Android7 Linux3.18设备休眠过程

在编写设备驱动的时候，我们会填写device_driver的pm方法, 这些方法是必须要有的，系统无法正常休眠，设置导致死机。驱动实现例如：

static const struct dev_pm_ops raydium_ts_pm_ops = {
    .suspend    = raydium_ts_suspend,
    .resume        = raydium_ts_resume,
};

static struct i2c_driver raydium_ts_driver = {
    .probe = raydium_ts_probe,
    .remove = raydium_ts_remove,
    .id_table = raydium_ts_id,
    .driver = {
        .name = RAYDIUM_NAME,
        .owner = THIS_MODULE,
        .of_match_table = raydium_match_table,
        .pm    = &raydium_ts_pm_ops,
    },
};

我们在device_driver->dev_pm_ops里填写了设备suspend和resume方法，这两个是设备被挂起到RAM是的PM.

应用层操作操作系统进入休眠时，应用会写入操作/sys/power/state节点，往节点写入字符串"mem"可是系统挂起到RAM.

设备挂起到RAM流程如下：

state_store ->  /kernel/kernel/power/main.c
  pm_suspend(state); -> /kernel/kernel/power/suspend.c
      enter_state(state);
        suspend_devices_and_enter(state); -> /kernel/kernel/power/suspend.c
      

详细看下suspend_devices_and_enter方法的实现：

int suspend_devices_and_enter(suspend_state_t state)
{
    error = platform_suspend_begin(state);
    
    error = dpm_suspend_start(PMSG_SUSPEND);
    if (error) {
		pr_err("PM: Some devices failed to suspend, or early wake event detected\n");
		log_suspend_abort_reason("Some devices failed to suspend, or early wake event detected");
		goto Recover_platform;
	}
}

我们关注dpm_suspend_start方法。如果改方法执行失败，此处将打印：

<3>[  350.999928] PM: Some devices failed to suspend, or early wake event detected
<6>[  351.081086] PM: resume of devices complete after 81.136 msecs

提示suspend失败。我们进入dpm_suspend_start方法，就可以看到suspend失败的原因了。

/kernel/drivers/base/power/main.c
int dpm_suspend_start(pm_message_t state)
{
    int error;
    // 执行设备 ->prepare，然后将设备键入到dpm_prepared_list链表中
    error = dpm_prepare(state);
    
    error = dpm_suspend(state);
    return error;
}

dpm_prepare函数将从dpm_list链表中取出各个设备，这些设备是在设备注册是键入的。取出后调用设备的parpare回调方法并回调执行。最后将设备加入到dpm_prepared_list链表中。

int dpm_suspend(pm_message_t state)
{
    cpufreq_suspend();
    
    while (!list_empty(&dpm_prepared_list)) {
        struct device *dev = to_device(dpm_prepared_list.prev);
        
        error = device_suspend(dev); // 执行休眠
        
        if (!list_empty(&dev->power.entry))
            list_move(&dev->power.entry, &dpm_suspended_list);
        put_device(dev); // 设备使用引用计数
    }
}

dpm_suspend调用device_suspend继续休眠操作。休眠后，将对设备引用计数做自减。

static int __device_suspend(struct device *dev, pm_message_t state, bool async)
{

    pm_callback_t callback = NULL;
    struct timer_list timer;
    
    init_timer_on_stack(&timer);
	timer.expires = jiffies + HZ * 12;
	timer.function = dpm_drv_timeout;
	timer.data = (unsigned long)&data;
	add_timer(&timer);
	
	if (dev->pm_domain) {
		info = "power domain ";
		callback = pm_op(&dev->pm_domain->ops, state);
		goto Run;
	}

	if (dev->type && dev->type->pm) {
		info = "type ";
		callback = pm_op(dev->type->pm, state);
		goto Run;
	}

	if (dev->class) {
		if (dev->class->pm) {
			info = "class ";
			callback = pm_op(dev->class->pm, state);
			goto Run;
		} else if (dev->class->suspend) {
			pm_dev_dbg(dev, state, "legacy class ");
			error = legacy_suspend(dev, state, dev->class->suspend,
						"legacy class ");
			goto End;
		}
	}

	if (dev->bus) {
		if (dev->bus->pm) {
			info = "bus ";
			callback = pm_op(dev->bus->pm, state);
		} else if (dev->bus->suspend) {
			pm_dev_dbg(dev, state, "legacy bus ");
			error = legacy_suspend(dev, state, dev->bus->suspend,
						"legacy bus ");
			goto End;
		}
	}

 Run:
	if (!callback && dev->driver && dev->driver->pm) {
		info = "driver ";
		callback = pm_op(dev->driver->pm, state);
	}

	error = dpm_run_callback(callback, dev, state, info);
	...
}

此处按照pm_domain->class->bus->driver的优先级，取得pm.suspend操作。对于SPI，I2C设备,pm操作在bus, class都是没有实现的，最终调用到device_driver中的suspend中。也就是我们最开始填写的dev_pm_ops中的resume和suspend函数。设备挂起到RAM操作就结束了。

除了挂起到RAM，dev_pm_ops中还有runtime resume和runtime suspend，也就是运行时PM。运行PM和挂起到RAM时的PM不太一样，运行时PM是针对单个设备，系统在非睡眠状态的情况下，某个设备在空闲时进入运行时挂起状态，而在不空闲时执行运行时回复，使得设备进入正常工作状态。如此，这个设备在运行时会更省电。