Hi all,

I recently spent some time with an OMAP4460 based Variscite VAR-STK-OM44 and
latest Linux releases and latest linux-omap Git. The evaluation kit runs pretty
well unless I try to enable cpufeq. This causes a crash with quite random
backtraces right on boot within a few seconds after entering userspace.

It turned out that this happens when the board switches to 920MHz. Commenting
out this operating point in the devicetree fixes this issue... but limits
performance quite a lot.

I did some research and I'd say its related to problems with voltage regulation.
I can watch the scaling driver changing the clock but for some reason the
voltage does not seem to get adjusted.
Might this behaviour be related to the OMAP infrastructure changes?

There are some messages at boot which might be related:

twl: not initialized
[    2.148742] twl6030_uv_to_vsel:OUT OF RANGE! non mapped vsel for 1375000 Vs
max 1316660
[    2.156799] twl6030_uv_to_vsel:OUT OF RANGE! non mapped vsel for 1375000 Vs
max 1316660
...

There is a little patch around that changes the initialisation order which makes
this messages go away but does not fix the odd behaviour. So maybe this is just
a cosmetic issue and the actual problem is somewhere else.

I have some kernel experience but not that much with the latest OMAP development
and therefore would be happy about some pointers what might be wrong. A boot log
is attached.

Greetings

Florian