linux启动内存分配器
来源：岁月联盟
时间：2012-01-05
一、数据结构介绍
1，保留区间
因为在建立启动内存分配器的时候，会涉及保留内存。也就是说，之前保留给页表、分配器本身（用于映射的位图）、io等得内存在分配器建立后，当用它来分配内存空间时，保留出来的那些部分就不能再分配了。linux中对保留内存空间的部分用下列数据结构表示
/*
* Early reserved memory areas.
*/
#define MAX_EARLY_RES 20/*保留空间最大块数*/
struct early_res {/*保留空间结构*/
u64 start, end;
char name[16];
char overlap_ok;
};
/*保留内存空间全局变量*/
static struct early_res early_res[MAX_EARLY_RES] __initdata = {
{ 0, PAGE_SIZE, "BIOS data page" }, /* BIOS data page */
{}
};
2，bootmem分配器
/*
* node_bootmem_map is a map pointer - the bits represent all physical
* memory pages (including holes) on the node.
*/
/*用于bootmem分配器的节点数据结构*/
typedef struct bootmem_data {
unsigned long node_min_pfn;/*存放bootmem位图的第一个页面（即内核映象结束处的第一个页面）。*/
unsigned long node_low_pfn;/*物理内存的顶点，最高不超过896MB。*/
void *node_bootmem_map;
unsigned long last_end_off;/*用来存放在前一次分配中所分配的最后一个字节相对于last_pos的位移量*/
unsigned long hint_idx;/*存放前一次分配的最后一个页面号*/
struct list_head list;
} bootmem_data_t;
全局链表
static struct list_head bdata_list __initdata = LIST_HEAD_INIT(bdata_list);
二、启动分配器的建立
启动分配器的建立主要的流程为初始化映射位图、活动内存区的映射位置0（表示可用）、保留内存区域处理，其中保留区存放在上面介绍的全局数组中，这里只是将分配器中对应映射位图值1，表示已经分配。
下面我们看内核中具体的初始化流程。
start_kernel()->setup_arch()->initmem_init()
void __init setup_arch(char **cmdline_p)
{
.......
<span style="white-space: pre; "> </span>/*此函数在开始对bootmem分配制度建立做些准备工作
然后调用相关函数建立bootmem分配制度*/
initmem_init(0, max_pfn);
.......
}
<span style="font-family: Arial, Verdana, sans-serif; "><span style="white-space: normal; "></span></span>
<span style="font-family: Arial, Verdana, sans-serif; "><span style="white-space: normal; "></span></span><pre name="code" class="cpp">void __init initmem_init(unsigned long start_pfn,
unsigned long end_pfn)
{
#ifdef CONFIG_HIGHMEM
highstart_pfn = highend_pfn = max_pfn;
if (max_pfn > max_low_pfn)
highstart_pfn = max_low_pfn;
/*将活动内存放到early_node_map中,前面已经分析过了*/
e820_register_active_regions(0, 0, highend_pfn);
/*设置上面变量中的内存为当前，在这里没有
设置相关的宏*/
sparse_memory_present_with_active_regions(0);
printk(KERN_NOTICE "%ldMB HIGHMEM available./n",
pages_to_mb(highend_pfn - highstart_pfn));
num_physpages = highend_pfn;
/*高端内存开始地址物理*/
high_memory = (void *) __va(highstart_pfn * PAGE_SIZE - 1) + 1;
#else
e820_register_active_regions(0, 0, max_low_pfn);
sparse_memory_present_with_active_regions(0);
num_physpages = max_low_pfn;
high_memory = (void *) __va(max_low_pfn * PAGE_SIZE - 1) + 1;
#endif
#ifdef CONFIG_FLATMEM
max_mapnr = num_physpages;
#endif
__vmalloc_start_set = true;
printk(KERN_NOTICE "%ldMB LOWMEM available./n",
pages_to_mb(max_low_pfn));
/*安装bootmem分配器，此分配器在伙伴系统起来之前
用来进行承担内存的分配等管理*/
setup_bootmem_allocator();
}
void __init setup_bootmem_allocator(void)
{
int nodeid;
unsigned long bootmap_size, bootmap;
/*
* Initialize the boot-time allocator (with low memory only):
*/
/*计算所需要的映射页面大小一个字节一位，
所以需要对总的页面大小除以8*/
bootmap_size = bootmem_bootmap_pages(max_low_pfn)<<PAGE_SHIFT;
/*直接中e820中找到一个大小合适的内存块，返回基址*/
bootmap = find_e820_area(0, max_pfn_mapped<<PAGE_SHIFT, bootmap_size,
PAGE_SIZE);
if (bootmap == -1L)
panic("Cannot find bootmem map of size %ld/n", bootmap_size);
/*将用于位图映射的页面保留*/
reserve_early(bootmap, bootmap + bootmap_size, "BOOTMAP");
printk(KERN_INFO " mapped low ram: 0 - %08lx/n",
max_pfn_mapped<<PAGE_SHIFT);
printk(KERN_INFO " low ram: 0 - %08lx/n", max_low_pfn<<PAGE_SHIFT);
/*对每一个在线的node*/
for_each_online_node(nodeid) {
unsigned long start_pfn, end_pfn;
#ifdef CONFIG_NEED_MULTIPLE_NODES/*not set*/
start_pfn = node_start_pfn[nodeid];
end_pfn = node_end_pfn[nodeid];
if (start_pfn > max_low_pfn)
continue;
if (end_pfn > max_low_pfn)
end_pfn = max_low_pfn;
#else
start_pfn = 0;
end_pfn = max_low_pfn;
#endif
/*对指定节点安装启动分配器*/
bootmap = setup_node_bootmem(nodeid, start_pfn, end_pfn,
bootmap);
}
/*bootmem的分配制度到这里就已经建立完成，把after_bootmem
变量置成1，标识*/
after_bootmem = 1;
}
static unsigned long __init setup_node_bootmem(int nodeid,
unsigned long start_pfn,
unsigned long end_pfn,
unsigned long bootmap)
{
unsigned long bootmap_size;
/* don't touch min_low_pfn */
/*初始化映射位图，将位图中的所有位置1*/
bootmap_size = init_bootmem_node(NODE_DATA(nodeid),
bootmap >> PAGE_SHIFT,
start_pfn, end_pfn);
printk(KERN_INFO " node %d low ram: %08lx - %08lx/n",
nodeid, start_pfn<<PAGE_SHIFT, end_pfn<<PAGE_SHIFT);
printk(KERN_INFO " node %d bootmap %08lx - %08lx/n",
nodeid, bootmap, bootmap + bootmap_size);
/*将活动内存区对应位图相关位置0，表示可被分配的*/
free_bootmem_with_active_regions(nodeid, end_pfn);
/*对置保留位的相关页面对应的位图设置为1，表示已经分配
或者不可用(不能被分配)*/
early_res_to_bootmem(start_pfn<<PAGE_SHIFT, end_pfn<<PAGE_SHIFT);
/*返回映射页面的最后地址，下次映射即可以从这里开始*/
return bootmap + bootmap_size;
}
对于初始化映射位图，最终调用init_bootmem_core()
/*
* Called once to set up the allocator itself.
*/
static unsigned long __init init_bootmem_core(bootmem_data_t *bdata,
unsigned long mapstart, unsigned long start, unsigned long end)
{
unsigned long mapsize;
mminit_validate_memmodel_limits(&start, &end);
bdata->node_bootmem_map = phys_to_virt(PFN_PHYS(mapstart));
bdata->node_min_pfn = start;
bdata->node_low_pfn = end;
/*添加bdata变量到链表中*/
link_bootmem(bdata);
/*
* Initially all pages are reserved - setup_arch() has to
* register free RAM areas explicitly.
*/
/*计算本bdata的mapsize，也就是内存页面大小的1/8*/
mapsize = bootmap_bytes(end - start);
/*将所有map置1*/
memset(bdata->node_bootmem_map, 0xff, mapsize);
bdebug("nid=%td start=%lx map=%lx end=%lx mapsize=%lx/n",
bdata - bootmem_node_data, start, mapstart, end, mapsize);
return mapsize;
}
/*
* link bdata in order
*/
/*添加到链表，由添加的代码可知
链表中的数据开始位置为递增的*/
static void __init link_bootmem(bootmem_data_t *bdata)
{
struct list_head *iter;
/*添加到全局链表bdata_list中*/
list_for_each(iter, &bdata_list) {
bootmem_data_t *ent;
ent = list_entry(iter, bootmem_data_t, list);
if (bdata->node_min_pfn < ent->node_min_pfn)
break;
}
list_add_tail(&bdata->list, iter);
}
/**
* free_bootmem_with_active_regions - Call free_bootmem_node for each active range
* @nid: The node to free memory on. If MAX_NUMNODES, all nodes are freed.
* @max_low_pfn: The highest PFN that will be passed to free_bootmem_node
*
* If an architecture guarantees that all ranges registered with
* add_active_ranges() contain no holes and may be freed, this
* this function may be used instead of calling free_bootmem() manually.
*/
/*用active_region来初始化bootmem分配器，基于低端内存区*/
void __init free_bootmem_with_active_regions(int nid,
unsigned long max_low_pfn)
{
int i;
/*对每个节点上得活动内存区*/
for_each_active_range_index_in_nid(i, nid) {
unsigned long size_pages = 0;
unsigned long end_pfn = early_node_map[i].end_pfn;
if (early_node_map[i].start_pfn >= max_low_pfn)
continue;
if (end_pfn > max_low_pfn)
end_pfn = max_low_pfn;
/*计算活动区的页面数*/
size_pages = end_pfn - early_node_map[i].start_pfn;
/*释放这部分内存，起始就是对应位图值0*/
free_bootmem_node(NODE_DATA(early_node_map[i].nid),
PFN_PHYS(early_node_map[i].start_pfn),
size_pages << PAGE_SHIFT);
}
}
/**
* free_bootmem_node - mark a page range as usable
* @pgdat: node the range resides on
* @physaddr: starting address of the range
* @size: size of the range in bytes
*
* Partial pages will be considered reserved and left as they are.
*
* The range must reside completely on the specified node.
*/
void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
unsigned long size)
{
unsigned long start, end;
/*相关宏进行控制,调试用*/
kmemleak_free_part(__va(physaddr), size);
start = PFN_UP(physaddr);/*取上界*/
end = PFN_DOWN(physaddr + size);/*取下界*/
/*调用此函数对相关bit位清0，表示没有分配，这里保留位为0*/
mark_bootmem_node(pgdat->bdata, start, end, 0, 0);
}
static int __init mark_bootmem_node(bootmem_data_t *bdata,
unsigned long start, unsigned long end,
int reserve, int flags)
{
unsigned long sidx, eidx;
bdebug("nid=%td start=%lx end=%lx reserve=%d flags=%x/n",
bdata - bootmem_node_data, start, end, reserve, flags);
BUG_ON(start < bdata->node_min_pfn);
BUG_ON(end > bdata->node_low_pfn);
/*此两个变量为到节点最小内存页面的偏移量*/
sidx = start - bdata->node_min_pfn;
eidx = end - bdata->node_min_pfn;
if (reserve)/*如果设置了保留位*/
return __reserve(bdata, sidx, eidx, flags);
else/*相关的map位清0*/
__free(bdata, sidx, eidx);
return 0;
}
/*bootmem分配器的保留操作*/
static int __init __reserve(bootmem_data_t *bdata, unsigned long sidx,
unsigned long eidx, int flags)
{
unsigned long idx;
int exclusive = flags & BOOTMEM_EXCLUSIVE;
bdebug("nid=%td start=%lx end=%lx flags=%x/n",
bdata - bootmem_node_data,
sidx + bdata->node_min_pfn,
eidx + bdata->node_min_pfn,
flags);
/*对连续的几个页面设置为保留*/
for (idx = sidx; idx < eidx; idx++)
if (test_and_set_bit(idx, bdata->node_bootmem_map)) {
if (exclusive) {
__free(bdata, sidx, idx);
return -EBUSY;
}
bdebug("silent double reserve of PFN %lx/n",
idx + bdata->node_min_pfn);
}
return 0;
}
/*bootmem分配器中释放内存*/
static void __init __free(bootmem_data_t *bdata,
unsigned long sidx, unsigned long eidx)
{
unsigned long idx;
bdebug("nid=%td start=%lx end=%lx/n", bdata - bootmem_node_data,
sidx + bdata->node_min_pfn,
eidx + bdata->node_min_pfn);
if (bdata->hint_idx > sidx)
bdata->hint_idx = sidx;/*更新变量hint_idx，用于分配*/
for (idx = sidx; idx < eidx; idx++)/*对应位清0*/
if (!test_and_clear_bit(idx, bdata->node_bootmem_map))
BUG();
}
void __init early_res_to_bootmem(u64 start, u64 end)
{
int i, count;
u64 final_start, final_end;
count = 0;
for (i = 0; i < MAX_EARLY_RES && early_res[i].end; i++)
count++;/*计算保留块的个数*/
printk(KERN_INFO "(%d early reservations) ==> bootmem [%010llx - %010llx]/n",
count, start, end);
for (i = 0; i < count; i++) {
struct early_res *r = &early_res[i];
printk(KERN_INFO " #%d [%010llx - %010llx] %16s", i,
r->start, r->end, r->name);
final_start = max(start, r->start);
final_end = min(end, r->end);
if (final_start >= final_end) {
printk(KERN_CONT "/n");
continue;
}
printk(KERN_CONT " ==> [%010llx - %010llx]/n",
final_start, final_end);
/*将指定区间置为保留*/
reserve_bootmem_generic(final_start, final_end - final_start,
BOOTMEM_DEFAULT);
}
}
上面的保留指定区间reserve_bootmem_generic()函数最终调用如下函数
/**
* reserve_bootmem - mark a page range as usable
* @addr: starting address of the range
* @size: size of the range in bytes
* @flags: reservation flags (see linux/bootmem.h)
*
* Partial pages will be reserved.
*
* The range must be contiguous but may span node boundaries.
*/
int __init reserve_bootmem(unsigned long addr, unsigned long size,
int flags)
{
unsigned long start, end;
start = PFN_DOWN(addr);/*下界*/
end = PFN_UP(addr + size);/*上界*/
return mark_bootmem(start, end, 1, flags);
}
/*保留指定内存区间*/
static int __init mark_bootmem(unsigned long start, unsigned long end,
int reserve, int flags)
{
unsigned long pos;
bootmem_data_t *bdata;
pos = start;
/*通过bdata_list链表找到在指定区间的bdata*/
list_for_each_entry(bdata, &bdata_list, list) {
int err;
unsigned long max;
if (pos < bdata->node_min_pfn ||
pos >= bdata->node_low_pfn) {
BUG_ON(pos != start);
continue;
}
max = min(bdata->node_low_pfn, end);
/*设置为保留*/
err = mark_bootmem_node(bdata, pos, max, reserve, flags);
if (reserve && err) {/*如果出错，递归调用*/
mark_bootmem(start, pos, 0, 0);
return err;
}
if (max == end)
return 0;
pos = bdata->node_low_pfn;
}
BUG();
}
三、内存的分配和释放
介绍了上面的初始化流程，对于分配和释放就简单了，分配就是将分配器映射位图中对应的位置1，释放过程相反。
/*分配size大小的空间*/
static void * __init alloc_bootmem_core(struct bootmem_data *bdata,
unsigned long size, unsigned long align,
unsigned long goal, unsigned long limit)
{
unsigned long fallback = 0;
unsigned long min, max, start, sidx, midx, step;
bdebug("nid=%td size=%lx [%lu pages] align=%lx goal=%lx limit=%lx/n",
bdata - bootmem_node_data, size, PAGE_ALIGN(size) >> PAGE_SHIFT,
align, goal, limit);
BUG_ON(!size);
BUG_ON(align & (align - 1));
BUG_ON(limit && goal + size > limit);
/*如果没有映射位图返回空，分配失败*/
if (!bdata->node_bootmem_map)
return NULL;
min = bdata->node_min_pfn;
max = bdata->node_low_pfn;
goal >>= PAGE_SHIFT;
limit >>= PAGE_SHIFT;
if (limit && max > limit)
max = limit;
if (max <= min)
return NULL;
/*step为需要对齐于页面数*/
step = max(align >> PAGE_SHIFT, 1UL);
/*计算起始页面*/
if (goal && min < goal && goal < max)
start = ALIGN(goal, step);
else
start = ALIGN(min, step);
/*计算分配页面区间*/
sidx = start - bdata->node_min_pfn;
midx = max - bdata->node_min_pfn;
/*前一次分配的页号比这次开始分配的页面号大
那么，如果第一次没有分配到，回退到这次的
开始重新试，因为第一次分配是从上一次分配
的位置开始的*/
if (bdata->hint_idx > sidx) {
* Handle the valid case of sidx being zero and still
* catch the fallback below.
*/
fallback = sidx + 1;
/*从上一次分配的位置开始，对齐与页面*/
sidx = align_idx(bdata, bdata->hint_idx, step);
}
while (1) {
int merge;
void *region;
unsigned long eidx, i, start_off, end_off;
find_block:
/*查找第一个为0的位*/
sidx = find_next_zero_bit(bdata->node_bootmem_map, midx, sidx);
sidx = align_idx(bdata, sidx, step);
eidx = sidx + PFN_UP(size);/*结束位置*/
if (sidx >= midx || eidx > midx)/*找到结束了*/
break;
for (i = sidx; i < eidx; i++)/*检查这段区域是否空闲*/
if (test_bit(i, bdata->node_bootmem_map)) {/*如果不是，将跳过这段继续查找*/
sidx = align_idx(bdata, i, step);
if (sidx == i)
sidx += step;
goto find_block;
}
if (bdata->last_end_off & (PAGE_SIZE - 1) &&/*如果为相邻的页面，也就是说上次分配的页面和这次分配的开始页面为相邻的*/
PFN_DOWN(bdata->last_end_off) + 1 == sidx)
start_off = align_off(bdata, bdata->last_end_off, align);
else
start_off = PFN_PHYS(sidx);
/*merge==1表示上次结束和这次开始不在同一个页面上*/
merge = PFN_DOWN(start_off) < sidx;
end_off = start_off + size;
/*更新数据*/
bdata->last_end_off = end_off;
bdata->hint_idx = PFN_UP(end_off);
/*
* Reserve the area now:
*/
/*设定新加入的页面为保留,就是将对应的映射位置1*/
if (__reserve(bdata, PFN_DOWN(start_off) + merge,
PFN_UP(end_off), BOOTMEM_EXCLUSIVE))
BUG();
/*对应开始地址的虚拟地址返回*/
region = phys_to_virt(PFN_PHYS(bdata->node_min_pfn) +
start_off);
memset(region, 0, size);/*分配的大小*/
/*
* The min_count is set to 0 so that bootmem allocated blocks
* are never reported as leaks.
*/
/*调试用*/
kmemleak_alloc(region, size, 0, 0);
return region;
}
if (fallback) {/*回退，重新查看*/
sidx = align_idx(bdata, fallback - 1, step);
fallback = 0;
goto find_block;
}
return NULL;
}
摘自 bullbat的专栏
