[20191207][Android10]高德地图导航功耗场景优化

1. 测试现象

1.1 EPRODUCING PROCEDURES:

1.进入高德 Map(data) wifi)，起点为自己所在位置，搜索一个位置进行导航；
2.等待30秒后开始记录电流，持续5分钟；
3.按Power键，灭屏待机；
4.手机灭屏15秒后开始记录电流，持续3分钟；

1.2 Test numberSummary

设备	高德地图导航
测试机	461.82 mA
对比机	369.13 mA

2.分析

2.1 拔屏以扣除LCD影响

设备	高德地图导航
测试机+450nit带屏幕导航	405.18 mA
对比机+450nit带屏幕导航	380.11
测试机+拔屏	174.45 mA
对比机+拔屏	131.76 mA

从上述看扣掉 LCD 同样有功耗差异，即LCD 关系不大

注意：导航场景：一般是带导航语音，即需要考虑Audio的影响

2.2 设计实验找出问题差异

以下是我自己设计的实验方法

1).对比机与测试机的GPS搜星实验

GPS puls 搜星查看，我apk放到附件中
只开启GPS+飞行模式
尽量2台机器相同环境下同时测试
4.放置5分钟后，apk主界面截图

GPS plus+工具	搜星情况
对比机无GPS信号增强器	观察卫星数量和信号强度
对比机有GPS信号增强器	观察卫星数量和信号强度
测试机无GPS信号增强器	观察卫星数量和信号强度
测试机有GPS信号增强器	观察卫星数量和信号强度

实验发现：基本无差异，故 GPS 单体应该是没问题

2).导航电流测试

最暗亮度测试，避免扣掉屏幕带来的影响
静音+插入耳机，去除导航语音带来的影响
尽量2台机器相同环境下同时测试

测试项目	测试机器(mA)	对比机(mA)
case 1 静态HOME界面固定墙纸_飞行模式	65.21	68.19
case 2网络定位条件，高德导航测试：只保持WiFi连接，飞行模式	87.63	88.95
case 3 GPS定位条件，高德导航测试：开启GPS+插卡+WiFi 连接	147.63	120.71
case 4 GPS定位条件，高德导航测试：GPS增强器+开启GPS+插卡+WiFi 连接	169.64	153.48

case 1 目的: 对比相同亮度下是否存在差异
case 2 目的: 查看是否网络定位有问题
case 3 目的：查看是否GPS定位存问题
case 3 目的: 查杀是否与GPS信号强度有关，因为室内是GPS信号比较弱

上述：

case 1 说明测试机和对比机亮屏下相同亮度的测试电流是基本一致，即基础的系统功耗是相同的
case 2 说明网络定位是OK的，说明相同APK在不同机器测试，运行是后台也基本保持一致,且说明如果GPS不开启则是导航是好的；
case 3、case 4 说明GPS开启，会带来导航功耗影响

3).GPS单体功耗

这块数据没有，故需要让硬件提供：

测试步骤：

Wi-Fi Off, GPS ON, BT OFF, NFC OFF, Fly mode ON.
灭屏，等待1min，记录平均电流，即是打开GPS时的电流。
在上述的条件时，记录下可以稳定复现的最小电流值，即是 GPS floor current.
Average current when GPS navigation working：
  	 Step1:通过下拉菜单，打开飞行模式，打开位置服务；
Step2:进入工程模式 adb shell am start -n com.mediatek.ygps/com.mediatek.ygps.YgpsActivity；
Step3:进入Location->YGPS->Enable in BG.(need restart)，重启后再次进入该界面;
Step4:关闭屏幕，待电流稳定，测试3分钟，记录电流A1；
Step5:通过下拉菜单，关闭位置服务；
Step6:进入工程模式，进入Location->YGPS，再回到YGPS界面；
Step7:关闭屏幕，待电流稳定，测试3分钟，记录电流A2；
GPS ON=A1-A2

由于不知道 ygps 源码，通过测试步骤基本猜到，主要是测试GPS开启电流和工作电流，相当于高德地图的定位场景，非导航场景

测试数据：

设备	平均电流
测试机	34mA
对比机	34mA

从上面数据看，GPS的工作电流是基本一致的，即GPS单体是OK的

这就奇怪了，GPS单体是好的，为什么导航一开GPS就存问题呢？还需要再排除干扰。例如以下干扰：

WiFi 环境因素
4G 插卡因素
应用启动后会进行页面、地图数据、配置文件下载
测试过程中，如果网络发生变化，热点不稳定

4).离线导航

Wi-Fi Off, GPS ON, BT OFF, NFC OFF, Fly mode ON，排除WiFi带来的干扰
最小亮度
使用 what_temp apk抓CPU数据
静音插耳机
使用高德离线地图导航

设备	测试机	对比机
高德地图定位界面	55.13mA	50.13mA
高德地图导航界面	83.00mA	60.49mA

上述查看：

现象复现，即排除了Wi-Fi Off, GPS ON, BT OFF, NFC OFF, Fly mode影响，即wifi、4G都没问题
定位场景差异不大，导航场景差异比较大。
需要查看高德地图定位细节

2.3 高德地图的dump信息

dumpsys location|grep -B 2 “com.autonavi.minimap”

Tokyo_Lite:/ $ dumpsys location|grep -B 2 "com.autonavi.minimap"
  Battery Saver Location Mode: NO_CHANGE
  Location Listeners:
    Reciever[35eed63 listener UpdateRecord[passive com.autonavi.minimap(10167 foreground) Request[POWER_NONE passive fastest=0] null], isNeedBgGpsRestrict false monitoring location: true]
    Reciever[fe54b41 listener UpdateRecord[passive android(1000 foreground) Request[POWER_NONE passive fastest=+30m0s0ms] null], isNeedBgGpsRestrict false monitoring location: true]
    Reciever[a523146 listener UpdateRecord[passive com.autonavi.minimap(10167 foreground) Request[POWER_NONE passive fastest=+1s0ms] null], isNeedBgGpsRestrict false monitoring location: true]
    Reciever[4698940 listener UpdateRecord[passive android(1000 foreground) Request[POWER_NONE passive fastest=0] null], isNeedBgGpsRestrict false monitoring location: false]
    Reciever[3e939ea listener monitoring location: false]
    Reciever[a629ca9 listener UpdateRecord[gps com.autonavi.minimap(10167 foreground) Request[ACCURACY_FINE gps requested=+1s0ms fastest=+1s0ms] null], isNeedBgGpsRestrict false monitoring location: true]
  Active Records by Provider:
    gps:
      UpdateRecord[gps com.autonavi.minimap(10167 foreground) Request[ACCURACY_FINE gps requested=+1s0ms fastest=+1s0ms] null], isNeedBgGpsRestrict false
--
      UpdateRecord[passive android(1000 foreground) Request[POWER_NONE passive fastest=0] null], isNeedBgGpsRestrict false
      UpdateRecord[passive android(1000 foreground) Request[POWER_NONE passive fastest=+30m0s0ms] null], isNeedBgGpsRestrict false
      UpdateRecord[passive com.autonavi.minimap(10167 foreground) Request[POWER_NONE passive fastest=0] null], isNeedBgGpsRestrict false
      UpdateRecord[passive com.autonavi.minimap(10167 foreground) Request[POWER_NONE passive fastest=+1s0ms] null], isNeedBgGpsRestrict false
--
  Active GnssNavigationMessage Listeners:
  Active GnssStatus Listeners:
    5600 10167 com.autonavi.minimap: false
    5600 10167 com.autonavi.minimap: false
    5600 10167 com.autonavi.minimap: false
    5766 10167 com.autonavi.minimap: false
  Historical Records by Provider:
    android: passive: Min interval 0 seconds: Max interval 1800 seconds: Duration requested 99 total, 99 foreground, out of the last 99 minutes: Currently active
    com.autonavi.minimap: passive: Min interval 0 seconds: Max interval 1 seconds: Duration requested 2 total, 2 foreground, out of the last 98 minutes: Currently active
    com.autonavi.minimap: gps: Interval 1 seconds: Duration requested 2 total, 2 foreground, out of the last 98 minutes: Currently active

查看高德地图的dump信息，我们基本知道高德地图申请了哪些定位：

GPS 定位，定位间隔1秒一个
Reciever[a629ca9 listener UpdateRecord[gps com.autonavi.minimap(10167 foreground) Request[ACCURACY_FINE gps requested=+1s0ms fastest=+1s0ms] null], isNeedBgGpsRestrict false monitoring location: true]
passive 定位
Reciever[a523146 listener UpdateRecord[passive com.autonavi.minimap(10167 foreground) Request[POWER_NONE passive fastest=+1s0ms] null], isNeedBgGpsRestrict false monitoring location: true]
监听了 GnssStatus 数据
Active GnssStatus Listeners:
5600 10167 com.autonavi.minimap: false
5600 10167 com.autonavi.minimap: false
5600 10167 com.autonavi.minimap: false
5766 10167 com.autonavi.minimap: false

2.4 分析高德地图定位和导航调用的函数

我们在dump locaiton已经知道高德地图会申请使用GPS，并且定位间隔为1秒，故不需要在

1) LocationManager 埋点

public class LocationManager {

    public boolean addNmeaListener(
        @NonNull OnNmeaMessageListener listener, @Nullable Handler handler) {
    boolean result;

    // huazhi.su
    if("true".equals(SystemProperties.get("persist.sys.tct.addNmeaListener.debug", "false"))) {
        Log.e(TAG, "skip addNmeaListener");
        return false;
    }
    // huazhi.su
    
    @RequiresPermission(ACCESS_FINE_LOCATION)
    public boolean registerGnssStatusCallback(
            @NonNull GnssStatus.Callback callback, @Nullable Handler handler) {
        boolean result;
        synchronized (mGnssStatusListeners) {
            // huazhi.su
            if("true".equals(SystemProperties.get("persist.sys.tct.registerGnssStatusCallback.debug", "false"))) {
                if("com.autonavi.minimap".equals(mContext.getPackageName())) {
                    Log.e(TAG, "skip registerGnssStatusCallback  packageName:" + mContext.getPackageName());
                }
                return false;
            }
            // huazhi.su

2) 查看导航场景高德地图使用的sdk情况

场景	API接口
定位场景	requestLocationUpdates、registerGnssStatusCallback
导航场景	requestLocationUpdates、registerGnssStatusCallback、addNmeaListener

即导航场景比定位场景多了 addNmeaListener

12-06 14:10:44.728  3716  3716 W System.err:    at android.location.LocationManager.addNmeaListener(LocationManager.java:2046)
12-06 14:10:44.736  3716  4034 W System.err:    at android.location.LocationManager.registerGnssStatusCallback(LocationManager.java:1944)

    locationManager.registerGnssStatusCallback(new GnssStatus.Callback() {
        @Override
        public void onSatelliteStatusChanged(GnssStatus status) {
            super.onSatelliteStatusChanged(status);
            Log.d(TAG, "onSatelliteStatusChanged");
        }
    });

    locationManager.addNmeaListener(new OnNmeaMessageListener() {
        @Override
        public void onNmeaMessage(String message, long timestamp) {
            Log.d(TAG, "onNmeaMessage message :" + message + ", timestamp :" + timestamp);
            sendRequestWithHttpClient();
        }
    });
    
    locationManager.requestLocationUpdates(GPS_PROVIDER, 1000, 0, mMyLocationListener);

之前的测试GPS单体的功耗基本是 requestLocationUpdates 这个接口的功能，故基本问题不大
且导航和定位场景中，导航场景存在和定位场景不同，故我们可以单独屏蔽掉 registerGnssStatusCallback、addNmeaListener

3. 问题根源研究

3.1 禁止监听 GnssStatus 和 Nmea

测试项	平均电流	备注
离线导航	133	异常电流波形
离线导航+禁止监听 GnssStatus	120	异常电流波形
离线导航+禁止监听 Nmea	110	异常电流波形
离线导航+禁止监听 Nmea + GnssStatus	100	正常电流波形

首先我们看下异常电流：特征是一秒钟一个波峰，且单个异常波形平均电流就有166mA，基本功耗不被拉大才怪

上述说明：监听 Nmea + GnssStatus 会带来功耗，但是肯定要给第三方应用正常使用这个数据

3.1 拦截 GnssStatus 和 Nmea

拦截位置

package com.android.server.location;

public class GnssLocationProvider ...{

    @NativeEntryPoint // libgnss.so上报
    private void reportNmea(long timestamp) {
        // 新增拦截
        ...
        // 对应上层的：public void onNmeaMessage(String message, long timestamp) 方法
        ...
    }
    
    @NativeEntryPoint // libgnss.so上报
    private void reportSvStatus(int svCount, int[] svidWithFlags, float[] cn0s,
        // 新增拦截
        ...
        // 对应上层的：public void onSatelliteStatusChanged(GnssStatus status) { 方法
        ...
    }

}

测试项	平均电流	备注
离线导航	133	异常电流波形
离线导航+拦截 GnssStatus 数据上报	123	异常电流波形
离线导航+拦截 Nmea 数据上报	110	异常电流波形
离线导航+拦截 Nmea + GnssStatus 数据上报	100	正常电流波形

拦截测试也说明 Nmea + GnssStatus 有影响，难道高德地图会收到数据做一些耗电操作，还是对比机有优化呢？
由于高德地图、对比机没有源码，所以我们无法知道，从源码一个函数一个函数地分析。看是否源代码有问题，即了解GPS的Nmea + GnssStatus上报给上层的源码

4. 源码分析并新增日志

4.1 源码数据通路新增日志

例如 Nmea 数据如何上报给上层apk，在这个数据传递的通路新增日志，每个函数调用的地方都加

1.数据源头-------------------------------------------------
package com.android.server.location;

    @NativeEntryPoint
    private void reportNmea(long timestamp) {
        if (!mItarSpeedLimitExceeded) {
            int length = native_read_nmea(mNmeaBuffer, mNmeaBuffer.length);
            String nmea = new String(mNmeaBuffer, 0 /* offset */, length);
            mGnssStatusListenerHelper.onNmeaReceived(timestamp, nmea);
        }
    }

2.-------------------------------------------------
public abstract class GnssStatusListenerHelper extends RemoteListenerHelper<IGnssStatusListener> {
        public void onNmeaReceived(final long timestamp, final String nmea) {
        // 遍历 addNmeaListener 监听数量，例如高德地图使用了2个，一个定位注册的，一个导航场景注册的
        foreach((IGnssStatusListener listener, CallerIdentity callerIdentity) -> {
            // 检查是否与定位权限
            if (hasPermission(mContext, callerIdentity)) {
                logPermissionDisabledEventNotReported(TAG, callerIdentity.mPackageName, "NMEA");
                return;
            }
            // 消息传递到上层
            listener.onNmeaReceived(timestamp, nmea);
        });
    }
}
3.-------------------------------------------------
public class LocationManager {

    @Override
    public void onNmeaReceived(long timestamp, String nmea) {
        ...
        mGnssHandler.obtainMessage(NMEA_RECEIVED).sendToTarget();
        ...
    }
    
    for (Nmea nmea : mNmeaBuffer) {
        // 数据给上层
        mGnssNmeaListener.onNmeaMessage(nmea.mNmea, nmea.mTimestamp);
    }

4.2 现象发现

1）日志大量打印，这里1秒钟70组数据从底层上报出来，都需要经过上述的函数，简直是太频繁

01-01 00:04:54.516  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GNGGA,160454.011,8960.0000,N,00000.0000,E,0,0,,137.0,M,13.0,M,,*5F
01-01 00:04:54.516  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GNGGA,160454.011,8960.0000,N,00000.0000,E,0,0,,137.0,M,13.0,M,,*5F
01-01 00:04:54.518  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GNGGA,160454.011,8960.0000,N,00000.0000,E,0,0,,137.0,M,13.0,M,,*5F
01-01 00:04:54.519  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GPGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*30
01-01 00:04:54.520  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GPGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*30
01-01 00:04:54.522  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GPGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*30
01-01 00:04:54.522  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GPGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*30
01-01 00:04:54.522  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GLGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*2C
01-01 00:04:54.524  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GLGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*2C
01-01 00:04:54.524  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GLGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*2C
01-01 00:04:54.526  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GLGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*2C
01-01 00:04:54.526  3740  4447 E LocationManager: skip onNmeaReceived: nmea$BDGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.527  3328  4051 E LocationManager: skip onNmeaReceived: nmea$BDGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.527  3328  4051 E LocationManager: skip onNmeaReceived: nmea$BDGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.527  3328  4561 E LocationManager: skip onNmeaReceived: nmea$BDGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.528  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GAGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.529  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GAGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.529  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GAGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.529  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GAGSA,A,1,,,,,,,,,,,,,99.99,99.99,99.99*21
01-01 00:04:54.531  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GPGSV,1,1,0*49
01-01 00:04:54.531  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GPGSV,1,1,0*49
01-01 00:04:54.531  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GPGSV,1,1,0*49
01-01 00:04:54.533  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GLGSV,1,1,0*55
01-01 00:04:54.533  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GLGSV,1,1,0*55
01-01 00:04:54.534  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GPGSV,1,1,0*49
01-01 00:04:54.534  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GLGSV,1,1,0*55
01-01 00:04:54.535  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GLGSV,1,1,0*55
01-01 00:04:54.535  3740  4447 E LocationManager: skip onNmeaReceived: nmea$BDGSV,1,1,0*58
01-01 00:04:54.536  3328  4051 E LocationManager: skip onNmeaReceived: nmea$BDGSV,1,1,0*58
01-01 00:04:54.536  3328  4561 E LocationManager: skip onNmeaReceived: nmea$BDGSV,1,1,0*58
01-01 00:04:54.536  3328  4561 E LocationManager: skip onNmeaReceived: nmea$BDGSV,1,1,0*58
01-01 00:04:54.537  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GAGSV,1,1,0*58
01-01 00:04:54.537  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GAGSV,1,1,0*58
01-01 00:04:54.538  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GAGSV,1,1,0*58
01-01 00:04:54.538  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GAGSV,1,1,0*58
01-01 00:04:54.539  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GNRMC,160454.011,V,8960.0000,N,00000.0000,E,0.000,0.00,311218,,,N*55
01-01 00:04:54.539  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GNRMC,160454.011,V,8960.0000,N,00000.0000,E,0.000,0.00,311218,,,N*55
01-01 00:04:54.539  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GNRMC,160454.011,V,8960.0000,N,00000.0000,E,0.000,0.00,311218,,,N*55
01-01 00:04:54.540  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GNVTG,0.00,T,,M,0.000,N,0.000,K,N*2C
01-01 00:04:54.541  3328  4561 E LocationManager: skip onNmeaReceived: nmea$GNVTG,0.00,T,,M,0.000,N,0.000,K,N*2C
01-01 00:04:54.542  3328  3345 E LocationManager: skip onNmeaReceived: nmea$GNVTG,0.00,T,,M,0.000,N,0.000,K,N*2C
01-01 00:04:54.542  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GNRMC,160454.011,V,8960.0000,N,00000.0000,E,0.000,0.00,311218,,,N*55
01-01 00:04:54.542  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GNVTG,0.00,T,,M,0.000,N,0.000,K,N*2C
01-01 00:04:54.542  3740  4447 E LocationManager: skip onNmeaReceived: nmea$PMTKTSX1,5012,326233.306,43.336,38.845,59000000,32.622746,32.623331,0.000000,-0.225000,0.000000,32.623*7C
01-01 00:04:54.543  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKTSX1,5012,326233.306,43.336,38.845,59000000,32.622746,32.623331,0.000000,-0.225000,0.000000,32.623*7C
01-01 00:04:54.544  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKTSX1,5012,326233.306,43.336,38.845,59000000,32.622746,32.623331,0.000000,-0.225000,0.000000,32.623*7C
01-01 00:04:54.544  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKTSX1,5012,326233.306,43.336,38.845,59000000,32.622746,32.623331,0.000000,-0.225000,0.000000,32.623*7C
01-01 00:04:54.544  3740  4447 E LocationManager: skip onNmeaReceived: nmea$PMTKGEPH,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*04
01-01 00:04:54.545  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKGEPH,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*04
01-01 00:04:54.545  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKGEPH,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*04
01-01 00:04:54.546  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKGEPH,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*04
01-01 00:04:54.546  3740  4447 E LocationManager: skip onNmeaReceived: nmea$PMTKGALM,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*19
01-01 00:04:54.547  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKGALM,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*19
01-01 00:04:54.547  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKGALM,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*19
01-01 00:04:54.548  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKGALM,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0*19
01-01 00:04:54.548  3740  4447 E LocationManager: skip onNmeaReceived: nmea$GNACCURACY,3162.1,180.0,3.5,500.0*29
01-01 00:04:54.548  3328  4051 E LocationManager: skip onNmeaReceived: nmea$GNACCURACY,3162.1,180.0,3.5,500.0*29
01-01 00:04:54.549  3328  3345 E LocationManager: skip onNmeaReceived: nmea$GNACCURACY,3162.1,180.0,3.5,500.0*29
01-01 00:04:54.549  3328  3345 E LocationManager: skip onNmeaReceived: nmea$GNACCURACY,3162.1,180.0,3.5,500.0*29
01-01 00:04:54.550  3740  4447 E LocationManager: skip onNmeaReceived: nmea$PMTKAGC,160454.011,6174,6241,7189,7273,0,0,0*68
01-01 00:04:54.550  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKAGC,160454.011,6174,6241,7189,7273,0,0,0*68
01-01 00:04:54.551  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKAGC,160454.011,6174,6241,7189,7273,0,0,0*68
01-01 00:04:54.551  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKAGC,160454.011,6174,6241,7189,7273,0,0,0*68
01-01 00:04:54.553  3740  4447 E LocationManager: skip onNmeaReceived: nmea$PMTKAG2,160454.011,6*1C
01-01 00:04:54.553  3328  4051 E LocationManager: skip onNmeaReceived: nmea$PMTKAG2,160454.011,6*1C
01-01 00:04:54.553  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKAG2,160454.011,6*1C
01-01 00:04:54.554  3328  3345 E LocationManager: skip onNmeaReceived: nmea$PMTKAG2,160454.011,6*1C

2）异常发现

发现只要去掉 hasPermission 的调用，功耗就能掉下来，没想到是这个函数搞得鬼

    public void onNmeaReceived(final long timestamp, final String nmea) {

        foreach((IGnssStatusListener listener, CallerIdentity callerIdentity) -> {
            // huazhi.su add start
            if("true".equals(SystemProperties.get("persist.sys.Helper.hasPermission", "false"))) {
                // 实测1秒中被调用了70次，去除后平均电流减少20mA
                // 这个判断作用:若用户定位的时候突然关闭GPS权限，则也要相应停止数据上报到上层
                // 出发点是好：但是1秒钟判断了70次就不厚道了，且不同应用注册的数量不同，故这里的次数也不一样
                if (!hasPermission(mContext, callerIdentity)) {
                    logPermissionDisabledEventNotReported(TAG, callerIdentity.mPackageName, "NMEA");
                    return;
                }
            }
            // huazhi.su add end
            listener.onNmeaReceived(timestamp, nmea);

        });
        
    protected boolean hasPermission(Context context, CallerIdentity callerIdentity) {
        if (LocationPermissionUtil.doesCallerReportToAppOps(context, callerIdentity)) {
            // The caller is identified as a location provider that will report location
            // access to AppOps. Skip noteOp but do checkOp to check for location permission.
            return mAppOps.checkOpNoThrow(AppOpsManager.OP_FINE_LOCATION, callerIdentity.mUid,
                    callerIdentity.mPackageName) == AppOpsManager.MODE_ALLOWED;
        }

        return mAppOps.noteOpNoThrow(AppOpsManager.OP_FINE_LOCATION, callerIdentity.mUid,
                callerIdentity.mPackageName) == AppOpsManager.MODE_ALLOWED;
    }

上述中 reportSvStatus 也使用到了 hasPermission 的判断，但是实测回调的次数没有 reportNmea多，故功耗电流小一些

    public void onSvStatusChanged (
        ...
        hasPermission(mContext, callerIdentity)
        ...
    }

5. 问题发现与解决

5.1. 功耗问题原因

去掉了 hasPermission 就好，说明系统也经不起频繁执行某个函数的方法

5.2. 分析

源码也存在问题，为什么这么频繁调用的地方，每次都是执行一个函数，干嘛不把这个变量的状态保持下来。读状态比去执行函数的效率高多了。
估计 Google 写这个函数的时候，没考虑的这里实际会被频繁调用到

5.3. 解决方案

监听权限变化事件，当权限改变的时候，更新下对应uid的权限，并保存起来

    context.getPackageManager().addOnPermissionsChangeListener(mOnPermissionsChangedListener);

    private OnPermissionsChangedListener mOnPermissionsChangedListener = new OnPermissionsChangedListener() {

        public void onPermissionsChanged(int uid) {
            updatePermissionsChanged(uid);
        }
    };

若当前保存了uid的权限状态，则读状态，而不是每次执行函数获取，提高效率

    private boolean isHasPermission(Context context, CallerIdentity callerIdentity) {
        if(mPermissionsList.containsKey(callerIdentity.mUid)) {
            return mPermissionsList.get(callerIdentity.mUid);
        } else {
            boolean isHasPermission = hasPermission(mContext, callerIdentity);
            mPermissionsList.put(callerIdentity.mUid, isHasPermission);
            return isHasPermission;
        }
    }

5.4 高德地图导航场景电流优化了 30mA

优化后电流由 133mA 下降到100mA, 1秒钟一个的异常波形也消失了。
由于这里数据是我自己测的