[SMAUG] Caldera Workstation 3.1

[Rick Moen]

(This reflects working through several messages in chronological order,
so please have patience.)

To recap Calvin Chu's problem of "hangs":

> Anyway, when I go to install, it reaches Lizard, opens up with the
> screen that says pick a language, and it hangs.  Doesn't read the
> mouse or keyboard. 

Just a brief comment (and, Calvin, this isn't trying to beat up on
you, just a _general_ lesson for all of us to note):  When people talk
about "hangs", they mean "I tried a few things, but couldn't regain
control."  

That's fine as far as it goes, but diagnosis requires specify what those
"few things" were, so we know what possibilities to eliminate and what
steps to suggest.  Therefore, in general, you'll want to just tell us
_explicitly_ what you did, and what the computer did, leading up to the
point where things went wrong -- and then also what you did to attempt
to recover.  

In other words, you should give an account in chronological order of
what exactly happened.  If you can't do this because you didn't bother to take
notes, please go back and try again, taking notes this time.  Why should you
bother?  Because otherwise, people will often ignore your request for help, as
being way too vague and likely to waste their time on wild goose chases.

(Again, Calvin, please consider the above to be _not_ directed at you.
It's a general observation.)


> The page did indicate that there are issues with the power management
> that go away when disabled at the BIOS level, which I did. [...]
> I have a suspicion that the power management daemon was causing the
> lockups....

I'm nobody's idea of an expert at this, but I still suspect some sort of
power management is disabled in the BIOS.  I _really_ doubt that running 
apmd has any effect (other than occupying a bit of RAM), if power
management is switched off at the BIOS level.

> When starting X11, it gives a grey background with the "X" pointer in
> the center, and it stops responding to any input whatsoever. 

By the way, what version of XFree86 is this?  You can tell by capturing
the raw X server's standard output and standard error streams, and
writing them to a file, like this:

   X > logfile 2>&1

"logfile" should then have a whole lot of mumblings from the X server's
startup, including its version.


Ah, I'm now catching up on Calvin's latest post:

> I went through and started disabling likely suspects and when gpm and
> apmd are out, the system begins to function stable.

Ah, well, it's not unheard of for gpm to mess up XFree86.  There are
ways to make XFree86 run compatibly with it, but, as a first-pass
step, you might want to disable startup of the gpm service.  ("gpm" is
the general purpose mouse driver, which lets you cut and paste
in/between console sessions, without X11 running.)

Similarly, maybe I'm wrong about apmd being inherently harmless.  And
results are what matter.  So, you probably want to disable it, too.

I realise my reference to "disabling services" sounds cryptic from your
newcomer's perspective.  File that as to-be-explained, for now.

> cardctl ident -- yields "No pcmcia driver in /procs/devices"

So, Card Services isn't running.  This suggests that possibly the
Caldera installer didn't recognise your laptop's PCMCIA chipset, or
something similar.  So, it didn't set the PCMCIA startup script to 
automatically run at boot-up.  (This is a surmise, on my part.)

> I tried popping the pcmcia card out, and back in.  No beeping.

This strongly supports my surmise.

> When I checked the process list for cardmgr it is not there.  How is
> it started?

OK, really briefly (sorry):  The kernel loads at boot time, process
number zero.  It splits (forks) off a second process, "init" (process
number one).  init is hard-coded to read /etc/inittab ("init table"),
which in turn specifies the rest of startup.  (You should look at it.)

/etc/inittab defines about seven "runlevels" (running states of the
machine), upon which some processes can be auto-started and others 
auto-stopped.  Rebooting is runlevel 6.  Halting the machine is runlevel
0.  Single-user is runlevel 1 (which is what my earlier suggestion of
going to single-user mode by typing "init 1" invokes).  Full multi-user
mode without X11 auto-startup is (usually) runlevel 3.  Full multi-user 
mode with X11 auto-startup is (usually) runlevel 5.

Each runlevel is implemented using a clever shell script named
/etc/rc.d/rc and a set of symbolic links for that runlevel in...

   /etc/rc.d/rc0.d/
   /etc/rc.d/rc1.d/
   /etc/rc.d/rc2.d/
   [...]
   /etc/rc.d/rc6.d/

The symbolic links in each runlevel specify what services are to be
started or stopped, any time the system enters that runlevel state.
(All such symlinks point to the actual _scripts_ that start or stop
individual services, which are in /etc/rc.d/init.d/ .)

If the symlink starts with capital-S, that signals that the service
is to be _started_.  If it starts with capital-K, that signals that the 
service is to be _stopped_ (K for killed).  Any other starting character
leads to the symlink being ignored (disabled), since /etc/rc.d/rc parses
_only_ scripts starting with "S" or "K".  Clear?

That means that, if you want to disable a service startup or shutdown
symlink, just lower-case the initial "S" or "K" of the symlink name.

Notice that /etc/inittabincludes a line like this:

   id:3:initdefault:

That means runlevel 3 is the default running state's runlevel.  So (if
yours is likewise), that runlevel (detailed below) will be most of
interest.

Now, probably the runlevel 3 directory (/etc/rc.d/rc3.d/) -- or whatever
is your initdefault runlevel -- has some script that includes startup of 
PCMCIA card services.  Unfortunately, different distributions do this 
differently.  Some have a symlink like this:

   S10network -> ../init.d/network

...that includes PCMCIA startup.  With Debian, by contrast, there's a
separate PCMCIA startup script.

But this is _supposed_ to get set up correctly for you by the installer.
That gets me back to the original point:  It seems likely that something
went wrong _during_ installation, at the point where either the
installer is supposed to auto-probe for PCMCIA functionality on the
motherboard, or you're supposed to inform it that it needs to add PCMCIA
support.  I unfortunately have no idea where in the Caldera installer
this would be, or what it looks like.

> Starting X11 still causes an instant crash -- that even
> control-alt-backspace and control-alt-delete or control-alt-f1 can not
> get it out of.

OK, this may be nitpicking, but "crash" is yet another one of those
vague terms like "freeze" that users assume are good explanations,
but in fact are really lousy, and entirely useless for diagnosis.  What
you mean -- again -- is that you can't figure out how to regain control.
It may be that, if you had inbound network access via telnet or SSH, you
would be able to get in via remote, do a "ps" listing to figure out
what's giving you a hard time, and kill that process.  

> I have /etc/sysconfig/mouse configured as a ps/2 mouse.

I vaguely recall that, on some systems, this is a file parsed by 
the gpm daemon for mouse information, when the latter starts.

> I disabled the trackpad from Bios and plugged in a ps/2 mouse and
> still no go.  

That was really not relevant to your problem.  I assure you, all PS/2
devices, including trackpads, are basically the same to gpm or X11 
mouse drivers.

> I tried running kxconfig, and interestingly, it loads up the graphical
> X configuration interface, and then locks up.

I assume this must be a Caldera-specific X configuration utility?  I've
not encountered it, before.

Two utilities are included with XFree86:  XF86Setup and xf86config.
Red Hat adds to that its Xconfigurator utility.  SuSE adds SaX, instead.
Debian adds anXious, instead.

On XFree86 v. 3.x, I have a way of using xf86config that is pretty much 
foolproof, but a little difficult to explain.  XFree86 v. 4.x's setup
just plain seems to work.

> I tried running against the SVGA version, and the 4.x version with
> similar results.

Beg pardon?  I'm not clear what you mean.  Do you mean that Caldera 3.1
includes both XFree86 3.x and 4.x, and that you've tried 3.x's SVGA
server _and_ XFree86 4.x?

> When the CD-ROM kernel is running.. i get a 800x600 full screen
> console.  When I'm booting off the hard drive, I get a 640x480
> resolution console..  What's going on here?

Extremely sub-optimal X configuration, during installation.

> When the system is not crashed, and I do a Fn-Screen off, the screen
> shuts off.  I suspect this is a bios level function?

Correct.

> Video Card
> Type: 2 the Max MAXColor S3 Trio64V+

> (does this correspond to the C&T ???)

No, it does not.

Try running SuperProbe (while logged in as the root user).  Its output
is pretty much definitive.  I strongly suspect that whoever told you it
had a C&T 65554 was on crack, and it's indeed an S3 "Trio64" AKA S3
model 764 (or possibly model M65).  See:
http://www.xfree86.org/3.3.6/S31.html . In which case, do _NOT_ use the 
SVGA X server (in XFree86 3.x).  Use the S3 one, instead.

I am _not_ at all sure that XFree86 4.x yet includes support for the 
Trio family.  I rather suspect it does not:  
http://www.xfree86.org/4.1.0/Status28.html
http://www.xfree86.org/4.1.0/RELNOTES4.html 

-- 
Cheers,                     Errors have been made.  Others will be blamed.
Rick Moen
rick@linuxmafia.com