From: Brad Peters <bpeters@redhat.com>
Date: Tue, 26 Aug 2008 14:41:06 -0400
Subject: [ppc] export LPAR CPU utilization stats for use by hv
Message-id: 20080826184106.12189.82753.sendpatchset@squad5-lp1.lab.bos.redhat.com
O-Subject: [PATCH RHEL5.3 439516] Export LPAR CPU utilization stats for use by hypervisor
Bugzilla: 439516
RH-Acked-by: David Howells <dhowells@redhat.com>

RHBZ#:
======
https://bugzilla.redhat.com/show_bug.cgi?id=439516

Description:
===========
This patch adds OS support to exporting CPU utilization statistics, with
the goal of permitting the hypervisor to adjust workloads and implement
energy efficiency policies.

The hypervisor can look at the value in the wait_state_cycles field of
the VPA for an estimate of how busy dedicated processors are.
Currently, as the kernel never touches this field, we appear to be
100% busy.

This patch records the time duration the kernel spends in powersave and
passes that to the hypervisor to provide a reasonable indication of
utilisation.

RHEL Version Found:
================
RHEL 5.2

kABI Status:
============
No symbols were harmed.

Brew:
=====
Built on all platforms.
http://brewweb.devel.redhat.com/brew/taskinfo?taskID=1371092

Upstream Status:
================
Committed:
http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=96366a8d3f17ee964802378d4a1f51fda86f070e

Test Status:
============
Tony Breeds <IBM> confirmed the patch works on the test kernel

===============================================================

Brad Peters 1-978-392-1000 x 23183
IBM on-site partner.

Proposed Patch:
===============
This patch is based on 2.6.18-105.el5

diff --git a/arch/powerpc/platforms/pseries/setup.c b/arch/powerpc/platforms/pseries/setup.c
index 32d59c3..acef34e 100644
--- a/arch/powerpc/platforms/pseries/setup.c
+++ b/arch/powerpc/platforms/pseries/setup.c
@@ -473,6 +473,7 @@ static void pseries_dedicated_idle_sleep(void)
 	unsigned int cpu = smp_processor_id();
 	unsigned long start_snooze;
 	unsigned long *smt_snooze_delay = &__get_cpu_var(smt_snooze_delay);
+	unsigned long in_purr, out_purr;
 
 	/*
 	 * Indicate to the HV that we are idle. Now would be
@@ -480,6 +481,7 @@ static void pseries_dedicated_idle_sleep(void)
 	 */
 	get_lppaca()->idle = 1;
 	get_lppaca()->cpuctls_task_attrs = 1;
+	in_purr = mfspr(SPRN_PURR);
 
 	/*
 	 * We come in with interrupts disabled, and need_resched()
@@ -513,6 +515,8 @@ static void pseries_dedicated_idle_sleep(void)
 out:
 	HMT_medium();
 	get_lppaca()->cpuctls_task_attrs = 0;
+	out_purr = mfspr(SPRN_PURR);
+	get_lppaca()->wait_state_cycles += out_purr - in_purr;
 	get_lppaca()->idle = 0;
 }