中关村村草 发表于 2012-01-05 16:27

linux启动内存分配器2。。。。。。。。。。。。

linux启动内存分配器2。。。。。。。。。。。。







view plaincopy to clipboardprint?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;
      end_pfn = node_end_pfn;
      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;
}
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;
                end_pfn = node_end_pfn;
                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;
}view plaincopy to clipboardprint?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;
}
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()


view plaincopy to clipboardprint?/*
* 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;
}
/*
* 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;
}
view plaincopy to clipboardprint?/*
* 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);
}
/*
* 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);
}view plaincopy to clipboardprint?/**
* 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.end_pfn;

      if (early_node_map.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.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.nid),
                PFN_PHYS(early_node_map.start_pfn),
                size_pages << PAGE_SHIFT);
    }
}
/**
* 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.end_pfn;

                if (early_node_map.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.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.nid),
                                PFN_PHYS(early_node_map.start_pfn),
                                size_pages << PAGE_SHIFT);
        }
}view plaincopy to clipboardprint?/**
* 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);
}
/**
* 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);
}view plaincopy to clipboardprint?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;
}
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;
}view plaincopy to clipboardprint?/*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;
}view plaincopy to clipboardprint?/*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();
}
/*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();
}view plaincopy to clipboardprint?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.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;
      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);
    }
}
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.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;
                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


]

第一夫人 发表于 2012-01-05 16:27

谢谢分享
页: [1]
查看完整版本: linux启动内存分配器2。。。。。。。。。。。。