amd_model_fxx_generate_powernow with dual-socket dual-core K8 system just handles first CPU(?)

[Oskar Enoksson]

I'm trying to make Cool'n'Quiet working on DL145G1 (dual-socket 940-pin 
K8 processors) but I just get FID/VID changes on one of the CPU's (under 
Linux 2.6.37).

Problem seems to be that amd_model_fxx_generate_powernow just generates 
two "Processor" objects in SSDT. I think there should be one for each core.

If I add a factor 2 on row cpu/amd/model_fxx/powernow_acpi.c I get four 
Processor objects in my SSDT and everything works:

-    for (index = 0; index < cmp_cap + 1; index++) {
+    for (index = 0; index < 2*(cmp_cap + 1); index++) {
         len += write_pstates_for_core(Pstate_num, Pstate_feq, Pstate_vid

cmp_cap is correctly detected further up as 1 or 0 depending on the 
dual-core-capability (of the first CPU).

Shouldn't there be a loop over all processor sockets around all this?

[Marc Jones]

On Wed, Oct 5, 2011 at 4:05 PM, Oskar Enoksson  wrote:
> I'm trying to make Cool'n'Quiet working on DL145G1 (dual-socket 940-pin K8
> processors) but I just get FID/VID changes on one of the CPU's (under Linux
> 2.6.37).
>
> Problem seems to be that amd_model_fxx_generate_powernow just generates two
> "Processor" objects in SSDT. I think there should be one for each core.
>
> If I add a factor 2 on row cpu/amd/model_fxx/powernow_acpi.c I get four
> Processor objects in my SSDT and everything works:
> for (index = 0; index < cmp_cap + 1; index++) {
> for (index = 0; index < 2*(cmp_cap + 1); index++) {
>
> len += write_pstates_for_core(Pstate_num, Pstate_feq, Pstate_vid
>
> cmp_cap is correctly detected further up as 1 or 0 depending on the
> dual-core-capability (of the first CPU).
>
> Shouldn't there be a loop over all processor sockets around all this?

Seems so.