Unstable kernel (dumbbell) crashes

[Jeffrey Bouquet]

On Sun, 20 Dec 2015 20:04:20 +0200, Beeblebrox <zaphod at berentweb.com> wrote:

> Hello.
> 
> I have been away from my FreeBSD system for over 7 months. The first on my return was to update world/kernel then do a full poudriere run for all my packages. The process had many problems; I'm reporting items below as FYI.
> I'm tracking https://github.com/dumbbell/freebsd.git for my Radeon card.
> 
> 1. installworld ran into same problem as below and I got around it by following the prescription therein: https://lists.freebsd.org/pipermail/freebsd-testing/2015-September/001094.html
> Reviewing the questions asked in the thread, I find nothing of significance in my particular case. Thus, still a mystery.
> 
> 2. The pkgng DB has become buggy & inconsistent. I am preparing to "pkg delete -a" and re-install all from a list after the poudriere run finishes.
> 
> 3. The newly built kernel keeps crashing and re-booting unpredictably. The poudriere run seems to have something to do with triggering this.
> 
> 4. Service dbus still fails to start from /etc/rc.conf (which prevents xorg login) and must be started manually.
> 
> 5. I built and booted into a DEBUG kernel. The system quickly becomes unstable and eventually the simplest commands (ls or cat) take over 60 secs to respond. File containing dmesg output and memory status is attached. However, Memory status seems to be a symptom and not a cause. 
> 
> Regards.
> 
> -- 
> FreeBSD_amd64_11-Current_RadeonKMS
> Please CC my email when responding, mail from list is not delivered.
> _______________________________________________
> freebsd-current at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe at freebsd.org"


Yesterday I updated most packages from a v11 that had not been updated in many months.  Strange pkg messages
were solved by wholesale deleting of particular mentions in the "pkg upgrade" output and deleting orphaned packages/ports
from the output of "pkg version -vRL=  " (or -vIL)" after make fetchindex and portsdb -u...

pkg wanting to remove files I was prepared for by running it all via
script log1 pkg upgrade
script log2 pkg upgrade
until it all was done.
One setback was continually wanting to download tex* (an hour download for one package) that I did not want
installed and the manpages are still terse as to how to prevent that.
Not a big concern; that is not my principal BSD machine. 

All that is sort of aside.

The reason for this reply is I was wondering if an UPDATING entry in /usr/src 
at the bottom, under COMMON ITEMS...
could be (something to the
effect that if an installworld fails, one could download a month-old .txz or .tar or something  ) and recover
by in effect rolling back all the system binaries and libraries.   )

That is somewhat easier than...
... the more-difficult
Even better if some script could be crafted ascertaining which files within the system may be at fault. 

Apologies for not answering anything in the mail above.  I've no recollection of any answer I could reply to each/any
of the questions...  sort of like a "me too" but only for a troubled installworld (often) when it builds fine.