Re: Contributing to Hurd

[Neal H. Walfield]

At Tue, 20 Apr 2010 15:43:36 +0530,
arnuld uttre wrote:
> I chose the "Hurd on modern microkernel"   as the path to walk on as I
> am neither interesting in web-side nor in outdated Mach. Any
> directions to start ?

Given your interest in this, I'd start by reviewing our experiences
with the Hurd.  In particular, there is the critique [1], which looks
at the structure of the Hurd and its most important shortcomings.

I choose to address these problems with a new microkernel, Viengoos,
which uses asynchronous communication, improves resource (in
particular, memory) management, and is capability based [2].  There
are a few papers about it.  They focus primarily on the memory
management issues.  The reference guide looks at capabilities and
addressing [3].

If you decide to work on Viengoos, an initial non-trivial project
would be to fix the way addressing is done.  Currently a thread object
has two name space: the thread's implicit name space and the thread
object's special slots.  When a thread object is invoked, the method
determines which name space to use.  Instead, these name spaces should
be unified by using the first bit of the address to distinguish the
name spaces.

Neal

[1] http://www.gnu.org/software/hurd/hurd/critique.html
[2] http://www.gnu.org/software/hurd/microkernel/viengoos.html
[3] 
http://www.gnu.org/software/hurd/microkernel/viengoos/documentation/reference-guide.pdf