linux内存管理slab算法之kmem_cache结构创建

kmem_cache是slab的核心结构体，主要描述slab的各种信息和链接空闲slab，还保存高速缓存的指针数组。所以要想使用slab分配得先创建kmem_cache结构体。

struct kmem_cache *
kmem_cache_create (const char *name, size_t size, size_t align,
	unsigned long flags, void (*ctor)(void *))
{
	size_t left_over, slab_size, ralign;
	struct kmem_cache *cachep = NULL, *pc;
	if (!name || (size < BYTES_PER_WORD) ||
	    size > KMALLOC_MAX_SIZE) {
		printf("slab create error\n");
	}
	 //size和字对齐，32位4字节对齐，64位8字节对齐
	if (size & (BYTES_PER_WORD - 1)) {
		size += (BYTES_PER_WORD - 1);
		size &= ~(BYTES_PER_WORD - 1);
	}


	//如果标记了cache对齐,cache行大小在现代的处理器中大部分是64B
	if (flags & SLAB_HWCACHE_ALIGN) {
		ralign = cache_line_size(); //获取cache行大小
		while (size <= ralign / 2)  //获取size最小对齐
			ralign /= 2;
	} else {
		ralign = BYTES_PER_WORD;
	}
    /* 特定体系结构最小的slab对象大小*/
	if (ralign < ARCH_SLAB_MINALIGN) {
		ralign = ARCH_SLAB_MINALIGN;
	}
	if (ralign < align) {
		ralign = align;    //取更大的对齐
	}
	/*
	 * 计算完成对齐
	 */
	align = ralign;

	//获取kmem_cache结构体指针，从cache_cache结构体中分配
	cachep = kmem_cache_zalloc(&cache_cache, GFP_KERNEL);
	if (!cachep)
		goto oops;


	 //如果申请的对象size大于等于512B，则slab管理结构不在本slab页面上,在早期初始化slab阶段所分配的cache都是onslab的
	 //早期初始化完成后可以offslab
	if ((size >= (PAGE_SIZE >> 3)) && !slab_early_init)
		/*
		 * Size is large, assume best to place the slab management obj
		 * off-slab (should allow better packing of objs).
		 */
		 //如果分配的对象很大，则最好将slab管理结构放在slab外面，也就是动态的从其它cache分配一块
		flags |= CFLGS_OFF_SLAB;

    //对齐申请的size
	size = ALIGN(size, align);

    //计算剩余字节和一个slab的页面的阶
	left_over = calculate_slab_order(cachep, size, align, flags);

	//如果对象个数是0则说明分配出错，内存不够了
	if (!cachep->num) {
		printf(
		       "kmem_cache_create: couldn't create cache %s.\n", name);
		cachep = NULL;
		goto oops;
	}
	//计算对齐后的slab管理结构的大小
	slab_size = ALIGN(cachep->num * sizeof(kmem_bufctl_t)
			  + sizeof(struct slab), align);
	
	 //如果slab管理结构在本cache外面，并且剩余的字节数大于等于管理结构体的
	 //大小则把标记位置为onslab，并且剩余字节数减去slab管理结构的大小
	 //这样做应该是为了节省空间，减少内部碎片，但是可能会造成比较多的cache miss
	if (flags & CFLGS_OFF_SLAB && left_over >= slab_size) {
		flags &= ~CFLGS_OFF_SLAB;
		left_over -= slab_size;
	}
    //如果slab管理结构不在本cache上，则slab管理结构的大小不需要对齐，只需要计算真实的slab管理结构大小
	if (flags & CFLGS_OFF_SLAB) {
		/* really off slab. No need for manual alignment */
		slab_size =
		    cachep->num * sizeof(kmem_bufctl_t) + sizeof(struct slab);
	}
    //着色偏移等于cache line的大小
	cachep->colour_off = cache_line_size();
	/* Offset must be a multiple of the alignment. */
	//如果指定的对齐大于cache line大小，则着色偏移等于指定对齐
	if (cachep->colour_off < align)
		cachep->colour_off = align;
    //着色颜色个数等于剩余字节数除以着色偏移
	cachep->colour = left_over / cachep->colour_off;
	//slab管理结构的大小
	cachep->slab_size = slab_size;
	//标记位
	cachep->flags = flags;
	cachep->gfpflags = 0;
	//cache中的object大小
	cachep->buffer_size = size;
	//cache对象大小的倒数
	cachep->reciprocal_buffer_size = reciprocal_value(size);

    //如果slab管理结构不在本cache，则需要给slab管理结构指定一个大小适合的kmem_cache
    //给slab管理结构分配内存
	if (flags & CFLGS_OFF_SLAB) {
		cachep->slabp_cache = kmem_find_general_cachep(slab_size, 0u);
		/*
		 * This is a possibility for one of the malloc_sizes caches.
		 * But since we go off slab only for object size greater than
		 * PAGE_SIZE/8, and malloc_sizes gets created in ascending order,
		 * this should not happen at all.
		 * But leave a BUG_ON for some lucky dude.
		 */
		//BUG_ON(ZERO_OR_NULL_PTR(cachep->slabp_cache));
	}
	//构造函数指针，分配完回调
	cachep->ctor = ctor;
	//kmem_cache的名字
	cachep->name = name;
	//设置cache的array缓存
	if (setup_cpu_cache(cachep)) {
		__kmem_cache_destroy(cachep);
		cachep = NULL;
		goto oops;
	}
	/* cache setup completed, link it into the list */
	//把分配的cache加入到kmem_cache链表
	list_add(&cachep->next, &cache_chain);
oops:
    //如果cachep指针为空，则说明没分配成功报错
	if (!cachep && (flags & SLAB_PANIC))
		printf("kmem_cache_create(): failed to create slab `%s'\n",
		      name);
	//mutex_unlock(&cache_chain_mutex);
//	put_online_cpus();
    //返回cachep
	return cachep;
}

接下来主要分析calculate_slab_order函数，此函数主要计算一个slab占用的页面个数，以最小对象个数为单位。

static size_t calculate_slab_order(struct kmem_cache *cachep,
			size_t size, size_t align, unsigned long flags)
{
	unsigned long offslab_limit;
	size_t left_over = 0;
	int gfporder;

	//计算一个slab占用页面的最小阶
	for (gfporder = 0; gfporder <= KMALLOC_MAX_ORDER; gfporder++) {
		unsigned int num;
		size_t remainder;

		cache_estimate(gfporder, size, align, flags, &remainder, &num);
		if (!num)
			continue;

		if (flags & CFLGS_OFF_SLAB) {
			/*
			 * Max number of objs-per-slab for caches which
			 * use off-slab slabs. Needed to avoid a possible
			 * looping condition in cache_grow().
			 */
			 /*
			  *如果slab管理结构不在本slab页面上则需要计算slab大小的限制，主要针对
			  *大内存对象
			  */
			offslab_limit = size - sizeof(struct slab);
			offslab_limit /= sizeof(kmem_bufctl_t);

 			if (num > offslab_limit)
				break;
		}

		//一个slab对象个数
		cachep->num = num;
		//一个slab占用的页面的个数的阶
		cachep->gfporder = gfporder;
		//内部碎片大小
		left_over = remainder;

		/*
		 * A VFS-reclaimable slab tends to have most allocations
		 * as GFP_NOFS and we really don't want to have to be allocating
		 * higher-order pages when we are unable to shrink dcache.
		 */
		//if (flags & SLAB_RECLAIM_ACCOUNT)
		//	break;

		/*
		 * Large number of objects is good, but very large slabs are
		 * currently bad for the gfp()s.
		 */
		 //一个slab按照最小的页面数计算，比如不超过4KB的对象，每次分配slab只需要
		 //一页即可
		if (gfporder >= slab_break_gfp_order)
			break;

		/*
		 * Acceptable internal fragmentation?
		 */
		 //内部碎片乘以8，要小于等于所分配的页大小
		 //比如此时只分配了一页，那么leftover的大小不能大于512B
		if (left_over * 8 <= (PAGE_SIZE << gfporder))
			break;
	}
	return left_over;
}

具体的计算函数cache_estimate

//计算一个slab中的object的数目和slab剩余的字节数
static void cache_estimate(unsigned long gfporder, size_t buffer_size,
			   size_t align, int flags, size_t *left_over,
			   unsigned int *num)
{
	int nr_objs;
	size_t mgmt_size;
	size_t slab_size = PAGE_SIZE << gfporder;

	//如果slab管理结构不在本slab页面上
	if (flags & CFLGS_OFF_SLAB) {
		mgmt_size = 0;
		nr_objs = slab_size / buffer_size; //一个slab上对象的个数

		if (nr_objs > SLAB_LIMIT)
			nr_objs = SLAB_LIMIT; //如果超过个数限制就等于最大的
	} else {//如果在本页面上
		
		//计算对象个数
		nr_objs = (slab_size - sizeof(struct slab)) /
			  (buffer_size + sizeof(kmem_bufctl_t));

		/*
		 * This calculated number will be either the right
		 * amount, or one greater than what we want.
		 */
		 //计算的结构可能超过slab_size大小，需要减去一个对象
		if (slab_mgmt_size(nr_objs, align) + nr_objs*buffer_size
		       > slab_size)
			nr_objs--;

		if (nr_objs > SLAB_LIMIT)
			nr_objs = SLAB_LIMIT;

		//slab管理结构的最终大小
		mgmt_size = slab_mgmt_size(nr_objs, align);
	}
	*num = nr_objs;
	//内部碎片大小
	*left_over = slab_size - nr_objs*buffer_size - mgmt_size;
}