ACPI / CPPC: Add a CPUFreq driver for use with CPPC
authorAshwin Chaugule <ashwin.chaugule@linaro.org>
Fri, 2 Oct 2015 14:04:01 +0000 (10:04 -0400)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Mon, 12 Oct 2015 21:04:31 +0000 (23:04 +0200)
This driver utilizes the methods introduced in a previous
patch titled - "ACPI: Introduce CPU performance controls using CPPC"
and enables usage with existing CPUFreq governors.

Signed-off-by: Ashwin Chaugule <ashwin.chaugule@linaro.org>
Reviewed-by: Al Stone <al.stone@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/cpufreq/Kconfig.arm
drivers/cpufreq/Makefile
drivers/cpufreq/cppc_cpufreq.c [new file with mode: 0644]

index cd0391e46c6dbc6163819b533ce225f3afe898e8..642fd49793b0a9284a83eddb3275ff627ca46763 100644 (file)
@@ -227,3 +227,20 @@ config ARM_PXA2xx_CPUFREQ
          This add the CPUFreq driver support for Intel PXA2xx SOCs.
 
          If in doubt, say N.
+
+config ACPI_CPPC_CPUFREQ
+       tristate "CPUFreq driver based on the ACPI CPPC spec"
+       depends on ACPI
+       select ACPI_CPPC_LIB
+       default n
+       help
+         This adds a CPUFreq driver which uses CPPC methods
+         as described in the ACPIv5.1 spec. CPPC stands for
+         Collaborative Processor Performance Controls. It
+         is based on an abstract continuous scale of CPU
+         performance values which allows the remote power
+         processor to flexibly optimize for power and
+         performance. CPPC relies on power management firmware
+         support for its operation.
+
+         If in doubt, say N.
index 41340384f11f291c4fe93e96e73559926f4f59b1..47a7ebbcf998b1ff9e13eedc6647bba428c3e44f 100644 (file)
@@ -76,6 +76,8 @@ obj-$(CONFIG_ARM_SPEAR_CPUFREQ)               += spear-cpufreq.o
 obj-$(CONFIG_ARM_TEGRA20_CPUFREQ)      += tegra20-cpufreq.o
 obj-$(CONFIG_ARM_TEGRA124_CPUFREQ)     += tegra124-cpufreq.o
 obj-$(CONFIG_ARM_VEXPRESS_SPC_CPUFREQ) += vexpress-spc-cpufreq.o
+obj-$(CONFIG_ACPI_CPPC_CPUFREQ) += cppc_cpufreq.o
+
 
 ##################################################################################
 # PowerPC platform drivers
diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
new file mode 100644 (file)
index 0000000..93c219f
--- /dev/null
@@ -0,0 +1,176 @@
+/*
+ * CPPC (Collaborative Processor Performance Control) driver for
+ * interfacing with the CPUfreq layer and governors. See
+ * cppc_acpi.c for CPPC specific methods.
+ *
+ * (C) Copyright 2014, 2015 Linaro Ltd.
+ * Author: Ashwin Chaugule <ashwin.chaugule@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+#define pr_fmt(fmt)    "CPPC Cpufreq:" fmt
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/delay.h>
+#include <linux/cpu.h>
+#include <linux/cpufreq.h>
+#include <linux/vmalloc.h>
+
+#include <acpi/cppc_acpi.h>
+
+/*
+ * These structs contain information parsed from per CPU
+ * ACPI _CPC structures.
+ * e.g. For each CPU the highest, lowest supported
+ * performance capabilities, desired performance level
+ * requested etc.
+ */
+static struct cpudata **all_cpu_data;
+
+static int cppc_cpufreq_set_target(struct cpufreq_policy *policy,
+               unsigned int target_freq,
+               unsigned int relation)
+{
+       struct cpudata *cpu;
+       struct cpufreq_freqs freqs;
+       int ret = 0;
+
+       cpu = all_cpu_data[policy->cpu];
+
+       cpu->perf_ctrls.desired_perf = target_freq;
+       freqs.old = policy->cur;
+       freqs.new = target_freq;
+
+       cpufreq_freq_transition_begin(policy, &freqs);
+       ret = cppc_set_perf(cpu->cpu, &cpu->perf_ctrls);
+       cpufreq_freq_transition_end(policy, &freqs, ret != 0);
+
+       if (ret)
+               pr_debug("Failed to set target on CPU:%d. ret:%d\n",
+                               cpu->cpu, ret);
+
+       return ret;
+}
+
+static int cppc_verify_policy(struct cpufreq_policy *policy)
+{
+       cpufreq_verify_within_cpu_limits(policy);
+       return 0;
+}
+
+static void cppc_cpufreq_stop_cpu(struct cpufreq_policy *policy)
+{
+       int cpu_num = policy->cpu;
+       struct cpudata *cpu = all_cpu_data[cpu_num];
+       int ret;
+
+       cpu->perf_ctrls.desired_perf = cpu->perf_caps.lowest_perf;
+
+       ret = cppc_set_perf(cpu_num, &cpu->perf_ctrls);
+       if (ret)
+               pr_debug("Err setting perf value:%d on CPU:%d. ret:%d\n",
+                               cpu->perf_caps.lowest_perf, cpu_num, ret);
+}
+
+static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
+{
+       struct cpudata *cpu;
+       unsigned int cpu_num = policy->cpu;
+       int ret = 0;
+
+       cpu = all_cpu_data[policy->cpu];
+
+       cpu->cpu = cpu_num;
+       ret = cppc_get_perf_caps(policy->cpu, &cpu->perf_caps);
+
+       if (ret) {
+               pr_debug("Err reading CPU%d perf capabilities. ret:%d\n",
+                               cpu_num, ret);
+               return ret;
+       }
+
+       policy->min = cpu->perf_caps.lowest_perf;
+       policy->max = cpu->perf_caps.highest_perf;
+       policy->cpuinfo.min_freq = policy->min;
+       policy->cpuinfo.max_freq = policy->max;
+
+       if (policy->shared_type == CPUFREQ_SHARED_TYPE_ANY)
+               cpumask_copy(policy->cpus, cpu->shared_cpu_map);
+       else {
+               /* Support only SW_ANY for now. */
+               pr_debug("Unsupported CPU co-ord type\n");
+               return -EFAULT;
+       }
+
+       cpumask_set_cpu(policy->cpu, policy->cpus);
+       cpu->cur_policy = policy;
+
+       /* Set policy->cur to max now. The governors will adjust later. */
+       policy->cur = cpu->perf_ctrls.desired_perf = cpu->perf_caps.highest_perf;
+
+       ret = cppc_set_perf(cpu_num, &cpu->perf_ctrls);
+       if (ret)
+               pr_debug("Err setting perf value:%d on CPU:%d. ret:%d\n",
+                               cpu->perf_caps.highest_perf, cpu_num, ret);
+
+       return ret;
+}
+
+static struct cpufreq_driver cppc_cpufreq_driver = {
+       .flags = CPUFREQ_CONST_LOOPS,
+       .verify = cppc_verify_policy,
+       .target = cppc_cpufreq_set_target,
+       .init = cppc_cpufreq_cpu_init,
+       .stop_cpu = cppc_cpufreq_stop_cpu,
+       .name = "cppc_cpufreq",
+};
+
+static int __init cppc_cpufreq_init(void)
+{
+       int i, ret = 0;
+       struct cpudata *cpu;
+
+       if (acpi_disabled)
+               return -ENODEV;
+
+       all_cpu_data = kzalloc(sizeof(void *) * num_possible_cpus(), GFP_KERNEL);
+       if (!all_cpu_data)
+               return -ENOMEM;
+
+       for_each_possible_cpu(i) {
+               all_cpu_data[i] = kzalloc(sizeof(struct cpudata), GFP_KERNEL);
+               if (!all_cpu_data[i])
+                       goto out;
+
+               cpu = all_cpu_data[i];
+               if (!zalloc_cpumask_var(&cpu->shared_cpu_map, GFP_KERNEL))
+                       goto out;
+       }
+
+       ret = acpi_get_psd_map(all_cpu_data);
+       if (ret) {
+               pr_debug("Error parsing PSD data. Aborting cpufreq registration.\n");
+               goto out;
+       }
+
+       ret = cpufreq_register_driver(&cppc_cpufreq_driver);
+       if (ret)
+               goto out;
+
+       return ret;
+
+out:
+       for_each_possible_cpu(i)
+               if (all_cpu_data[i])
+                       kfree(all_cpu_data[i]);
+
+       kfree(all_cpu_data);
+       return -ENODEV;
+}
+
+late_initcall(cppc_cpufreq_init);
This page took 0.026874 seconds and 5 git commands to generate.