live and let live.与民争利的政府必亡
Saturday, 30 July 2016
The Haskell Lightweight Virtual Machine (HaLVM): GHC running on Xen.
1. What is the HaLVM?
The Haskell Lightweight Virtual Machine, or HaLVM, is a port of the Glasgow Haskell Compiler toolsuite to enable developers to write high-level, lightweight virtual machines that can run directly on the Xen hypervisor.
While Galois initially designed the HaLVM to allow for quick and easy prototyping of operating systems components, the HaLVM has grown over time to allow for a much wider variety of use cases. When connected with the appropriate libraries, the HaLVM can, for example, operate as a network appliance.
Writing for the Haskell Lightweight Virtual Machine is just like writing normal Haskell, and many pure Haskell libraries port to the HaLVM with little or no difficulty. In fact, we include the standard Haskell Cabal toolset in order to more easily facilitate the integration of outside Haskell libraries. However, instead of running on top of a typical operating system, HaLVM programs run at a very low level, directly on the Xen hypervisor. This allows for very lightweight, single purpose Xen domains with minimal resource requirements.
We develop the HaLVM almost exclusively in Fedora Linux, running a slightly modified version of their versions of Xen. Our modified versions simply add the flag "verbose=y" to their build specification, on the lines that build and install the hypervisor (search for dist-xen and install-xen). We try to keep the latest version we're using available, in src/misc/xen.spec. In addition, I will try to keep recent binary and source RPMs available at:
It may provide some information that you'd find useful, or may automate portions of this process for you.
If you plan to do development work on the HaLVM itself, please fork the HaLVM. This allows us to more easily tell who is working on the HaLVM, and GitHub's tools make merging your changes much more easy.
If GHC 7.8.4 is not present, a version of it will be downloaded for use during the build process.
Once checked out, the HaLVM builds as follows:
git submodule update --init --recursive
The configure system will accept and honor the "--prefix" flag as per normal. We also strongly suggest using the "--enable-gmp" flag, in order to enable the (much faster) GMP library for large integer math.
If you intend on using inter-domain communication (specified in Communication.Rendezvous and Communication.IVC), run the mkrenddir script (located wherever you installed the HaLVM, /usr/local/bin by default) as root. This creates a top-level /rendezvous directory in the XenStore with appropriate permissions.
3. Where To Look Next
The HaLVM comes with a number of examples / test cases, located in the folder examples. We suggest taking a look at these to see a wide variety of HaLVM programs you might use as a starting point.
Further information about using the HaLVM is available on the HaLVM wiki. Developers interested in improving the HaLVM should also take a look at the current list of HaLVM bugs, and submit any new feature requests to the HaLVM bug system.
4. How To Contact Us
If you have any questions or suggestions for the HaLVM, please feel free to join the HaLVM mailing list and send them to us. Please note that in order to stop a great deal of spam, you must be a member of the mailing list in order to send messages to it.
If you have bugs you would like to file or patches you'd like to send, we'd strongly prefer you utilize GitHub's interfaces for both these functions. Please see GitHub for more information.