From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Rafael J. Wysocki" Subject: [PATCH v3] cpuidle: poll_state: Add time limit to poll_idle() Date: Wed, 14 Mar 2018 15:08:57 +0100 Message-ID: <3111105.SmgpqUHPkp@aspire.rjw.lan> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7Bit Return-path: Sender: linux-kernel-owner@vger.kernel.org To: Linux PM Cc: Peter Zijlstra , Frederic Weisbecker , Thomas Gleixner , Paul McKenney , Thomas Ilsche , Doug Smythies , Rik van Riel , Aubrey Li , Mike Galbraith , LKML List-Id: linux-pm@vger.kernel.org From: Rafael J. Wysocki If poll_idle() is allowed to spin until need_resched() returns 'true', it may actually spin for a much longer time than expected by the idle governor, since set_tsk_need_resched() is not always called by the timer interrupt handler. If that happens, the CPU may spend much more time than anticipated in the "polling" state. To prevent that from happening, limit the time of the spinning loop in poll_idle(). Suggested-by: Peter Zijlstra Signed-off-by: Rafael J. Wysocki --- v2 -> v3: Use local_clock() for time measurements and drop the counter, since that should be lightweight enough (as suggested by Peter). --- drivers/cpuidle/poll_state.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) Index: linux-pm/drivers/cpuidle/poll_state.c =================================================================== --- linux-pm.orig/drivers/cpuidle/poll_state.c +++ linux-pm/drivers/cpuidle/poll_state.c @@ -6,15 +6,24 @@ #include #include +#include #include +#define POLL_IDLE_TIME_LIMIT (TICK_NSEC / 16) + static int __cpuidle poll_idle(struct cpuidle_device *dev, struct cpuidle_driver *drv, int index) { + u64 time_start = local_clock(); + local_irq_enable(); if (!current_set_polling_and_test()) { - while (!need_resched()) + while (!need_resched()) { cpu_relax(); + + if (local_clock() - time_start > POLL_IDLE_TIME_LIMIT) + break; + } } current_clr_polling();