From patchwork Wed Dec 2 21:19:16 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ben Gamari X-Patchwork-Id: 7753431 Return-Path: X-Original-To: patchwork-linux-samsung-soc@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork2.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork2.web.kernel.org (Postfix) with ESMTP id 860A8BEEE1 for ; Wed, 2 Dec 2015 21:33:37 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 5247420546 for ; Wed, 2 Dec 2015 21:33:36 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id C29812053F for ; Wed, 2 Dec 2015 21:33:34 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757842AbbLBVdC (ORCPT ); Wed, 2 Dec 2015 16:33:02 -0500 Received: from mail.smart-cactus.org ([54.187.36.80]:50827 "EHLO mail.smart-cactus.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752536AbbLBV3X (ORCPT ); Wed, 2 Dec 2015 16:29:23 -0500 Received: from localhost.localdomain (HSI-KBW-149-172-43-78.hsi13.kabel-badenwuerttemberg.de [149.172.43.78]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-SHA256 (128/128 bits)) (Client did not present a certificate) (Authenticated sender: ben@smart-cactus.org) by mail.smart-cactus.org (Postfix) with ESMTPSA id B192F64B4F; Wed, 2 Dec 2015 21:07:16 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=mail.smart-cactus.org; s=mail; t=1449090442; bh=7+/HZEbQy9cG644kHo2qb5+6/w+53lvsY+TfnfOjCy4=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=Dz2t6xvw3Yt4NwfPENdx1UTGX/RUiyFzWNr1Oz+Oii8SIXiZpefaSFxBqTUlUGKZS vIJHhtAWhzNA8ooDQgUjsCr20BzgTNuUBInkGP4DvscuoOQ37/yhGzuInKTAnvsR1k /c+vT6zIT1l8kQWga+hSiVVbTsuNyUTQlk4o3pyU= From: Ben Gamari To: Thomas Abraham , Sylwester Nawrocki , Michael Turquette , Kukjin Kim , Kukjin Kim , Viresh Kumar , Krzysztof Kozlowski Cc: Tomasz Figa , Lukasz Majewski , Heiko Stuebner , Chanwoo Choi , Kevin Hilman , Javier Martinez Canillas , Tobias Jakobi , Anand Moon , linux-samsung-soc@vger.kernel.org, linux-clk@vger.kernel.org, linux-pm@vger.kernel.org, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, b.zolnierkie@samsung.com, Doug Anderson , Javier Martinez Canillas , Andreas Faerber , Sachin Kamat , Ben Gamari Subject: [PATCH 01/12] cpufreq: arm_big_little: add cluster regulator support Date: Wed, 2 Dec 2015 22:19:16 +0100 Message-Id: <1449091167-20758-2-git-send-email-ben@smart-cactus.org> X-Mailer: git-send-email 2.6.2 In-Reply-To: <1449091167-20758-1-git-send-email-ben@smart-cactus.org> References: <1449091167-20758-1-git-send-email-ben@smart-cactus.org> Sender: linux-samsung-soc-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-samsung-soc@vger.kernel.org X-Spam-Status: No, score=-6.8 required=5.0 tests=BAYES_00,DKIM_SIGNED, RCVD_IN_DNSWL_HI,T_DKIM_INVALID,T_RP_MATCHES_RCVD,UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP From: Bartlomiej Zolnierkiewicz Add cluster regulator support as a preparation to adding generic arm_big_little_dt cpufreq_dt driver support for ODROID-XU3 board. This allows arm_big_little[_dt] driver to set not only the frequency but also the voltage (which is obtained from operating point's voltage value) for CPU clusters. Cc: Kukjin Kim Cc: Doug Anderson Cc: Javier Martinez Canillas Cc: Andreas Faerber Cc: Sachin Kamat Cc: Thomas Abraham Signed-off-by: Bartlomiej Zolnierkiewicz Signed-off-by: Ben Gamari --- .../bindings/cpufreq/arm_big_little_dt.txt | 4 + drivers/cpufreq/arm_big_little.c | 155 ++++++++++++++++++--- 2 files changed, 142 insertions(+), 17 deletions(-) diff --git a/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt index 0715695..8ca4a12 100644 --- a/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt +++ b/Documentation/devicetree/bindings/cpufreq/arm_big_little_dt.txt @@ -18,6 +18,10 @@ Required properties: Optional properties: - clock-latency: Specify the possible maximum transition latency for clock, in unit of nanoseconds. +- cpu-cluster.0-supply: Provides the regulator node supplying voltage to CPU + cluster 0. +- cpu-cluster.1-supply: Provides the regulator node supplying voltage to CPU + cluster 1. Examples: diff --git a/drivers/cpufreq/arm_big_little.c b/drivers/cpufreq/arm_big_little.c index c5d256c..855599b 100644 --- a/drivers/cpufreq/arm_big_little.c +++ b/drivers/cpufreq/arm_big_little.c @@ -31,6 +31,7 @@ #include #include #include +#include #include "arm_big_little.h" @@ -57,6 +58,9 @@ static bool bL_switching_enabled; static struct cpufreq_arm_bL_ops *arm_bL_ops; static struct clk *clk[MAX_CLUSTERS]; +static struct regulator *reg[MAX_CLUSTERS]; +static struct device *cpu_devs[MAX_CLUSTERS]; +static int transition_latencies[MAX_CLUSTERS]; static struct cpufreq_frequency_table *freq_table[MAX_CLUSTERS + 1]; static atomic_t cluster_usage[MAX_CLUSTERS + 1]; @@ -125,6 +129,75 @@ static unsigned int bL_cpufreq_get_rate(unsigned int cpu) } } +static int +bL_cpufreq_set_rate_cluster(u32 cpu, u32 cluster, u32 new_rate) +{ + unsigned long volt = 0, volt_old = 0; + long freq_Hz; + u32 old_rate; + int ret; + + freq_Hz = new_rate * 1000; + old_rate = clk_get_rate(clk[cluster]) / 1000; + + if (!IS_ERR(reg[cluster])) { + struct dev_pm_opp *opp; + unsigned long opp_freq; + + rcu_read_lock(); + opp = dev_pm_opp_find_freq_ceil(cpu_devs[cluster], &freq_Hz); + if (IS_ERR(opp)) { + rcu_read_unlock(); + pr_err("%s: cpu %d, cluster: %d, failed to find OPP for %ld\n", + __func__, cpu, cluster, freq_Hz); + return PTR_ERR(opp); + } + volt = dev_pm_opp_get_voltage(opp); + opp_freq = dev_pm_opp_get_freq(opp); + rcu_read_unlock(); + volt_old = regulator_get_voltage(reg[cluster]); + pr_debug("%s: cpu %d, cluster: %d, Found OPP: %ld kHz, %ld uV\n", + __func__, cpu, cluster, opp_freq / 1000, volt); + } + + pr_debug("%s: cpu %d, cluster: %d, %u MHz, %ld mV --> %u MHz, %ld mV\n", + __func__, cpu, cluster, + old_rate / 1000, (volt_old > 0) ? volt_old / 1000 : -1, + new_rate / 1000, volt ? volt / 1000 : -1); + + /* scaling up? scale voltage before frequency */ + if (!IS_ERR(reg[cluster]) && new_rate > old_rate) { + ret = regulator_set_voltage_tol(reg[cluster], volt, 0); + if (ret) { + pr_err("%s: cpu: %d, cluster: %d, failed to scale voltage up: %d\n", + __func__, cpu, cluster, ret); + return ret; + } + } + + ret = clk_set_rate(clk[cluster], new_rate * 1000); + if (WARN_ON(ret)) { + pr_err("%s: clk_set_rate failed: %d, cluster: %d\n", + __func__, cluster, ret); + if (!IS_ERR(reg[cluster]) && volt_old > 0) + regulator_set_voltage_tol(reg[cluster], volt_old, 0); + return ret; + } + + /* scaling down? scale voltage after frequency */ + if (!IS_ERR(reg[cluster]) && new_rate < old_rate) { + ret = regulator_set_voltage_tol(reg[cluster], volt, 0); + if (ret) { + pr_err("%s: cpu: %d, cluster: %d, failed to scale voltage down: %d\n", + __func__, cpu, cluster, ret); + clk_set_rate(clk[cluster], old_rate * 1000); + return ret; + } + } + + return 0; +} + static unsigned int bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate) { @@ -148,7 +221,7 @@ bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate) pr_debug("%s: cpu: %d, old cluster: %d, new cluster: %d, freq: %d\n", __func__, cpu, old_cluster, new_cluster, new_rate); - ret = clk_set_rate(clk[new_cluster], new_rate * 1000); + ret = bL_cpufreq_set_rate_cluster(cpu, new_cluster, new_rate); if (!ret) { /* * FIXME: clk_set_rate hasn't returned an error here however it @@ -160,22 +233,15 @@ bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate) */ if (clk_get_rate(clk[new_cluster]) != new_rate * 1000) ret = -EIO; - } - - if (WARN_ON(ret)) { - pr_err("clk_set_rate failed: %d, new cluster: %d\n", ret, - new_cluster); - if (bLs) { - per_cpu(cpu_last_req_freq, cpu) = prev_rate; - per_cpu(physical_cluster, cpu) = old_cluster; - } + } else if (ret && bLs) { + per_cpu(cpu_last_req_freq, cpu) = prev_rate; + per_cpu(physical_cluster, cpu) = old_cluster; + } - mutex_unlock(&cluster_lock[new_cluster]); + mutex_unlock(&cluster_lock[new_cluster]); + if(ret) return ret; - } - - mutex_unlock(&cluster_lock[new_cluster]); /* Recalc freq for old cluster when switching clusters */ if (old_cluster != new_cluster) { @@ -195,9 +261,10 @@ bL_cpufreq_set_rate(u32 cpu, u32 old_cluster, u32 new_cluster, u32 rate) pr_debug("%s: Updating rate of old cluster: %d, to freq: %d\n", __func__, old_cluster, new_rate); - if (clk_set_rate(clk[old_cluster], new_rate * 1000)) - pr_err("%s: clk_set_rate failed: %d, old cluster: %d\n", - __func__, ret, old_cluster); + if (bL_cpufreq_set_rate_cluster(cpu, old_cluster, new_rate)) { + pr_err("%s: bL_cpufreq_set_rate_cluster failed\n", + __func__); + } } mutex_unlock(&cluster_lock[old_cluster]); } @@ -304,6 +371,8 @@ static void _put_cluster_clk_and_freq_table(struct device *cpu_dev) return; clk_put(clk[cluster]); + if (!IS_ERR(reg[cluster])) + regulator_put(reg[cluster]); dev_pm_opp_free_cpufreq_table(cpu_dev, &freq_table[cluster]); if (arm_bL_ops->free_opp_table) arm_bL_ops->free_opp_table(cpu_dev); @@ -337,7 +406,9 @@ static void put_cluster_clk_and_freq_table(struct device *cpu_dev) static int _get_cluster_clk_and_freq_table(struct device *cpu_dev) { + unsigned long min_uV = ~0, max_uV = 0; u32 cluster = raw_cpu_to_cluster(cpu_dev->id); + char name[14] = "cpu-cluster."; int ret; if (freq_table[cluster]) @@ -350,6 +421,51 @@ static int _get_cluster_clk_and_freq_table(struct device *cpu_dev) goto out; } + name[12] = cluster + '0'; + reg[cluster] = regulator_get_optional(cpu_dev, name); + if (!IS_ERR(reg[cluster])) { + unsigned long opp_freq = 0; + + dev_dbg(cpu_dev, "%s: reg: %p, cluster: %d\n", + __func__, reg[cluster], cluster); + cpu_devs[cluster] = cpu_dev; + + /* + * Disable any OPPs where the connected regulator isn't able to + * provide the specified voltage and record minimum and maximum + * voltage levels. + */ + while (1) { + struct dev_pm_opp *opp; + unsigned long opp_uV; + + rcu_read_lock(); + opp = dev_pm_opp_find_freq_ceil(cpu_dev, &opp_freq); + if (IS_ERR(opp)) { + rcu_read_unlock(); + break; + } + opp_uV = dev_pm_opp_get_voltage(opp); + rcu_read_unlock(); + + if (regulator_is_supported_voltage(reg[cluster], opp_uV, + opp_uV)) { + if (opp_uV < min_uV) + min_uV = opp_uV; + if (opp_uV > max_uV) + max_uV = opp_uV; + } else { + dev_pm_opp_disable(cpu_dev, opp_freq); + } + + opp_freq++; + } + + ret = regulator_set_voltage_time(reg[cluster], min_uV, max_uV); + if (ret > 0) + transition_latencies[cluster] = ret * 1000; + } + ret = dev_pm_opp_init_cpufreq_table(cpu_dev, &freq_table[cluster]); if (ret) { dev_err(cpu_dev, "%s: failed to init cpufreq table, cpu: %d, err: %d\n", @@ -483,6 +599,11 @@ static int bL_cpufreq_init(struct cpufreq_policy *policy) else policy->cpuinfo.transition_latency = CPUFREQ_ETERNAL; + if (cur_cluster < MAX_CLUSTERS && + policy->cpuinfo.transition_latency != CPUFREQ_ETERNAL) + policy->cpuinfo.transition_latency + += transition_latencies[cur_cluster]; + if (is_bL_switching_enabled()) per_cpu(cpu_last_req_freq, policy->cpu) = clk_get_cpu_rate(policy->cpu);