The Prognostic Lab | Computer Science Department | University of Pittsburgh

XEMEM: Cross Enclave Shared Memory

The XEMEM shared memory architecture supports application-level shared memory communication across enclaves (co-kernels and/or Palacios VMs). XEMEM exports a user-level API that is backwards compatible with the API exported by SGI/Cray’s XPMEM shared memory implementation for Linux systems, which allows processes to selectively export regions of their address space to be mapped by other processes. XEMEM allows any application targeting the API can be deployed across any multi-enclave topology without modification.

XEMEM provides a single global shared memory address space through the use of globally unique memory segment IDs managed by a global name service. In addition to naming, the name service also provides global discovery of shared memory regions allowing applications to transparently map memory regions from any other OS/R instance.

Download XEMEM

Related Publications

HPDC [2015] (pdf) B. Kocoloski and J. Lange,
XEMEM: Efficient Shared Memory for Composed Applications on Multi-OS/R Exascale Systems,
Proceedings of the 24th International ACM Symposium on High Performance Parallel and Distributed Computing, (HPDC 2015)