kern/45928: Random freeze and interrupt/system storms (netbsd-5/amd64 and -current/amd64)

[mm_lists%pulsar-zone.net@localhost]

>Number:         45928
>Category:       kern
>Synopsis:       Random freeze and interrupt/system storms (netbsd-5/amd64 and 
>-current/amd64)
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Feb 05 07:20:00 +0000 2012
>Originator:     Matthew Mondor
>Release:        netbsd-5, -current
>Organization:
>Environment:
NetBSD ninja.xisop 5.1_STABLE NetBSD 5.1_STABLE (GENERIC_MM) #1: Fri Feb  3 
06:04:22 EST 2012  root@ninja.xisop:/usr/obj/sys/arch/amd64/compile/GENERIC_MM 
amd64

>Description:
Since I installed NetBSD on an i5 2500 system I have I experience
instability issues, involving resumable freezes and interrupt/system
time storms reported on the cores.

Suddenly, the system becomes unresponsive and top shows two of the 4
cores in 100% interrupt usage and the two others in 100% system usage.
The system becomes reusable again but then it shortly does this again
and it gets worse, where I have to reset.

This sometimes happens after several hours or can also occur a minute
or two after I rebooted.  The system does not have to be notoriously
busy for this to happen, it seems to occur anytime.

Unfortunatly I'm not sure how to diagnose this further.  I think that
it happens less often, or that it's less serious since I applied the
patch from kern/45160, but it still happens.  Also, it seems to happen
more quickly if I use a hard drive in PIO mode (see kern/45917), but
even when I'm not using those this happens (in which case I see no
abnormal storms reported by vmstat/systat, so I suspect it's soft
interrupts).

This also happens on -current it seems.
>How-To-Repeat:

>Fix: