[Prev][Next][Index][Thread]
Re: Bug in OSKit's linux block I/O glue [PATCH]
> Notes on recent experience with OSKit 20000202
>
> I can build oskit fine in Linux as long as all three of the libc options
> are enabled.
> Using a parallel make it takes maybe 10-15 mins. The sample kernels all work
> although the smp test kernel prints some gibberish after shutting down
> the cpu's.
>
> My problems happen when I try and compile it natively under Hurd.
There is no need to compile oskit under Hurd. First the object file
format is the same for Hurd and Linux and the oskit libs are statically
linked into the kernel binary anyway.
> I built oskit-mach using Debian's oskit deb, it compiled well enough
> under Hurd,
> except the resulting kernel reboots my system shortly after it says
> OSKit-Mach!
> Not sure if oskit was compiled with smp and then oskit-mach tries to go
> smp and
> loses it.
I have been able to get oskit-mach to work with oskit-20000202, but
not later versions (possibly 20000505 would work as well).
> I'm working on getting the serial debugging setup going, maybe that will
> help get
> this thing working.
If you have problems with oskit-mach or Hurd, you're more the welcome,
even encouraged to report them to the bug-hurd@gnu.org mailing list.
Check the recent archives at http://mail.gnu.org/pipermail/bug-hurd/
for updates on oskit-mach.
Igor