Re: libl4 test cases

[Neal H. Walfield]

At Wed, 9 Feb 2005 12:59:44 +0100,
Espen Skoglund wrote:
> 
> [Neal H Walfield]
> >>> I was browsing the L4Ka site earlier, and they have listed but not
> >>> linked the PowerPC/64 architecture as being either supported or
> >>> under development. Assuming that the L4 kernel is made to run on
> >>> Cell, what are issues and challenges to making L4/Hurd also work?
> >> 
> >> If you were to use an IDL compiler for Hurd: absolutely nothing.
> 
> > How does an IDL compiler encapsulate the architectural differences
> > which is what I think Douglas is referring to?  For instance, how
> > would an IDL compiler eliminate the need to write low-level locking
> > primitives, memory barriers and TLS support?
> 
> I would think that these things are covered somewhere deep down in
> glibc.  But then I'm not very familiar with glibc internals, so I
> couldn't say whether there is a clear separation in glibc regarding
> hardware and the underlying OS.

That may be true but without that support, the Hurd on L4 won't run on
Cell.  So, Hurd on L4 has as dependencies not only Pistachio but also
libl4, glibc and the device driver framework.