John Sonnenschein wrote:
> Anyone had problems with building polaris recently? I've got the new
> tools & i just checked out from SVN about 10 minutes ago, but the
> build fails in a couple places that I can correct ( duplication of a
> function probably because of errant c&p, and an incorrectly defined
> struct ), but even after fixing them dies with screen after screen of
> linker errors ( offset <foo> is non-aligned)
>   
This is why several weeks ago I had queried whether anyone had even 
tried.  Since then
one (other than yourself) person outside of Sun has tried pulling from 
the external svn
repository and has reported a few problems.  Tom has been addressing 
them as they
are reported.  He expects to have everything fixed up and working by the 
end of this
week (or that is what he told me :-).  With respect to the linker 
errors.... well, just guessing
that you don't have the GCC tool chain fixes were the compiler DOES 
generate misaligned
relocations.  I fixed that in our copy, but why I queried how to get 
this back into the "official"
code release I got nowhere.  I believe that all the problems are related 
to dwarf section relocations, so
if you remove the -gdwarf2 from Makefile.master (I think that is where 
the usage exists) those problem
will go away.  If you wish to keep those I can try and help you fix the 
GCC tools.
> Also, any chance when the kernel build I can boot it off a floppy or
> such yet, I've had no luck getting this rs/6000 booting off the
> network
>
>   
Does the RS/6000 box you have support OpenFirmware (I would think it 
would, but I have
no idea whether or not it works properly, given that I've never used a 
RS/6000 box)?

Right now the port has only been tested with using an NFS root, though I 
don't know of any
reason why a UFS root would not work.  I'm encouraged you are trying 
though.  Please keep
us (all of us) apprising of your results.



Reply via email to