kern/43631: pckbc(4) failure during setroot() on SMP

[jakllsch%kollasch.net@localhost]

>Number:         43631
>Category:       kern
>Synopsis:       pckbc(4) failure during setroot() on SMP
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Jul 18 02:00:00 +0000 2010
>Originator:     Jonathan A. Kollasch
>Release:        NetBSD 5.99.36
>Organization:
>Environment:
System: NetBSD trantor.kollasch.net 5.99.36 NetBSD 5.99.36 (TRANTOR) #4: Sat 
Jul 17 15:00:17 CDT 2010 
jakllsch%trantor.kollasch.net@localhost:/local/jakllsch/nbsd-head/build/objdir/sys/arch/amd64/compile/TRANTOR
 amd64
Architecture: x86_64
Machine: amd64
>Description:
Two processors try to access one pckbc(4) during setroot() in RB_ASKNAME case.
>How-To-Repeat:
`boot -a` on a modern quad-core x86 box.
Note that the keyboard input doesn't show up when it asks for the root device,
and that using a keyboard LED toggling key (numlock) repeatedly soon causes an
access fault.  If you happen to get a working ddb at this point, you may find
the keyboard interrput handler trying to toggle the LED on one cpu while
setroot() is polling the same keyboard for input on another.
It should also be noted that boot -d and boot -c have working keyboard,
as does ddb when entered while the system is multiuser.
>Fix:
better locking?