From mboxrd@z Thu Jan 1 00:00:00 1970 From: Preeti U Murthy Subject: Re: [PATCH v2] cpuidle: powernv/pseries: Auto-promotion of snooze to deeper idle state Date: Fri, 19 Jun 2015 11:07:10 +0530 Message-ID: <5583AA86.9010203@linux.vnet.ibm.com> References: <1434626591-14509-1-git-send-email-shilpa.bhat@linux.vnet.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: Received: from e17.ny.us.ibm.com ([129.33.205.207]:44185 "EHLO e17.ny.us.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751777AbbFSFhS (ORCPT ); Fri, 19 Jun 2015 01:37:18 -0400 Received: from /spool/local by e17.ny.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Fri, 19 Jun 2015 01:37:17 -0400 In-Reply-To: <1434626591-14509-1-git-send-email-shilpa.bhat@linux.vnet.ibm.com> Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Shilpasri G Bhat , rjw@rjwysocki.net Cc: linux-pm@vger.kernel.org, daniel.lezcano@linaro.org, linux-kernel@vger.kernel.org, linuxppc-dev@lists.ozlabs.org On 06/18/2015 04:53 PM, Shilpasri G Bhat wrote: > The idle cpus which stay in snooze for a long period can degrade the > perfomance of the sibling cpus. If the cpu stays in snooze for more > than target residency of the next available idle state, then exit from > snooze. This gives a chance to the cpuidle governor to re-evaluate the > last idle state of the cpu to promote it to deeper idle states. > > Signed-off-by: Shilpasri G Bhat > --- Reviewed-by: Preeti U Murthy > Changes from v1: > -Modified commit message > > drivers/cpuidle/cpuidle-powernv.c | 12 ++++++++++++ > drivers/cpuidle/cpuidle-pseries.c | 11 +++++++++++ > 2 files changed, 23 insertions(+) > > diff --git a/drivers/cpuidle/cpuidle-powernv.c b/drivers/cpuidle/cpuidle-powernv.c > index 5937207..1e3ef5e 100644 > --- a/drivers/cpuidle/cpuidle-powernv.c > +++ b/drivers/cpuidle/cpuidle-powernv.c > @@ -29,18 +29,25 @@ struct cpuidle_driver powernv_idle_driver = { > > static int max_idle_state; > static struct cpuidle_state *cpuidle_state_table; > +static u64 snooze_timeout; > +static bool snooze_timeout_en; > > static int snooze_loop(struct cpuidle_device *dev, > struct cpuidle_driver *drv, > int index) > { > + u64 snooze_exit_time; > + > local_irq_enable(); > set_thread_flag(TIF_POLLING_NRFLAG); > > + snooze_exit_time = get_tb() + snooze_timeout; > ppc64_runlatch_off(); > while (!need_resched()) { > HMT_low(); > HMT_very_low(); > + if (snooze_timeout_en && get_tb() > snooze_exit_time) > + break; > } > > HMT_medium(); > @@ -252,6 +259,11 @@ static int powernv_idle_probe(void) > cpuidle_state_table = powernv_states; > /* Device tree can indicate more idle states */ > max_idle_state = powernv_add_idle_states(); > + if (max_idle_state > 1) { > + snooze_timeout_en = true; > + snooze_timeout = powernv_states[1].target_residency * > + tb_ticks_per_usec; > + } > } else > return -ENODEV; > > diff --git a/drivers/cpuidle/cpuidle-pseries.c b/drivers/cpuidle/cpuidle-pseries.c > index bb9e2b6..07135e0 100644 > --- a/drivers/cpuidle/cpuidle-pseries.c > +++ b/drivers/cpuidle/cpuidle-pseries.c > @@ -27,6 +27,8 @@ struct cpuidle_driver pseries_idle_driver = { > > static int max_idle_state; > static struct cpuidle_state *cpuidle_state_table; > +static u64 snooze_timeout; > +static bool snooze_timeout_en; > > static inline void idle_loop_prolog(unsigned long *in_purr) > { > @@ -58,14 +60,18 @@ static int snooze_loop(struct cpuidle_device *dev, > int index) > { > unsigned long in_purr; > + u64 snooze_exit_time; > > idle_loop_prolog(&in_purr); > local_irq_enable(); > set_thread_flag(TIF_POLLING_NRFLAG); > + snooze_exit_time = get_tb() + snooze_timeout; > > while (!need_resched()) { > HMT_low(); > HMT_very_low(); > + if (snooze_timeout_en && get_tb() > snooze_exit_time) > + break; > } > > HMT_medium(); > @@ -244,6 +250,11 @@ static int pseries_idle_probe(void) > } else > return -ENODEV; > > + if (max_idle_state > 1) { > + snooze_timeout_en = true; > + snooze_timeout = cpuidle_state_table[1].target_residency * > + tb_ticks_per_usec; > + } > return 0; > } > >