some notes when studying suspend/resume

/sys/devices/.../power/control  

drivers/base/power/sysfs.c

auto: to allow the devices to be power managed by run time

on: to prevent the devices from being power managed by run time, the realization way is to add usage_count to prevent power_down/runtime_suspend

/sys/devices/.../power/wakeup

drivers/base/power/sysfs.c

drivers/base/power/wakeup.c

/sys/kernel/debug/wakeup_sources

enable:

disable:

not find the ture useness.

async_synchronize_full()

async_synchronize_full - synchronize all asynchronous function calls

This function waits until all asynchronous function calls have been done.

device_enable_async_suspend()

to run suspend function in no dependacy and calling sleep/block

/sys/kernel/debug/suspend_stats

kernel/power/main.c

used to monitor the suspend/resume states.

kernel/power/suspend.c

suspend_set_ops() is used to set platform_suspend_ops *suspend_ops, it contains the "enter" operation, which is used to set S3 state.

kernel/power/suspend.c

pm_suspend() is the entrance of suspend() function.

drivers/base/power/main.c

device_pm_add() is used to add the device which will run suspend/resume to the dpm_list.

platform_device_register()->platform_device_register()->dev_set_name()

power.0 is a plantform device registered by kernel/power/wakelock.c

arch/x86/kernel/smpboot.c

arch/x86/kernel/process_32.c

drivers/idle/intel_idle.c

start_secondary()->cpu_idle()-> pm_idle()->cpuidle_idle_call()-> target_state->enter()

drivers/cpuidle/sysfs.c

drivers/cpuidle/governors/ladder.c

drivers/cpuidle/governors/menu.c   //default setting for idle governor

/sys/devices/system/cpu/cpuidle/current_governor_ro

next_state = cpuidle_curr_governor->select(dev); /* ask the governor for the next state */

drivers/cpufreq/cpufreq.c

drivers/cpufreq/cpufreq_userspace.c

echo userspace > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor

echo 1600000   > /sys/devices/system/cpu/cpu1/cpufreq/scaling_setspeed