iOS 获取APP的CPU、内存等信息
目标是开发一个SDK，嵌入到APP里面，用来统计当前APP的实时CPU、内存等信息
2015.11.17
http://stackoverflow.com/questions/12889422/ios-cpu-usage-for-each-process-using-sysctl
这是第一个找到，采用的是sysctl函数
但是出来的CPU数据和instrument、GT的数据对不上（后两者数据比较接近）
2015.11.19
https://github.com/TianJIANG/ios_monitor
从guithub上搜到的，利用的主要是#import <mach/mach.h> 里面的task_info 等，打开了一道新的大门，后续找到不少类似的方法
http://stackoverflow.com/questions/8223348/ios-get-cpu-usage-from-application
这个答案也是给的这个方法，末尾额外加了一行代码， vm_dealloc，解决leaking问题
补充几个相关的：
http://stackoverflow.com/questions/5182924/where-is-my-ipad-runtime-memory-going%E2%80%8C%E2%80%8B
http://blog.sina.com.cn/s/blog_693de6100101ffwm.html
http://www.zhihu.com/question/22992491
Github 搜 “Activity Monitor”
https://github.com/AndrewBarba/ActivityMonitor
https://github.com/vsnrain/ActivityMonitor
此算法是获取当前APP的CPU，数值与Instrument、GT接近
- (void)GetCpuUsage { kern_return_t kr; task_info_data_t tinfo; mach_msg_type_number_t task_info_count; task_info_count = TASK_INFO_MAX; kr = task_info(mach_task_self(), TASK_BASIC_INFO, (task_info_t)tinfo, &task_info_count); if (kr != KERN_SUCCESS) { return; } task_basic_info_t basic_info; thread_array_t thread_list; mach_msg_type_number_t thread_count; thread_info_data_t thinfo; mach_msg_type_number_t thread_info_count; thread_basic_info_t basic_info_th; uint32_t stat_thread = 0; // Mach threads basic_info = (task_basic_info_t)tinfo; // get threads in the task kr = task_threads(mach_task_self(), &thread_list, &thread_count); if (kr != KERN_SUCCESS) { return; } if (thread_count > 0) stat_thread += thread_count; long tot_sec = 0; long tot_usec = 0; float tot_cpu = 0; int j; for (j = 0; j < thread_count; j++) { thread_info_count = THREAD_INFO_MAX; kr = thread_info(thread_list[j], THREAD_BASIC_INFO, (thread_info_t)thinfo, &thread_info_count); if (kr != KERN_SUCCESS) { return; } basic_info_th = (thread_basic_info_t)thinfo; if (!(basic_info_th->flags & TH_FLAGS_IDLE)) { tot_sec = tot_sec + basic_info_th->user_time.seconds + basic_info_th->system_time.seconds; tot_usec = tot_usec + basic_info_th->system_time.microseconds + basic_info_th->system_time.microseconds; tot_cpu = tot_cpu + basic_info_th->cpu_usage / (float)TH_USAGE_SCALE * 100.0; } } // for each thread kr = vm_deallocate(mach_task_self(), (vm_offset_t)thread_list, thread_count * sizeof(thread_t)); assert(kr == KERN_SUCCESS); NSLog(@"CPU Usage: %f \n", tot_cpu); }
此算法是获取当前APP的内存，数值与GT的一致，与Instrument不一致
- (void)GetCurrentTaskUsedMemory { task_basic_info_data_t taskInfo; mach_msg_type_number_t infoCount = TASK_BASIC_INFO_COUNT; kern_return_t kernReturn = task_info(mach_task_self(), TASK_BASIC_INFO, (task_info_t)&taskInfo, &infoCount); if(kernReturn != KERN_SUCCESS) { return; } NSLog(@"Memory Usage: %f", taskInfo.resident_size / 1024.0 / 1024.0); }
2015.11.20
XNU
https://en.wikipedia.org/wiki/XNU
MACH
https://en.wikipedia.org/wiki/Mach_(kernel)
Kernel Programming Guide
https://developer.apple.com/library/mac/documentation/Darwin/Conceptual/KernelProgramming/About/About.html
github 上搜 mach_msg_type_number_t
https://github.com/search?l=objective-c&q=mach_msg_type_number_t&type=Code&utf8=✓
2015.11.22
用VM Tracker查看内存，有那么几项
Resident Size｜Dirty Size｜Virtual Size
http://stackoverflow.com/questions/5176074/what-do-dirty-and-resident-mean-in-relation-to-virtual-memory
这篇解释了三者的差别，我理解我们应该跟踪的是Resident Size，但是数值上与VM Tracker上对不上
- (void)GetMemoryStatistics { // Get Page Size int mib[2]; int page_size; size_t len; mib[0] = CTL_HW; mib[1] = HW_PAGESIZE; len = sizeof(page_size); // // 方法一: 16384 // int status = sysctl(mib, 2, &page_size, &len, NULL, 0); // if (status < 0) { // perror("Failed to get page size"); // } // // 方法二: 16384 // page_size = getpagesize(); // 方法三: 4096 if( host_page_size(mach_host_self(), &page_size)!= KERN_SUCCESS ){ perror("Failed to get page size"); } printf("Page size is %d bytes\n", page_size); // Get Memory Size mib[0] = CTL_HW; mib[1] = HW_MEMSIZE; long ram; len = sizeof(ram); if (sysctl(mib, 2, &ram, &len, NULL, 0)) { perror("Failed to get ram size"); } printf("Ram size is %f MB\n", ram / (1024.0) / (1024.0)); // Get Memory Statistics // vm_statistics_data_t vm_stats; // mach_msg_type_number_t info_count = HOST_VM_INFO_COUNT; vm_statistics64_data_t vm_stats; mach_msg_type_number_t info_count64 = HOST_VM_INFO64_COUNT; // kern_return_t kern_return = host_statistics(mach_host_self(), HOST_VM_INFO, (host_info_t)&vm_stats, &info_count); kern_return_t kern_return = host_statistics64(mach_host_self(), HOST_VM_INFO64, (host_info64_t)&vm_stats, &info_count64); if (kern_return != KERN_SUCCESS) { printf("Failed to get VM statistics!"); } double vm_total = vm_stats.wire_count + vm_stats.active_count + vm_stats.inactive_count + vm_stats.free_count; double vm_wire = vm_stats.wire_count; double vm_active = vm_stats.active_count; double vm_inactive = vm_stats.inactive_count; double vm_free = vm_stats.free_count; double unit = (1024.0) * (1024.0); NSLog(@"Total Memory: %f", vm_total * page_size / unit); NSLog(@"Wired Memory: %f", vm_wire * page_size / unit); NSLog(@"Active Memory: %f", vm_active * page_size / unit); NSLog(@"Inactive Memory: %f", vm_inactive * page_size / unit); NSLog(@"Free Memory: %f", vm_free * page_size / unit); }
1、关于Ram大小，用HW_MEMSIZE计算得到1000MB，是准确的
2、关于page size，上面提供了三种方法
其中方法一、二在64位机器上返回了16384，只有第三种方法返回了4096
http://stackoverflow.com/questions/21552747/strange-behavior-on-64bit-ios-devices-when-retrieving-vm-statistics/33574804#33574804
这篇文章提出了此疑问，但是没有特别明确的解释
我认为page size应该是4096，用VM Tracker运行，截图如下：
以第一项_LINKEDIT为例，13692*4096/1024/1024=53.48M，与Vitual Size吻合
3、关于vm_total、vm_wire、vm_active、vm_inactive、vm_free这几个值
其中一组运行结果如下：
Page size is 4096 bytes Ram size is 1000.000000 MB 2015-11-22 20:59:41.191 CompSDKDemo[1602:410503] Total Memory: 777.519531 2015-11-22 20:59:41.191 CompSDKDemo[1602:410503] Wired Memory: 205.484375 2015-11-22 20:59:41.192 CompSDKDemo[1602:410503] Active Memory: 374.941406 2015-11-22 20:59:41.192 CompSDKDemo[1602:410503] Inactive Memory: 175.710938 2015-11-22 20:59:41.192 CompSDKDemo[1602:410503] Free Memory: 21.382812
可以看到，Total Memory不是1000MB，这个如何解释呢？
a. 1000M应该是实际的RAM大小，而Total Memory，应该是Virtual Memory，两者是否一回事，有待商榷？
b. 从APP Store上下了一个System Monitor，如下：
可以看到，Wired、Active、Inactive的值都对得上，唯独Free的值对不上
不负责任的猜测，他的Free是通过1000M减其它三项得到的
Source Code : Get Hardware Info of iPhone
http://blog.sina.com.cn/s/blog_4a04a3c90100r9gn.html
How to determine CPU and memory consumption from inside a process?
http://stackoverflow.com/questions/63166/how-to-determine-cpu-and-memory-consumption-from-inside-a-process
总结下来，关于内存的有两个问题：
1. active、inactive、wired、free加起来不等于1000M
这个可以先放一放，我们可以先不用管这部分
2. 当前app消耗的内存，目前的算法与 Debug Gauges的值有偏差，与GT吻合
不过发现点击页面增长的值和发回释放的值，与Debug Gauges基本一致，因此可以使用
关于CPU，上面算法给出的值符合要求，可以使用；
接下来是：耗电量、网速、帧率
2015.11.23
耗电量，目前没有找到合适的工具
有两种获取电池电量信息的方法
方法一：
这个方法需要导入 IOKit 库，但是不知从什么时候开始，iOS系统不允许用户导入库
http://www.cocoachina.com/bbs/read.php?tid=268692
这篇文章提供了野路子方法，但是实施起来颇为不便，考虑到要做sdk，不适合
CFTypeRef blob = IOPSCopyPowerSourcesInfo();
CFArrayRef sources = IOPSCopyPowerSourcesList(blob);
方法二：
[UIDevice currentDevice].batteryLevel
据说精度达到1%
找到一篇文章，提供了三种方法
http://www.jianshu.com/p/11c1afdf5415
网络流量
https://github.com/QbsuranAlang/GetNetworkFlow/blob/master/GetNetworkFlow/GetNetworkFlow/ViewController.m
2015.11.24
经测试，mach方法获取的内存值与top命令拿到的RSS、VSS是一致的
