On Sun, Jul 01, 2007 at 08:12:52AM +0200, Bernd Ernesti wrote:
> 
> I added the attached patch to amdpm where I ended up with:
> 

Yes, I did very much the same thing with amdpm.c (I think your code is
nicer though :) on my nforce4 board and I get:

amdpm0 at pci0 dev 1 function 1: NVIDIA nForce4 SMBus (rev. 0xa2)
amdpm_smbus_attach...
LSMB: Picked IRQ 20 with weight 0
linkdev LSMB returned ACPI global int 20
amdpm0: interrupting at ioapic0 pin 20 (irq 0)
iic0 at amdpm0: I2C bus
amdpm0: random number generator enabled (apprx. 0ms)

I tried building with AMDPM_RND_COUNTERS on but I don't seem to be
getting back the array of counters that I should when I do a vmstat -e
so I don't know if the rng is working correctly or not.  I wrote a
driver for a sensor chip that was supposed to be on the I2C bus
according to the motherboard docs.  It _seemed_ to sort of work, I
would get semi believable numbers back from the device but it was not
quite right - I suspect because something else was polling the device
too, messing up my requests.

IMHO it would be better if we fix up amdpm.c to handle the nforce2/3/4
smbus function and update the man page than add another driver which
overlaps functionality with an existing one.

-- 
Brett Lymn