星空探索

进程创建 fork函数运行。copy_mm()--> dup_mmstatic struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p, struct user_namespace *user_ns){ mm->mmap = NULL; mm->mm_rb = RB_ROOT;

主要还是两个函数free_bootmem_node和free_bootmem这里主要是标记相应比特位为0，空闲。/** * free_bootmem_node - mark a page range as usable * @pgdat: node the range resides on * @physaddr: starting address of the ran

主要提供reserve_bootmem_node和reserve_bootmem/** * reserve_bootmem_node - mark a page range as reserved * @pgdat: node the range resides on * @physaddr: starting address of the range * @size:

alloc_bootmem_bdata和alloc_bootmem_corestatic void * __init alloc_bootmem_bdata(struct bootmem_data *bdata, unsigned long size, unsigned long align, unsigned long goal, unsigned

主要是提供__free_pages_boot_core()函数，对外提供__free_pages_bootmem()和deferred_free_range()__free_pages_bootmem()1. free_bootmem_late2. free_all_bootmem_core 《3. __memblock_free_late/* * f

/** * free_all_bootmem - release free pages to the buddy allocator * * Returns the number of pages actually released. */unsigned long __init free_all_bootmem(void){ unsigned long total_p

__free_pages_boot_corestatic void __init __free_pages_boot_core(struct page *page, unsigned int order){ unsigned int nr_pages = 1 << order; struct page *p = page; unsigned int loop; pr

主要是free_one_page()对外提供__free_pages_ok和free_hot_cold_page/* * Freeing function for a buddy system allocator. * * The concept of a buddy system is to maintain direct-mapped table * (co

memblock， Logical memory blocks.管理struct memblock { bool bottom_up;  /* is bottom up direction? */ phys_addr_t current_limit; struct memblock_type memory; struct memblock_type reserved;#if

这里主要提供两个函数memblock_add()和memblock_add_node()，如果标记内存则调用memblock_reserve()。这些函数核心调用memblock_add_range()/** * memblock_add_range - add new memblock region * @type: memblock type to add ne

static int __init_memblock memblock_remove_range(struct memblock_type *type,  phys_addr_t base, phys_addr_t size){ int start_rgn, end_rgn; int i, ret; ret = memblock_isolate_range(t

主要有mark_bootmem_node和mark_bootmem两个函数static int __init mark_bootmem_node(bootmem_data_t *bdata, unsigned long start, unsigned long end, int reserve, int flags){ unsigned long sidx,

内存管理之bootmem管理之初始化内存管理之bootmem管理之内存分配bootmem管理中，内存分配的核心是alloc_bootmem_bdata()函数void * __init alloc_bootmem_bdata(struct bootmem_data *bdata, unsigned long size, unsigned long align,

前面我们讨论了页描述符对象分配，以及初始化。接下来我们再来了解内核页表建立初始化处理。针对ARM64进行。start_kernel --> setup_arch --> paging_init()void __init paging_init(void){ phys_addr_t pgd_phys = early_pgtable_alloc(); 分配页全局目录表地址

在前面我们讨论了内存节点node_mem_map成员的初始化。了解到此成员其实是以struct page为单位的数组，数组大小跟此节点内存大小相关，那么这些描述符是何时初始化呢，当前主要是在mem_init函数进行设置。start_kernel() --> mm_init --> mem_init()需要注意的是mem_init是跟体系结构相关的函数，比如针对ARM64的实现函数：

在linux内核中，所有的物理内存都用struct page结构来描述，这些对象以数组形式存放，而这个数组的地址就是mem_map。内核以节点node为单位，每个node下的物理内存统一管理，也就是说在表示内存node的描述类型struct pglist_data中，有node_mem_map这个成员，其针对平坦型内存进行描述(CONFIG_FLAT_NODE_MEM_MAP)，与此相反的是S

static inline void set_page_order(struct page *page, unsigned int order){ set_page_private(page, order); __SetPageBuddy(page);}static inline void rmv_page_order(struct page *page){ __Clear

zone 空闲页管理：/* Free memory management - zoned buddy allocator.  */#ifndef CONFIG_FORCE_MAX_ZONEORDER#define MAX_ORDER 11#else#define MAX_ORDER CONFIG_FORCE_MAX_ZONEORDER#endif#define MAX_

Memory compaction The longstanding memory fragmentation problem has been covered many times in these pages. In short: as the system runs, pages tend to be scattered between users, making

/** * cma_alloc() - allocate pages from contiguous area * @cma:   Contiguous memory region for which the allocation is performed. * @count: Requested number of pages. * @align: Requested align

linux kernel中的CMA即，连续内存区管理，其提供配置为CONFIG_CMA和CONFIG_CMA_DEBUG毋庸置疑，其管理的是一块块连续内存块。这个在物理地址上是连续的。这点跟我们使用的伙伴算法以及虚拟地址有点不一样。尽管伙伴算法中使用kmalloc申请连续物理内存也可以，但是在长时间测试环境下，连续物理内存可能申请不到。因此，内核设计者设计了CMA，即连续物理内存管理。

每个物理页关联一个page描述符，这样通过软件形式维护物理内存：struct page {第1个 unsigned long flags;第2个 union {  struct address_space *mapping;  void *s_mem;  atomic_t compound_mapcount; };第3个 union {  pgoff_t ind

对于一段连续内存块，我们这里假设为1024个页[0, 1023]。 |------------------------------------------------------| |0-------234-------------345------------678--------1023| 我们首先看看查找buddy算法，查找buddy需要一个pfn和其order。查找函数

memblock_removeint __init_memblock memblock_remove(phys_addr_t base, phys_addr_t size){ return memblock_remove_range(&memblock.memory, base, size);}

memblock_freeint memblock_free(phys_addr_t base, phys_addr_t size){ memblock_dbg("   memblock_free: [%#016llx-%#016llx] %pF\n",     (unsigned long long)base,     (unsigned long long)ba

memblock_reserveint __init_memblock memblock_reserve(phys_addr_t base, phys_addr_t size){ memblock_dbg("memblock_reserve: [%#016llx-%#016llx] flags %#02lx %pF\n",     (unsigned long long)bas

int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned intorder){       int node_id;       int ret;        if (zone_pagecache_reclaimable(zone)min_unmapped_pages &&          zone_pag

static inline struct page *__alloc_pages_slowpath(gfp_t gfp_mask, unsigned intorder,                                          structalloc_context *ac){       boolcan_direct_reclaim = gfp_mask

isolate_lru_pageint isolate_lru_page(struct page *page){int ret = -EBUSY;VM_BUG_ON_PAGE(!page_count(page), page);WARN_RATELIMIT(PageTail(page), "trying to isolate tail page");i

Discontiguous memory support，即非连续内存管理主要数据结构 struct bootmem_data#ifndef CONFIG_NO_BOOTMEM/* * node_bootmem_map is a map pointer - the bits represent all physical  * memory pages (including

sched_class

linux kernel000000000000

在前面介绍的mm_init中，其调用函数mem_init()，在这个函数中会把bootmem中空闲内存释放到伙伴系统。我们下面看bootmem中一个释放内存的函数free_all_bootmem unsigned long __init free_all_bootmem(void){ unsigned long total_pages = 0; bootmem_data_t

memblock

sched_class

static unsigned long isolate_lru_pages(unsigned longnr_to_scan,              struct lruvec *lruvec, struct list_head*dst,              unsignedlong *nr_scanned, struct scan_control *sc,

static void shrink_active_list(unsignedlong nr_to_scan,  struct lruvec *lruvec,  struct scan_control *sc,  enum lru_list lru){       unsignedlong nr_taken;       unsignedlong nr_scanned;      

phys_addr_t __init_memblock memblock_start_of_DRAM(void){ return memblock.memory.regions[0].base;}phys_addr_t __init_memblock memblock_end_of_DRAM(void){ int idx = memblock.memory.cnt

/* * pgdat_balanced() is used when checking if anode is balanced. * * For order-0, all zones must be balanced! * * For high-order allocations only zones thatmeet watermarks and are in a *

/* * kswapd shrinks the zone by the number ofpages required to reach * the high watermark. * * Returns true if kswapd scanned at least therequested number of pages to * reclaim or if the lac