youtube.com/watch?v=nAbCKa0FzjQ), and Rust is no exception.
Example: Why doesn't `Vec::new()` go to the allocator?