Android中关于cpu/cpuset/schedtune的应用都是基于进程优先级的,根据不同优先级划分进程类型。AMS(ActivityManagerService)和PMS(PackageManagerService)等通过class Process设置进程优先级、调度策略等;android/osProcess JNI通过调用libcutils.so/libutils.so执行getpriority/setpriority/sched_setscheduler/sched_getschedler系统调用或者直接操作CGroup文件节点以达到设置优先级,限制进程CPU资源的目的。
根据优先级,通过设置CGroup的cpu/cpuset/stune控制进程获得CPU执行时间、可调度CPU范围等,以达到对不同优先级进程的控制。
Android关于cpu/cpuset/schedtune的框架结构
进程优先级和调度策略从上到下贯穿其中,但是在不同的层级的名称有一些变化。下面逐一介绍。
class Process以及android/os/Process JNI
frameworks/base/core/java/android/os/Process.java
其他服务通过class Process来设置进程优先级、调度侧率等。
class Process中优先级划分:
public static final int THREAD_PRIORITY_DEFAULT = 0; 应用的默认优先级 /* public static final int THREAD_PRIORITY_LOWEST = 19; 线程的最低优先级 public static final int THREAD_PRIORITY_BACKGROUND = 10; 后台线程的默认优先级 public static final int THREAD_PRIORITY_FOREGROUND = -2; 前台进程的标准优先级 public static final int THREAD_PRIORITY_DISPLAY = -4; 系统用于显示功能的优先级 public static final int THREAD_PRIORITY_URGENT_DISPLAY = -8; 系统用于重要显示功能的优先级 public static final int THREAD_PRIORITY_AUDIO = -16; 音频线程默认优先级 public static final int THREAD_PRIORITY_URGENT_AUDIO = -19; 重要音频线程默认优先级 |
调度策略划分:
public static final int SCHED_OTHER = 0; 默认调度策略,对应CFS调度类 public static final int SCHED_FIFO = 1; FIFO调度策略,对应RT调度类 public static final int SCHED_RR = 2; RR调度策略,对应RT调度类 public static final int SCHED_BATCH = 3; 批调度策略,对应CFS调度类 public static final int SCHED_IDLE = 5; idle调度策略 |
class Process相关API,主要用于:
public static final native void setThreadPriority(int tid, int priority) public static final native void setThreadScheduler(int tid, int policy, int priority) public static final native void setThreadPriority(int tid, int priority) public static final native int getThreadPriority(int tid) public static final native int getThreadScheduler(int tid) public static final native void setThreadGroup(int tid, int group) public static final native void setProcessGroup(int pid, int group) |
frameworks/base/core/jni/android_util_Process.cpp
对应JNINativeMethod如下:
static const JNINativeMethod methods[] = { |
scheduler相关API直接调用sched_setscheduler/sched_getscheduler。
libcutils.so/libutils.so
在介绍这个函数之前先介绍一下此处所使用的优先级定义,可以看出和class Process中是完全的对应关系:
ANDROID_PRIORITY_LOWEST = 19, /* use for background tasks */ /* most threads run at normal priority */ /* threads currently running a UI that the user is interacting with */ /* the main UI thread has a slightly more favorable priority */ /* ui service treads might want to run at a urgent display (uncommon) */ /* all normal audio threads */ /* service audio threads (uncommon) */ /* should never be used in practice. regular process might not ANDROID_PRIORITY_DEFAULT = ANDROID_PRIORITY_NORMAL, |
还需要在研究一下,Sched Policy中使用的优先级映射关系:
/* Keep in sync with THREAD_GROUP_* in frameworks/base/core/java/android/os/Process.java */ |
Threads.cpp中定义了androidSetThreadPriority用于设置线程的优先级。
int androidSetThreadPriority(pid_t tid, int pri) if (pri >= ANDROID_PRIORITY_BACKGROUND) { 如果priority大于等于BACKGROUND,则设置为BACKGROUND类型的调度策略。 if (rc) { if (setpriority(PRIO_PROCESS, tid, pri) < 0) { 设置优先级 return rc; |
set_cpuset_policy根据SchedPolicy类型将tid写入cpuset和schedtune子系统中。
有下面的函数可以得出cpuset、schedtune和不同类型SchedPolicy之间的对应关系:
/dev/cpuset/foreground/tasks SP_FOREGROUND SP_AUDIO_APP SP_AUDIO_SYS
/dev/cpuset/background/tasks SP_BACKGROUND
/dev/cpuset/system-background/tasks SP_SYSTEM
/dev/cpuset/top-app/tasks SP_TOP_APP
/dev/stune/top-app/tasks SP_TOP_APP
/dev/stune/foreground/tasks SP_FOREGROUND SP_AUDIO_APP SP_AUDIO_SYS
/dev/stune/background/tasks SP_BACKGROUND
int set_cpuset_policy(int tid, SchedPolicy policy) int fd = -1; if (add_tid_to_cgroup(tid, fd) != 0) { #ifdef USE_SCHEDBOOST return 0; |
set_sched_policy设置cpu/schedtune两个子系统,子系统节点和SchedPolicy类型对应如下:
/dev/cpuctl/tasks SP_FOREGROUND SP_AUDIO_APP SP_AUDIO_SYS
/dev/cpuctl/bg_non_interactive/tasks SP_BACKGROUND
/dev/stune/top-app/tasks SP_TOP_APP
/dev/stune/foreground/tasks SP_FOREGROUND SP_AUDIO_APP SP_AUDIO_SYS
/dev/stune/background/tasks SP_BACKGROUND
int set_sched_policy(int tid, SchedPolicy policy) #if POLICY_DEBUG snprintf(statfile, sizeof(statfile), "/proc/%d/stat", tid); int fd = open(statfile, O_RDONLY | O_CLOEXEC); for (p = statline; *p != '('; p++); strncpy(thread_name, p, (q-p)); if (__sys_supports_schedgroups) { 是否使能schedtune CGroup
#ifdef USE_SCHEDBOOST param.sched_priority = 0; if (__sys_supports_timerslack) { return 0; |
上面的一系列转换可以用下图表示:
system\core\libcutils\Sched_policy.c中,对SP_*系列SchedPolicy转换成使用不同cpuctl、cpuset、stune句柄,将对应的pid、tid写入tasks中。
SP_BACKGROUND对应SCHED_BACH调度策略,其他对应SCHED_NORMAL。
SCHED_NORMAL:默认的调度策略,在旧版中为SCHED_OTHER。SCHED_BATCH:针对批处理进程。SCHED_IDLE:使用此调度侧率的进程优先级最低。
SCHED_NORMAL和SCHED_BATCH区别只是再唤醒时有区别,唤醒较频繁的进程不适合SCHED_BATCH。
如果使能__sys_supports_schedgroups,就不会调用sched_setscheduler去设置SchedulePolicy。
SCHED_NORMAL和SCHED_BACH区别
SP_BACKGROUND对应SCHED_BACH调度策略,其他对应SCHED_NORMAL。
SCHED_NORMAL:默认的调度策略,在旧版中为SCHED_OTHER。SCHED_BATCH:针对批处理进程。SCHED_IDLE:使用此调度侧率的进程优先级最低。
SCHED_NORMAL和SCHED_BATCH区别只是再唤醒时有区别,唤醒较频繁的进程不适合SCHED_BATCH。
如果使能__sys_supports_schedgroups,就不会调用sched_setscheduler去设置SchedulePolicy。
Android中的一个应用
cpu子系统:
# Create cgroup mount points for process groups mkdir /dev/cpuctl mount cgroup none /dev/cpuctl cpu chown system system /dev/cpuctl chown system system /dev/cpuctl/tasks chmod 0666 /dev/cpuctl/tasks write /dev/cpuctl/cpu.shares 1024 write /dev/cpuctl/cpu.rt_runtime_us 800000 write /dev/cpuctl/cpu.rt_period_us 1000000 mkdir /dev/cpuctl/bg_non_interactive chown system system /dev/cpuctl/bg_non_interactive/tasks chmod 0666 /dev/cpuctl/bg_non_interactive/tasks # 5.0 % write /dev/cpuctl/bg_non_interactive/cpu.shares 52 write /dev/cpuctl/bg_non_interactive/cpu.rt_runtime_us 700000 write /dev/cpuctl/bg_non_interactive/cpu.rt_period_us 1000000 |
cpuset子系统:
# sets up initial cpusets for ActivityManager mkdir /dev/cpuset mount cpuset none /dev/cpuset # this ensures that the cpusets are present and usable, but the device's # init.rc must actually set the correct cpus mkdir /dev/cpuset/foreground write /dev/cpuset/foreground/cpus 0 write /dev/cpuset/foreground/mems 0 mkdir /dev/cpuset/foreground/boost write /dev/cpuset/foreground/boost/cpus 0 write /dev/cpuset/foreground/boost/mems 0 mkdir /dev/cpuset/background write /dev/cpuset/background/cpus 0 write /dev/cpuset/background/mems 0 # system-background is for system tasks that should only run on # little cores, not on bigs # to be used only by init, so don't change system-bg permissions mkdir /dev/cpuset/system-background write /dev/cpuset/system-background/cpus 0 write /dev/cpuset/system-background/mems 0 mkdir /dev/cpuset/top-app write /dev/cpuset/top-app/cpus 0 write /dev/cpuset/top-app/mems 0 # change permissions for all cpusets we'll touch at runtime chown system system /dev/cpuset chown system system /dev/cpuset/foreground chown system system /dev/cpuset/foreground/boost chown system system /dev/cpuset/background chown system system /dev/cpuset/system-background chown system system /dev/cpuset/top-app chown system system /dev/cpuset/tasks chown system system /dev/cpuset/foreground/tasks chown system system /dev/cpuset/foreground/boost/tasks chown system system /dev/cpuset/background/tasks chown system system /dev/cpuset/system-background/tasks chown system system /dev/cpuset/top-app/tasks # set system-background to 0775 so SurfaceFlinger can touch it chmod 0775 /dev/cpuset/system-background chmod 0664 /dev/cpuset/foreground/tasks chmod 0664 /dev/cpuset/foreground/boost/tasks chmod 0664 /dev/cpuset/background/tasks chmod 0664 /dev/cpuset/system-background/tasks chmod 0664 /dev/cpuset/top-app/tasks chmod 0664 /dev/cpuset/tasks |
schedtune子系统:
# Create energy-aware scheduler tuning nodes mkdir /dev/stune mount cgroup none /dev/stune schedtune mkdir /dev/stune/foreground chown system system /dev/stune chown system system /dev/stune/foreground chown system system /dev/stune/tasks chown system system /dev/stune/foreground/tasks chmod 0664 /dev/stune/tasks chmod 0664 /dev/stune/foreground/tasks |