Ending this year, Ron G. Minnich has got Harvey running in RISC-V architecture, booting Harvey on Spike (ISA Simulator) and running rc shell on it. But he never rests and now is working on bringing it to QEMU and to FPGA. It’s a big step for Harvey because we fixed some multiarch issues across the source and Ron found some bugs in timer interrupts in the hardware, so we all learned something.
What is Harvey OS?
Harvey is an effort to provide a modern, distributed, 64 bit operating system. A different environment for researching and finding new lines of work. It can be built with gcc and clang and has an ANSI/POSIX compliant subsystem.
Two news items about alternative operating systems in a row?
The year’s off to a good start.
The name is a hint.
Harvey is a invisible giant white rabbit in the 1950s movie of the same name.
Glenda, the Plan9 mascot, is a white bunny …
Their logo as well.
Can’t do better than this on OSNews
Harvey is a 64 bit distributed operating system under GPL.
For now it runs in x86_64 (amd64) machines, but we’re working at RISCV and ARM64 (aarch64) support. We’re improving now the kernel and userland, trying to bring up a full usable operating system, modern and, at the same time, keeping ideas and concepts from other platforms what we consider very interesting. We collected some code and ideas of Plan 9 from Bell Labs, in addition to researching code from NIX and NxM, with some influences from Linux, BSD and in general from Unix world.
Harvey is a general purpose operating system, so every idea you can think could be welcome here. Come and see, and do it by yourself.
How does it compare to Genode?
not at all – it’s like apples and oranges … or in this case rabbits and horses.
Genode is a framework. It aims to allow you to use any (micro)kernel with whatever userland you choose.
Harvey aims to be a full OS with own kernel and own (posix) userland.
Edited 2017-01-03 19:21 UTC
So this does not seems like apples and oranges at all. If Genode as you say can combine any kernel with any userland then how would an OS built using Genode and utilizing Harvey’s kernel and Harvey’s userland compare to Harvey the original? Would you still call it apples and oranges?
Genode is concentrating its efforts towards microkernels mainly. Linux as a development platform is a exception to that.
But as you ask:
Yes you can run Genode with a Linux kernel as host and a second paravirtualized Linux for the userland. Same would theoretically be possible for Harvey if someone is willing to to the work.
But still this does not make it comparable by any useful means.
Genode is a framework or api-layer between kernel and userland.
Harvey is a operating system on its own.
Edited 2017-01-03 23:09 UTC
As much as I love compiling programs, I think an entire OS, must be available for downloading as an ISO file (and optionally for USB stick).
Happy New Year to everyone!!! ๐
G.
I dunno, I compiled an entire OS on a dual 66MHz PowerPC 603 based system once. It took over 8 hours. What’s wrong with that?? ๐
Oh nothing! Nothing at all!!!
Well…except….of the 8h compile process. ๐
I’m getting a bit lazy, these days! ๐
G.
As long as it spits out chars to stdout at tremendous speed, any length of compilation is satisfying. I hate those that just sit at the command line and do nothing before returning success. I kind of need that satisfaction that the code is being processed and run
Bill Shooter of Bul,
Yeah, as long as the build finishes without error, I’m pretty happy just to see text flying by. But when a build takes an hour only to throw out an error forcing you to restart it…oh god why have thou forsaken me?
I could certainly do without all the legacy build systems that are prevalent with C/C++ development. Having autoconf tools that test for strtol and other nonsense for the Nth time over and over again only to error out with missing dependencies is so frustrating. Then even the compilation step is slow compared to other languages because of C’s include file cascade problem. This has probably wasted billions of hours of developer & cpu time through to course of our history.
http://voices.canonical.com/jussi.pakkanen/2012/10/01/building-cc-w…
And then there are all the problems caused by the unsafe code generation. We’re so set in our legacy approaches to technology that we’ve become the biggest impediment to improving things for the future. This seems to be a cross-disciplinary problem: programming languages, ipv4/ipv6, 1500 MTU. This resistance to engineering upgrades puts us decades behind where we ought to be.
Linking GCC 4.x+ on a 50Mhz 128MB SparcStation LX… X.x It takes 10s of hours and probably will fail. Bear in mind the microsparc is alot wimpier than the SuperSparcs as well… probably equivalent to a 25Mhz supersparc.