Home

Awesome

Echidna: A Fast Smart Contract Fuzzer <a href="https://raw.githubusercontent.com/crytic/echidna/master/echidna.png"><img src="https://raw.githubusercontent.com/crytic/echidna/master/echidna.png" width="75"/></a>

Build Status

Echidna is a weird creature that eats bugs and is highly electrosensitive (with apologies to Jacob Stanley)

More seriously, Echidna is a Haskell program designed for fuzzing/property-based testing of Ethereum smart contracts. It uses sophisticated grammar-based fuzzing campaigns based on a contract ABI to falsify user-defined predicates or Solidity assertions. We designed Echidna with modularity in mind, so it can be easily extended to include new mutations or test specific contracts in specific cases.

Features

.. and a beautiful high-resolution handcrafted logo.

<a href="https://i.imgur.com/saFWti4.png"><img src="https://i.imgur.com/saFWti4.png" width="650"/></a>

Usage

Executing the test runner

The core Echidna functionality is an executable called echidna, which takes a contract and a list of invariants (properties that should always remain true) as input. For each invariant, it generates random sequences of calls to the contract and checks if the invariant holds. If it can find some way to falsify the invariant, it prints the call sequence that does so. If it can't, you have some assurance the contract is safe.

Writing invariants

Invariants are expressed as Solidity functions with names that begin with echidna_, have no arguments, and return a boolean. For example, if you have some balance variable that should never go below 20, you can write an extra function in your contract like this one:

function echidna_check_balance() public returns (bool) {
    return(balance >= 20);
}

To check these invariants, run:

$ echidna myContract.sol

An example contract with tests can be found tests/solidity/basic/flags.sol. To run it, you should execute:

$ echidna tests/solidity/basic/flags.sol

Echidna should find a call sequence that falsifies echidna_sometimesfalse and should be unable to find a falsifying input for echidna_alwaystrue.

Collecting and visualizing coverage

After finishing a campaign, Echidna can save a coverage maximizing corpus in a special directory specified with the corpusDir config option. This directory will contain two entries: (1) a directory named coverage with JSON files that can be replayed by Echidna and (2) a plain-text file named covered.txt, a copy of the source code with coverage annotations.

If you run tests/solidity/basic/flags.sol example, Echidna will save a few files serialized transactions in the coverage directory and a covered.$(date +%s).txt file with the following lines:

*r  |  function set0(int val) public returns (bool){
*   |    if (val % 100 == 0)
*   |      flag0 = false;
  }

*r  |  function set1(int val) public returns (bool){
*   |    if (val % 10 == 0 && !flag0)
*   |      flag1 = false;
  }

Our tool signals each execution trace in the corpus with the following "line marker":

Support for smart contract build systems

Echidna can test contracts compiled with different smart contract build systems, including Truffle or hardhat using crytic-compile. To invoke echidna with the current compilation framework, use echidna ..

On top of that, Echidna supports two modes of testing complex contracts. Firstly, one can describe an initialization procedure with Truffle and Etheno and use that as the base state for Echidna. Secondly, Echidna can call into any contract with a known ABI by passing in the corresponding Solidity source in the CLI. Use allContracts: true in your config to turn this on.

Crash course on Echidna

Our Building Secure Smart Contracts repository contains a crash course on Echidna, including examples, lessons and exercises.

Using Echidna in a GitHub Actions workflow

There is an Echidna action which can be used to run echidna as part of a GitHub Actions workflow. Please refer to the crytic/echidna-action repository for usage instructions and examples.

Configuration options

Echidna's CLI can be used to choose the contract to test and load a configuration file.

$ echidna contract.sol --contract TEST --config config.yaml

The configuration file allows users to choose EVM and test generation parameters. An example of a complete and annotated config file with the default options can be found at tests/solidity/basic/default.yaml. More detailed documentation on the configuration options is available in our wiki.

Echidna supports three different output drivers. There is the default text driver, a json driver, and a none driver, which should suppress all stdout output. The JSON driver reports the overall campaign as follows.

Campaign = {
  "success"      : bool,
  "error"        : string?,
  "tests"        : [Test],
  "seed"         : number,
  "coverage"     : Coverage,
  "gas_info"     : [GasInfo]
}
Test = {
  "contract"     : string,
  "name"         : string,
  "status"       : string,
  "error"        : string?,
  "testType"     : string,
  "transactions" : [Transaction]?
}
Transaction = {
  "contract"     : string,
  "function"     : string,
  "arguments"    : [string]?,
  "gas"          : number,
  "gasprice"     : number
}

Coverage is a dict describing certain coverage-increasing calls. Each GasInfo entry is a tuple that describes how maximal gas usage was achieved, and is also not too important. These interfaces are subject to change to be slightly more user-friendly at a later date. testType will either be property or assertion, and status always takes on either fuzzing, shrinking, solved, passed, or error.

Debugging Performance Problems

One way to diagnose Echidna's performance issues is to run echidna with profiling on. To run Echidna with basic profiling, add +RTS -p -s to your original echidna command:

$ nix develop # alternatively nix-shell
$ cabal --enable-profiling run echidna -- ... +RTS -p -s
$ less echidna.prof

This produces a report file (echidna.prof), that shows which functions take up the most CPU and memory usage.

If the basic profiling doesn't help, you can use more advanced profiling techniques.

Common causes for performance issues that we observed:

Checking for these is a good place to start. If you suspect some comuptation is too lazy and leaks memory, you can use force from Control.DeepSeq to make sure it gets evaluated.

Limitations and known issues

EVM emulation and testing are hard. Echidna has some limitations in the latest release. Some of these are inherited from hevm while some are results from design/performance decisions or simply bugs in our code. We list them here including their corresponding issue and the status ("wont fix", "on hold", "in review", "fixed"). Issues that are "fixed" are expected to be included in the next Echidna release.

DescriptionIssueStatus
Vyper support is limited#652wont fix
Limited library support for testing#651wont fix

Installation

Precompiled binaries

Before starting, make sure Slither is installed (pip3 install slither-analyzer --user). If you want to quickly test Echidna in Linux or MacOS, we provide statically linked Linux binaries built on Ubuntu and mostly static MacOS binaries on our releases page. You can also grab the same type of binaries from our CI pipeline, just click the commit to find binaries for Linux or MacOS.

Homebrew (macOS / Linux)

If you have Homebrew installed on your Mac or Linux machine, you can install Echidna and all of its dependencies (Slither, crytic-compile) by running brew install echidna.

You can also compile and install the latest master branch code by running brew install --HEAD echidna

You can get further information in the echidna Homebrew Formula page. The formula itself is maintained as part of the homebrew-core repository

Docker container

If you prefer to use a pre-built Docker container, check out our docker package, which is auto-built via GitHub Actions. The echidna container is based on ubuntu:focal and it is meant to be a small yet flexible enough image to use Echidna on. It provides a pre-built version of echidna, as well as slither, crytic-compile, solc-select and nvm under 200 MB.

Note that the container images currently only build on x86 systems. Running them on ARM devices, such as Mac M1 systems, is not recommended due to the performance loss incurred by the CPU emulation.

Different tags are available for the Docker container image:

TagBuild in tag
vx.y.zBuild corresponding to release vx.y.z
latestLatest Echidna tagged release.
edgeMost recent commit on the default branch.
testing-fooTesting build based on the foo branch.

To run the container with the latest Echidna version interactively, you can use something like the following command. It will map the current directory as /src inside the container, and give you a shell where you can use echidna:

$ docker run --rm -it -v `pwd`:/src ghcr.io/crytic/echidna/echidna

Otherwise, if you want to locally build the latest version of Echidna, we recommend using Docker. From within a clone of this repository, run the following command to build the Docker container image:

$ docker build -t echidna -f docker/Dockerfile --target final-ubuntu .

Then, you can run the echidna image locally. For example, to install solc 0.5.7 and check tests/solidity/basic/flags.sol, you can run:

$ docker run -it -v `pwd`:/src echidna bash -c "solc-select install 0.5.7 && solc-select use 0.5.7 && echidna /src/tests/solidity/basic/flags.sol"

Building using Stack

If you'd prefer to build from source, use Stack. stack install should build and compile echidna in ~/.local/bin. You will need to link against libreadline and libsecp256k1 (built with recovery enabled), which should be installed with the package manager of your choosing. You also need to install the latest release of libff. Refer to our CI tests for guidance.

Some Linux distributions do not ship static libraries for certain things that Haskell needs, e.g. Arch Linux, which will cause stack build to fail with linking errors because we use the -static flag. In that case, use --flag echidna:-static to produce a dynamically linked binary.

If you're getting errors building related to linking, try tinkering with --extra-include-dirs and --extra-lib-dirs.

Building using Nix (works natively on Apple M1 systems)

Nix users can install the latest Echidna with:

$ nix-env -i -f https://github.com/crytic/echidna/tarball/master

With flakes enabled, you can run Echidna straight from this repo:

$ nix run github:crytic/echidna # master
$ nix run github:crytic/echidna/v2.1.1 # specific ref (tag/branch/commit)

To build a standalone release for non-Nix macOS systems, the following will bundle Echidna and all linked dylibs:

$ nix build .#echidna-bundle

Nix will automatically install all the dependencies required for development including crytic-compile and solc. A quick way to start developing Echidna:

$ git clone https://github.com/crytic/echidna
$ cd echidna
$ nix develop # alternatively nix-shell
[nix-shell]$ cabal run echidna
[nix-shell]$ cabal run tests
[nix-shell]$ cabal new-repl

Public use of Echidna

Property testing suites

This is a partial list of smart contracts projects that use Echidna for testing:

Trophies

The following security vulnerabilities were found by Echidna. If you found a security vulnerability using our tool, please submit a PR with the relevant information.

ProjectVulnerabilityDate
0x ProtocolIf an order cannot be filled, then it cannot be canceledOct 2019
0x ProtocolIf an order can be partially filled with zero, then it can be partially filled with one tokenOct 2019
0x ProtocolThe cobbdouglas function does not revert when valid input parameters are usedOct 2019
Balancer CoreAn attacker cannot steal assets from a public poolJan 2020
Balancer CoreAn attacker cannot generate free pool tokens with joinPoolJan 2020
Balancer CoreCalling joinPool-exitPool does not lead to free pool tokensJan 2020
Balancer CoreCalling exitswapExternAmountOut does not lead to free assetsJan 2020
Liquity DollarClosing troves require to hold the full amount of LUSD mintedDec 2020
Liquity DollarTroves can be improperly removedDec 2020
Liquity DollarInitial redeem can revert unexpectedlyDec 2020
Liquity DollarRedeem without redemptions might still return successDec 2020
Origin DollarUsers are allowed to transfer more tokens that they haveNov 2020
Origin DollarUser balances can be larger than total supplyNov 2020
Yield ProtocolArithmetic computation for buying and selling tokens is impreciseAug 2020

Research

We can also use Echidna to reproduce research examples from smart contract fuzzing papers to show how quickly it can find the solution. All of these can be solved, in a few seconds to one or two minutes on a laptop computer.

SourceCode
Using automatic analysis tools with MakerDAO contractsSimpleDSChief
Integer precision bug in Sigma PrimeVerifyFunWithNumbers
Learning to Fuzz from Symbolic Execution with Application to Smart ContractsCrowdsale
Harvey: A Greybox Fuzzer for Smart ContractsFoo, Baz

Academic Publications

Paper TitleVenuePublication Date
echidna-parade: Diverse multicore smart contract fuzzingISSTA 2021July 2021
Echidna: Effective, usable, and fast fuzzing for smart contractsISSTA 2020July 2020
Echidna: A Practical Smart Contract FuzzerFC 2020Feb 2020

If you are using Echidna for academic work, consider applying to the Crytic $10k Research Prize.

Getting help

Feel free to stop by our #ethereum slack channel in Empire Hacking for help using or extending Echidna.

License

Echidna is licensed and distributed under the AGPLv3 license.