Linux进程调度与切换
2016-04-15
张超《Linux内核分析》MOOC课程http://mooc.study.163.com/course/USTC-1000029000
一、分析
进程调度的时机与进程切换
操作系统原理中介绍了大量进程调度算法，这些算法从实现的角度看仅仅是从运行队列中选择一个新进程，选择的过程中运用了不同的策略而已。对于理解操作系统的工作机制，反而是进程的调度时机与进程的切换机制更为关键。
进程调度的时机：
schedule()是个内核函数，不是内核函数。所以用户态的进程不能直接调用，只能间接调用。内核线程是只有内核态没有用户态的特殊进程。
1.中断处理过程（包括时钟中断、I/O中断、系统调用和异常）中，直接调用schedule()，或者返回用户态时根据need_resched标记调用schedule()；
2.内核线程可以直接调用schedule()进行进程切换，也可以在中断处理过程中进行调度，也就是说内核线程作为一类的特殊的进程可以主动调度，也可以被动调度；
3.用户态进程无法实现主动调度，仅能通过陷入内核态后的某个时机点进行调度，即在中断处理过程中进行调度。
进程切换：
1.为了控制进程的执行，内核必须有能力挂起正在CPU上执行的进程，并恢复以前挂起的某个进程的执行，这叫做进程切换、任务切换、上下文切换；
2.挂起正在CPU上执行的进程，与中断时保存现场是不同的，中断前后是在同一个进程上下文中，只是由用户态转向内核态执行；
3.进程上下文包含了进程执行需要的所有信息
I 用户地址空间： 包括程序代码，数据，用户堆栈等 II 控制信息 ：进程描述符，内核堆栈等
III 硬件上下文（注意中断也要保存硬件上下文只是保存的方法不同）
4.schedule()函数选择一个新的进程来运行，并调用context_switch进行上下文的切换，这个宏调用switch_to来进行关键上下文切换
schedule 在/linux-3.18.6/kernel/sched/core.c
2733/* 2734 * __schedule() is the main scheduler function. 2735 * 2736 * The main means of driving the scheduler and thus entering this function are: 2737 * 2738 * 1. Explicit blocking: mutex, semaphore, waitqueue, etc. 2739 * 2740 * 2. TIF_NEED_RESCHED flag is checked on interrupt and userspace return 2741 * paths. For example, see arch/x86/entry_64.S. 2742 * 2743 * To drive preemption between tasks, the scheduler sets the flag in timer 2744 * interrupt handler scheduler_tick(). 2745 * 2746 * 3. Wakeups don't really cause entry into schedule(). They add a 2747 * task to the run-queue and that's it. 2748 * 2749 * Now, if the new task added to the run-queue preempts the current 2750 * task, then the wakeup sets TIF_NEED_RESCHED and schedule() gets 2751 * called on the nearest possible occasion: 2752 * 2753 * - If the kernel is preemptible (CONFIG_PREEMPT=y): 2754 * 2755 * - in syscall or exception context, at the next outmost 2756 * preempt_enable(). (this might be as soon as the wake_up()'s 2757 * spin_unlock()!) 2758 * 2759 * - in IRQ context, return from interrupt-handler to 2760 * preemptible context 2761 * 2762 * - If the kernel is not preemptible (CONFIG_PREEMPT is not set) 2763 * then at the next: 2764 * 2765 * - cond_resched() call 2766 * - explicit schedule() call 2767 * - return from syscall or exception to user-space 2768 * - return from interrupt-handler to user-space 2769 */ 2770static void __sched __schedule(void) 2771{ 2772 struct task_struct *prev, *next; 2773 unsigned long *switch_count; 2774 struct rq *rq; 2775 int cpu; 2776 2777need_resched: 2778 preempt_disable(); 2779 cpu = smp_processor_id(); 2780 rq = cpu_rq(cpu); 2781 rcu_note_context_switch(cpu); 2782 prev = rq->curr; 2783 2784 schedule_debug(prev); 2785 2786 if (sched_feat(HRTICK)) 2787 hrtick_clear(rq); 2788 2789 /* 2790 * Make sure that signal_pending_state()->signal_pending() below 2791 * can't be reordered with __set_current_state(TASK_INTERRUPTIBLE) 2792 * done by the caller to avoid the race with signal_wake_up(). 2793 */ 2794 smp_mb__before_spinlock(); 2795 raw_spin_lock_irq(&rq->lock); 2796 2797 switch_count = &prev->nivcsw; 2798 if (prev->state && !(preempt_count() & PREEMPT_ACTIVE)) { 2799 if (unlikely(signal_pending_state(prev->state, prev))) { 2800 prev->state = TASK_RUNNING; 2801 } else { 2802 deactivate_task(rq, prev, DEQUEUE_SLEEP); 2803 prev->on_rq = 0; 2804 2805 /* 2806 * If a worker went to sleep, notify and ask workqueue 2807 * whether it wants to wake up a task to maintain 2808 * concurrency. 2809 */ 2810 if (prev->flags & PF_WQ_WORKER) { 2811 struct task_struct *to_wakeup; 2812 2813 to_wakeup = wq_worker_sleeping(prev, cpu); 2814 if (to_wakeup) 2815 try_to_wake_up_local(to_wakeup); 2816 } 2817 } 2818 switch_count = &prev->nvcsw; 2819 } 2820 2821 if (task_on_rq_queued(prev) || rq->skip_clock_update < 0) 2822 update_rq_clock(rq); 2823 2824 next = pick_next_task(rq, prev); 2825 clear_tsk_need_resched(prev); 2826 clear_preempt_need_resched(); 2827 rq->skip_clock_update = 0; 2828 2829 if (likely(prev != next)) { 2830 rq->nr_switches++; 2831 rq->curr = next; 2832 ++*switch_count; 2833 2834 context_switch(rq, prev, next); /* unlocks the rq */ 2835 /* 2836 * The context switch have flipped the stack from under us 2837 * and restored the local variables which were saved when 2838 * this task called schedule() in the past. prev == current 2839 * is still correct, but it can be moved to another cpu/rq. 2840 */ 2841 cpu = smp_processor_id(); 2842 rq = cpu_rq(cpu); 2843 } else 2844 raw_spin_unlock_irq(&rq->lock); 2845 2846 post_schedule(rq); 2847 2848 sched_preempt_enable_no_resched(); 2849 if (need_resched()) 2850 goto need_resched; 2851}
我们看其中的两个，第一是第2824行的next = pick_next_task(rq, prev); //完成找到下一个进程
第二是第2834行的context_switch(rq, prev, next); /* unlocks the rq */ //完成切换
I next = pick_next_task(rq, prev);//进程调度算法都封装这个函数内部
pick_next_stack在/linux-3.18.6/kernel/sched/core.c
2694/* 2695 * Pick up the highest-prio task: 2696 */ 2697static inline struct task_struct * 2698pick_next_task(struct rq *rq, struct task_struct *prev) 2699{ 2700 const struct sched_class *class = &fair_sched_class; 2701 struct task_struct *p; 2702 2703 /* 2704 * Optimization: we know that if all tasks are in 2705 * the fair class we can call that function directly: 2706 */ 2707 if (likely(prev->sched_class == class && 2708 rq->nr_running == rq->cfs.h_nr_running)) { 2709 p = fair_sched_class.pick_next_task(rq, prev); 2710 if (unlikely(p == RETRY_TASK)) 2711 goto again; 2712 2713 /* assumes fair_sched_class->next == idle_sched_class */ 2714 if (unlikely(!p)) 2715 p = idle_sched_class.pick_next_task(rq, prev); 2716 2717 return p; 2718 } 2719 2720again: 2721 for_each_class(class) { 2722 p = class->pick_next_task(rq, prev); 2723 if (p) { 2724 if (unlikely(p == RETRY_TASK)) 2725 goto again; 2726 return p; 2727 } 2728 } 2729 2730 BUG(); /* the idle class will always have a runnable task */ 2731}
II context_switch(rq, prev, next);//进程上下文切换，切换到新的内存和新的寄存器状态
context_switch在 /linux-3.18.6/kernel/sched/core.c
2331/* 2332 * context_switch - switch to the new MM and the new 2333 * thread's register state. 2334 */ 2335static inline void 2336context_switch(struct rq *rq, struct task_struct *prev, 2337 struct task_struct *next) 2338{ 2339 struct mm_struct *mm, *oldmm; 2340 2341 prepare_task_switch(rq, prev, next); 2342 2343 mm = next->mm; 2344 oldmm = prev->active_mm; 2345 /* 2346 * For paravirt, this is coupled with an exit in switch_to to 2347 * combine the page table reload and the switch backend into 2348 * one hypercall. 2349 */ 2350 arch_start_context_switch(prev); 2351 2352 if (!mm) { 2353 next->active_mm = oldmm; 2354 atomic_inc(&oldmm->mm_count); 2355 enter_lazy_tlb(oldmm, next); 2356 } else 2357 switch_mm(oldmm, mm, next); 2358 2359 if (!prev->mm) { 2360 prev->active_mm = NULL; 2361 rq->prev_mm = oldmm; 2362 } 2363 /* 2364 * Since the runqueue lock will be released by the next 2365 * task (which is an invalid locking op but in the case 2366 * of the scheduler it's an obvious special-case), so we 2367 * do an early lockdep release here: 2368 */ 2369 spin_release(&rq->lock.dep_map, 1, _THIS_IP_); 2370 2371 context_tracking_task_switch(prev, next); 2372 /* Here we just switch the register state and the stack. */ 2373 switch_to(prev, next, prev); 2374 2375 barrier(); 2376 /* 2377 * this_rq must be evaluated again because prev may have moved 2378 * CPUs since it called schedule(), thus the 'rq' on its stack 2379 * frame will be invalid. 2380 */ 2381 finish_task_switch(this_rq(), prev); 2382}
其中的第2341行的prepare_task_switch(rq, prev, next); //完成切换前的准备工作
第2373行的switch_to(prev, next, prev); //完成切换
III switch_to利用了prev和next两个参数：prev指向当前进程，next指向被调度的进程
在/linux-3.18.6/arch/x86/include/asm/switch_to.h
1#ifndef _ASM_X86_SWITCH_TO_H 2#define _ASM_X86_SWITCH_TO_H 3 4struct task_struct; /* one of the stranger aspects of C forward declarations */ 5__visible struct task_struct *__switch_to(struct task_struct *prev, 6 struct task_struct *next); 7struct tss_struct; 8void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, 9 struct tss_struct *tss); 10 11#ifdef CONFIG_X86_32 12 13#ifdef CONFIG_CC_STACKPROTECTOR 14#define __switch_canary \ 15 "movl %P[task_canary](%[next]), %%ebx\n\t" \ 16 "movl %%ebx, "__percpu_arg([stack_canary])"\n\t" 17#define __switch_canary_oparam \ 18 , [stack_canary] "=m" (stack_canary.canary) 19#define __switch_canary_iparam \ 20 , [task_canary] "i" (offsetof(struct task_struct, stack_canary)) 21#else /* CC_STACKPROTECTOR */ 22#define __switch_canary 23#define __switch_canary_oparam 24#define __switch_canary_iparam 25#endif /* CC_STACKPROTECTOR */ 26 27/* 28 * Saving eflags is important. It switches not only IOPL between tasks, 29 * it also protects other tasks from NT leaking through sysenter etc. 30 */ 31#define switch_to(prev, next, last) \ 32do { \ 33 /* \ 34 * Context-switching clobbers all registers, so we clobber \ 35 * them explicitly, via unused output variables. \ 36 * (EAX and EBP is not listed because EBP is saved/restored \ 37 * explicitly for wchan access and EAX is the return value of \ 38 * __switch_to()) \ 39 */ \ 40 unsigned long ebx, ecx, edx, esi, edi; \ 41 \ 42 asm volatile("pushfl\n\t" /* save flags */ \ 43 "pushl %%ebp\n\t" /* save EBP */ \ 44 "movl %%esp,%[prev_sp]\n\t" /* save ESP */ \ 45 "movl %[next_sp],%%esp\n\t" /* restore ESP */ \ 46 "movl $1f,%[prev_ip]\n\t" /* save EIP */ \ 47 "pushl %[next_ip]\n\t" /* restore EIP */ \ 48 __switch_canary \ 49 "jmp __switch_to\n" /* regparm call */ \ 50 "1:\t" \ 51 "popl %%ebp\n\t" /* restore EBP */ \ 52 "popfl\n" /* restore flags */ \ 53 \ 54 /* output parameters */ \ 55 : [prev_sp] "=m" (prev->thread.sp), \ 56 [prev_ip] "=m" (prev->thread.ip), \ 57 "=a" (last), \ 58 \ 59 /* clobbered output registers: */ \ 60 "=b" (ebx), "=c" (ecx), "=d" (edx), \ 61 "=S" (esi), "=D" (edi) \ 62 \ 63 __switch_canary_oparam \ 64 \ 65 /* input parameters: */ \ 66 : [next_sp] "m" (next->thread.sp), \ 67 [next_ip] "m" (next->thread.ip), \ 68 \ 69 /* regparm parameters for __switch_to(): */ \ 70 [prev] "a" (prev), \ 71 [next] "d" (next) \ 72 \ 73 __switch_canary_iparam \ 74 \ 75 : /* reloaded segment registers */ \ 76 "memory"); \ 77} while (0) 78 79#else /* CONFIG_X86_32 */ 80 81/* frame pointer must be last for get_wchan */ 82#define SAVE_CONTEXT "pushf ; pushq %%rbp ; movq %%rsi,%%rbp\n\t" 83#define RESTORE_CONTEXT "movq %%rbp,%%rsi ; popq %%rbp ; popf\t" 84 85#define __EXTRA_CLOBBER \ 86 , "rcx", "rbx", "rdx", "r8", "r9", "r10", "r11", \ 87 "r12", "r13", "r14", "r15" 88 89#ifdef CONFIG_CC_STACKPROTECTOR 90#define __switch_canary \ 91 "movq %P[task_canary](%%rsi),%%r8\n\t" \ 92 "movq %%r8,"__percpu_arg([gs_canary])"\n\t" 93#define __switch_canary_oparam \ 94 , [gs_canary] "=m" (irq_stack_union.stack_canary) 95#define __switch_canary_iparam \ 96 , [task_canary] "i" (offsetof(struct task_struct, stack_canary)) 97#else /* CC_STACKPROTECTOR */ 98#define __switch_canary 99#define __switch_canary_oparam 100#define __switch_canary_iparam 101#endif /* CC_STACKPROTECTOR */ 102 103/* Save restore flags to clear handle leaking NT */ 104#define switch_to(prev, next, last) \ 105 asm volatile(SAVE_CONTEXT \ 106 "movq %%rsp,%P[threadrsp](%[prev])\n\t" /* save RSP */ \ 107 "movq %P[threadrsp](%[next]),%%rsp\n\t" /* restore RSP */ \ 108 "call __switch_to\n\t" \ 109 "movq "__percpu_arg([current_task])",%%rsi\n\t" \ 110 __switch_canary \ 111 "movq %P[thread_info](%%rsi),%%r8\n\t" \ 112 "movq %%rax,%%rdi\n\t" \ 113 "testl %[_tif_fork],%P[ti_flags](%%r8)\n\t" \ 114 "jnz ret_from_fork\n\t" \ 115 RESTORE_CONTEXT \ 116 : "=a" (last) \ 117 __switch_canary_oparam \ 118 : [next] "S" (next), [prev] "D" (prev), \ 119 [threadrsp] "i" (offsetof(struct task_struct, thread.sp)), \ 120 [ti_flags] "i" (offsetof(struct thread_info, flags)), \ 121 [_tif_fork] "i" (_TIF_FORK), \ 122 [thread_info] "i" (offsetof(struct task_struct, stack)), \ 123 [current_task] "m" (current_task) \ 124 __switch_canary_iparam \ 125 : "memory", "cc" __EXTRA_CLOBBER) 126 127#endif /* CONFIG_X86_32 */ 128 129#endif /* _ASM_X86_SWITCH_TO_H */ 130
完成进程切换
二、分析进程切换：我们用switch_to中的部分代码分析
27/* 28 * Saving eflags is important. It switches not only IOPL between tasks, 29 * it also protects other tasks from NT leaking through sysenter etc. 30 */ 31#define switch_to(prev, next, last) \ 32do { \ 33 /* \ 34 * Context-switching clobbers all registers, so we clobber \ 35 * them explicitly, via unused output variables. \ 36 * (EAX and EBP is not listed because EBP is saved/restored \ 37 * explicitly for wchan access and EAX is the return value of \ 38 * __switch_to()) \ 39 */ \ 40 unsigned long ebx, ecx, edx, esi, edi; \ 41 \ 42 asm volatile("pushfl\n\t" /* save flags */ \ 43 "pushl %%ebp\n\t" /* save EBP */ \ 44 "movl %%esp,%[prev_sp]\n\t" /* save ESP */ \ 45 "movl %[next_sp],%%esp\n\t" /* restore ESP */ \ 46 "movl $1f,%[prev_ip]\n\t" /* save EIP */ \ 47 "pushl %[next_ip]\n\t" /* restore EIP */ \ 48 __switch_canary \ 49 "jmp __switch_to\n" /* regparm call */ \ 50 "1:\t" \ 51 "popl %%ebp\n\t" /* restore EBP */ \ 52 "popfl\n" /* restore flags */ \ 53 \ 54 /* output parameters */ \ 55 : [prev_sp] "=m" (prev->thread.sp), \ 56 [prev_ip] "=m" (prev->thread.ip), \ 57 "=a" (last), \ 58 \ 59 /* clobbered output registers: */ \ 60 "=b" (ebx), "=c" (ecx), "=d" (edx), \ 61 "=S" (esi), "=D" (edi) \ 62 \ 63 __switch_canary_oparam \ 64 \ 65 /* input parameters: */ \ 66 : [next_sp] "m" (next->thread.sp), \ 67 [next_ip] "m" (next->thread.ip), \ 68 \ 69 /* regparm parameters for __switch_to(): */ \ 70 [prev] "a" (prev), \ 71 [next] "d" (next) \ 72 \ 73 __switch_canary_iparam \ 74 \ 75 : /* reloaded segment registers */ \ 76 "memory"); \ 77} while (0)
利用了prev和next两个参数：prev指向当前进程，当前进程用X表示。next指向被调度的进程，即下一个进程，用Y表示。至于如何实现调度，看pick_next_task。
看第42行：把flags压入到当前进程X的栈里面，保存flags。
看第43行：把当前的ebp压入当前进程X的栈里，保存ebp。
看第44行：把当前的esp保存到当前进程X的thread.sp里面。其中[prev_sp]是个标识，他在第55行，代替的是prev->thread.sp。
看第45行：把下一个进行Y的thread.sp赋值给esp，这一步实现把本来指向X的栈指针esp，现在指向了Y。其中[next_sp]如上所述，在第66行。
看第46行：把50行的位置存到X进程的thread_ip里面，保存eip。下一次可以从50行开始执行。其中[prev_ip]如上所述，在第56行。
看第47行：把下一个进程Y的threat.ip压入Y进程的栈里面。其中[next_ip]如上所示，在第67行。
看第49行：跳转到__swap_to
看第51行：Y进程里面出栈操作，放到ebp里面。
看第52行：把Y进程里面的出栈，弹出flags
第51，52行正好和第42，43行操作互逆。
三、实验：用gdb跟踪分析一个schedule()函数
四、Linux系统的一般执行过程
最一般情况：正在运行的用户态进程X切换到运行用户态进程Y的过程
1.正在运行的用户态进程X
2.发生中断——save cs:eip/esp/eflags(current) to kernel stack,then load cs:eip(entry of a specific ISR) and ss:esp(point to kernel stack).
3. SAVE_ALL //保存现场
4. 中断处理过程中或中断返回前调用了schedule()，其中的switch_to做了关键的进程上下文切换
5. 标号1之后开始运行用户态进程Y(这里Y曾经通过以上步骤被切换出去过因此可以从标号1继续执行)
6. restore_all //恢复现场
7. iret - pop cs:eip/ss:esp/eflags from kernel stack
8. 继续运行用户态进程Y
几种特殊的情况：
1. 通过中断处理过程中的调度时机，用户态进程与内核线程之间互相切换和内核线程之间互相切换，与最一般的情况非常类似，只是内核线程运行过程中发生中断没有进程用户态和内核态的转换；
2. 内核线程主动调用schedule()，只有进程上下文的切换，没有发生中断上下文的切换，与最一般的情况略简略；
3. 创建子进程的系统调用在子进程中的执行起点及返回用户态，如fork；
4. 加载一个新的可执行程序后返回到用户态的情况，如execve；
