Re: debian on ruffian
On Mon, 14 Sep 1998, David Monro wrote:
> Well, I just got a ruffian (or rather my SO did for work) and I've attempted to
> install debian-unstable on it. It now seems to be mostly working, but I did
> stumble a few times on the way.
FINALLY! We got a Ruffian around! I've been dying for one of these
babies to test our stuff on because of the newness of it all. Glad to
hear from you.
> 1a) The machine would crash after a while unless I told milo that there was ony
> 224 rather than 256 M of memory (after reading about this problem on axp-list).
That's odd, but will probably be solved by a new Milo when it comes out.
> 2) X server wouldn't start with errors about IOPL. Making /etc/alpha_systype ->
> Miata cured it (after a couple of guesses). I suppose this will have to wait
> for a fixed libc. Could the install process at least make a guess or ask the
I can fix glibc if necessary, but try rolling your own kernel first to see
if it fixes the problem. I doubt it will, but it's worth a shot. Anyway,
if you can dig up the glibc patch I need, I'd be happy to do it.
> 3) hwclock still complains about IOPL after the above fix. Is this normal?
Unknown. AFAIK, you're the only Ruffian around here so far.
> 4) During the install stage some of the dependencies seemed to cause problems;
> in particular the tex packages wouldn't install the first time round. Rerunning
> the install stage cured it. Unfortunately I didn't catch the reason for the
> failure before it scrolled off the screen.
Doh! I'll look at that soon when my access to master speeds up a bit.
> The machine is now working (and I even got OSF/1 netscape running). A few
> things are still kooky:
> The X server generates unaligned accesses when running netscape.
Normal. I wouldn't worry too much about unaligned accesses (just a sign
of poor coding and non-portability). You can even comment out the lines
in the kernel source that report that stuff if you REALLY don't want to be
> The X font server segfaults and dies when contacted by the X server.
> fdisk generates unaligned accesses.
> update-menus segfaulted during the install; looked like a near-null pointer
> dereference (00000010).
Damn...fdisk again!?!?! I've got a new way of fixing fdisk now, so I'm
gonna give that a shot. Either that or rewrite the backend like I've been
talking about for a year now.
As for the font server and such...I may have to defer that to any
interested party. I don't run X, so I can't test this stuff.
> Anyway, looks like it is working, and I'm really glad not to have to run Red
> Hat :-)
And we're really glad to finally get a Ruffian user to run Debian! :)