diff --git a/study/kernel/01-process/05-schedule/04-stop/README.md b/study/kernel/01-process/05-schedule/04-stop/README.md new file mode 100644 index 0000000..3c8a8cd --- /dev/null +++ b/study/kernel/01-process/05-schedule/04-stop/README.md @@ -0,0 +1,58 @@ +Linux进程退出详解 +======= + + +| 日期 | 内核版本 | 架构| 作者 | GitHub| CSDN | +| ------- |:-------:|:-------:|:-------:|:-------:|:-------:| +| 2016-06-14 | [Linux-4.6](http://lxr.free-electrons.com/source/?v=4.6) | X86 & arm | [gatieme](http://blog.csdn.net/gatieme) | [LinuxDeviceDrivers](https://github.com/gatieme/LDD-LinuxDeviceDrivers) | [Linux进程管理与调度](http://blog.csdn.net/gatieme/article/category/6225543) | + +1. 调度器概述introduction +2. 调度器演变develop +3. 调度器的设计design +4. 周期性调度器periodic_scheduler +5. 主调度器main_schedulrt +6. 优先级priority +7. 抢占-内核抢占和用户抢占以及抢占的时机preempt +8. 进程切换context_switch + +9. 完全公平调度器cfs-设计 +10. cfs-负荷权重 +11. cfs-虚拟时钟 +12. cfs-延迟跟踪 +13. cfs-操作pick_next_task +14. rt实时调度器 +15. dl时调度器 +16. stop调度 +17. idle调度器 +18. SMP调度 +19. 调度域和控制组 + +http://blog.csdn.net/b02042236/article/details/6076473 +http://www.tuicool.com/articles/MjyANr +http://iamzhongyong.iteye.com/blog/1895728 +http://blog.csdn.net/xiaofei0859/article/details/8113211 + + + +BFS +https://en.wikipedia.org/wiki/Brain_Fuck_Scheduler + +http://blog.csdn.net/u201017971/article/details/50511511 + + +http://baike.baidu.com/link?url=qO-044OZarVCuMDuioyhbYLswbB7MkyVwW3vPbWzHGE6j2-2X3IKIiXCUecABqkg9KCXSPCQ3Kc6IP26uCT0JK + +Con Kolivas + +http://www.ibm.com/developerworks/cn/linux/l-cn-bfs/ + + +rifs进程调度比起bfs cfs好在哪差在哪 + +交互性极佳，特别是在大负载，那种交互性差距很明显 +吞吐量比bfs和cfs低一半，但还是比windows高1/3 + +要测试交互性的差距请用mplayer测， +要体验交互性大可以开make -j512然后看网页，听音乐，移动窗口。 + + diff --git a/study/kernel/01-process/05-schedule/05-deadline/README.md b/study/kernel/01-process/05-schedule/05-deadline/README.md new file mode 100644 index 0000000..3c8a8cd --- /dev/null +++ b/study/kernel/01-process/05-schedule/05-deadline/README.md @@ -0,0 +1,58 @@ +Linux进程退出详解 +======= + + +| 日期 | 内核版本 | 架构| 作者 | GitHub| CSDN | +| ------- |:-------:|:-------:|:-------:|:-------:|:-------:| +| 2016-06-14 | [Linux-4.6](http://lxr.free-electrons.com/source/?v=4.6) | X86 & arm | [gatieme](http://blog.csdn.net/gatieme) | [LinuxDeviceDrivers](https://github.com/gatieme/LDD-LinuxDeviceDrivers) | [Linux进程管理与调度](http://blog.csdn.net/gatieme/article/category/6225543) | + +1. 调度器概述introduction +2. 调度器演变develop +3. 调度器的设计design +4. 周期性调度器periodic_scheduler +5. 主调度器main_schedulrt +6. 优先级priority +7. 抢占-内核抢占和用户抢占以及抢占的时机preempt +8. 进程切换context_switch + +9. 完全公平调度器cfs-设计 +10. cfs-负荷权重 +11. cfs-虚拟时钟 +12. cfs-延迟跟踪 +13. cfs-操作pick_next_task +14. rt实时调度器 +15. dl时调度器 +16. stop调度 +17. idle调度器 +18. SMP调度 +19. 调度域和控制组 + +http://blog.csdn.net/b02042236/article/details/6076473 +http://www.tuicool.com/articles/MjyANr +http://iamzhongyong.iteye.com/blog/1895728 +http://blog.csdn.net/xiaofei0859/article/details/8113211 + + + +BFS +https://en.wikipedia.org/wiki/Brain_Fuck_Scheduler + +http://blog.csdn.net/u201017971/article/details/50511511 + + +http://baike.baidu.com/link?url=qO-044OZarVCuMDuioyhbYLswbB7MkyVwW3vPbWzHGE6j2-2X3IKIiXCUecABqkg9KCXSPCQ3Kc6IP26uCT0JK + +Con Kolivas + +http://www.ibm.com/developerworks/cn/linux/l-cn-bfs/ + + +rifs进程调度比起bfs cfs好在哪差在哪 + +交互性极佳，特别是在大负载，那种交互性差距很明显 +吞吐量比bfs和cfs低一半，但还是比windows高1/3 + +要测试交互性的差距请用mplayer测， +要体验交互性大可以开make -j512然后看网页，听音乐，移动窗口。 + + diff --git a/study/kernel/01-process/05-schedule/06-rt/README.md b/study/kernel/01-process/05-schedule/06-rt/README.md new file mode 100644 index 0000000..19984ec --- /dev/null +++ b/study/kernel/01-process/05-schedule/06-rt/README.md @@ -0,0 +1,231 @@ +Linux实时进程调度 +======= + + +| 日期 | 内核版本 | 架构| 作者 | GitHub| CSDN | +| ------- |:-------:|:-------:|:-------:|:-------:|:-------:| +| 2016-07-12 | [Linux-4.6](http://lxr.free-electrons.com/source/?v=4.6) | X86 & arm | [gatieme](http://blog.csdn.net/gatieme) | [LinuxDeviceDrivers](https://github.com/gatieme/LDD-LinuxDeviceDrivers) | [Linux进程管理与调度](http://blog.csdn.net/gatieme/article/details/51456569) | + + + +按照POSIX的标准的强制要求, 除了"普通"进程之外, Linux还支持两种实时调度类. + +调度器结构使得实时进程可以平滑集成到内核中, 而无需修改核心调度器, 这显然是调度类带来的好处. + +实时进程与普通进程有一个根本的不同之处: 如果系统中有一个实时进程且可运行, 那么调度器总是会选中它运行, 除非有另外一个优先级更高的实时进程. + +现有的两种实时类, 不同之处如下所示: + +| 名称 | 调度策略 | 描述 | +|:-------:|:-------:|:-------:| +| 循环进程 | SCHED_RR | 有时间片, 其值在进程运行时会减少, 就像是普通进程. 在所有的时间段到期后, 则该值重置为初始值, 而进程则置与队列的末尾, 这保证了在有几个相同优先级相同的SCHED_RR进程的情况下, 他们总是依次执行 | +| 先进先出进程 | SCHED_FIFO | 没有时间片, 在被调度器选择执行后, 可以运行任意长时间. | + +很明显, 如果实时进程编写的比较差, 系统可能长时间无法使用. 最简单的例子, 只要写一个无限循环, 循环体内不进入休眠即可. 因而我们在编写实时应用程序时, 应该格外小心. + + +#实时进程的数据结构 +------- + +##实时调度器类 +------- + +实时进程的调度器类用rt_sched_class来表示, 其定义在[kernel/sched/rt.c, line 2326](http://lxr.free-electrons.com/source/kernel/sched/rt.c?v=4.6#L2326)中, 如下所示. + + + +```c +const struct sched_class rt_sched_class = +{ + .next = &fair_sched_class, + .enqueue_task = enqueue_task_rt, + .dequeue_task = dequeue_task_rt, + .yield_task = yield_task_rt, + + .check_preempt_curr = check_preempt_curr_rt, + + .pick_next_task = pick_next_task_rt, + .put_prev_task = put_prev_task_rt, + +#ifdef CONFIG_SMP + .select_task_rq = select_task_rq_rt, + + .set_cpus_allowed = set_cpus_allowed_common, + .rq_online = rq_online_rt, + .rq_offline = rq_offline_rt, + .task_woken = task_woken_rt, + .switched_from = switched_from_rt, +#endif + + .set_curr_task = set_curr_task_rt, + .task_tick = task_tick_rt, + + .get_rr_interval = get_rr_interval_rt, + + .prio_changed = prio_changed_rt, + .switched_to = switched_to_rt, + + .update_curr = update_curr_rt, +}; +``` +实时调度器类的实现比完全调度器简单. + +##实时进程的优先级 +------- + + +实时进程的特点在于其优先级比普通进程高, 对应地, 其static_prio值总是普通进程低. + +task_has_rt_policy则检测进程是否关联到实时进程策略. + + +**linux优先级的表示** + + +>在用户空间通过nice命令设置进程的静态优先级, 这在内部会调用nice系统调用, 进程的nice值在-20~+19之间. 值越低优先级越高. +> +>setpriority系统调用也可以用来设置进程的优先级. 它不仅能够修改单个线程的优先级, 还能修改进程组中所有进程的优先级, 或者通过制定UID来修改特定用户的所有进程的优先级 + + +内核使用一些简单的数值范围0~139表示内部优先级, 数值越低, 优先级越高。 + +从0~99的范围专供实时进程使用, nice的值[-20,19]则映射到范围100~139, 即普通进程的优先级 + + + + +| 优先级范围 | 描述 | +| ------------- |:-------------:| +| 0——99 | 实时进程 | +| 100——139 | 非实时进程 | + +![内核的优先级标度](../images/priority.jpg) + + + +内核表示优先级的所有信息基本都放在[include/linux/sched/prio.h](http://lxr.free-electrons.com/source/include/linux/sched/prio.h?v=4.6)中, 其中定义了一些表示优先级的宏和函数. + + +而MAX_RT_PRIO - 1 = 99指定了实时进程的最大优先级, 而MAX_PRIO - 1 = 139则是普通进程的最大优先级数值 + +```c +// http://lxr.free-electrons.com/source/include/linux/sched/prio.h?v=4.6#L22 +#define MAX_USER_RT_PRIO 100 +#define MAX_RT_PRIO MAX_USER_RT_PRIO + +#define MAX_PRIO (MAX_RT_PRIO + NICE_WIDTH) +#define DEFAULT_PRIO (MAX_RT_PRIO + NICE_WIDTH / 2) +``` + + +| 宏 | 值 | 描述 | +| ------------- |:-------------:|:-------------:| +| MIN_NICE | -20 | 对应于优先级100, 可以使用NICE_TO_PRIO和PRIO_TO_NICE转换 | +| MAX_NICE | 19 | 对应于优先级139, 可以使用NICE_TO_PRIO和PRIO_TO_NICE转换 | +| NICE_WIDTH | 40 | nice值得范围宽度, 即[-20, 19]共40个数字的宽度 | +| MAX_RT_PRIO, MAX_USER_RT_PRIO | 100 | 实时进程的最大优先级 | +| MAX_PRIO | 140 | 普通进程的最大优先级 | +| DEFAULT_PRIO | 120 | 进程的默认优先级, 对应于nice=0 | +| MAX_DL_PRIO | 0 | 使用EDF最早截止时间优先调度算法的实时进程最大的优先级 | +` + + +##实时就绪队列 +------- + +核心调度器的就绪队列也包含了用于实时进程的子就绪队列, 是一个嵌入的struct rt_rq实例. + +参见[struct rq的定义](http://lxr.free-electrons.com/source/kernel/sched/sched.h?v4.6#L602) + + +```c +struct rq +{ + /* ...... */ + struct rt_rq rt; + /* ...... */ +} +``` +实时rt进程的就绪队列定义在[kernel/sched/sched.h, line 450](http://lxr.free-electrons.com/source/kernel/sched/sched.h?v=4.6#L450)中 + +```c +/* Real-Time classes' related field in a runqueue: */ +struct rt_rq { + struct rt_prio_array active; + unsigned int rt_nr_running; + unsigned int rr_nr_running; +#if defined CONFIG_SMP || defined CONFIG_RT_GROUP_SCHED + struct { + int curr; /* highest queued rt task prio */ +#ifdef CONFIG_SMP + int next; /* next highest */ +#endif + } highest_prio; +#endif +#ifdef CONFIG_SMP + unsigned long rt_nr_migratory; + unsigned long rt_nr_total; + int overloaded; + struct plist_head pushable_tasks; +#ifdef HAVE_RT_PUSH_IPI + int push_flags; + int push_cpu; + struct irq_work push_work; + raw_spinlock_t push_lock; +#endif +#endif /* CONFIG_SMP */ + int rt_queued; + + int rt_throttled; + u64 rt_time; + u64 rt_runtime; + /* Nests inside the rq lock: */ + raw_spinlock_t rt_runtime_lock; + +#ifdef CONFIG_RT_GROUP_SCHED + unsigned long rt_nr_boosted; + + struct rq *rq; + struct task_group *tg; +#endif +``` + +其主要包含了rt_prio_array + + + +```c +/* + * This is the priority-queue data structure of the RT scheduling class: + */ +struct rt_prio_array { + DECLARE_BITMAP(bitmap, MAX_RT_PRIO+1); /* include 1 bit for delimiter */ + struct list_head queue[MAX_RT_PRIO]; +}; +``` +具有相同优先级的所有实时进程都保存在一个链表中, 表头为active.queue[prio], 而active.bitmap位图中每个比特位对应于一个链表, 凡包含了进程的链表, 对应的比特位置位. 如果连表中没有进程, 则对应的比特位不置位. + + + + +##实时进程的判断 +------- + + +内核可以通过rt_task宏可以通过检查其优先级来检测给定进程是否是实时进程, 该函数定义在[include/linux/sched/rt.h, line 13](http://lxr.free-electrons.com/source/include/linux/sched/rt.h?v=4.6#L13) + + +```c +// http://lxr.free-electrons.com/source/include/linux/sched/rt.h?v4.6#L6 +static inline int rt_prio(int prio) +{ + if (unlikely(prio < MAX_RT_PRIO)) + return 1; + return 0; +} + +static inline int rt_task(struct task_struct *p) +{ + return rt_prio(p->prio); +} +``` \ No newline at end of file diff --git a/study/kernel/01-process/05-schedule/04-cfs/01-cfs/README.md b/study/kernel/01-process/05-schedule/07-cfs/01-cfs/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/01-cfs/README.md rename to study/kernel/01-process/05-schedule/07-cfs/01-cfs/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/01-cfs/fair.c b/study/kernel/01-process/05-schedule/07-cfs/01-cfs/fair.c similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/01-cfs/fair.c rename to study/kernel/01-process/05-schedule/07-cfs/01-cfs/fair.c diff --git a/study/kernel/01-process/05-schedule/04-cfs/02-load_weight/README.md b/study/kernel/01-process/05-schedule/07-cfs/02-load_weight/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/02-load_weight/README.md rename to study/kernel/01-process/05-schedule/07-cfs/02-load_weight/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/03-vruntime/README.md b/study/kernel/01-process/05-schedule/07-cfs/03-vruntime/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/03-vruntime/README.md rename to study/kernel/01-process/05-schedule/07-cfs/03-vruntime/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/03-vruntime/update_curr.c b/study/kernel/01-process/05-schedule/07-cfs/03-vruntime/update_curr.c similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/03-vruntime/update_curr.c rename to study/kernel/01-process/05-schedule/07-cfs/03-vruntime/update_curr.c diff --git a/study/kernel/01-process/05-schedule/04-cfs/03-vruntime/update_min_vruntime.c b/study/kernel/01-process/05-schedule/07-cfs/03-vruntime/update_min_vruntime.c similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/03-vruntime/update_min_vruntime.c rename to study/kernel/01-process/05-schedule/07-cfs/03-vruntime/update_min_vruntime.c diff --git a/study/kernel/01-process/05-schedule/04-cfs/04-queue/README.md b/study/kernel/01-process/05-schedule/07-cfs/04-queue/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/04-queue/README.md rename to study/kernel/01-process/05-schedule/07-cfs/04-queue/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/04-queue/dequeue_task_fair.c b/study/kernel/01-process/05-schedule/07-cfs/04-queue/dequeue_task_fair.c similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/04-queue/dequeue_task_fair.c rename to study/kernel/01-process/05-schedule/07-cfs/04-queue/dequeue_task_fair.c diff --git a/study/kernel/01-process/05-schedule/04-cfs/04-queue/enqueue_task_fair.c b/study/kernel/01-process/05-schedule/07-cfs/04-queue/enqueue_task_fair.c similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/04-queue/enqueue_task_fair.c rename to study/kernel/01-process/05-schedule/07-cfs/04-queue/enqueue_task_fair.c diff --git a/study/kernel/01-process/05-schedule/04-cfs/05-pick_next/README.md b/study/kernel/01-process/05-schedule/07-cfs/05-pick_next/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/05-pick_next/README.md rename to study/kernel/01-process/05-schedule/07-cfs/05-pick_next/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/05-pick_next/images/last_next_left.png b/study/kernel/01-process/05-schedule/07-cfs/05-pick_next/images/last_next_left.png similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/05-pick_next/images/last_next_left.png rename to study/kernel/01-process/05-schedule/07-cfs/05-pick_next/images/last_next_left.png diff --git a/study/kernel/01-process/05-schedule/04-cfs/06-task_tick_fair/README.md b/study/kernel/01-process/05-schedule/07-cfs/06-task_tick_fair/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/06-task_tick_fair/README.md rename to study/kernel/01-process/05-schedule/07-cfs/06-task_tick_fair/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/07-task_new_fair/README.md b/study/kernel/01-process/05-schedule/07-cfs/07-task_new_fair/README.md similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/07-task_new_fair/README.md rename to study/kernel/01-process/05-schedule/07-cfs/07-task_new_fair/README.md diff --git a/study/kernel/01-process/05-schedule/04-cfs/07-task_new_fair/task_fork_fair.c b/study/kernel/01-process/05-schedule/07-cfs/07-task_new_fair/task_fork_fair.c similarity index 100% rename from study/kernel/01-process/05-schedule/04-cfs/07-task_new_fair/task_fork_fair.c rename to study/kernel/01-process/05-schedule/07-cfs/07-task_new_fair/task_fork_fair.c diff --git a/study/kernel/01-process/05-schedule/08-idle/README.md b/study/kernel/01-process/05-schedule/08-idle/README.md new file mode 100644 index 0000000..3c8a8cd --- /dev/null +++ b/study/kernel/01-process/05-schedule/08-idle/README.md @@ -0,0 +1,58 @@ +Linux进程退出详解 +======= + + +| 日期 | 内核版本 | 架构| 作者 | GitHub| CSDN | +| ------- |:-------:|:-------:|:-------:|:-------:|:-------:| +| 2016-06-14 | [Linux-4.6](http://lxr.free-electrons.com/source/?v=4.6) | X86 & arm | [gatieme](http://blog.csdn.net/gatieme) | [LinuxDeviceDrivers](https://github.com/gatieme/LDD-LinuxDeviceDrivers) | [Linux进程管理与调度](http://blog.csdn.net/gatieme/article/category/6225543) | + +1. 调度器概述introduction +2. 调度器演变develop +3. 调度器的设计design +4. 周期性调度器periodic_scheduler +5. 主调度器main_schedulrt +6. 优先级priority +7. 抢占-内核抢占和用户抢占以及抢占的时机preempt +8. 进程切换context_switch + +9. 完全公平调度器cfs-设计 +10. cfs-负荷权重 +11. cfs-虚拟时钟 +12. cfs-延迟跟踪 +13. cfs-操作pick_next_task +14. rt实时调度器 +15. dl时调度器 +16. stop调度 +17. idle调度器 +18. SMP调度 +19. 调度域和控制组 + +http://blog.csdn.net/b02042236/article/details/6076473 +http://www.tuicool.com/articles/MjyANr +http://iamzhongyong.iteye.com/blog/1895728 +http://blog.csdn.net/xiaofei0859/article/details/8113211 + + + +BFS +https://en.wikipedia.org/wiki/Brain_Fuck_Scheduler + +http://blog.csdn.net/u201017971/article/details/50511511 + + +http://baike.baidu.com/link?url=qO-044OZarVCuMDuioyhbYLswbB7MkyVwW3vPbWzHGE6j2-2X3IKIiXCUecABqkg9KCXSPCQ3Kc6IP26uCT0JK + +Con Kolivas + +http://www.ibm.com/developerworks/cn/linux/l-cn-bfs/ + + +rifs进程调度比起bfs cfs好在哪差在哪 + +交互性极佳，特别是在大负载，那种交互性差距很明显 +吞吐量比bfs和cfs低一半，但还是比windows高1/3 + +要测试交互性的差距请用mplayer测， +要体验交互性大可以开make -j512然后看网页，听音乐，移动窗口。 + +