In this repository, we provide a model of the EVM in K.
bash <(curl https://kframework.org/install)
: install kup package manager.kup install kevm
: install KEVM.kup list kevm
: list available KEVM versions.kup update kevm
: update to latest KEVM version.
NOTE: The first run will take longer to fetch all the libraries and compile sources. (30m to 1h)
These may be useful for learning KEVM and K (newest to oldest):
- K, KEVM and Foundry Integration overview
- Jello Paper, a nice presentation of this repository.
- 20 minute tour of the semantics at 2017 Devcon3.
- KEVM 1.0 technical report, especially sections 3 and 5.
- KEVM Paper at CSF'18/FLoC.
To get support for KEVM, please join our Discord Channel.
If you want to start proving with KEVM, refer to VERIFICATION.md.
The following files constitute the KEVM semantics:
- network.md provides the status codes reported to an Ethereum client on execution exceptions.
- json-rpc.md is an implementation of JSON RPC in K.
- evm-types.md provides the (functional) data of EVM (256-bit words, wordstacks, etc...).
- serialization.md provides helpers for parsing and unparsing data (hex strings, recursive-length prefix, Merkle trees, etc.).
- evm.md is the main KEVM semantics, containing EVM’s configuration and transition rules.
- gas.md contains all information relevant to gas.
- schedule.md contains all information relevant to EVM schedules.
These additional files extend the semantics to make the repository more useful:
- buf.md defines the
#buf
byte-buffer abstraction for use during symbolic execution. - abi.md defines the Contract ABI Specification for use in proofs and easy contract/function specification.
- hashed-locations.md defines the
#hashedLocation
abstraction used to specify Solidity-generated storage layouts. - edsl.md combines the previous three abstractions for ease-of-use.
These files are used for testing the semantics itself:
- state-utils.md provides functionality for EVM initialization, setup, and querying.
- driver.md is an execution harness for KEVM, providing a simple language for describing tests/programs.
There are two backends of K available: LLVM for concrete execution and Haskell for symbolic execution.
This repository generates the build-products for each backend in $XDG_CACHE_HOME/evm-semantics-<digest>
.
Run install-build-deps
to install the required OS-supplied dependencies.
There are some additional notes for specific systems:
Ubuntu:
- The script works for ≥ 22.04.
- On Ubuntu < 18.04, you'll need to skip
libsecp256k1-dev
and instead build it from source (via ourMakefile
) usingmake libsecp256k1
.
Arch:
- No issues known.
MacOS:
-
After installing the Command Line Tools, Homebrew, and getting the blockchain plugin, run:
-
NOTE: It is recommended to use the homebrew version of
flex
and XCode. -
If you are building on an Apple Silicon machine, ensure that your
PATH
is set up correctly before runningmake deps
ormake k-deps
. You can do so usingdirenv
by copyingmacos-envrc
to.envrc
, then runningdirenv allow
. -
If the build on macOS still fails, you can also try adding the following lines to the top of your
Makefile
underUNAME_S
:ifeq ($(UNAME_S), Darwin) SHELL := /usr/local/bin/bash PATH := $(pwd)/.build/usr/bin:$(PATH) endif
- Haskell Stack.
Note that the version of the
stack
tool provided by your package manager might not be recent enough. Please follow installation instructions from the Haskell Stack website linked above.
To upgrade stack
(if needed):
stack upgrade
export PATH=$HOME/.local/bin:$PATH
You need to install the K Framework on your system, see the instructions there. The fastest way is via the kup package manager, with which you can do to get the correct version of K:
kup install k.openssl --version v$(cat deps/k_release)
You can also drop into a single development shell with the correct version of K on path by doing:
kup shell k.openssl --version v$(cat deps/k_release)
First you need to set up a virtual environment using Poetry with the prerequisites python 3.8.*
, pip >= 20.0.2
, poetry >= 1.3.2
:
make poetry
You also need to get the blockchain plugin submodule and install it.
git submodule update --init --recursive
poetry -C kevm-pyk run kdist --verbose build evm-semantics.plugin
To change the default compiler:
CXX=clang++-15 poetry -C kevm-pyk run kdist --verbose build evm-semantics.plugin
On Apple silicon:
APPLE_SILICON=true poetry -C kevm-pyk run kdist --verbose build evm-semantics.plugin
Finally, you can build the semantics.
poetry -C kevm-pyk run kdist --verbose build -j6
You can build specific targets using options evm-semantics.{llvm,kllvm,kllvm-runtime,haskell,haskell-standalone,plugin}
, e.g.:
poetry -C kevm-pyk run kdist build -j2 evm-semantics.llvm evm-semantics.haskell
Targets can be cleaned with
poetry -C kevm-pyk run kdist clean
For more information, refer to kdist --help
and the dist.py module.
To execute tests from the Ethereum Test Set, the submodule needs to be fetched first.
git submodule update --init --recursive -- tests/ethereum-tests
The tests are run using the supplied Makefile
.
The following subsume all other tests:
make test
: All of the quick tests.make test-all
: All of the quick and slow tests.
These are the individual test-suites (all of these can be suffixed with -all
to also run slow tests):
make test-vm
: VMTests from the Ethereum Test Set.make test-bchain
: Subset of BlockchainTests from the Ethereum Test Set.make test-proof
: Proofs from the Verified Smart Contracts.make test-interactive
: Tests of thekevm
command.
All these targets call pytest
under the hood. You can pass additional arguments to the call by appending them to variable PYTEST_ARGS
. E.g. run
make test-vm PYTEST_ARGS+=-vv
to execute VMTests with increased verbosity, and
make test-vm PYTEST_ARGS+=-n0
to execute them on a single worker.
Files produced by test runs, e.g. kompiled definition and logs, can be found in /tmp/pytest-of-<user>/
.
If built from the source, the kevm-pyk
executable will be installed in a virtual environment handled by Poetry.
You can call kevm-pyk --help
to get a quick summary of how to use the script.
Run the file tests/ethereum-tests/BlockchainTests/GeneralStateTests/VMTests/vmArithmeticTest/add0.json
:
poetry -C kevm-pyk run kevm-pyk run tests/ethereum-tests/BlockchainTests/GeneralStateTests/VMTests/vmArithmeticTest/add0.json --schedule DEFAULT --mode VMTESTS
To enable the debug symbols for the llvm backend, build using this command:
poetry -C kevm-pyk run kdist build evm-semantics.llvm --arg enable-llvm-debug=true
To debug a conformance test, add the --debugger
flag to the command:
poetry -C kevm-pyk run kevm-pyk run tests/ethereum-tests/BlockchainTests/GeneralStateTests/stExample/add11.json --target llvm --mode NORMAL --schedule SHANGHAI --chainid 1 --debugger
Always have your build up-to-date.
- If using the kup package manager, run
kup install kevm --version .
to install the local version. - If building from source:
make poetry
needs to be re-run if you touch any of thekevm-pyk
code.poetry -C kevm-pyk run kdist build <target> --force
needs to be re-run if you touch any of this repos files.poetry -C kevm-pyk run kdist clean
is a safe way to remove the target directory
We now support building KEVM using nix flakes.
To set up nix flakes you will need to be on nix
2.4 or higher and follow the instructions here.
For example, if you are on a standard Linux distribution, such as Ubuntu, first install nix
and then enable flakes by editing either ~/.config/nix/nix.conf
or /etc/nix/nix.conf
and adding:
experimental-features = nix-command flakes
This is needed to expose the Nix 2.0 CLI and flakes support that are hidden behind feature-flags.
By default, Nix will build the project and its transitive dependencies from source, which can take up to an hour. We recommend setting up the binary cache to speed up the build process significantly.
To build KEVM, run:
nix build .#kevm
This will build all of KEVM and K and put a link to the resulting binaries in the result/
folder.
NOTE: Mac users, especially those running M1/M2 Macs may find nix segfaulting on occasion. If this happens, try running the nix command like this: GC_DONT_GC=1 nix build .
If you want to temporarily add the kevm
binary to the current shell, run
nix shell .#kevm
Nix can also be used to quickly profile different versions of the Haskell backend. Simply run:
nix build github:runtimeverification/evm-semantics#profile \
--override-input k-framework/haskell-backend github:runtimeverification/haskell-backend/<HASH> \
-o prof-<HASH>
replacing <HASH>
with the commit you want to run profiling against.
If you want to profile against a working version of the Haskell backend repository, simply cd
into the root of the repo and run:
nix build github:runtimeverification/evm-semantics#profile \
--override-input k-framework/haskell-backend $(pwd) \
-o prof-my-feature
To compare profiles, you can use:
nix run github:runtimeverification/evm-semantics#compare-profiles -- prof-my-feature prof-<HASH>
This will produce a nice table with the times for both versions of the haskell-backend.
Note that #profile
pre-pends the output of kore-exec --version
to the profile run, which is then used as a tag by the #compare-profiles
script.
Therefore, any profiled local checkout of the haskell-backend will report as dirty-ghc8107
in the resulting table.
This repository can build two pieces of documentation for you, the Jello Paper and the 2017 Devcon3 presentation.
For the presentations in the media
directory, you'll need pdflatex
, commonly provided with texlive-full
, and pandoc
.
sudo apt install texlive-full pandoc
To build all the PDFs (presentations and reports) available in the media/
directory, use:
make media
- EVM Yellowpaper: Original specification of EVM.
- LEM Semantics of EVM
- EVM Opcode Interactive Reference
- Solidity ABI Encoding
For more information about the K Framework, refer to these sources:
- The K Tutorial
- Semantics-Based Program Verifiers for All Languages
- Reachability Logic Resources
- Matching Logic Resources
- Logical Frameworks: Discussion of logical frameworks.