kern/165381: powerd(8) eats CPUs for breakfast

[Jamie Gritton]

>Number:         165381
>Category:       kern
>Synopsis:       powerd(8) eats CPUs for breakfast
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Feb 21 23:10:10 UTC 2012
>Closed-Date:
>Last-Modified:
>Originator:     Jamie Gritton
>Release:        9.0
>Organization:
>Environment:
FreeBSD glorfindel.gritton.org 9.0-RELEASE FreeBSD 9.0-RELEASE #1: Sat Jan 28 23:11:49 MST 2012     gritton at glorfindel.gritton.org:/usr/obj/usr/src/sys/GLORFINDEL  amd64

>Description:
I have a desktop amd64 system with an Athlon 7750 CPU (don't know the chipset offhand) that I decided to try powerd on, as I'm fond of quiet fans.  I set powerd_enable="YES" in rc.conf and ran "sh /etc/rc.d/powerd start".  After one or two minutes, my display froze.  After resetting the box, I got nothing.  After a hard power-cycle, nothing.  After waiting ten minutes for everything to get nice and cool and another power cycle, nothing.  Not even the first BIOS screen.  It was pretty clear something was fried, most likely the CPU.

That's OK, I had another currently unused system (don't we all?) and I swapped CPUs...
>How-To-Repeat:
Swap CPUs and turn the computer on again, without remembering to removed the powerd_enable="YES" like from rc.conf.  Sure enough, two minutes later I fried my second CPU.  After a quick eBay purchase of a third 7750, this times I remembered to fix my rc.conf and everything works again.

I apologize for not having more information, but with no CPU it's hard to get a good crash dump.
>Fix:


>Release-Note:
>Audit-Trail:
>Unformatted: