Qualcomm thermal的介绍

                                                                   Qualcomm thermal的介绍

      随着智能设备的发展，很多产品越做越小，在一块很小的板子和结构内会附带很多功能，加上很多应用需要处理数据很大，必定造成设备很容易发热。且发热不仅会带来体验的不足还会带来大量的电量消耗。所以在移动消费产品，对热量的解决方案还是很重要的。今天我们就介绍下高通410C开发板上热量管理的机制，软件如何控制。

       首先在软件架构上我们分为以下部分。

一：驱动部分

驱动代码位置：

共有部分：kernel/deriver/thermal_core.c

                    kernel/driver/user_space.c

platform部分：

                mitigation部分：

                 kernel/driver/msm_thermal-dev.c

                 kernel/driver/msm_thermal.c

                sensor部分：

                qpnp-temp-alarm.c

                msm8974-tsens.c

简要叙述驱动文件关键位置，基本理解驱动构架以及如何运作。

msm8974-tsens.c

1：sensor的注册，将thermal注册为class类：

2：thermal_zone_device_ops接口的填充，以给thermal_core的调用。

3：触发中断与通知上层。

thermal_core.c

1：thermal_zone_devices_register（）提供相应sensor register接口，按先后顺序注册/sys/class/thermal/thermal_zoneN。

2：为上层提供trip_temp、type等接口,如：   

      root@msm8916_32:/ # cat /sys/class/thermal/thermal_zone5/trip_point_0_type     

      configurable_hi

      root@msm8916_32:/ # cat /sys/class/thermal/thermal_zone5/trip_point_0_temp     

      85

这里trip_point_0_temp为high temp thresthold,那么trip_point_1_temp为low temp thresthold  

3：调用sensor相关接口：

关于tpye name与thermal_zone的对应关系：

以8016为例:

                                   type                         name                                  

                          thermal_zone0           battery

                          thermal_zone1           tsens_tz_sensor0

                          thermal_zone2           tsens_tz_sensor1

                          thermal_zone3           tsens_tz_sensor2        pop_mem

                          thermal_zone4           tsens_tz_sensor4        cpu2-3

                          thermal_zone5           tsens_tz_sensor5        cpu0-1

                          thermal_zone6           pm8916_tz

                          thermal_zone7           bms

msm_thermal.c

mitigation分为轮训与中断，首先看下poll部分：

每隔指定的时间就会check温度，当超过范围会降频,关闭cpu，重设thsreshold等操作。

interrupt部分，默认启动的是中断方式，这里是interrupt的初始化。

再以hotplug细说interrupt的运行，这里会根据dts里面的参数先设置，后通过set_threshold调用之前sensor填充的接口,写入到寄存器里面，当达到温度后，触发中断后会通知上层，再温度非常高的话会运行起kernel mitigation，通过hotplug notify来做相应处理

二：上层thermal-engine部分：

 

       thermal工作机制：首先上层设置trig point，包括恢复温度与临界温度，thermal engine主要有algo_monitor运作sensor monitor起辅助作用够成。algo_monitor依赖sensor_monitor的运作,而sensor_monitor的运行依赖中断的触发。

1：首先初始化engine,设置临界温度（85度）其它参数等信息。

2：当温度升高达到临界温度，底层中断触发同时通知上层，algo_monitor由block状态到运行起来，同时启动50ms的timer来采样sensor的温度:

a:当温度大于trig point时，cpu开始降频，每次温度下降一个等级。

b:当温度小于trig point时，cpu可以恢复到max状态，当时取消timer同时将trig point恢复到85度。

关于set_point部分的设定。

ss-data.c中，关于temp threshold的深入。

下面讨论高温threshold

1：首先上层set_point（默认85度）会写入/sys/class/thermal/thermal_zoneX/trip_point_0_temp。

2：当sensor温度达到85度后，触发中断通知上层与kernel mitigation

流程：tsens_isr->thermal_sensor_trip->thermal_sensor_trip->__update_sensor_thresholds   

由于85度，并没有达到kernel mitigation与cpu hotplug的95度thresthold，那里这里的pos->notify()并不会跑进mitigation与hotplug。

在__update_sensor_thresholds（）里面会将trip_point_0_temp由85度改为95度。这时中断不再触发，需要等到高温95度触发中断。此时thermal-engine已经在运转，超过85度已经启用cpu降频。很快cpu温度会从85度以上降低下来，假如降低到84度时，那么trip_point_0_temp是不是应该重新置为85度呢？

下面贴出重新设回threshold的方法：

在ss_algorithm.c的handle_thresh_sig（）里面会去处理同时重新设定cpu可以跑到最高freq。

假如机器温度实在是太高超过了95度，那么kernel mitigation与cpu hotplug将会运转起来。

利用msm_thermal.c提供的cpus_offlined接口关闭指定的cpu。

禁用cpu3 :

       echo 8 >  sys/module/msm_thermal/core_control/cpus_offlined   

CPU升温：

       cat /dev/zero > /dev/null &    执行30次

打开feature:

       stop thermal-engine 

       thermal-engine --debug & 
       logcat -v time -s ThermalEngine 

logcat现场分析，第一段logcat分析：

下面的由于温度达到了pop_mem的70度，那么algo_monitor就运作起来了，

03-07 22:01:44.619 D/ThermalEngine( 2110): tsens_uevent: tsens_tz_sensor2

03-07 22:01:44.619 D/ThermalEngine( 2110): sensor_monitor: tsens_tz_sensor2 Reading 70000 .

03-07 22:01:44.619 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:01:44.619 D/ThermalEngine( 2110): thresh_notify: SS Id SS-POPMEM, Update recieved pop_mem 70000  //底层中断激活

03-07 22:01:44.619 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(0), Hi(0) 2147483647, Lo(0) -2147483648, Interval(0) 2147483647

03-07 22:01:44.619 D/ThermalEngine( 2110): sensor_monitor: tsens_tz_sensor2 Wait for client request.

03-07 22:01:44.619 D/ThermalEngine( 2110): algo_monitor: Thresh EVT

03-07 22:01:44.619 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC  //pop_mem 70度

03-07 22:01:44.619 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM, Read pop_mem 70000mC

03-07 22:01:44.619 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM Transition State 2

03-07 22:01:44.619 D/ThermalEngine( 2110): sensors_manager_set_thresh_lvl: tsens_tz_sensor2 Hi(0) 0, Lo(1) 45000, Interval(0) 0

03-07 22:01:44.619 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(1), Hi(0) 2147483647, Lo(1) 45000, Interval(0) 2147483647    //设置低温point45度触发中断

03-07 22:01:44.619 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_0_type)

03-07 22:01:44.619 D/ThermalEngine( 2110): TSENS threshold at 0 enabled: 0

03-07 22:01:44.619 D/ThermalEngine( 2110): Setting up TSENS thresholds low: 45

03-07 22:01:44.619 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_1_type)

03-07 22:01:44.619 D/ThermalEngine( 2110): TSENS threshold at 1 enabled: 1 //写入低有效 45度触发

//上面的中断触发激活了timer

03-07 22:01:44.619 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

03-07 22:01:44.619 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:01:44.619 D/ThermalEngine( 2110): sensor_monitor: tsens_tz_sensor2 Sensor wait.

03-07 22:01:45.619 D/ThermalEngine( 2110): algo_monitor: Timer EVT

03-07 22:01:45.619 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:69000 mC

03-07 22:01:45.619 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM Read pop_mem 69000mC, Err 1000mC, SampleCnt 1//此时读取到的69度

03-07 22:01:45.619 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM, E0 1000mC, E1 0mC

03-07 22:01:45.619 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:69000 mC

03-07 22:01:45.619 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM, Read pop_mem 69000mC

03-07 22:01:45.619 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM Transition State 3  //stop采样，重新设置70度高有效

03-07 22:01:45.619 D/ThermalEngine( 2110): device_clnt_cancel_request: DEV cpu

03-07 22:01:45.619 D/ThermalEngine( 2110): sensors_manager_set_thresh_lvl: tsens_tz_sensor2 Hi(1) 70000, Lo(0) 0, Interval(0) 0

03-07 22:01:45.619 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(1), Hi(1) 70000, Lo(0) -2147483648, Interval(0) 2147483647

03-07 22:01:45.619 D/ThermalEngine( 2110): Setting up TSENS thresholds high: 70

03-07 22:01:45.619 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_0_type)

03-07 22:01:45.619 D/ThermalEngine( 2110): TSENS threshold at 0 enabled: 1

03-07 22:01:45.619 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_1_type)

03-07 22:01:45.619 D/ThermalEngine( 2110): TSENS threshold at 1 enabled: 0

//温度在70以下取消timer

03-07 22:01:45.619 D/ThermalEngine( 2110): settimer: Start timer 0.000(sec)

//下一轮的中断激活

03-07 22:01:45.619 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:01:49.629 D/ThermalEngine( 2110): tsens_uevent: tsens_tz_sensor2

03-07 22:01:49.629 D/ThermalEngine( 2110): sensor_monitor: tsens_tz_sensor2 Reading 70000 .

03-07 22:01:49.629 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:01:49.629 D/ThermalEngine( 2110): thresh_notify: SS Id SS-POPMEM, Update recieved pop_mem 70000

03-07 22:01:49.629 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(0), Hi(0) 2147483647, Lo(0) -2147483648, Interval(0) 2147483647

03-07 22:01:49.629 D/ThermalEngine( 2110): sensor_monitor: tsens_tz_sensor2 Wait for client request.

03-07 22:01:49.629 D/ThermalEngine( 2110): algo_monitor: Thresh EVT

03-07 22:01:49.629 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:01:49.629 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM, Read pop_mem 70000mC

03-07 22:01:49.629 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM Transition State 2

03-07 22:01:49.629 D/ThermalEngine( 2110): sensors_manager_set_thresh_lvl: tsens_tz_sensor2 Hi(0) 0, Lo(1) 45000, Interval(0) 0

03-07 22:01:49.629 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(1), Hi(0) 2147483647, Lo(1) 45000, Interval(0) 2147483647

03-07 22:01:49.629 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_0_type)

03-07 22:01:49.629 D/ThermalEngine( 2110): TSENS threshold at 0 enabled: 0

03-07 22:01:49.629 D/ThermalEngine( 2110): Setting up TSENS thresholds low: 45

03-07 22:01:49.629 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_1_type)

03-07 22:01:49.629 D/ThermalEngine( 2110): TSENS threshold at 1 enabled: 1

//设定1s timer

03-07 22:01:49.629 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

03-07 22:01:49.629 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:01:49.629 D/ThermalEngine( 2110): sensor_monitor: tsens_tz_sensor2 Sensor wait.

03-07 22:01:50.629 D/ThermalEngine( 2110): algo_monitor: Timer EVT

03-07 22:01:50.629 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:01:50.629 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM Read pop_mem 70000mC, Err 0mC, SampleCnt 1

03-07 22:01:50.629 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM, E0 0mC, E1 0mC

第二段logcat分析：

采样到71度启动降频,这时底层会设置高位point为95度

03-07 22:04:15.669 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

03-07 22:04:15.669 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:04:16.669 D/ThermalEngine( 2110): algo_monitor: Timer EVT

03-07 22:04:16.669 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:71000 mC

03-07 22:04:16.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM Read pop_mem 71000mC, Err -1000mC, SampleCnt 1

03-07 22:04:16.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM, E0 -1000mC, E1 0mC

03-07 22:04:16.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu, op_value 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu0, op_value 1152000

03-07 22:04:16.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[0] to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): CPU[0] frequency limited to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu1, op_value 1152000

03-07 22:04:16.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[1] to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): CPU[1] frequency limited to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu2, op_value 1152000

03-07 22:04:16.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[2] to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): CPU[2] frequency limited to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu3, op_value 1152000

03-07 22:04:16.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[3] to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): CPU[3] frequency limited to 1152000

03-07 22:04:16.669 D/ThermalEngine( 2110): increase_mitigation_lvl: SS Id SS-POPMEM, Device cpu, Req. Freq 1152000Khz, Applied freq 1152000kHz, TC delay 1

03-07 22:04:16.669 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

下一轮timer:

刚才降频了，这次采样到温度恢复到了70度，那么cpu恢复最大值，同时设置低有效45度。

03-07 22:04:16.669 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:04:17.669 D/ThermalEngine( 2110): algo_monitor: Timer EVT

03-07 22:04:17.669 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:04:17.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM Read pop_mem 70000mC, Err 0mC, SampleCnt 1

03-07 22:04:17.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM, E0 0mC, E1 -1000mC

03-07 22:04:17.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu, op_value 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu0, op_value 1209600

03-07 22:04:17.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[0] to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): CPU[0] frequency limited to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu1, op_value 1209600

03-07 22:04:17.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[1] to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): CPU[1] frequency limited to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu2, op_value 1209600

03-07 22:04:17.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[2] to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): CPU[2] frequency limited to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): devices_manager_set_op_value: DEV cpu3, op_value 1209600

03-07 22:04:17.669 I/ThermalEngine( 2110): ACTION: CPU - Setting CPU[3] to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): CPU[3] frequency limited to 1209600

03-07 22:04:17.669 D/ThermalEngine( 2110): decrease_mitigation_lvl: SS Id SS-POPMEM, Device cpu, Req. Freq 1209600Khz, Applied freq 1209600kHz, TC delay 0

03-07 22:04:17.669 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:04:17.669 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM, Read pop_mem 70000mC

03-07 22:04:17.669 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM Transition State 2

03-07 22:04:17.669 D/ThermalEngine( 2110): sensors_manager_set_thresh_lvl: tsens_tz_sensor2 Hi(0) 0, Lo(1) 45000, Interval(0) 0

03-07 22:04:17.669 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(1), Hi(0) 2147483647, Lo(1) 45000, Interval(0) 2147483647

03-07 22:04:17.669 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_0_type)

03-07 22:04:17.669 D/ThermalEngine( 2110): TSENS threshold at 0 enabled: 0

03-07 22:04:17.669 D/ThermalEngine( 2110): Setting up TSENS thresholds low: 45

03-07 22:04:17.669 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_1_type)

03-07 22:04:17.669 D/ThermalEngine( 2110): TSENS threshold at 1 enabled: 1

03-07 22:04:17.669 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

03-07 22:04:17.669 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:04:18.669 D/ThermalEngine( 2110): algo_monitor: Timer EVT

03-07 22:04:18.669 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:04:18.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM Read pop_mem 70000mC, Err 0mC, SampleCnt 1

03-07 22:04:18.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM, E0 0mC, E1 0mC

03-07 22:04:18.669 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

03-07 22:04:18.669 D/ThermalEngine( 2110): algo_monitor: Wait for EV

03-07 22:04:19.669 D/ThermalEngine( 2110): algo_monitor: Timer EVT

03-07 22:04:19.669 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:04:19.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM Read pop_mem 70000mC, Err 0mC, SampleCnt 1

03-07 22:04:19.669 D/ThermalEngine( 2110): handle_timer_sig: SS Id SS-POPMEM, E0 0mC, E1 0mC

03-07 22:04:19.669 I/ThermalEngine( 2110): Sensor:tsens_tz_sensor2:70000 mC

03-07 22:04:19.669 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM, Read pop_mem 70000mC

03-07 22:04:19.669 D/ThermalEngine( 2110): handle_thresh_sig: SS Id SS-POPMEM Transition State 2

03-07 22:04:19.669 D/ThermalEngine( 2110): sensors_manager_set_thresh_lvl: tsens_tz_sensor2 Hi(0) 0, Lo(1) 45000, Interval(0) 0

03-07 22:04:19.669 D/ThermalEngine( 2110): update_active_thresh: tsens_tz_sensor2 Active(1), Hi(0) 2147483647, Lo(1) 45000, Interval(0) 2147483647

03-07 22:04:19.669 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_0_type)

03-07 22:04:19.669 D/ThermalEngine( 2110): TSENS threshold at 0 enabled: 0

03-07 22:04:19.669 D/ThermalEngine( 2110): Setting up TSENS thresholds low: 45

03-07 22:04:19.669 D/ThermalEngine( 2110): enable_threshold: tsens_tz_sensor2 (/sys/devices/virtual/thermal/thermal_zone3/trip_point_1_type)

03-07 22:04:19.669 D/ThermalEngine( 2110): TSENS threshold at 1 enabled: 1

03-07 22:04:19.669 D/ThermalEngine( 2110): settimer: Start timer 1.000(sec)

03-07 22:04:19.669 D/ThermalEngine( 2110): algo_monitor: Wait for EV

总结：

        以上就是高通的thermal机制的介绍，它可以在温度过高时，通过软件对多核CPU的降频和关闭，达到降低温度一个目的。希望通过本文对大家能有一定帮助，对高通的thermal机制有个大概的了解，并对自己的产品做一个热量控制调试。