David T. McWherter (dtm@vramp.net)
Mon, 7 Dec 1998 09:39:55 -0500 (EST)
The funny thing is, the vmlinux we're currently using on
our ANS700-based machine works fine booting up from the
internal drive. We've not been able to compile another one
that works similarly for months. Very weird...
But it might be useful info in tracking down the bug...I'll
have to snoop around the SCSI code some more, I guess :-)
-David
On Sat, 5 Dec
1998, Robert I Leichtman wrote:
> A long time ago...almost a year we were told that the real problem was 1 bit
> involving the SCSI controller in that it thought it was a 'raid' when in fact no
> 'raid' was present. The writer of that comment was going to pursue the issue with
> Apple or and Adaptec and IBM, since AIX was perhaps looking for that hardware
> assigned bit. I hope I could be more helpful but that was the issue we have faced
> since we tried to install Linux on the internal drive to no avail... only on an
> external which obviously sits on another bus with a different controller. Hope I've
> been of some help. If someone comes up with a solution it would help all of us, in
> that it's cheaper to us internal drives [as well as more secure] than externals
> which need there own power supplies and cases.
>
> Bob Leichtman
> Weblur Internet Services
> Computer Integrated Devices
> 248-626-7742
> weblur.com
> cid.weblur.com
>
> "David T. McWherter" wrote:
>
> > A bit of interesting news, it seems that moderately new kernels (125,129,
> > etc) from vger and/or ftp://samba.anu.edu.au/pub/linux-pmac/ work pretty
> > well. The biggest problems are with a funny video console (I've yet to
> > get enough time to figure out how to port our video driver to the new
> > style in the new kernelss) and with the internal SCSI bus on the machines.
> >
> > Recently, I debugged, to some degree, some oddities with the internal
> > SCSI bus. It had regularly seemed that new kernels that we booted would
> > panic early in the boot process because of an inability to mount the
> > root filesystem. The filesystem was a std. ext2 filesystem, but the SCSI
> > driver/fs code was reading in the wrong data for the superblock, causing
> > it to be not recognized by the kernel's ext2 fs driver code.
> >
> > Well, it seems that the problem was ssolely due to booting the vmlinux
> > from the internal SCSI bus. Putting an external drive on the machine,
> > putting a kernel on an ext2fs partition on it, with Quik living on the
> > internal hard drive, along with a root/usr/etc filesystem on it, making
> > Quik boot up from OF on the internal drive, then load the kernel from
> > the external hard drive, giving it arguments to use the internal hard
> > drive as the root filesystem caused things to work "right."
> >
> > So, I'm guessing that somehow the OF is hosing the internal SCSI bus
> > in such a way that the new SCSI driver code isn't recovering from
> > properly. Unfortunately, the SCSI code is <sarcasm>moderately</sarcasm>
> > complicated, so I don't expect to be fixing tis problem soon (I don't
> > even have an idea as to where to begin at the moment). But, thanks
> > to that new knowledge, my linux-ppc work for ANS's will revolve mostly
> > around bringing our video driver into line with the newer kernels...
> >
> > -David
> >
> > > I'm sorry that we can't be more helpful, but both of us (Dave McWherter
> > > and Dave Beck) are students who are at the same time trying to complete
> > > important research grants. In defense of our "lack" of support, I
> > > have to point out that, well, it works for both of our ANS 700s (a
> > > 150 and a 200). I'm aware (as I am sure Dave McWherter is) that others
> > > are having trouble, but we both have sent out our best ideas. The main
> > > point of this list was to enable the ANS people interested in runing
> > > Linux to support each other, not for Dave McWherter and myself to become
> > > the sole source of ANS "support." What I would recommend, for you, at
> > > this point, is to post, in full, your successes and failures with all the
> > > permutations you have tried and _why_ you have tried them. If everyone
> > > were to do this, we could assemble a large volume of data on what works
> > > and more importantly what doesn't.
> > >
> > > Quoting Nic Olinsky (olinskyn@shs.sunnyside.wednet.edu):
> > > > Aren't the people who posted the ANS files on hellfire the same people
> > > > who run this list?
> > > >
> > > > Isn't anyone from hellfire monitoring this list?
> > > >
> > > > Doesn't the hellfire people know the trouble people are having with the
> > > > files?
> > > >
> > > > Do you see a pattern emerging? ;-)
> > > >
> > > > It would be nice to hear from someone from hellfire who could tell all
> > > > of us why the files aren't working, or at least give us directions on
> > > > how to make them work. I assume they did work for *someone* or they
> > > > wouldn't have posted them with a pointer to this list for support
> > > > questions.
> > > >
> > > > Any listmoms out there?
> > >
> > > --
> > > Dave Beck
> > > dave@arginine.umdnj.edu Sites of interest (set 3):
> > > Computer Science and Biology http://selene.biochem.uga.edu/tutorial/
> > > Drexel University, Philadelphia PA http://www.cold.org/
> > >
> >
> > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> > David T. McWherter dtm@waterw.com
>
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
David T. McWherter dtm@waterw.com
This archive was generated by hypermail 2.0b3 on Thu Dec 10 1998 - 16:55:55 PST