linux的页面分配程序采用,Linux内存页面分配

深藏阁楼爱情的钟 2022-10-20 12:13 179阅读 0赞

在Linux中，对物理内存的管理是怎么实现的呢？对页面的分配和回收是如何实现的呢？

Linux中对物理内存的管理是通过zone来管理的，以X86为例，16MB以下的物理内存为DMA zone；896MB以下的区域为normal zone；然后896MB以上的区域统称为HighMemory zone；然后在分配内存的时候，可以指定区域来实现在某个特定的zone分配页面；

那么zone又是如何管理物理内存的呢？Linux同样是使用伙伴系统来管理物理内存，来尽量避免碎片的产生；对伙伴系统的原理应该很清楚，就不说了

struct zone \{

/\* Fields commonly accessed by the page allocator \*/

/\* zone watermarks, access with \*\_wmark\_pages(zone) macros \*/

unsigned long watermark\[NR\_WMARK\];

/\*

\* We don't know if the memory that we're going to allocate will be freeable

\* or/and it will be released eventually, so to avoid totally wasting several

\* GB of ram we must reserve some of the lower zone memory (otherwise we risk

\* to run OOM on the lower zones despite there's tons of freeable ram

\* on the higher zones). This array is recalculated at runtime if the

\* sysctl\_lowmem\_reserve\_ratio sysctl changes.

\*/

unsigned longlowmem\_reserve\[MAX\_NR\_ZONES\];

struct per\_cpu\_pageset \_\_percpu \*pageset;

/\*

\* free areas of different sizes

\*/

spinlock\_tlock;

struct free\_area free\_area\[MAX\_ORDER\];

/\* Fields commonly accessed by the page reclaim scanner \*/

spinlock\_tlru\_lock;

struct lruveclruvec;

unsigned longpages\_scanned; /\* since last reclaim \*/

unsigned longflags; /\* zone flags, see below \*/

wait\_queue\_head\_t\* wait\_table;

unsigned longwait\_table\_hash\_nr\_entries;

unsigned longwait\_table\_bits;

/\*

\* spanned\_pages is the total pages spanned by the zone, including

\* holes, which is calculated as:

\* spanned\_pages = zone\_end\_pfn - zone\_start\_pfn;

\* present\_pages is physical pages existing within the zone, which

\* is calculated as:

\*present\_pages = spanned\_pages - absent\_pages(pages in holes);

\* managed\_pages is present pages managed by the buddy system, which

\* is calculated as (reserved\_pages includes pages allocated by the

\* bootmem allocator):

\*managed\_pages = present\_pages - reserved\_pages;

\* So present\_pages may be used by memory hotplug or memory power

\* management logic to figure out unmanaged pages by checking

\* (present\_pages - managed\_pages). And managed\_pages should be used

\* by page allocator and vm scanner to calculate all kinds of watermarks

\* and thresholds.

\*/

unsigned longspanned\_pages;

unsigned longpresent\_pages;

unsigned longmanaged\_pages;

/\*

\* rarely used fields:

\*/

const char\*name;

\};

其中和伙伴系统有关的是：

struct free\_area free\_area\[MAX\_ORDER\]; //MAX\_ORDER=11

11个free\_area 分别为 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 和 1024个页面，所以最大的为4MB； free\_area的结构体为

struct free\_area \{

struct list\_headfree\_list\[MIGRATE\_TYPES\];

unsigned longnr\_free;

\};所有的空闲页面都挂在free\_list下；伙伴系统的回收和分配方式很容易理解，就不说了，至于代码就不贴在这里了

伙伴系统是物理内存页分配的基础，在分配页面的时候，最主要的函数是\_ \_alloc\_pages( )，而它的大体过程为

for (i = 0; (z=zonelist->zones\[i\]) != NULL; i++) \{

if (zone\_watermark\_ok(z, order, ...)) \{

page = buffered\_rmqueue(z, order, gfp\_mask);

if (page)

return page;

zone\_watermark\_ok(z, order, ...)函数，与zone的临界值有关，用来检测内存分配的安全性，以后还会详细的讨论。

buffered\_rmqueue()函数是在给定的zone分配页面，它继续调用\_ \_rmqueue( )到伙伴系统中分配页面；

struct free\_area \*area;

unsigned int current\_order;

for (current\_order=order; current\_order<11; ++current\_order) \{

area = zone->free\_area + current\_order;

if (!list\_empty(&area->free\_list))

goto block\_found;

block\_found:

page = list\_entry(area->free\_list.next, struct page, lru);

list\_del(&page->lru);

ClearPagePrivate(page);

page->private = 0;

area->nr\_free--;

zone->free\_pages -= 1UL << order;

size = 1 << curr\_order;

while (curr\_order > order) \{

area--;

curr\_order--;

size >>= 1;

buddy = page + size;

/\* insert buddy as first element in the list \*/

list\_add(&buddy->lru, &area->free\_list);

area->nr\_free++;

buddy->private = curr\_order;

SetPagePrivate(buddy);

return page;

return NULL; 这样就可以完成对物理页面的分配了，所有的物理页面的分配，例如slab的分配，都是基于上面的过程实现的！

linux的页面分配程序采用,Linux内存页面分配

发表评论取消回复

还没有评论，来说两句吧...

相关阅读

相关 Linux学习--内存分配算法

相关 Linux 如何进行内存分配

相关 Linux应用内存分配函数

相关 linux内核内存分配函数,linux内核编程内存分配

相关嵌入式linux 内存分配,嵌入式linux内核中常见内存分配

相关 linux的页面分配程序采用,Linux内存页面分配

相关 linux内存分配及内存管理

相关 Linux中的内存分配

相关 linux内存分配与回收

相关 linux内核的内存分配函数