Panacea: Non-interactive and Stateless Oblivious RAM

This is an implementation of Panacea. The source code is based on SortingHat.

WARNING:

This is proof-of-concept implementation. It may contain bugs and security issues. Please do not use in production systems.

Building and testing

Our implementation is written in the Rust programming language. Please refer to The Rust Programming Language Website to install the required tools (e.g., rustup, cargo and rustc). The main library we use is concrete-core, but this is imported automatically with cargo.

The stable toolchain can be used for Linux and Mac computers with Intel processors.

Compiling for Windows is not supported at the moment.

To Compile

cargo build --release

To Run Tests

cargo test --release

Special Instructions Mac computers with ARM Chips

To build natively on Mac computers with ARM chips (e.g., M1 and M2), the Rust nightly toolchain (1.68 or above) must be used. This can be installed by running

rustup toolchain install nightly

Additionally, .cargo/config.toml should be modified by swapping the commented lines so that it looks like this:

[build]
# rustflags = ["-C", "target-cpu=native"]
rustflags = ["-C", "target-cpu=apple-m1"]

For convenience, you can execute the following command so that the nightly toolchain is always used under the repo's directory

rustup override set nightly 

Then you can compile and test using the commands above.

Otherwise, you can use the +nightly modifier with cargo every time to specify the desired toolchain

# To install:
cargo +nightly build --release
# To test:
cargo +nightly test --release

Running

Getting the CLI options

cargo run --release -- --help
# or
./target/release/panacea --help

Running

cargo run --release -- --params params.json
# or
./target/release/panacea --params params.json

Some parameters used in the paper are given in params.json. The executable runs the ORAM protocol, acting both as the client and the server.

Note that larger parameters have higher memory requirement (batched mode with rows=384 and cols=4096 needs 1 TB of memory). Please refer to the paper to for the exact memory requirements.

Setting the number of threads

# for example, 8 threads
RAYON_NUM_THREADS=8 ./target/release/panacea --params params.json