leaktracer

---

• leaktracer
  • Introduction
    • Why do I need this?
  • Usage
    • Cargo.toml
    • Setup
    • Accessing the stats
  • Example
  • Debug only
  • Support the developer
  • Changelog
  • License

---

Introduction

A Rust allocator to trace memory allocations in Rust programs, by intercepting the allocations.

The library provides the LeaktracerAllocator, which is an allocator that for each allocation, it stores the memory allocated and the allocation count for each function that allocated memory.

It's extremely easy to setup and it was designed to have something really plug-and-play.

Why do I need this?

You may ask why you would need this library in a language like Rust, which is known for its memory safety. The answer is that even in Rust, memory leaks can occur, especially when storing data in maps or vectors along time without cleaning them up.

Sometimes it can happen that you don't know where the huge memory usage is coming from, because either the cleanup method is not working, or you forgot to clean up the data. In complex applications, this can be a nightmare to debug, so that's why I created this library.

Usage

Cargo.toml

Add the following to your Cargo.toml:

[dependencies]
leaktracer = "0.1"

Setup

Then, in your main.rs you need to set the allocator to LeaktracerAllocator and initialize the symbol table:

use leaktracer::LeaktracerAllocator;

#[global_allocator]
static ALLOCATOR: LeaktracerAllocator = LeaktracerAllocator::init();

and then at the beginning of your main function, initialize the symbol table:

leaktracer::init_symbol_table(&["my_crate_name"]);

The init_symbol_table function takes a slice of strings, which are the names of the crates you want to trace. This is useful if you have multiple crates in your project and you want to trace only specific ones.

Why is this necessary? Because the library use the backtrace to get the current call stack, but unfortunately the backtrace, is quite polluted by other non-relevant calls (such as std::alloc, std::vec, etc.), so you need to specify which crates you want to trace.

Accessing the stats

Of course, once initialized you want to access the stats, to see how many allocations were made, and where they were made.

You can do this by accessing the symbol_table like this:

leaktracer::with_symbol_table(|table| {
    for (name, symbol) in table.iter() {
        println!(
            "Symbol: {name}, Allocated: {}, Count: {}",
            symbol.allocated(),
            symbol.count()
        );
    }
})?;

You can also access the full amount of memory allocated and the total count of allocations by using the LeaktracerAllocator methods:

println!(
    "Allocated {} bytes",
    ALLOCATOR.allocated()
);

Example

You can find an example in the examples folder at examples/tracing.rs.

You can run the example with:

cargo run --example tracing

Debug only

The LeaktracerAllocator is meant to be used in debug mode only, as it uses the backtrace crate to get the call stack, which is not available in release mode and it's extremely slow and expensive. Therefore, it is not possible to use it in release mode.

Support the developer

If you like leaktracer, please consider a little donation 🥳

---

Changelog

View Changelog here

---

License

Licensed under MIT license (SEE LICENSE or http://opensource.org/licenses/MIT)