[ltp] Re: no C3 state in X - but i do get to C3 with X running but not displayed on screen

[Jeff Lessem]

Uwe Walter <walter-l@tm.uka.de> writes:

> Indeed, I can observe the same behaviour: My T42p CPU does fall into C3
> if uhci_hci is loaded and no external USB (1.1) device is attached.

Not to hijack the thread, but I have the opposite problem.  I would
like to selectively prevent my laptop, a T43, from entering C3 or C4.
Currently, running 2.6.11.5 with the stock acpi + ibm-acpi-0.11, the
laptop will happily enter C3 and C4 when running X (x.org opensource
radeon driver) with the uhci_hcd and such loaded.

The problem is that the clock in vmware (running a Windows XP guest)
runs extremely slowly when the processor enters C3 or C4 (this is the
current theory).  CPU Mhz doesn't effect this, rather the C-state.  I
would like to be able to do something to lock the processor into C1 or
C2, so that vmware is useful.  Currently I run some hi-cpu usage
program at nice +20 when I use vmware, and then the clock in the guest
runs correctly.

I tried "plugging in" the internal bluetooth device, which uses USB
1.1 and is attached to the uhci port, but the laptop still goes into
C3 and C4.  I then tried running hcitool scan, to actually use the
bluetooth radio, and the laptop will still enter C3 or C4 while
running the scan.