port-arm/48096: arm/BEAGLEBONE machdep.frequency.current sysctl node missing

[dennis.c.ferguson%gmail.com@localhost]

>Number:         48096
>Category:       port-arm
>Synopsis:       arm/BEAGLEBONE machdep.frequency.current sysctl node missing
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    port-arm-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Jul 29 17:55:00 +0000 2013
>Originator:     Dennis Ferguson
>Release:        6.99.23
>Organization:
>Environment:
NetBSD beagle4.to.akit-ferguson.com 6.99.23 NetBSD 6.99.23 (BEAGLEBONE) #0: Fri 
Jul 26 11:32:33 EDT 2013  
keenan%ts1.to.mistimed.ca@localhost:/usr/obj/sys/arch/evbarm/compile/BEAGLEBONE 
evbarm
>Description:
Early in the boot of a BEAGLEBONE kernel on a BeagleBone Black I see this:

NetBSD 6.99.23 (BEAGLEBONE) #0: Fri Jul 26 11:32:33 EDT 2013
        
keenan%ts1.to.mistimed.ca@localhost:/usr/obj/sys/arch/evbarm/compile/BEAGLEBONE
total memory = 512 MB
avail memory = 498 MB
sysctl_createv: sysctl_locate(current) returned 2
timecounter: Timecounters tick every 10.000 msec

The sysctl_createv: line indicates that a call to sysctl_createv() has failed.  
The missing node turns out to be machdep.frequency.current, i.e. sysctl -a 
returns:

...
machdep.powersave = 1
machdep.frequency.available = 550 600 650 700 720 800 900 1000
machdep.cpu_arch = 7A
...

This is inconvenient since, while the BeagleBone Black CPU should run at 1000 
MHz (independent of how the power is supplied), it comes out of the bootstrap 
at 550 MHz and the missing node means there is no way to change that.

Note that eyeball inspection suggests that 
mpu_current_frequency_sysctl_helper() in sys/arch/arm/omap/am335x_prcm.c may 
never have been run in its current form.  The 1000 in the first KASSERT() 
should probably be 1000000.
>How-To-Repeat:
Boot a BEAGLEBONE kernel on appropriate hardware (a BB Black, at least).
Notice the sysctl_createv: message.
Notice the missing machdep.frequency.current sysctl node.
>Fix:
I don't know.  I stared at the offending call to sysctl_createv() for a very 
long time and couldn't see anything wrong with it.