CPUFreq boost代码分析

参考4.9内核，还是先看文档描述：

Introduction

Some CPUs support a functionality to raise the operating frequency of
some cores in a multi-core package if certain conditions apply, mostly
if the whole chip is not fully utilized and below it’s intended thermal
budget. The decision about boost disable/enable is made either at hardware
(e.g. x86) or software (e.g ARM).
On Intel CPUs this is called “Turbo Boost”, AMD calls it “Turbo-Core”,
in technical documentation “Core performance boost”. In Linux we use
the term “boost” for convenience.

有些多核CPU系统中支持提高操作频率， 由硬件or软件来决定，Linux用boost来表示这个功能。一般ARM(Exynos)使用的是SW，x86使用的是HW。

有个boost开关用户可以控制禁用使能：

To allow the user to toggle the boosting functionality, the cpufreq core
driver exports a sysfs knob to enable or disable it. There is a file:
/sys/devices/system/cpu/cpufreq/boost

ok，来看相关代码，路径在drivers/cpufreq/：

static bool cpufreq_boost_supported(void)
{
	return likely(cpufreq_driver) && cpufreq_driver->set_boost;
}

如果支持提高频率，drv就要配置->set_boost，可以看到acpi-cpufreq.c配了，还有一处是cpufreq.c里也配了，分别看下：

static void __init acpi_cpufreq_boost_init(void)
{
        if (boot_cpu_has(X86_FEATURE_CPB) || boot_cpu_has(X86_FEATURE_IDA)) {
                msrs = msrs_alloc();

                if (!msrs)
                        return;

                acpi_cpufreq_driver.set_boost = set_boost;
                acpi_cpufreq_driver.boost_enabled = boost_state(0);

可见是x86 hw本身就支持，只要满足feature: X86_FEATURE_CPB or X86_FEATURE_IDA，默认禁用(boost_enabled=0)。

static int __init acpi_cpufreq_init(void)
{
	...
        acpi_cpufreq_boost_init();

        ret = cpufreq_register_driver(&acpi_cpufreq_driver);
        if (ret) {
	...
}

late_initcall(acpi_cpufreq_init);

acpi drv在注册到core前就定下了。还有一处在core里：

int cpufreq_enable_boost_support(void)
{
        if (!cpufreq_driver)
                return -EINVAL;

        if (cpufreq_boost_supported())
                return 0;

        cpufreq_driver->set_boost = cpufreq_boost_set_sw; //tj: here

        /* This will get removed on driver unregister */
        return create_boost_sysfs_file();
}
EXPORT_SYMBOL_GPL(cpufreq_enable_boost_support);

cpufreq-dt.c会call：

static int cpufreq_init(struct cpufreq_policy *policy)
{
	...
        /* Support turbo/boost mode */
        if (policy_has_boost_freq(policy)) {
                /* This gets disabled by core on driver unregister */
                ret = cpufreq_enable_boost_support();
                if (ret)

static struct cpufreq_driver dt_cpufreq_driver = {
        .flags = CPUFREQ_STICKY | CPUFREQ_NEED_INITIAL_FREQ_CHECK,
        .verify = cpufreq_generic_frequency_table_verify,
        .target_index = set_target,
        .get = cpufreq_generic_get,
        .init = cpufreq_init,

是挂到了->init中去设置->set_boost，从名字看应该是sw boost。判断是否支持：

bool policy_has_boost_freq(struct cpufreq_policy *policy)
{
	struct cpufreq_frequency_table *pos, *table = policy->freq_table;

	if (!table)
		return false;

	cpufreq_for_each_valid_entry(pos, table)
		if (pos->flags & CPUFREQ_BOOST_FREQ)
			return true;

	return false;
}

和频率表的生成(CPUFREQ_BOOST_FREQ)有关。跟下这个flag：cpufreq目录下没有，opengrok搜了下有：

/kernel/msm-4.9/drivers/base/power/opp/
H A D	cpu.c	85 freq_table[i].flags = CPUFREQ_BOOST_FREQ;

opp是啥？和pm有关：

/**
 * dev_pm_opp_init_cpufreq_table() - create a cpufreq table for a device
...
int dev_pm_opp_init_cpufreq_table(struct device *dev,
                                  struct cpufreq_frequency_table **table)
{
	...
                /* Is Boost/turbo opp ? */
                if (dev_pm_opp_is_turbo(opp))
                        freq_table[i].flags = CPUFREQ_BOOST_FREQ;
        }

dev_pm_opp_is_turbo():

bool dev_pm_opp_is_turbo(struct dev_pm_opp *opp)
{
        struct dev_pm_opp *tmp_opp;

        opp_rcu_lockdep_assert();

        tmp_opp = rcu_dereference(opp);
        if (IS_ERR_OR_NULL(tmp_opp) || !tmp_opp->available) {
                pr_err("%s: Invalid parameters\n", __func__);
                return false;
        }    

        return tmp_opp->turbo;
}

取决于->turbo，是从设备树而来:

new_opp->turbo = of_property_read_bool(np, "turbo-mode");

搜了下三星的Exynos会用到：

arch/arm/boot/dts/exynos4412.dtsi:137:			turbo-mode;

翻了下git log原来就是三星加的software boost for exynos arch，只不过后来放到了cpufreq-dt.c中，流程上放到了register之后。

dt_cpufreq_probe
     |-> cpufreq_register_driver
               |-> cpufreq_driver->init(cpufreq_init)

register里还有个创建sysfs boost的，可见是给前文的hardware(x86) boost用。

int cpufreq_register_driver(struct cpufreq_driver *driver_data)
{
	...
	if (cpufreq_boost_supported()) {
		ret = create_boost_sysfs_file();
		if (ret)
			goto err_null_driver;
	}

另外如果boost使能了，boost freq也是min/max的备选：

int cpufreq_frequency_table_cpuinfo(struct cpufreq_policy *policy,
				    struct cpufreq_frequency_table *table)
{
	...
	cpufreq_for_each_valid_entry(pos, table) {
		freq = pos->frequency;

		if (!cpufreq_boost_enabled() //tj: here
		    && (pos->flags & CPUFREQ_BOOST_FREQ))
			continue;

		pr_debug("table entry %u: %u kHz\n", (int)(pos - table), freq);
		if (freq < min_freq)
			min_freq = freq;
		if (freq > max_freq)
			max_freq = freq;
	}

	policy->min = policy->cpuinfo.min_freq = min_freq;
	policy->max = policy->cpuinfo.max_freq = max_freq;

sysfs的scaling_available是不包含boost freq的：

/**
 * show_scaling_available_frequencies - show available normal frequencies for
 * the specified CPU
 */
static ssize_t scaling_available_frequencies_show(struct cpufreq_policy *policy,
						  char *buf)
{
	return show_available_freqs(policy, buf, false);
}
cpufreq_attr_available_freq(scaling_available);
EXPORT_SYMBOL_GPL(cpufreq_freq_attr_scaling_available_freqs);

scaling_boost是show boost freq:

/**
 * show_available_boost_freqs - show available boost frequencies for
 * the specified CPU
 */
static ssize_t scaling_boost_frequencies_show(struct cpufreq_policy *policy,
					      char *buf)
{
	return show_available_freqs(policy, buf, true);
}
cpufreq_attr_available_freq(scaling_boost);

来看下show_available_freqs:

static ssize_t show_available_freqs(struct cpufreq_policy *policy, char *buf,
				    bool show_boost)
{
	ssize_t count = 0;
	struct cpufreq_frequency_table *pos, *table = policy->freq_table;

	if (!table)
		return -ENODEV;

	cpufreq_for_each_valid_entry(pos, table) {
		/*
		 * show_boost = true and driver_data = BOOST freq
		 * display BOOST freqs
		 *
		 * show_boost = false and driver_data = BOOST freq
		 * show_boost = true and driver_data != BOOST freq
		 * continue - do not display anything
		 *
		 * show_boost = false and driver_data != BOOST freq
		 * display NON BOOST freqs
		 */
		if (show_boost ^ (pos->flags & CPUFREQ_BOOST_FREQ))
			continue;

		count += sprintf(&buf[count], "%d ", pos->frequency);
	}

这里driver_data指的就是CPUFREQ_BOOST_FREQ这个flag了，所以逻辑就是：

• 当显示boost freq时看cpufreq table有没有boost freq，如果有就显示；
• 当显示normal freq时就是从cpufreq table中过滤boost freq。

Done.