powernowd vs. ondemand governor

[Matthew Garrett]

On Wed, Feb 18, 2009 at 02:15:59AM +0000, Scott James Remnant wrote:
> On Tue, 2009-02-17 at 18:40 +0000, Matthew Garrett wrote:
> > No, acpi-cpufreq should be preferred over any hardware specific drivers 
> > at this point. This wasn't always the case.
> > 
> So we should always try and load acpi-cpufreq first, if that fails,
> maybe fall back on hardware drivers?

Yeah, with the possibility that you may have to work through a set of 
hardware drivers to find one that binds. It can be done with them all 
linked in, but the link order would have to be fixed - something like 
this:

diff --git a/arch/x86/kernel/cpu/cpufreq/Makefile b/arch/x86/kernel/cpu/cpufreq/Makefile
index 560f776..35fffcd 100644
--- a/arch/x86/kernel/cpu/cpufreq/Makefile
+++ b/arch/x86/kernel/cpu/cpufreq/Makefile
@@ -1,3 +1,6 @@
+# Link order matters
+
+obj-$(CONFIG_X86_ACPI_CPUFREQ)		+= acpi-cpufreq.o
 obj-$(CONFIG_X86_POWERNOW_K6)		+= powernow-k6.o
 obj-$(CONFIG_X86_POWERNOW_K7)		+= powernow-k7.o
 obj-$(CONFIG_X86_POWERNOW_K8)		+= powernow-k8.o
@@ -10,7 +13,6 @@ obj-$(CONFIG_X86_GX_SUSPMOD)		+= gx-suspmod.o
 obj-$(CONFIG_X86_SPEEDSTEP_ICH)		+= speedstep-ich.o
 obj-$(CONFIG_X86_SPEEDSTEP_LIB)		+= speedstep-lib.o
 obj-$(CONFIG_X86_SPEEDSTEP_SMI)		+= speedstep-smi.o
-obj-$(CONFIG_X86_ACPI_CPUFREQ)		+= acpi-cpufreq.o
 obj-$(CONFIG_X86_SPEEDSTEP_CENTRINO)	+= speedstep-centrino.o
 obj-$(CONFIG_X86_P4_CLOCKMOD)		+= p4-clockmod.o
 obj-$(CONFIG_X86_CPUFREQ_NFORCE2)	+= cpufreq-nforce2.o

would probably do the job.

-- 
Matthew Garrett | mjg59 at srcf.ucam.org