This is the code repository for our Usenix Security 2020 paper EthBMC.
The introduction of smart contracts has significantly advanced the state-of-the-art in cryptocurrencies. Smart contracts are programs who live on the blockchain, governing the flow of money. However, the promise of monetary gain has attracted miscreants, resulting in spectacular hacks which resulted in the loss of millions worth of currency. In response, several powerful static analysis tools were developed to address these problems. We surveyed eight recently proposed static analyzers for Ethereum smart contracts and found that none of them captures all relevant features of the Ethereum ecosystem. For example, we discovered that a precise memory model is missing and inter-contract analysis is only partially supported.
Based on these insights, we present the design and implementation of, a bounded model checker based on symbolic execution which provides a precise model of the Ethereum network. We demonstrate its capabilities in a series of experiments. First, we compare against the eight aforementioned tools, showing that even relatively simple toy examples can obstruct other analyzers. Further proving that precise modeling is indispensable, we leverage ETHBmc capabilities for automatic vulnerability scanning. We perform a large-scale analysis of roughly 2.2 million accounts currently active on the blockchain and automatically generate 5,905 valid inputs which trigger a vulnerability. From these, 1,989 can destroy a contract at will (so called suicidal contracts) and the rest can be used by an adversary to arbitrarily extract money. Finally, we compare our large-scale analysis against two previous analysis runs, finding significantly more inputs (22.8%) than previous approaches.
To compile and use EthBMC you will need:
- Rust nightly (tested with version 1.44.0)
- evm stand alone executable; make sure it is in your
$PATH(originally with version 1.8.17; also tested with 1.10.4 - available in the Geth & Tools bundle) - We offer support for three different smt solvers (tested with yices2 version 2.6.1. Note we recommend Yices2; see Section 6.5 in the paper):
When using the webservice-version (see esvm/bin):
- Configuration is done using a Rocket.toml
- When using the Service with timeouts the service expects a ethbmc binary in the PATH
- ethbmc: Simple stand alone binary
- ethbmc-service: Webservice which accepts accounts and returns the result as json
- ethbmc-scheduler: Accepts two lists, an account list and a server list, then distributes the accounts between the backing servers running a service instance
cargo build --release
Note when analyzing big contracts you might have to increase Rusts stack size, see here.
EthBMC 1.0.0
EthBMC: A Bounded Model Checker for Smart Contracts
USAGE:
ethbmc [FLAGS] [OPTIONS] <INPUT>
FLAGS:
-x, --acc The input is an Ethereum account address, must be used with parity backend, mainnet only
--concrete-copy Use concrete calldatacopy
-d, --debug-grap Dump debug graph after analysis
--no-optimizations Disable all optimizations
--dump-solver Dump all solver queries to ./queries
-h, --help Prints help information
--json Output json without logging
--list The input is a list of Ethereum account address, writes the result to a csv file in the
output folder
--no-verify Skip verification phase.
-V, --version Prints version information
OPTIONS:
--block <block> The blocknumber to evaluate, defaults to latest block. Note you will need
an archive node for anaylzing blocks which are not the latest.
-c, --call-bound <call_bound> Set bound for calls
--cores <cores> Set the amount of cores the se can use
--ip <ip> The ip of a running node.
-b, --loop-bound <loop_bound> Set bound for loops
-m, --message-bound <message_bound> Set bound for message iteration
--port <port> The port of a running node.
--solver <solver> The SMT solver to use: z3, boolector, yices2 [yices2]
--solver-timeout <solver-timeout> Set solver timeout in milliseconds
ARGS:
<INPUT> Set input file / address
The easiest way to use EthBMC is to create an input yml file (see examples):
./target/release/ethbmc examples/rubixi/rubixi.yml
When you have a running Ethereum node you can also use it to analyze an on-chain account:
./target/release/ethbmc -x --ip ip.to.your.node 0xAccount_Address
When you have an archive node you can also use it to analyze an account at a specific block:
./target/release/ethbmc -x --ip ip.to.your.node --block block_number_to_analyze 0xAccount_Address
Note when executing the parity example (examples/parity) we recommend limiting the loop execution to 1 and use concrete-copy. The bug can still be found without these restrictions, but it takes a long time.
./target/release/ethbmc -b1 --concrete-copy examples/parity/parity.yml
The yaml format allows to easily initialise a multi-account environment offline. Under state you list all the accounts which should be present in the environment. Each account gets its address as a key. Additionally you can set the balance of the accounts, the nonce and the code field. Optionally you can supply storage. These are key-value pairs which get loaded as the initial storage of an account, otherwise it is assumed empty. Additionally you must supply a victim address. This is the account from which the analysis is started. See for example the example for analysing the parity hack:
state:
# Wallet
0xad62f08b3b9f0ecc7251befbeff80c9bb488fe9:
balance: 0x8AC7230489E80000
nonce: 0x0
code: 606060...
storage:
0x0: 0xcafecafecafecafecafecafecafecafecafecafe # "owner"
# Receiver
0xcafecafecafecafecafecafecafecafecafecafe:
balance: 0x0
nonce: 0x0
code: 60606...
victim: 0xad62f08b3b9f0ecc7251befbeff80c9bb488fe9
We initialise two accounts, the wallet account, which holds the stub for forwarding all requests to the library, and the Receiver account, the parity library. We additionally supply some funds to it, to simulate a hack of the wallet, as well as setting the first storage variable (0x0) (the variable holding the address of the library contract) to the second account in the environment.
For the on-chain analysis (Section 6.2) we used a patched parity node to obtain the storage of an account (the code can be found here). You can still use a normal node which supports the web3 API. However, the analysis then does not take storage variables into account (all variables are assumed to be zero). Note to analyze blocks at an earlier time you will need to use a patched archive node.
- Test can only be run one at a time at the moment (cargo test -- --test-threads=1)
- Integration tests take a long time and should allways be run with optimizations (cargo test integra --release -- --ignored)
- Some tests need a deployed parity instance, set PARITY_IP env var to the corresponding ip (PARITY_IP=127.0.0.1 cargo test parity -- --test-threads=1 --ignored)
You can run all test with the supplied shell script (requires archive patched parity node):
PARITY_IP=127.0.0.1 ./run_all_test.sh
When you cite our work please use the following bibtex entry:
@inproceedings {frank2020ethbmc,
title={ETHBMC: A Bounded Model Checker for Smart Contracts},
author={Frank, Joel and Aschermann, Cornelius and Holz, Thorsten},
booktitle = {USENIX Security Symposium (USENIX Security)},
year = {2020},
}
We would like to thank the parity team for open-sourcing their code.