Skip to content

PLSysSec/cargo-scan

Repository files navigation

Cargo Scan

Cargo Scan is a tool for auditing Rust crates.

⚠️ Cargo Scan is under active development. Some interfaces may be subject to change.

Installation

  1. Clone this repository.
  2. Run rustup update to ensure you have the latest version of Rust (or install it via the official website).
  3. Run cargo build.

Known working builds:

  • August 22, 2024 on Mac OS (Sonoma) using Rust 1.80.1.

  • August 22, 2024 on Linux (Arch Linux) using Rust 1.80.1.

Quick-start

To use Cargo Scan you first need a Rust crate somewhere on your system (any Rust code with a Cargo.toml file). There are three ways to run Cargo Scan:

  1. Using the basic command-line interface (easiest: cargo run --bin scan <path to crate>);

  2. Using interactive auditing mode (cargo run --bin audit <path to crate> -i); or

  3. Using the VSCode extension.

Instructions for these are included below.

Method 1: Using the basic command-line interface

To scan a crate, you run the binary (from this repository), and provide it a path to the crate:

cargo run --bin scan <path to crate>

This is the simplest available output, which is returned in CSV format as a list of side effects, one per line. The beginning of the line gives the crate name, the function body and callee that contains the effect, and the effect type or pattern that it matches. The last four items on each line give the directory, file, line, and column where the effect occurs. Example output is as follows:

$ cargo run --bin scan data/packages/rand
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.14s
     Running `target/debug/scan data/packages/rand`
crate, fn_decl, callee, effect, dir, file, line, col
rand, rand::distributions::other::<Standard as Distribution<char>>::sample, core::char::from_u32_unchecked, [UnsafeCall], data/packages/rand/src/distributions, other.rs, 94, 17
rand, rand::distributions::other::<Alphanumeric as DistString>::append_string, alloc::string::String::as_mut_vec, [UnsafeCall], data/packages/rand/src/distributions, other.rs, 136, 27
...
rand, rand::distributions::uniform::<Uniform<char> as super::DistString>::append_string, core::char::methods::len_utf8, [FnPtrCreation], data/packages/rand/src/distributions, uniform.rs, 860, 50

num_effects, total, loc_lb, loc_ub, macros, loc_lb, loc_ub, conditional_code, loc_lb, loc_ub, skipped_calls, loc_lb, loc_ub, skipped_fn_ptrs, loc_lb, loc_ub, skipped_other, loc_lb, loc_ub, unsafe_trait, loc_lb, loc_ub, unsafe_impl, loc_lb, loc_ub, pub_fns, pub_fns_with_effects, pub_total_effects, audited_fns, audited_loc
12, 25, 8708, 8708, 127, 1117, 1185, 21, 2410, 2410, 0, 0, 0, 0, 0, 0, 25, 3, 25, 0, 0, 0, 0, 0, 0, 53, 23, 52, 58, 540

If you don't want the last two lines (totals at the bottom), suppress them with the -s option.

For additional usage options, run help:

cargo run --bin scan -- --help

What's a side effect?

See Wikipedia. Basically, it represents some behavior of a function that might or might not be dangerous (e.g., operating system calls, memory unsafety, or filesystem/network access). These are behaviors that you may want to look into further when auditing a crate.

If you don't have a crate on your system

You can run one of our provided example crates in data/test-packages:

cargo run --bin scan data/test-packages/permissions-ex

Method 2: Interactive mode

Interactive mode is a variant of the command-line interface that allows you to mark audits as safe or unsafe as you go through them. Interactive mode works by maintaining an audit file of the effects you have marked so far, which are maintained dynamically as you progress through an audit.

cargo run --bin audit <path to crate> -i

At each effect, the audit will ask you whether you want to mark it safe or unsafe, skip it or provide more context, or mark it caller-checked, which means that the caller of the function is responsible for ensuring safety. The interactive mode looks like this:

help[Audit location]:
   ┌─ data/packages/rand/src/distributions/other.rs:94:1
   │
79 │   ╭       fn sample<R: Rng + ?Sized>(&self, rng: &mut R) -> char {
80 │   │           // A valid `char` is either in the interval `[0, 0xD800)` or
   │   ╰──'
   ·   │
90 │     ╭         let mut n = range.sample(rng);
91 │     │         if n <= 0xDFFF {
92 │     │             n -= GAP_SIZE;
93 │     │         }
94 │ ╭   │         unsafe { char::from_u32_unchecked(n) }
95 │ │   │     }
   │ ╰───│^ unsafe call: core::char::from_u32_unchecked
96 │     │ }
   │     ╰'

? Select how to mark this effect:
  (s)afe, (u)nsafe, (c)aller checked, (e)xpand context, ask me (l)ater, e(x)it tool

If the command is run a second time, it continues the existing audit. To review the audit, use -r.

For additional usage options, run help:

cargo run --bin scan -- --help

Method 3: Using the VSCode extension

Installing the extension in VSCode

  1. Build the extension by running make in the editor-plugin directory. This will produce a .vsix file in that location.
  2. Launch VSCode and install the extension from the .vsix file:
    • Press Ctrl + Shift + P (Windows/Linux) or Cmd + Shift + P (macOS) to open the Command Palette.
    • Type "Extensions: Install from VSIX..." and select it.
    • Navigate to the folder where the .vsix file is located and open it.

The extension is automatically activated when a Rust package is detected in the workspace.

Performing an audit in VSCode

There are two types of audits you can perform with Cargo Scan in VSCode

  1. Single Crate Audit that scans only the currently open package for effects.
  • To run a single crate audit type cargo-scan: Audit Crate in the Command Palette.
  1. Chain Audit that performs a full audit and also scans the transitive dependencies of the open package.
  • To run a chain audit, create the chain by typing cargo-scan: Create Chain in the Command Palette and then perform the actual audit by running the command cargo-scan: Audit Chain.

The set of effects that Cargo Scan identifies are shown in the Effects view of the Explorer side bar.

Other usage

Running the unit tests

  • Run cargo test to run Rust unit tests

  • Run make test to re-run the tool on all our test packages, whose results are in data/results and placed under version control to check for any regressions.

Running an experiment

You can also run ./scripts/scan.py -h to see options for running an experiment; this is useful for running a scan on a large list of crates, e.g. the top 100 crates on crates.io or your own provided list. Alternatively, see Makefile for some pre-defined experiments to run, such as make top10.

Running in tandem with other tools

Cargo Scan can also be used in tandem with other Rust supply chain auditing tools, such as cargo vet. We are interested in exploring other use cases for integration; if you think Cargo Scan would be useful for your project, let us know!

Additional information

Repository structure

Most of the source code of Cargo Scan can be found in src/, with miscellaneous data and scripts in data/ and scripts/. The VS Code extension is developed in lang_server and editor-plugin.

Current maintainers

Cargo Scan is open source. It is currently maintained by Caleb Stanford and Lydia Zoghbi. It includes past contributions from other authors at UC Davis PL and UC San Diego PLSysSec. For copyright information, see the LICENSE.

Issues

We welcome bug reports, suggestions, and pull requests! Please raise an issue or send us an email about your use case.

Citing this repository

If you want to cite this repository, please consider citing our technical report. The current draft (in submission) is available below:

  • Auditing Rust Crates Effectively. Lydia Zoghbi, David Thien, Ranjit Jhala, Deian Stefan, and Caleb Stanford. Preprint (2024). Draft Link

You may also cite the repository directly:

  • Cargo Scan: a tool for auditing Rust crates. Lydia Zoghbi, David Thien, Ranjit Jhala, Deian Stefan, and Caleb Stanford. GitHub repository (2024). GitHub Link