主要介绍partial和cpu partial的产生.
内核未定义 CONFIG_NUMA

partial没有指明是node partial还是cpu partial时, 则指的是node partial.

1. node partial的产生

在cpu0上执行newslabobjects -> new_slab, 由于可能睡眠, 之后可能运行在cpu1上. 这时若cpu1的c->page非空, 则根据情况, 可能将其放入node partial中.

linux-3.10.86/mm/slub.c

__slab_alloc
|--local_irq_save(flags);
|--freelist =new_slab_objects //buddy的page的首个obj
|    |--page = new_slab(s, flags, node); //在cpu0上
|    |--c = __this_cpu_ptr(s->cpu_slab) //进程可能迁移到cpu1上
|    |--if (c->page) flush_slab(s, c);

flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c)
|-- deactivate_slab(s, c->page, c->freelist)


static void deactivate_slab(struct kmem_cache *s, struct page *page, void *freelist)
{

    /*
     while循环前:
     obj_t2 -> obj_t1 -> NULL
     ^
     @page->freelist

     obj_03 -> obj_02 -> obj_01 -> NULL
     ^
     @freelist


     循环单次后:   
     obj_03   ->  obj_t2  -> obj_t1 -> NULL
     ^
     page->freelist 


     obj_02 -> obj_01 -> NULL
     ^
     freelist

     再来一回:
     obj_02 -> obj_03   ->  obj_t2  -> obj_t1 -> NULL
     ^
     page->freelist


     obj_01 -> NULL
     ^
     freelist

     问题:为何不直接修改指针把两个链表串起来, 而是要一个一个object的放到链表中?
     答:因为要修改counters, 所以要一个一个数.
    */
    while (freelist && (nextfree = get_freepointer(s, freelist))) {
        ...
    }

    ....

    /*
    如果这里freelist为NULL, 说明@freelist传入时就为NULL
    */
    /*把the last one也接上*/
    if (freelist) {
        new.inuse--;
        /* 
        new->freelist
        v
        obj_01 -> obj_02 -> obj_03   ->  obj_t2  -> obj_t1 -> NULL
        ^
        freelist
        */
        set_freepointer(s, freelist, old.freelist);
        new.freelist = freelist;
    } else
        ...

    new.frozen = 0;

    //有M_FREE, M_PARTIAL等情况, 这里只看M_PARTIAL
    ...

    if (m == M_PARTIAL) {
        /*问题:为何不是put_cpu_partial()?*/
        add_partial(n, page, tail);

    }

}

1. tid

linux-3.10.86/include/linux/slub_def.h

struct kmem_cache_cpu {
    ...
    unsigned long tid;  /* Globally unique transaction id */
    ..
};

linux-3.10.86/mm/slub.c

#ifdef CONFIG_PREEMPT
/*
 * Calculate the next globally unique transaction for disambiguiation
 * during cmpxchg. The transactions start with the cpu number and are then
 * incremented by CONFIG_NR_CPUS.

问题:一次不是加1, 而是加CONFIGNRCPUS, 为啥?
答:这个设计是为了让任何时刻每个cpu的tid值都不一样. 不过上面的注释有点老, 因为实际并不是加CONFIGNRCPUS, 而是TID_STEP.

文来自本人旧博客: blog.163.com/awaken_ing/blog/static/1206131972016316114114855

平台 ARM Versatile Express for Cortex-A9 SMP
内核版本 3.10.86 (未定义 CONFIG_NUMA)

1. 概览

The SLUB allocator 相比SLAB, 试图 remove the metadata overhead inside slabs, reduce the number of caches, and so on. The only metadata present in the SLUB allocator is the in-object “next-free-object” pointer, which allows us to link free objects together. struct kmem_cache 的成员 int offset 用来指明 指针在 object 中的偏移量, 这个指针指向的是下一个可用的object. How does the allocator manage to find the first free object? The answer lies in the approach of saving a pointer to such an object inside each page struct associated with the slab page. SLUB allocator 没有SLAB的full list 和 empty list.

图片来自 http://events.linuxfoundation.org/sites/events/files/slides/slaballocators.pdf

linux-3.10.86/mm/slub.c

slab_alloc() -> slab_alloc_node()
slab_alloc_node(struct kmem_cache *s, ...
{
    struct kmem_cache_cpu *c=__this_cpu_ptr(s->cpu_slab);
    object = c->freelist;
    if fastpath {
        void *next_object = get_freepointer_safe(s, object);
        c->freelist=next_object; //via  this_cpu_cmpxchg_double()
    }else
        ...
    return object;
}
static inline void *get_freepointer(struct kmem_cache *s, void *object)
{
    return *(void **)(object + s->offset);
}

文来自本人的旧博客 blog.163.com/awaken_ing/blog/static/1206131972016124113539444/

0. 引言

The lost wake-up problem 请参考 http://www.linuxjournal.com/article/8144
本篇主要解释为何修改后的代码没有问题. 修改后的代码为:

1  set_current_state(TASK_INTERRUPTIBLE);
2  spin_lock(&list_lock);
3  if(list_empty(&list_head)) {
4         spin_unlock(&list_lock); //如果这里面让出cpu?
//如果在这个点被生产者唤醒会如何?
5         schedule();
6         spin_lock(&list_lock);
7  }
8  set_current_state(TASK_RUNNING);
9
10 /* Rest of the code ... */
11 spin_unlock(&list_lock);

1. 何时设置 TIFNEEDRESCHED

任务task01还没有完成, 但调度器认为 其运行的够久了, 就设置TIFNEEDRESCHED, 这样, 在合适的时机, task01就会被调度走, 由其他task使用cpu.

除了schedulertick会设置外, 当一个优先级高的进程进入可执行状态的时候, trytowakeup()也会设置这个标志。

2. 不检查preempt_count?

例 linux-3.10.86/arch/arm/kernel/entry-armv.S

__vector_irq()
|--__irq_usr()  @entry-armv.S
|   |--usr_entry()
|   |--irq_handler()
|   |--ret_to_user_from_irq() @entry-common.S
|   |   |--work_pending()   
|   |   |   |--do_work_pending @signal.c

do_work_pending
{
    do {
        if (likely(thread_flags & _TIF_NEED_RESCHED))
            schedule();
        else
           ...
    }
}