Odd problem with ofw_console on Ultra-10

Garance A Drosihn drosih at rpi.edu
Mon Aug 2 19:04:34 PDT 2004

Back on July 21/2004,
in the thread "Installworld problem on sparc64?"
        I (Garance) wrote:
>I did a buildworld/buildkernel/installkernel yesterday, and it
>seemed to work fine.  I then went to reboot, to do the installworld.
>If I try to boot in single-user mode, the startup process hangs
>right after:  Mounting root from ufs:/dev/ad0a
>If I just boot up normally (not bothering with single-user), it
>comes up fine.  The line that shows up after "Mounting root" is:
>    Pre-seeding PRNG: kickstart.
>So I do the normal boot-up, and then type `shutdown' to get into
>single-user mode.  Again the machine hangs.  I have to power-off
>and back on to get it's attention.
>Eventually I booted up multi-user, just killed a few daemons,
>and ran installworld.  That seems to have gone okay.  I was able
>to reboot after doing the installworld, and it comes up pretty
>much okay (a few minor messages from ntpd, but otherwise nothing
>seems unusual).  A `shutdown -r' also works fine.  However, if I
>try to do a plain 'shutdown now', the machine again hangs.  It
>stops after writing out:
>    Stopping cron.
>    Shutting down local daemons:.
>    Writing entropy file:.
>    Terminated
>    .
>and never comes up asking me for which shell I want to run (where
>/bin/sh would be the default).  Also, now that I have completely
>upgraded, I still can not seem to boot in single-user mode, as I
>described above.

>My install is working off the regular console (keyboard/monitor
>setup), not a serial-console setup, on an Ultra-10.

 From the onslaught of replies to this (aka "Zero"), I assume I am
still the only person seeing this.  Well, I still do not quite
understand all the details of the problem.  But after much time,
several blind alleys, and many buildworlds later, I have narrowed
it down to the following commit:

    obrien      2004-06-24 02:57:12 UTC
    FreeBSD src repository

    Better OFW console support on Sun Ultra2 machines.
    Ultra2 users may want to set OFWCONS_POLL_HZ to a value of '20'.
    I have left default value at '4' as higher values can consume
    a more than is acceptable amount of CPU, and we don't have a
    consensus yet what is an optimal value.

    Submitted by:   Pyun YongHyeon <yongari at kt-is.co.kr>

    Revision  Changes    Path
    1.11      +1 -0      src/sys/conf/options.sparc64
    1.25      +17 -9     src/sys/dev/ofw/ofw_console.c
    1.7       +2 -0      src/sys/sparc64/conf/NOTES

If I take my non-working system, and backout revision 1.25 of
ofw_console.c, then it goes back to working fine in single-user
mode.  Before doing that, I also tried adding the line:

option	OFWCONS_POLL_HZ=30	# 20 or more works best on Ultra2

(to increase it up from the default of '4'), but that didn't
seem to do any good.  I suspect my main problem is that I run
my ultra-10 from a regular keyboard/console, while everyone else
seems to be doing it via a serial console.

Now that I think I have a way around the single-user problem, I'm
going to try to bring my system up-to-date (minus the above change),
and see how well that works.  That'll take me another day, though...

There is much about the behavior I see from this bug which does
not make sense to me.  Given the behavior I see, real issue might
be something pretty simple.  Like maybe something that needs to be
reset in /sbin/init .

Garance Alistair Drosehn            =   gad at gilead.netel.rpi.edu
Senior Systems Programmer           or  gad at freebsd.org
Rensselaer Polytechnic Institute    or  drosih at rpi.edu

More information about the freebsd-sparc64 mailing list