Re: [PATCH] tools: add x86_energy_perf_policy to program MSR_IA32_ENERGY_PERF_BIAS

[Ingo Molnar <mingo@elte.hu>]

* Andrew Morton <akpm@linux-foundation.org> wrote:

> On Sat, 23 Oct 2010 00:40:18 -0400 (EDT) Len Brown <lenb@kernel.org> wrote:
> 
> > MSR_IA32_ENERGY_PERF_BIAS first became available on Westmere Xeon.
> > It is implemented in all Sandy Bridge processors -- mobile, desktop and server.
> > It is expected to become increasingly important in subsequent generations.
> > 
> > x86_energy_perf_policy is a user-space utility to set this
> > hardware energy vs performance policy hint in the processor.
> > Most systems would benefit from "x86_energy_perf_policy normal"
> > at system startup, as the hardware default is maximum performance
> > at the expense of energy efficiency.  See the comments
> > in the source code for more information.
> > 
> > Linux-2.6.36 added "epb" to /proc/cpuinfo to indicate
> > if an x86 processor supports MSR_IA32_ENERGY_PERF_BIAS,
> > though the kernel does not actually program the MSR.
> > 
> > In March, Venkatesh Pallipadi proposed a small driver
> > that programmed MSR_IA32_ENERGY_PERF_BIAS, based on
> > the cpufreq governor in use.  It also offered
> > a boot-time cmdline option to override.
> > http://lkml.org/lkml/2010/3/4/457
> > But hiding the hardware policy behind the
> > governor choice was deemed "kinda icky".
> > 
> > So in June, I proposed a generic user/kernel API to
> > consolidate the power/performance policy trade-off.
> > "RFC: /sys/power/policy_preference"
> > http://lkml.org/lkml/2010/6/16/399
> > That is my preference for implementing this capability,
> > but I received no support on the list.
> > 
> > So in September, I sent x86_energy_perf_policy.c to LKML,
> > a user-space utility that scribbles directly to the MSR.
> > http://lkml.org/lkml/2010/9/28/246
> > 
> > Here is the same utility re-sent, this time proposed
> > to reside in the kernel tools directory.
> > 
> > Signed-off-by: Len Brown <len.brown@intel.com>
> > ---
> >  tools/power/x86/x86_energy_perf_policy/Makefile    |    7 +
> >  .../x86_energy_perf_policy.c                       |  358 ++++++++++++++++++++
> >  2 files changed, 365 insertions(+), 0 deletions(-)
> >  create mode 100644 tools/power/x86/x86_energy_perf_policy/Makefile
> >  create mode 100644 tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c
> 
> tools/power/x86, eh?  It seems a better place than under
> Documentation/, where such things have thus far landed!
> 
> I looked briefly, wondering about the kbuild situation.  It doesn't
> appear to be wired up, so one has to manually enter that directory and
> type `make'?
> 
> I guess that's OK as an interim thing but longer-term I suppose we
> should have some more complete build and deployment system.  So
> (thinking out loud) a `make' would invoke a `make tools', and that
> `make tools' would build the tools which are specific to the target
> arch[*], and any generic ones.  And a `make tools_install' would install
> those tools in, I guess, /lib/modules/$(uname -r)/bin.

In terms of build and documentation environment, tools/perf/ has one 
cloned/inherited from Git, which is rather good and functional.

Sharing it with the kernel's build system depends on the kbuild developers being 
interested in it.

Thanks,

	Ingo