cpufreq: qcom-hw: cleanup the driver for HW driver

Move the current code to support as per the upstream code and bring in
fixes required to support the EPSS driver. Fix em_register_perf_domain
as new API convention.

Change-Id: I267731a89a130d1fca5b1fb25aab95430a9099c7
Signed-off-by: Taniya Das <tdas@codeaurora.org>
This commit is contained in:
Taniya Das 2019-11-04 15:32:21 +05:30
commit b23d5d6083

View file

@ -3,19 +3,24 @@
* Copyright (c) 2018-2019, The Linux Foundation. All rights reserved.
*/
#include <linux/bitfield.h>
#include <linux/cpufreq.h>
#include <linux/cpu_cooling.h>
#include <linux/energy_model.h>
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/of_address.h>
#include <linux/of_platform.h>
#include <linux/pm_opp.h>
#include <linux/energy_model.h>
#include <linux/sched.h>
#include <linux/cpu_cooling.h>
#include <linux/slab.h>
#define LUT_MAX_ENTRIES 40U
#define CORE_COUNT_VAL(val) (((val) & (GENMASK(18, 16))) >> 16)
#define LUT_SRC GENMASK(31, 30)
#define LUT_L_VAL GENMASK(7, 0)
#define LUT_CORE_COUNT GENMASK(18, 16)
#define LUT_VOLT GENMASK(11, 0)
#define LUT_ROW_SIZE 32
#define CLK_HW_DIV 2
@ -23,24 +28,23 @@
(acc_count ? ((c - cpumask_first(m) + 1) * 4) : 0)
enum {
REG_ENABLE,
REG_FREQ_LUT_TABLE,
REG_VOLT_LUT_TABLE,
REG_FREQ_LUT,
REG_VOLT_LUT,
REG_PERF_STATE,
REG_CYCLE_CNTR,
REG_ARRAY_SIZE,
};
static unsigned long cpu_hw_rate, xo_rate;
static const u16 *offsets;
static unsigned int lut_row_size = LUT_ROW_SIZE;
static bool accumulative_counter;
struct cpufreq_qcom {
struct cpufreq_frequency_table *table;
void __iomem *reg_bases[REG_ARRAY_SIZE];
void __iomem *base;
cpumask_t related_cpus;
unsigned int max_cores;
unsigned long xo_rate;
unsigned long cpu_hw_rate;
};
struct cpufreq_counter {
@ -51,40 +55,43 @@ struct cpufreq_counter {
static const u16 cpufreq_qcom_std_offsets[REG_ARRAY_SIZE] = {
[REG_ENABLE] = 0x0,
[REG_FREQ_LUT_TABLE] = 0x110,
[REG_VOLT_LUT_TABLE] = 0x114,
[REG_FREQ_LUT] = 0x110,
[REG_VOLT_LUT] = 0x114,
[REG_PERF_STATE] = 0x920,
[REG_CYCLE_CNTR] = 0x9c0,
};
static const u16 cpufreq_qcom_epss_std_offsets[REG_ARRAY_SIZE] = {
[REG_ENABLE] = 0x0,
[REG_FREQ_LUT_TABLE] = 0x100,
[REG_VOLT_LUT_TABLE] = 0x200,
[REG_FREQ_LUT] = 0x100,
[REG_VOLT_LUT] = 0x200,
[REG_PERF_STATE] = 0x320,
[REG_CYCLE_CNTR] = 0x3c4,
};
static struct cpufreq_counter qcom_cpufreq_counter[NR_CPUS];
static struct cpufreq_qcom *qcom_freq_domain_map[NR_CPUS];
static struct cpufreq_counter qcom_cpufreq_counter[NR_CPUS];
static u64 qcom_cpufreq_get_cpu_cycle_counter(int cpu)
{
struct cpufreq_counter *cpu_counter;
struct cpufreq_qcom *cpu_domain;
struct cpufreq_policy *policy;
u64 cycle_counter_ret;
unsigned long flags;
u16 offset;
u32 val;
cpu_domain = qcom_freq_domain_map[cpu];
policy = cpufreq_cpu_get_raw(cpu);
if (!policy)
return 0;
cpu_counter = &qcom_cpufreq_counter[cpu];
spin_lock_irqsave(&cpu_counter->lock, flags);
offset = CYCLE_CNTR_OFFSET(cpu, &cpu_domain->related_cpus,
offset = CYCLE_CNTR_OFFSET(cpu, policy->related_cpus,
accumulative_counter);
val = readl_relaxed_no_log(cpu_domain->reg_bases[REG_CYCLE_CNTR] +
offset);
val = readl_relaxed_no_log(policy->driver_data +
offsets[REG_CYCLE_CNTR] + offset);
if (val < cpu_counter->prev_cycle_counter) {
/* Handle counter overflow */
@ -106,9 +113,9 @@ static int
qcom_cpufreq_hw_target_index(struct cpufreq_policy *policy,
unsigned int index)
{
struct cpufreq_qcom *c = policy->driver_data;
void __iomem *base = policy->driver_data;
writel_relaxed(index, c->reg_bases[REG_PERF_STATE]);
writel_relaxed(index, base + offsets[REG_PERF_STATE]);
arch_set_freq_scale(policy->related_cpus,
policy->freq_table[index].frequency,
policy->cpuinfo.max_freq);
@ -118,17 +125,17 @@ qcom_cpufreq_hw_target_index(struct cpufreq_policy *policy,
static unsigned int qcom_cpufreq_hw_get(unsigned int cpu)
{
struct cpufreq_qcom *c;
struct cpufreq_policy *policy;
void __iomem *base;
unsigned int index;
policy = cpufreq_cpu_get_raw(cpu);
if (!policy)
return 0;
c = policy->driver_data;
base = policy->driver_data;
index = readl_relaxed(c->reg_bases[REG_PERF_STATE]);
index = readl_relaxed(base + offsets[REG_PERF_STATE]);
index = min(index, LUT_MAX_ENTRIES - 1);
return policy->freq_table[index].frequency;
@ -152,7 +159,6 @@ qcom_cpufreq_hw_fast_switch(struct cpufreq_policy *policy,
static int qcom_cpufreq_hw_cpu_init(struct cpufreq_policy *policy)
{
struct em_data_callback em_cb = EM_DATA_CB(of_dev_pm_opp_get_cpu_power);
struct cpufreq_qcom *c;
struct device *cpu_dev;
int ret;
@ -176,12 +182,12 @@ static int qcom_cpufreq_hw_cpu_init(struct cpufreq_policy *policy)
if (ret <= 0)
dev_err(cpu_dev, "OPP table is not ready\n");
policy->fast_switch_possible = true;
policy->freq_table = c->table;
policy->driver_data = c;
policy->driver_data = c->base;
policy->fast_switch_possible = true;
policy->dvfs_possible_from_any_cpu = true;
em_register_perf_domain(policy->cpus, ret, &em_cb);
dev_pm_opp_of_register_em(policy->cpus);
return 0;
}
@ -236,11 +242,10 @@ static struct cpufreq_driver cpufreq_qcom_hw_driver = {
};
static int qcom_cpufreq_hw_read_lut(struct platform_device *pdev,
struct cpufreq_qcom *c)
struct cpufreq_qcom *c, u32 max_cores)
{
struct device *dev = &pdev->dev;
void __iomem *base_freq, *base_volt;
u32 data, src, lval, i, core_count, prev_cc, prev_freq, cur_freq, volt;
u32 data, src, lval, i, core_count, prev_cc, prev_freq, freq, volt;
unsigned long cpu;
c->table = devm_kcalloc(dev, LUT_MAX_ENTRIES + 1,
@ -248,57 +253,63 @@ static int qcom_cpufreq_hw_read_lut(struct platform_device *pdev,
if (!c->table)
return -ENOMEM;
base_freq = c->reg_bases[REG_FREQ_LUT_TABLE];
base_volt = c->reg_bases[REG_VOLT_LUT_TABLE];
cpu = cpumask_first(&c->related_cpus);
for (i = 0; i < LUT_MAX_ENTRIES; i++) {
data = readl_relaxed(base_freq + i * lut_row_size);
src = (data & GENMASK(31, 30)) >> 30;
lval = data & GENMASK(7, 0);
core_count = CORE_COUNT_VAL(data);
data = readl_relaxed(c->base + offsets[REG_FREQ_LUT] +
i * lut_row_size);
src = FIELD_GET(LUT_SRC, data);
lval = FIELD_GET(LUT_L_VAL, data);
core_count = FIELD_GET(LUT_CORE_COUNT, data);
data = readl_relaxed(base_volt + i * lut_row_size);
volt = (data & GENMASK(11, 0)) * 1000;
data = readl_relaxed(c->base + offsets[REG_VOLT_LUT] +
i * lut_row_size);
volt = FIELD_GET(LUT_VOLT, data) * 1000;
if (src)
c->table[i].frequency = c->xo_rate * lval / 1000;
freq = xo_rate * lval / 1000;
else
c->table[i].frequency = c->cpu_hw_rate / 1000;
freq = cpu_hw_rate / 1000;
cur_freq = c->table[i].frequency;
dev_dbg(dev, "index=%d freq=%d, core_count %d\n",
i, c->table[i].frequency, core_count);
if (core_count != c->max_cores)
cur_freq = CPUFREQ_ENTRY_INVALID;
if (freq != prev_freq && core_count == max_cores) {
c->table[i].frequency = freq;
dev_pm_opp_add(get_cpu_device(cpu), freq * 1000, volt);
dev_dbg(dev, "index=%d freq=%d, core_count %d\n",
i, c->table[i].frequency, core_count);
} else {
c->table[i].frequency = CPUFREQ_ENTRY_INVALID;
}
/*
* Two of the same frequencies with the same core counts means
* end of table.
*/
if (i > 0 && c->table[i - 1].frequency ==
c->table[i].frequency && prev_cc == core_count) {
if (i > 0 && prev_freq == freq && prev_cc == core_count) {
struct cpufreq_frequency_table *prev = &c->table[i - 1];
if (prev_freq == CPUFREQ_ENTRY_INVALID)
if (prev_cc != max_cores) {
prev->frequency = prev_freq;
prev->flags = CPUFREQ_BOOST_FREQ;
dev_pm_opp_add(get_cpu_device(cpu),
prev_freq * 1000, volt);
}
break;
}
prev_cc = core_count;
prev_freq = cur_freq;
cur_freq *= 1000;
for_each_cpu(cpu, &c->related_cpus)
dev_pm_opp_add(get_cpu_device(cpu), cur_freq, volt);
prev_cc = core_count;
prev_freq = freq;
freq *= 1000;
}
c->table[i].frequency = CPUFREQ_TABLE_END;
dev_pm_opp_set_sharing_cpus(get_cpu_device(cpu), &c->related_cpus);
return 0;
}
static int qcom_get_related_cpus(int index, struct cpumask *m)
static void qcom_get_related_cpus(int index, struct cpumask *m)
{
struct device_node *cpu_np;
struct of_phandle_args args;
@ -318,25 +329,17 @@ static int qcom_get_related_cpus(int index, struct cpumask *m)
if (index == args.args[0])
cpumask_set_cpu(cpu, m);
}
return 0;
}
static int qcom_cpu_resources_init(struct platform_device *pdev,
unsigned int cpu, int index,
unsigned int max_cores,
unsigned long xo_rate,
unsigned long cpu_hw_rate)
unsigned int max_cores)
{
struct cpufreq_qcom *c;
struct resource *res;
struct device *dev = &pdev->dev;
const u16 *offsets;
int ret, i, cpu_r;
void __iomem *base;
if (qcom_freq_domain_map[cpu])
return 0;
int ret, cpu_r;
c = devm_kzalloc(dev, sizeof(*c), GFP_KERNEL);
if (!c)
@ -351,12 +354,9 @@ static int qcom_cpu_resources_init(struct platform_device *pdev,
if (IS_ERR(base))
return PTR_ERR(base);
for (i = REG_ENABLE; i < REG_ARRAY_SIZE; i++)
c->reg_bases[i] = base + offsets[i];
if (!of_property_read_bool(dev->of_node, "qcom,skip-enable-check")) {
/* HW should be in enabled state to proceed */
if (!(readl_relaxed(c->reg_bases[REG_ENABLE]) & 0x1)) {
if (!(readl_relaxed(base + offsets[REG_ENABLE]) & 0x1)) {
dev_err(dev, "Domain-%d cpufreq hardware not enabled\n",
index);
return -ENODEV;
@ -364,22 +364,16 @@ static int qcom_cpu_resources_init(struct platform_device *pdev,
}
accumulative_counter = !of_property_read_bool(dev->of_node,
"qcom,no-accumulative-counter");
"qcom,no-accumulative-counter");
c->base = base;
ret = qcom_get_related_cpus(index, &c->related_cpus);
if (ret) {
qcom_get_related_cpus(index, &c->related_cpus);
if (!cpumask_weight(&c->related_cpus)) {
dev_err(dev, "Domain-%d failed to get related CPUs\n", index);
return ret;
return -ENONET;
}
c->max_cores = max_cores;
if (!c->max_cores)
return -ENOENT;
c->xo_rate = xo_rate;
c->cpu_hw_rate = cpu_hw_rate;
ret = qcom_cpufreq_hw_read_lut(pdev, c);
ret = qcom_cpufreq_hw_read_lut(pdev, c, max_cores);
if (ret) {
dev_err(dev, "Domain-%d failed to read LUT\n", index);
return ret;
@ -397,7 +391,6 @@ static int qcom_resources_init(struct platform_device *pdev)
struct of_phandle_args args;
struct clk *clk;
unsigned int cpu;
unsigned long xo_rate, cpu_hw_rate;
int ret;
clk = devm_clk_get(&pdev->dev, "xo");
@ -405,16 +398,14 @@ static int qcom_resources_init(struct platform_device *pdev)
return PTR_ERR(clk);
xo_rate = clk_get_rate(clk);
devm_clk_put(&pdev->dev, clk);
clk_put(clk);
clk = devm_clk_get(&pdev->dev, "alternate");
if (IS_ERR(clk))
return PTR_ERR(clk);
cpu_hw_rate = clk_get_rate(clk) / CLK_HW_DIV;
devm_clk_put(&pdev->dev, clk);
clk_put(clk);
of_property_read_u32(pdev->dev.of_node, "qcom,lut-row-size",
&lut_row_size);
@ -428,13 +419,17 @@ static int qcom_resources_init(struct platform_device *pdev)
}
ret = of_parse_phandle_with_args(cpu_np, "qcom,freq-domain",
"#freq-domain-cells", 0, &args);
if (ret < 0)
"#freq-domain-cells", 0,
&args);
of_node_put(cpu_np);
if (ret)
return ret;
if (qcom_freq_domain_map[cpu])
continue;
ret = qcom_cpu_resources_init(pdev, cpu, args.args[0],
args.args[1], xo_rate,
cpu_hw_rate);
args.args[1]);
if (ret)
return ret;
}
@ -471,12 +466,17 @@ static int qcom_cpufreq_hw_driver_probe(struct platform_device *pdev)
return rc;
}
dev_dbg(&pdev->dev, "QCOM CPUFreq HW driver initialized\n");
of_platform_populate(pdev->dev.of_node, NULL, NULL, &pdev->dev);
dev_dbg(&pdev->dev, "QCOM CPUFreq HW driver initialized\n");
return 0;
}
static int qcom_cpufreq_hw_driver_remove(struct platform_device *pdev)
{
return cpufreq_unregister_driver(&cpufreq_qcom_hw_driver);
}
static const struct of_device_id qcom_cpufreq_hw_match[] = {
{ .compatible = "qcom,cpufreq-hw", .data = &cpufreq_qcom_std_offsets },
{ .compatible = "qcom,cpufreq-hw-epss",
@ -486,6 +486,7 @@ static const struct of_device_id qcom_cpufreq_hw_match[] = {
static struct platform_driver qcom_cpufreq_hw_driver = {
.probe = qcom_cpufreq_hw_driver_probe,
.remove = qcom_cpufreq_hw_driver_remove,
.driver = {
.name = "qcom-cpufreq-hw",
.of_match_table = qcom_cpufreq_hw_match,
@ -498,4 +499,11 @@ static int __init qcom_cpufreq_hw_init(void)
}
subsys_initcall(qcom_cpufreq_hw_init);
MODULE_DESCRIPTION("QCOM firmware-based CPU Frequency driver");
static void __exit qcom_cpufreq_hw_exit(void)
{
platform_driver_unregister(&qcom_cpufreq_hw_driver);
}
module_exit(qcom_cpufreq_hw_exit);
MODULE_DESCRIPTION("QCOM CPUFREQ HW Driver");
MODULE_LICENSE("GPL v2");