wujf/gallocy

A distributed shared memory infrastructure.

gallocy

The gallocy library implements a memory allocator that transparently allocates memory across many machines. Combined with a gallocy-aware implementation of the POSIX threads library, this framework can make any pthreads application a distributed program! The gallocy library exists entirely in userspace, so not kernel modifications are necessary.

The gallocy library is under active development and is not ready for use. The gallocy library is in a pre-alpha state, and probably will be for a very, very long time. Reach out to us if you're interested in helping.

requirements

You need to make sure your local environment has a few things before you can get started. Here is a list of those things and our suggestions.

• cmake - we use cmake to build our code
• docker - we use docker to run distributed tests
• gcc/g++ 4.9+ - we make thorough use of C++14
• linux - we strongly suggest using with a 3.16+ kernel
• python - we use python to build a distributed test framework

You're probably already using most of these things if you're running on an operating system released after 2015, but if not, setup one in a virtual machine and remote in for your development work (hint: we develop on OS X and remote into a Debian 8.2 (glibc 2.19) virtual machine running in VirtualBox). You can use environment project to automate the bootstrapping of your machine.

This project uses cthulhu as a distributed testing framework.

getting started

Get the code using git.

git clone --recursive https://github.com/sholsapp/gallocy

Get started right away with the project helper.

Usage: project <subcommand> [options]
Subcommands:
    build - invoke cmake/make to build the code
    clean - clean things up
    coverage - run the system test suite and collect coverage data
    docker - build a docker image named gallocy-example
    integration - run the system test suite
    leakcheck - run the unit test suite under valgrind
    stylecheck - run the style checker
    test - run the unit test suite

Building is enough to get you started.

./project build

This will build Unix Makefiles, build library code, build sample application code, build the google-test test driver, and then run the library code unit tests. Output files will be placed into the install directory.

documentation

• NUTSHELL good for learning about allocators
• IMPLEMENTATION good for learning about distributed shared memory
• DEVELOPERS good for testing and debugging
• CREDITS about the contributors