[LV] Success! was: Re: Status of the toolchain build scripts?

[Jan-Benedict Glaw]

On Mon, 2007-07-09 12:07:24 +0200, Mikkel Lauritsen <mikkel-vax at tala.dk> wrote:
> "Peter Lund" <firefly at vax64.dk> wrote:
> > > So - I guess the current question is which areas (kernel, libc etc.) are
> > > most in need of attention?
> >
> > How about:
> > 1) an easy way to get a toolchain up and running
> > 2) an easy way to get a kernel compiled
> > 3) an easy way to get a userland of sorts compiled
> >
> > as your recent experience shows, there are some obstacles along the
> > way ;)

Hacking would be good, all over all and any topics.

> > I can perhaps be of some assistance in the packaging of a precompiled
> > version of such a toolchain in the form of a socalled "Autopackage", an
> > executable file containing a self-extracting installer which should work
> > satisfactorily on most recent Linux distributions and some older ones,
> > making short thrift of an otherwise somewhat painful installation
> > experience.
> 
> I think having a prepackaged toolchain would be a really good thing, as I'd
> naively assume that anybody interested in linux on Vax would tend to focus
> on the kernel itself instead of mucking around with GCC and its friends.
> 
> Would it make sense to take a recent toolchain build and bundle it up in an
> autopackage or equivalent format, and then make it available from the
> website?

I've got a script running that builds the toolchain and creates a
tarball six times a day. In theory, it'll build a userspace-capable
cross-compiler, but currently it breaks in uClibc and thus it'll only
build a kernel-only cross-compiler. I can post the tarballs, of
course.

There's also an older version of this tarball on the SF.net web pages,
at
http://linux-vax.sourceforge.net/download/vax-linux-gcc-4.1.0-20050618-for-i686.tar.gz

> With respect to the kernel things are in a reasonably good shape, I think.

Needs some merging of Linus's tree...

> A working default config is obviously a good thing to have, but other than
> that building a kernel is no more difficult than on all other platforms.
> Still, we could consider documenting which parts of the kernel are in
> working order so new people for example would know that including the
> iso9660 filesystem is a bad idea.
> 
> Userland I haven't got any comments on yet :-)

We've got some old uClibc-based userland. That basically works, but is
old. I've done some uClibc hacking, but there are some (minor) issues
so it doesn't complete the build. (As noted above, I regulary get
mails from the build script...)

However, in the long-term, uClibc is probably *not* the route to go.
GNU libc should be taken. The basic port is quite straight-forward,
but there are some interesting pieces:

  * Dynamic loader
	  I was told that generating PIC code currently doesn't work
	due to GCC bugs/missing features/...  The NetBSD people worked
	on that and have patches in their CVS repo. Once these go
	upstream (ask Mark!), the first issue should be fixed.
	  The second issue is that you need to have an idea of the
	process of dynamic linking.

  * TLS
	  Since GNU libc abandoned linuxthreads (and uses NPTL now),
	working TLS simply is requires. This'll require at least
	binutils hacking for the relocs (and probably also some GCC
	hackery.)

  * NPTL
	  Long-term, NPTL is *the* threading library for linux.
	This'll need some additional handling functions to be
	implemented.

> On a completely unrelated note, does anybody happen to know if there are
> any kinks involved in MOP booting a VXT2000? It apparently obtains an
> image from the MOP server without problems, but after that absolutely
> nothing happens other than the chevron is displayed again. It reports a
> number that I guess is the size of the downloaded file, but it doesn't
> quite fit with the size of the actual file on the server; it's about 14K
> too large.

I played with my VXT2000+ a bit. I made it "boot" somewhat, but some
more work is required to make it useable. The serial ports are on some
funny chip, SGEC (network) probably won't handle the NIC at all and I
didn't ever start to disassemble the ROMs to figure out how the
framebuffer works...

MfG, JBG

-- 
      Jan-Benedict Glaw      jbglaw at lug-owl.de              +49-172-7608481
  Signature of:                        Lauf nicht vor Deinem Glück davon:
  the second  :                             Es könnte hinter Dir stehen!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lug-owl.de/pipermail/vax-linux/attachments/20070709/21052c6d/attachment.pgp>
-------------- next part --------------
_______________________________________________
Linux-Vax mailing list
Linux-Vax at pergamentum.com
http://www.pergamentum.com/mailman/listinfo/linux-vax