/pyth-client

Client API for on-chain pyth programs

Primary LanguageC++Apache License 2.0Apache-2.0

pyth-client

Pyth oracle program and off-chain client API.

Oracle Program

The Pyth oracle program lives in the program/ directory. It consists of both C and Rust code, but everything can be built and tested using cargo.

Build Instructions

First, make sure you have the solana tool suite installed on your machine. (The build depends on some C makefiles that are in the tool suite.) Make sure you have installed the same solana version that is being used in CI

Then, simply run cargo build to compile the oracle program as a native binary, or cargo build-bpf to build a BPF binary that can be uploaded to the blockchain. This step will produce a program binary target/deploy/pyth_oracle.so.

Testing

Simply run cargo test. This command will run several sets of tests:

  • Unit tests of individual functions
  • Simulated transaction tests against the BPF binary running on a solana simulator
  • Exhaustive / randomized test batteries for core oracle functionality

Rust tests live in the tests/ module of the rust code, and C tests are named something like test_*.c. The C tests are linked into the rust binary so they run as part of cargo test as well (see tests/test_c_code.rs).

You can also run cargo test-bpf, which runs the same tests as cargo test, though it's slightly slower and the UX is worse.

pre-commit hooks

pre-commit is a tool that checks and fixes simple issues (formatting, ...) before each commit. You can install it by following their website. In order to enable checks for this repo run pre-commit install from command-line in the root of this repo.

The checks are also performed in the CI to ensure the code follows consistent formatting. Formatting is only currently enforced in the program/ directory. You might also need to install the nightly toolchain to run the formatting by running rustup toolchain install nightly.

pythd (off-chain client API)

⚠️ pythd is deprecated and has been replaced by pyth-agent. This new client is backward compatible with pythd, but more stable and configurable.

pythd provides exposes a web API for interacting with the on-chain oracle program.

Build Instructions

# depends on openssl
apt install libssl-dev

# depends on libz
apt install zlib1g zlib1g-dev

# depends on libzstd
apt install libzstd-dev

# uses cmake to build
apt install cmake

# default is release build
./scripts/build.sh

You may want to build this repository inside a linux docker container for various reasons, e.g., if building on an M1 mac, or to run BPF binaries. You can run the following command to open a shell in a linux docker container with the pyth-client directory mounted:

export PYTH_REPO=/path/to/host/pyth-client

// Use the latest pythd image from https://hub.docker.com/r/pythfoundation/pyth-client/tags?name=pythd
export IMAGE=...

docker run -it \
  --volume "${HOME}:/home/pyth/home" \
  --volume "${HOME}/.config:/home/pyth/.config" \
  --mount "type=bind,src=${PYTH_REPO},target=/home/pyth/pyth-client" \
  --userns=host \
  --user="$( id -ur ):$( id -gr )" \
  --platform linux/amd64 \
  $IMAGE \
  /bin/bash -l

This command runs a recent pyth-client docker image that already has the necessary dependencies installed. Therefore, once the container is running, all you have to do is run cd pyth-client && ./scripts/build.sh. Note that updates to the pyth-client directory made inside the docker container will be persisted to the host filesystem (which is probably desirable).

Alternatively, you can run ./scripts/start-docker-build-env.sh && cd pyth-client && cargo build --release

Fuzzing

Build a docker image for running fuzz tests:

docker build . --platform linux/amd64 -f docker/fuzz/Dockerfile -t pyth-fuzz

Each fuzz test is invoked via an argument to the fuzz command-line program, and has a corresponding set of test cases in the subdirectory with the same name as the test. You can run these tests using a command like:

docker run -t \
  --platform linux/amd64 \
  -v "$(pwd)"/findings:/home/pyth/pyth-client/findings \
  pyth-fuzz \
  sh -c "./afl/afl-fuzz -i ./pyth-client/pyth/tests/fuzz/add/testcases -o ./pyth-client/findings ./pyth-client/build/fuzz add"

This command will run the add fuzz test on the tests cases in pyth/tests/fuzz/add/testcases, saving any outputs to findings/. Note that findings/ is shared between the host machine and the docker container, so you can inspect any error cases by looking in that subdirectory on the host.

If you find an error case that you want to investigate further, you can run the program on the failing input using something like:

docker run -t \
  --platform linux/amd64 \
  -v "$(pwd)"/findings:/home/pyth/pyth-client/findings \
  pyth-fuzz \
  sh -c "./pyth-client/build/fuzz add < ./pyth-client/findings/crashes/id\:000000\,sig\:06\,src\:000000\,op\:flip1\,pos\:0"

in this example, id\:000000\,sig\:06\,src\:000000\,op\:flip1\,pos\:0 is the file containing the failing input.

Development Setup Using VS Code

First create a docker container in daemon as your working container (IMAGE and PYTH_REPO same as above):

docker run --name pyth-dev -d \\
  --volume "${HOME}:/home/pyth/home" \\
  --volume "${HOME}/.config:/home/pyth/.config" \\
  --volume "${HOME}/.ssh:/home/pyth/.ssh" \\ # Github access
  --mount "type=bind,src=${PYTH_REPO},target=/home/pyth/pyth-client" \\
  --platform linux/amd64 \\
  $IMAGE \\
  /bin/bash -c "while [ true ]; do sleep 1000; done"

Default user in the image is pyth which may not have access to your directories. Assign your user id and group id to it to enable access.

host@host$ id $USER # Shows user_id, group_id, and group names
host@host$ docker exec -ti pyth-dev bash
pyth@pyth-dev$ sudo su
root@pyth-dev# groupadd -g 1004 1004
root@pyth-dev# usermod -u 1002 -g 1004 -s /bin/bash pyth

Finally, in docker extension inside VS Code click right and choose "Attach VS Code". If you're using a remote host in VS Code make sure to let this connection be open.

Deployment

Oracle program upgrades are managed by the Pythian Council multisig. The steps to deploy a new version are:

  1. Create a release branch from main. This should include the binary for the Pythnet oracle program (pyth_oracle_pythnet.so).
  2. Install Solana CLI if not already installed.
  3. Set Solana config for the target network, e.g., devnet: solana config set --url https://api.devnet.solana.com.
  4. Execute solana program write-buffer pyth_oracle_pythnet.so to obtain the buffer address.
  5. Run solana program show <ORACLE_PROGRAM_PUBKEY> to obtain the authority of the current program.
  6. Use solana program set-buffer-authority <BUFFER_PUBKEY> --new-buffer-authority <NEW_BUFFER_AUTHORITY> to assign the upgrade authority from the previous step to the buffer address.
  7. Submit a proposal with xc-admin for program upgrade using the upgrade-program command.
  8. Verify the buffer by running solana program dump <BUFFER> temp_file && shasum -a 256 temp_file && rm temp_file, comparing the hash with the one from build-bpf.sh#L35.
  9. Once the proposal secures enough signatures, it will be automatically relayed to the target network, upgrading the program.