From mboxrd@z Thu Jan 1 00:00:00 1970 From: Viresh Kumar Subject: [PATCH 0/3] cpufreq: schedutil: Fix 4.12 regressions Date: Fri, 9 Jun 2017 15:45:53 +0530 Message-ID: Return-path: Received: from mail-pg0-f46.google.com ([74.125.83.46]:34840 "EHLO mail-pg0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751524AbdFIKQG (ORCPT ); Fri, 9 Jun 2017 06:16:06 -0400 Received: by mail-pg0-f46.google.com with SMTP id k71so25273000pgd.2 for ; Fri, 09 Jun 2017 03:16:06 -0700 (PDT) Sender: linux-pm-owner@vger.kernel.org List-Id: linux-pm@vger.kernel.org To: Rafael Wysocki Cc: Viresh Kumar , linaro-kernel@lists.linaro.org, linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org, Vincent Guittot , Juri Lelli , Ingo Molnar , Peter Zijlstra , patrick.bellasi@arm.com, john.ettedgui@gmail.com, Srinivas Pandruvada , Joel Fernandes , Morten Rasmussen Hi Rafael, I have identified some regressions with the schedutil governor which happen due to one of your patches that got merged in 4.12-rc1. This series fixes all the drivers which provide a ->target_index() callback but doesn't fix the drivers which provide ->target() callback. Such platforms need to implement the ->resolve_freq() callback in order to get this fixed and I only had hardware for testing intel_pstate, which I fixed in this series. I am wondering if there is another way to fix this issue (than what I tried) or if we should revert the offending commit (39b64aa1c007) and look for other solutions. Anyway, this series has the necessary patches to fix it. Viresh Kumar (3): cpufreq: schedutil: Restore cached_raw_freq behavior cpufreq: schedutil: Fix selection algorithm while reducing frequency cpufreq: intel_pstate: Provide resolve_freq() to fix regression drivers/cpufreq/intel_pstate.c | 14 +++++++++++++ kernel/sched/cpufreq_schedutil.c | 45 +++++++++++++++++++++++++++++++++------- 2 files changed, 52 insertions(+), 7 deletions(-) -- 2.13.0.70.g6367777092d9