Merge back cpufreq material for 4.19.

Documentation/devicetree/bindings/arm/marvell/armada-37xx.txt

@@ -33,3 +33,18 @@ nb_pm: syscon@14000 {
 	compatible = "marvell,armada-3700-nb-pm", "syscon";
 	reg = <0x14000 0x60>;
 }
+
+AVS
+---
+
+For AVS an other component is needed:
+
+Required properties:
+- compatible     : should contain "marvell,armada-3700-avs", "syscon";
+- reg            : the register start and length for the AVS
+
+Example:
+avs: avs@11500 {
+	compatible = "marvell,armada-3700-avs", "syscon";
+	reg = <0x11500 0x40>;
+}

Documentation/trace/events-power.rst

@@ -27,6 +27,7 @@ cpufreq.
 
   cpu_idle		"state=%lu cpu_id=%lu"
   cpu_frequency		"state=%lu cpu_id=%lu"
+  cpu_frequency_limits	"min=%lu max=%lu cpu_id=%lu"
 
 A suspend event is used to indicate the system going in and out of the
 suspend mode:

drivers/cpufreq/armada-37xx-cpufreq.c

@@ -51,6 +51,16 @@
 #define  ARMADA_37XX_DVFS_LOAD_2	2
 #define  ARMADA_37XX_DVFS_LOAD_3	3
 
+/* AVS register set */
+#define ARMADA_37XX_AVS_CTL0		0x0
+#define	 ARMADA_37XX_AVS_ENABLE		BIT(30)
+#define	 ARMADA_37XX_AVS_HIGH_VDD_LIMIT	16
+#define	 ARMADA_37XX_AVS_LOW_VDD_LIMIT	22
+#define	 ARMADA_37XX_AVS_VDD_MASK	0x3F
+#define ARMADA_37XX_AVS_CTL2		0x8
+#define	 ARMADA_37XX_AVS_LOW_VDD_EN	BIT(6)
+#define ARMADA_37XX_AVS_VSET(x)	    (0x1C + 4 * (x))
+
 /*
  * On Armada 37xx the Power management manages 4 level of CPU load,
  * each level can be associated with a CPU clock source, a CPU
@@ -58,6 +68,17 @@
  */
 #define LOAD_LEVEL_NR	4
 
+#define MIN_VOLT_MV 1000
+
+/*  AVS value for the corresponding voltage (in mV) */
+static int avs_map[] = {
+	747, 758, 770, 782, 793, 805, 817, 828, 840, 852, 863, 875, 887, 898,
+	910, 922, 933, 945, 957, 968, 980, 992, 1003, 1015, 1027, 1038, 1050,
+	1062, 1073, 1085, 1097, 1108, 1120, 1132, 1143, 1155, 1167, 1178, 1190,
+	1202, 1213, 1225, 1237, 1248, 1260, 1272, 1283, 1295, 1307, 1318, 1330,
+	1342
+};
+
 struct armada37xx_cpufreq_state {
 	struct regmap *regmap;
 	u32 nb_l0l1;
@@ -71,6 +92,7 @@ static struct armada37xx_cpufreq_state *armada37xx_cpufreq_state;
 struct armada_37xx_dvfs {
 	u32 cpu_freq_max;
 	u8 divider[LOAD_LEVEL_NR];
+	u32 avs[LOAD_LEVEL_NR];
 };
 
 static struct armada_37xx_dvfs armada_37xx_dvfs[] = {
@@ -148,6 +170,128 @@ static void __init armada37xx_cpufreq_dvfs_setup(struct regmap *base,
 	clk_set_parent(clk, parent);
 }
 
+/*
+ * Find out the armada 37x supported AVS value whose voltage value is
+ * the round-up closest to the target voltage value.
+ */
+static u32 armada_37xx_avs_val_match(int target_vm)
+{
+	u32 avs;
+
+	/* Find out the round-up closest supported voltage value */
+	for (avs = 0; avs < ARRAY_SIZE(avs_map); avs++)
+		if (avs_map[avs] >= target_vm)
+			break;
+
+	/*
+	 * If all supported voltages are smaller than target one,
+	 * choose the largest supported voltage
+	 */
+	if (avs == ARRAY_SIZE(avs_map))
+		avs = ARRAY_SIZE(avs_map) - 1;
+
+	return avs;
+}
+
+/*
+ * For Armada 37xx soc, L0(VSET0) VDD AVS value is set to SVC revision
+ * value or a default value when SVC is not supported.
+ * - L0 can be read out from the register of AVS_CTRL_0 and L0 voltage
+ *   can be got from the mapping table of avs_map.
+ * - L1 voltage should be about 100mv smaller than L0 voltage
+ * - L2 & L3 voltage should be about 150mv smaller than L0 voltage.
+ * This function calculates L1 & L2 & L3 AVS values dynamically based
+ * on L0 voltage and fill all AVS values to the AVS value table.
+ */
+static void __init armada37xx_cpufreq_avs_configure(struct regmap *base,
+						struct armada_37xx_dvfs *dvfs)
+{
+	unsigned int target_vm;
+	int load_level = 0;
+	u32 l0_vdd_min;
+
+	if (base == NULL)
+		return;
+
+	/* Get L0 VDD min value */
+	regmap_read(base, ARMADA_37XX_AVS_CTL0, &l0_vdd_min);
+	l0_vdd_min = (l0_vdd_min >> ARMADA_37XX_AVS_LOW_VDD_LIMIT) &
+		ARMADA_37XX_AVS_VDD_MASK;
+	if (l0_vdd_min >= ARRAY_SIZE(avs_map))  {
+		pr_err("L0 VDD MIN %d is not correct.\n", l0_vdd_min);
+		return;
+	}
+	dvfs->avs[0] = l0_vdd_min;
+
+	if (avs_map[l0_vdd_min] <= MIN_VOLT_MV) {
+		/*
+		 * If L0 voltage is smaller than 1000mv, then all VDD sets
+		 * use L0 voltage;
+		 */
+		u32 avs_min = armada_37xx_avs_val_match(MIN_VOLT_MV);
+
+		for (load_level = 1; load_level < LOAD_LEVEL_NR; load_level++)
+			dvfs->avs[load_level] = avs_min;
+
+		return;
+	}
+
+	/*
+	 * L1 voltage is equal to L0 voltage - 100mv and it must be
+	 * larger than 1000mv
+	 */
+
+	target_vm = avs_map[l0_vdd_min] - 100;
+	target_vm = target_vm > MIN_VOLT_MV ? target_vm : MIN_VOLT_MV;
+	dvfs->avs[1] = armada_37xx_avs_val_match(target_vm);
+
+	/*
+	 * L2 & L3 voltage is equal to L0 voltage - 150mv and it must
+	 * be larger than 1000mv
+	 */
+	target_vm = avs_map[l0_vdd_min] - 150;
+	target_vm = target_vm > MIN_VOLT_MV ? target_vm : MIN_VOLT_MV;
+	dvfs->avs[2] = dvfs->avs[3] = armada_37xx_avs_val_match(target_vm);
+}
+
+static void __init armada37xx_cpufreq_avs_setup(struct regmap *base,
+						struct armada_37xx_dvfs *dvfs)
+{
+	unsigned int avs_val = 0, freq;
+	int load_level = 0;
+
+	if (base == NULL)
+		return;
+
+	/* Disable AVS before the configuration */
+	regmap_update_bits(base, ARMADA_37XX_AVS_CTL0,
+			   ARMADA_37XX_AVS_ENABLE, 0);
+
+
+	/* Enable low voltage mode */
+	regmap_update_bits(base, ARMADA_37XX_AVS_CTL2,
+			   ARMADA_37XX_AVS_LOW_VDD_EN,
+			   ARMADA_37XX_AVS_LOW_VDD_EN);
+
+
+	for (load_level = 1; load_level < LOAD_LEVEL_NR; load_level++) {
+		freq = dvfs->cpu_freq_max / dvfs->divider[load_level];
+
+		avs_val = dvfs->avs[load_level];
+		regmap_update_bits(base, ARMADA_37XX_AVS_VSET(load_level-1),
+		    ARMADA_37XX_AVS_VDD_MASK << ARMADA_37XX_AVS_HIGH_VDD_LIMIT |
+		    ARMADA_37XX_AVS_VDD_MASK << ARMADA_37XX_AVS_LOW_VDD_LIMIT,
+		    avs_val << ARMADA_37XX_AVS_HIGH_VDD_LIMIT |
+		    avs_val << ARMADA_37XX_AVS_LOW_VDD_LIMIT);
+	}
+
+	/* Enable AVS after the configuration */
+	regmap_update_bits(base, ARMADA_37XX_AVS_CTL0,
+			   ARMADA_37XX_AVS_ENABLE,
+			   ARMADA_37XX_AVS_ENABLE);
+
+}
+
 static void armada37xx_cpufreq_disable_dvfs(struct regmap *base)
 {
 	unsigned int reg = ARMADA_37XX_NB_DYN_MOD,
@@ -216,7 +360,7 @@ static int __init armada37xx_cpufreq_driver_init(void)
 	struct platform_device *pdev;
 	unsigned long freq;
 	unsigned int cur_frequency;
-	struct regmap *nb_pm_base;
+	struct regmap *nb_pm_base, *avs_base;
 	struct device *cpu_dev;
 	int load_lvl, ret;
 	struct clk *clk;
@@ -227,6 +371,14 @@ static int __init armada37xx_cpufreq_driver_init(void)
 	if (IS_ERR(nb_pm_base))
 		return -ENODEV;
 
+	avs_base =
+		syscon_regmap_lookup_by_compatible("marvell,armada-3700-avs");
+
+	/* if AVS is not present don't use it but still try to setup dvfs */
+	if (IS_ERR(avs_base)) {
+		pr_info("Syscon failed for Adapting Voltage Scaling: skip it\n");
+		avs_base = NULL;
+	}
 	/* Before doing any configuration on the DVFS first, disable it */
 	armada37xx_cpufreq_disable_dvfs(nb_pm_base);
 
@@ -270,16 +422,21 @@ static int __init armada37xx_cpufreq_driver_init(void)
 
 	armada37xx_cpufreq_state->regmap = nb_pm_base;
 
+	armada37xx_cpufreq_avs_configure(avs_base, dvfs);
+	armada37xx_cpufreq_avs_setup(avs_base, dvfs);
+
 	armada37xx_cpufreq_dvfs_setup(nb_pm_base, clk, dvfs->divider);
 	clk_put(clk);
 
 	for (load_lvl = ARMADA_37XX_DVFS_LOAD_0; load_lvl < LOAD_LEVEL_NR;
 	     load_lvl++) {
+		unsigned long u_volt = avs_map[dvfs->avs[load_lvl]] * 1000;
 		freq = cur_frequency / dvfs->divider[load_lvl];
-
-		ret = dev_pm_opp_add(cpu_dev, freq, 0);
+		ret = dev_pm_opp_add(cpu_dev, freq, u_volt);
 		if (ret)
 			goto remove_opp;
+
+
 	}
 
 	/* Now that everything is setup, enable the DVFS at hardware level */

drivers/cpufreq/cppc_cpufreq.c

@@ -296,10 +296,62 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	return ret;
 }
 
+static inline u64 get_delta(u64 t1, u64 t0)
+{
+	if (t1 > t0 || t0 > ~(u32)0)
+		return t1 - t0;
+
+	return (u32)t1 - (u32)t0;
+}
+
+static int cppc_get_rate_from_fbctrs(struct cppc_cpudata *cpu,
+				     struct cppc_perf_fb_ctrs fb_ctrs_t0,
+				     struct cppc_perf_fb_ctrs fb_ctrs_t1)
+{
+	u64 delta_reference, delta_delivered;
+	u64 reference_perf, delivered_perf;
+
+	reference_perf = fb_ctrs_t0.reference_perf;
+
+	delta_reference = get_delta(fb_ctrs_t1.reference,
+				    fb_ctrs_t0.reference);
+	delta_delivered = get_delta(fb_ctrs_t1.delivered,
+				    fb_ctrs_t0.delivered);
+
+	/* Check to avoid divide-by zero */
+	if (delta_reference || delta_delivered)
+		delivered_perf = (reference_perf * delta_delivered) /
+					delta_reference;
+	else
+		delivered_perf = cpu->perf_ctrls.desired_perf;
+
+	return cppc_cpufreq_perf_to_khz(cpu, delivered_perf);
+}
+
+static unsigned int cppc_cpufreq_get_rate(unsigned int cpunum)
+{
+	struct cppc_perf_fb_ctrs fb_ctrs_t0 = {0}, fb_ctrs_t1 = {0};
+	struct cppc_cpudata *cpu = all_cpu_data[cpunum];
+	int ret;
+
+	ret = cppc_get_perf_ctrs(cpunum, &fb_ctrs_t0);
+	if (ret)
+		return ret;
+
+	udelay(2); /* 2usec delay between sampling */
+
+	ret = cppc_get_perf_ctrs(cpunum, &fb_ctrs_t1);
+	if (ret)
+		return ret;
+
+	return cppc_get_rate_from_fbctrs(cpu, fb_ctrs_t0, fb_ctrs_t1);
+}
+
 static struct cpufreq_driver cppc_cpufreq_driver = {
 	.flags = CPUFREQ_CONST_LOOPS,
 	.verify = cppc_verify_policy,
 	.target = cppc_cpufreq_set_target,
+	.get = cppc_cpufreq_get_rate,
 	.init = cppc_cpufreq_cpu_init,
 	.stop_cpu = cppc_cpufreq_stop_cpu,
 	.name = "cppc_cpufreq",

drivers/cpufreq/cpufreq.c

@@ -923,7 +923,12 @@ static ssize_t store(struct kobject *kobj, struct attribute *attr,
 	struct freq_attr *fattr = to_attr(attr);
 	ssize_t ret = -EINVAL;
 
-	cpus_read_lock();
+	/*
+	 * cpus_read_trylock() is used here to work around a circular lock
+	 * dependency problem with respect to the cpufreq_register_driver().
+	 */
+	if (!cpus_read_trylock())
+		return -EBUSY;
 
 	if (cpu_online(policy->cpu)) {
 		down_write(&policy->rwsem);
@@ -2236,6 +2241,7 @@ static int cpufreq_set_policy(struct cpufreq_policy *policy,
 
 	policy->min = new_policy->min;
 	policy->max = new_policy->max;
+	trace_cpu_frequency_limits(policy);
 
 	policy->cached_target_freq = UINT_MAX;
 

drivers/cpufreq/imx6q-cpufreq.c

@@ -9,6 +9,7 @@
 #include <linux/clk.h>
 #include <linux/cpu.h>
 #include <linux/cpufreq.h>
+#include <linux/cpu_cooling.h>
 #include <linux/err.h>
 #include <linux/module.h>
 #include <linux/of.h>
@@ -50,6 +51,7 @@ static struct clk_bulk_data clks[] = {
 };
 
 static struct device *cpu_dev;
+static struct thermal_cooling_device *cdev;
 static bool free_opp;
 static struct cpufreq_frequency_table *freq_table;
 static unsigned int max_freq;
@@ -191,6 +193,16 @@ static int imx6q_set_target(struct cpufreq_policy *policy, unsigned int index)
 	return 0;
 }
 
+static void imx6q_cpufreq_ready(struct cpufreq_policy *policy)
+{
+	cdev = of_cpufreq_cooling_register(policy);
+
+	if (!cdev)
+		dev_err(cpu_dev,
+			"running cpufreq without cooling device: %ld\n",
+			PTR_ERR(cdev));
+}
+
 static int imx6q_cpufreq_init(struct cpufreq_policy *policy)
 {
 	int ret;
@@ -202,13 +214,22 @@ static int imx6q_cpufreq_init(struct cpufreq_policy *policy)
 	return ret;
 }
 
+static int imx6q_cpufreq_exit(struct cpufreq_policy *policy)
+{
+	cpufreq_cooling_unregister(cdev);
+
+	return 0;
+}
+
 static struct cpufreq_driver imx6q_cpufreq_driver = {
 	.flags = CPUFREQ_NEED_INITIAL_FREQ_CHECK,
 	.verify = cpufreq_generic_frequency_table_verify,
 	.target_index = imx6q_set_target,
 	.get = cpufreq_generic_get,
 	.init = imx6q_cpufreq_init,
+	.exit = imx6q_cpufreq_exit,
 	.name = "imx6q-cpufreq",
+	.ready = imx6q_cpufreq_ready,
 	.attr = cpufreq_generic_attr,
 	.suspend = cpufreq_generic_suspend,
 };

drivers/cpufreq/intel_pstate.c

@@ -657,21 +657,18 @@ static ssize_t store_energy_performance_preference(
 {
 	struct cpudata *cpu_data = all_cpu_data[policy->cpu];
 	char str_preference[21];
-	int ret, i = 0;
+	int ret;
 
 	ret = sscanf(buf, "%20s", str_preference);
 	if (ret != 1)
 		return -EINVAL;
 
-	while (energy_perf_strings[i] != NULL) {
-		if (!strcmp(str_preference, energy_perf_strings[i])) {
-			intel_pstate_set_energy_pref_index(cpu_data, i);
-			return count;
-		}
-		++i;
-	}
+	ret = match_string(energy_perf_strings, -1, str_preference);
+	if (ret < 0)
+		return ret;
 
-	return -EINVAL;
+	intel_pstate_set_energy_pref_index(cpu_data, ret);
+	return count;
 }
 
 static ssize_t show_energy_performance_preference(
@@ -2072,6 +2069,15 @@ static int __intel_pstate_cpu_init(struct cpufreq_policy *policy)
 			cpu->pstate.max_pstate : cpu->pstate.turbo_pstate;
 	policy->cpuinfo.max_freq *= cpu->pstate.scaling;
 
+	if (hwp_active) {
+		unsigned int max_freq;
+
+		max_freq = global.turbo_disabled ?
+			cpu->pstate.max_freq : cpu->pstate.turbo_freq;
+		if (max_freq < policy->cpuinfo.max_freq)
+			policy->cpuinfo.max_freq = max_freq;
+	}
+
 	intel_pstate_init_acpi_perf_limits(policy);
 
 	policy->fast_switch_possible = true;

drivers/cpufreq/pcc-cpufreq.c

@@ -593,6 +593,15 @@ static int __init pcc_cpufreq_init(void)
 		return ret;
 	}
 
+	if (num_present_cpus() > 4) {
+		pcc_cpufreq_driver.flags |= CPUFREQ_NO_AUTO_DYNAMIC_SWITCHING;
+		pr_err("%s: Too many CPUs, dynamic performance scaling disabled\n",
+		       __func__);
+		pr_err("%s: Try to enable another scaling driver through BIOS settings\n",
+		       __func__);
+		pr_err("%s: and complain to the system vendor\n", __func__);
+	}
+
 	ret = cpufreq_register_driver(&pcc_cpufreq_driver);
 
 	return ret;

drivers/cpufreq/qcom-cpufreq-kryo.c

@@ -109,6 +109,7 @@ static int qcom_cpufreq_kryo_probe(struct platform_device *pdev)
 	speedbin_nvmem = of_nvmem_cell_get(np, NULL);
 	of_node_put(np);
 	if (IS_ERR(speedbin_nvmem)) {
+		if (PTR_ERR(speedbin_nvmem) != -EPROBE_DEFER)
 			dev_err(cpu_dev, "Could not get nvmem cell: %ld\n",
 				PTR_ERR(speedbin_nvmem));
 		return PTR_ERR(speedbin_nvmem);

drivers/thermal/imx_thermal.c

@@ -3,6 +3,7 @@
 // Copyright 2013 Freescale Semiconductor, Inc.
 
 #include <linux/clk.h>
+#include <linux/cpu.h>
 #include <linux/cpufreq.h>
 #include <linux/cpu_cooling.h>
 #include <linux/delay.h>
@@ -644,6 +645,27 @@ static const struct of_device_id of_imx_thermal_match[] = {
 };
 MODULE_DEVICE_TABLE(of, of_imx_thermal_match);
 
+/*
+ * Create cooling device in case no #cooling-cells property is available in
+ * CPU node
+ */
+static int imx_thermal_register_legacy_cooling(struct imx_thermal_data *data)
+{
+	struct device_node *np = of_get_cpu_node(data->policy->cpu, NULL);
+	int ret;
+
+	if (!np || !of_find_property(np, "#cooling-cells", NULL)) {
+		data->cdev = cpufreq_cooling_register(data->policy);
+		if (IS_ERR(data->cdev)) {
+			ret = PTR_ERR(data->cdev);
+			cpufreq_cpu_put(data->policy);
+			return ret;
+		}
+	}
+
+	return 0;
+}
+
 static int imx_thermal_probe(struct platform_device *pdev)
 {
 	struct imx_thermal_data *data;
@@ -724,12 +746,10 @@ static int imx_thermal_probe(struct platform_device *pdev)
 		return -EPROBE_DEFER;
 	}
 
-	data->cdev = cpufreq_cooling_register(data->policy);
-	if (IS_ERR(data->cdev)) {
-		ret = PTR_ERR(data->cdev);
+	ret = imx_thermal_register_legacy_cooling(data);
+	if (ret) {
 		dev_err(&pdev->dev,
 			"failed to register cpufreq cooling device: %d\n", ret);
-		cpufreq_cpu_put(data->policy);
 		return ret;
 	}
 

include/linux/cpu.h

@@ -103,6 +103,7 @@ extern void cpus_write_lock(void);
 extern void cpus_write_unlock(void);
 extern void cpus_read_lock(void);
 extern void cpus_read_unlock(void);
+extern int  cpus_read_trylock(void);
 extern void lockdep_assert_cpus_held(void);
 extern void cpu_hotplug_disable(void);
 extern void cpu_hotplug_enable(void);
@@ -115,6 +116,7 @@ static inline void cpus_write_lock(void) { }
 static inline void cpus_write_unlock(void) { }
 static inline void cpus_read_lock(void) { }
 static inline void cpus_read_unlock(void) { }
+static inline int  cpus_read_trylock(void) { return true; }
 static inline void lockdep_assert_cpus_held(void) { }
 static inline void cpu_hotplug_disable(void) { }
 static inline void cpu_hotplug_enable(void) { }

include/trace/events/power.h

@@ -5,6 +5,7 @@
 #if !defined(_TRACE_POWER_H) || defined(TRACE_HEADER_MULTI_READ)
 #define _TRACE_POWER_H
 
+#include <linux/cpufreq.h>
 #include <linux/ktime.h>
 #include <linux/pm_qos.h>
 #include <linux/tracepoint.h>
@@ -148,6 +149,30 @@ DEFINE_EVENT(cpu, cpu_frequency,
 	TP_ARGS(frequency, cpu_id)
 );
 
+TRACE_EVENT(cpu_frequency_limits,
+
+	TP_PROTO(struct cpufreq_policy *policy),
+
+	TP_ARGS(policy),
+
+	TP_STRUCT__entry(
+		__field(u32, min_freq)
+		__field(u32, max_freq)
+		__field(u32, cpu_id)
+	),
+
+	TP_fast_assign(
+		__entry->min_freq = policy->min;
+		__entry->max_freq = policy->max;
+		__entry->cpu_id = policy->cpu;
+	),
+
+	TP_printk("min=%lu max=%lu cpu_id=%lu",
+		  (unsigned long)__entry->min_freq,
+		  (unsigned long)__entry->max_freq,
+		  (unsigned long)__entry->cpu_id)
+);
+
 TRACE_EVENT(device_pm_callback_start,
 
 	TP_PROTO(struct device *dev, const char *pm_ops, int event),

kernel/cpu.c

@@ -290,6 +290,12 @@ void cpus_read_lock(void)
 }
 EXPORT_SYMBOL_GPL(cpus_read_lock);
 
+int cpus_read_trylock(void)
+{
+	return percpu_down_read_trylock(&cpu_hotplug_lock);
+}
+EXPORT_SYMBOL_GPL(cpus_read_trylock);
+
 void cpus_read_unlock(void)
 {
 	percpu_up_read(&cpu_hotplug_lock);