Re: Decode dimms on dual socket machines

[Jean Delvare <khali-PUYAD+kWke1g9hUCZPvPmw@public.gmane.org>]

On Wed, 13 Apr 2011 08:13:54 +0200, Michael Lawnick wrote:
> Jean Delvare said the following:
> >> Now I loaded the module
> >> 
> >> test24:/media/i2c-tools/tools # lsmod |grep pca
> >> pca954x                 3288  0
> >> i2c_mux                 2672  1 pca954x
> >> test24:/media/i2c-tools/tools # ll /sys/bus/i2c/devices/
> >> total 0
> >> lrwxrwxrwx 1 root root 0 Apr 12 14:26 0-0050 -> 
> >> ../../../devices/pci0000:00/0000:00:1f.3/i2c-0/0-0050
> >> lrwxrwxrwx 1 root root 0 Apr 12 14:26 0-0052 -> 
> >> ../../../devices/pci0000:00/0000:00:1f.3/i2c-0/0-0052
> >> lrwxrwxrwx 1 root root 0 Apr 12 14:26 0-0054 -> 
> >> ../../../devices/pci0000:00/0000:00:1f.3/i2c-0/0-0054
> >> lrwxrwxrwx 1 root root 0 Apr 12 13:54 i2c-0 -> 
> >> ../../../devices/pci0000:00/0000:00:1f.3/i2c-0
> > 
> > Loading the pca954x driver doesn't do anything. It needs an
> > instantiated device with proper platform data before it will do
> > anything. This can only be done from within the kernel, using some sort
> > of platform initialization code.
> 
> I would expect a
> #> echo "<deviceName> 0x70" > /sys/bus/i2c/devices/i2c-0/new_device
> to span new buses, their numbers taken by random, even without platform
> data (not tried by myself yet, but will have to do it soon).

Hmm, you're right, my memory that we has made platform data mandatory
was wrong, sorry.

> The challenge will be to detect which of the 8 supported devices (
> pca_9540, pca_9542, pca_9543, pca_9544, pca_9545, pca_9546, pca_9547,
> pca_9548) is actually to be used.

In Michael's case it is irrelevant anyway, as the information he found
meanwhile clearly indicates that bus multiplexing is achieved by GPIOs
(most certainly on the ICH10) and not an PCA954x chip. The chip at 0x70
is probably the Intel 5500 IOH.

-- 
Jean Delvare