一、系统调用mmap
虚拟内存区域使用起始地址和结束地址描述,链表按起始地址递增排序。两系统调用区别:mmap指定的偏移的单位是字节,而mmap2指定的偏移的单位是页。ARM64架构实现系统调用mmap。
二、系统调用munmap
系统调用munmap用来删除内存映射,它有两个参数:起始地址和长度即可。它的主要工作委托给内核源码文件处理“mm/mmap.c”中的函数do_munmap。
vm_munmap -->do_munmap -->vma=find_vma(mm,start) -->error=__split_vma(mm,vma,start,0) -->last=find_vma(mm,end) -->interror=__split_vma(mm,last,end,1) -->munlock_vma_pages_all -->detach_vmas_to_be_unmapped -->unmap_region -->arch_unmap -->remove_vma_list
vma = find_vma(mm,start);//根据起始地址找到要删除的第一个虚拟内存区域vma
error = __split_vma(mm,vma,start,0);//如果只删除虚拟内存区域vma的部分,那么分裂虚拟内存区域vma
last = find_vma(mm,end);//根据结束地址找到要删除的最后一个虚拟内存区域vma
int error = __split_vma(mm,last,end,1);//如果只删除虚拟内存区域last的一部分,那么分裂虚拟内存区域vma
munlock_vma_pages_all;//针对所有删除目标,如果虚拟内存区域被锁定在内存中(不允许换出到交换区),调用函数解除锁定
detach_vmas_to_be_unmapped;//调用此函数,把所有删除目标从进程虚拟内存区域链表和树中删除,单独组成一条临时链表
unmap_region;//调用此函数,针对所有删除目标,在进程的页表中删除映射,并且从处理器的页表缓存中删除映射
arch_unmap;//调用此函数执行处理器架构特定的处理操作
remove_vma_list;//调用此函数,删除所有目标
三、物理内存组织结构
1.体系结构
目前多处理器系统有两种体系结构:
非一致内存访问(Non-Unit Memory Access,NUMA):指内存被划分成多个内存节点的多处理器系统。访问一个内存节点花费的时间取决于处理器和内存节点的距离。
对称多处理器(Sysmmetric Muti-Processor,SMP):即一致内存访问(Uniform Memory Access,UMA),所有处理器访问内存花费的时间是相同的。
2.内存模型
内存模型是从处理器角度看到的物理内存分布,内核管理不同内存模型的方式存在差异。内存管理子系统支持3种内存模型:
平坦内存(Flat Memory):内存的物理地址空间是连续的,没有空洞。
不连续内存(Discontiguous Memory):内存的物理地址空间存在空洞,这种模型可以高效地处理空洞。
稀疏内存(Space Memory):内存物理地址空间存在空洞,如果要支持内存热插拔,只能选择稀疏内存模型。
3.三级结构
内存管理子系统使用节点(node)、区域(zone)、页(page)三级结构描述物理内存。
3.1 内存节点--->分为两种情况
NUMA体系的内存节点,根据处理器和内存距离划分;
在具有不连续内存的NUMA系统中,表示比区域的级别更高的内存区域,根据物理地址是否连续,每块物理地址连续的内存是一个内存节点。
内存节点使用结构体pglist_data描述内存布局
Linux内核源码如下:
typedefstructpglist_data{ structzonenode_zones[MAX_NR_ZONES];//内存区域数组 structzonelistnode_zonelists[MAX_ZONELISTS];//备用区域列表 intnr_zones;//该节点包含内存区域数量 #ifdefCONFIG_FLAT_NODE_MEM_MAP/*means!SPARSEMEM*///除了稀疏内存模型以外 structpage*node_mem_map;//页描述符数组 #ifdefCONFIG_PAGE_EXTENSION structpage_ext*node_page_ext;//页的扩展属性 #endif #endif #ifndefCONFIG_NO_BOOTMEM structbootmem_data*bdata; #endif #ifdefCONFIG_MEMORY_HOTPLUG /* *Mustbeheldanytimeyouexpectnode_start_pfn,node_present_pages *ornode_spanned_pagesstayconstant.Holdingthiswillalso *guaranteethatanypfn_valid()staysthatway. * *pgdat_resize_lock()andpgdat_resize_unlock()areprovidedto *manipulatenode_size_lockwithoutcheckingforCONFIG_MEMORY_HOTPLUG. * *Nestsabovezone->lockandzone->span_seqlock */ spinlock_tnode_size_lock; #endif unsignedlongnode_start_pfn;//该节点的起始物理页号 unsignedlongnode_present_pages;/*物理页总数*/ unsignedlongnode_spanned_pages;/*物理页范围总的长度,包括空间*/ intnode_id;//节点标识符 wait_queue_head_tkswapd_wait; wait_queue_head_tpfmemalloc_wait; structtask_struct*kswapd;/*Protectedby mem_hotplug_begin/end()*/ intkswapd_max_order; enumzone_typeclasszone_idx; #ifdefCONFIG_NUMA_BALANCING /*Lockserializingthemigrateratelimitingwindow*/ spinlock_tnumabalancing_migrate_lock; /*Ratelimitingtimeinterval*/ unsignedlongnumabalancing_migrate_next_window; /*Numberofpagesmigratedduringtheratelimitingtimeinterval*/ unsignedlongnumabalancing_migrate_nr_pages; #endif #ifdefCONFIG_DEFERRED_STRUCT_PAGE_INIT /* *Ifmemoryinitialisationonlargemachinesisdeferredthenthis *isthefirstPFNthatneedstobeinitialised. */ unsignedlongfirst_deferred_pfn; #endif/*CONFIG_DEFERRED_STRUCT_PAGE_INIT*/ }pg_data_t;
node_mem_map此成员指向页描述符数组,每个物理页对应一个页描述符。
Node是内存管理最顶层的结构,在NUMA架构下,CPU平均划分为多个Node,每个Node有自己的内存控制器及内存插槽。CPU访问自己Node上内存速度快,而访问其他CPU所关联Node的内存速度慢。UMA被当作只一个Node的NUMA系统。
3.2 内存区域(zone)
内存节点被划分为内存区域。Linux内核源码分析:include/linux/mmzone.h
enumzone_type{ #ifdefCONFIG_ZONE_DMA /* *ZONE_DMAisusedwhentherearedevicesthatarenotable *todoDMAtoallofaddressablememory(ZONE_NORMAL).Thenwe *carveouttheportionofmemorythatisneededforthesedevices. *Therangeisarchspecific. * *Someexamples * *ArchitectureLimit *--------------------------- *parisc,ia64,sparc<4G * s390 <2G * arm Various * alpha Unlimited or 0-16MB. * * i386, x86_64 and multiple other arches * <16M. */ ZONE_DMA, /*Direct Memory Access,直接内存访问。如果有些设备不能直接访问所有内存,需要使用DMA区域。ISA*/ #endif #ifdef CONFIG_ZONE_DMA32 /* * x86_64 needs two ZONE_DMAs because it supports devices that are * only able to do DMA to the lower 16M but also 32 bit devices that * can only do DMA areas below 4G. */ ZONE_DMA32, /* 64位系统,如果既要支持能直接访问16MB以下内存设备,又要支持能直接访问4GB以下内存的32设备,必须使用此DMA32区域*/ #endif /* * Normal addressable memory is in ZONE_NORMAL. DMA operations can be * performed on pages in ZONE_NORMAL if the DMA devices support * transfers to all addressable memory. */ /*普通内存区域: 直接映射到内核虚拟地址空间的内存区域,又称为普通区域,又称为直接映射区域,又称为线性映射区域*/ ZONE_NORMAL, #ifdef CONFIG_HIGHMEM /* * A memory area that is only addressable by the kernel through * mapping portions into its own address space. This is for example * used by i386 to allow the kernel to address the memory beyond * 900MB. The kernel will set up special mappings (page * table entries on i386) for each page that the kernel needs to * access. */ /*高端内存区域: 此区域是32位时代的产物,内核和用户地址空间按1:3划分, 内核地址空间只有1GB,不能把1GB以上内存直接映射到该地址*/ ZONE_HIGHMEM, #endif /*可移动区域: 它是一个伪内存区域,用来防止内存碎片*/ ZONE_MOVABLE, #ifdef CONFIG_ZONE_DEVICE /*设备区域: 为支持持久内存热插拔增加的内存区域,每一个内存区域用一个zone结构体来描述*/ ZONE_DEVICE, #endif __MAX_NR_ZONES };
每个内存区域使用一个zone结构体描述,如下为主要成员:
structzone{ /*Read-mostlyfields*/ /*zonewatermarks,accesswith*_wmark_pages(zone)macros*/ unsignedlongwatermark[NR_WMARK];//页分配器使用的水线 unsignedlongnr_reserved_highatomic; /* *Wedon'tknowifthememorythatwe'regoingtoallocatewillbe *freeableor/anditwillbereleasedeventually,sotoavoidtotally *wastingseveralGBoframwemustreservesomeofthelowerzone *memory(otherwisewerisktorunOOMonthelowerzonesdespite *therebeingtonsoffreeableramonthehigherzones).Thisarrayis *recalculatedatruntimeifthesysctl_lowmem_reserve_ratiosysctl *changes. */ longlowmem_reserve[MAX_NR_ZONES];//页分配器使用,当前区域保留多少页不能借给高的区域类型 #ifdefCONFIG_NUMA intnode; #endif /* *ThetargetratioofACTIVE_ANONtoINACTIVE_ANONpageson *thiszone'sLRU.Maintainedbythepageoutcode. */ unsignedintinactive_ratio; structpglist_data*zone_pgdat;//指向内存节点的pglist_data实例 structper_cpu_pageset__percpu*pageset;//每处理页集合 /* *Thisisaper-zonereserveofpagesthatshouldnotbe *considereddirtyablememory. */ unsignedlongdirty_balance_reserve; #ifndefCONFIG_SPARSEMEM /* *Flagsforapageblock_nr_pagesblock.Seepageblock-flags.h. *InSPARSEMEM,thismapisstoredinstructmem_section */ unsignedlong*pageblock_flags; #endif/*CONFIG_SPARSEMEM*/ #ifdefCONFIG_NUMA /* *zonereclaimbecomesactiveifmoreunmappedpagesexist. */ unsignedlongmin_unmapped_pages; unsignedlongmin_slab_pages; #endif/*CONFIG_NUMA*/ /*zone_start_pfn==zone_start_paddr>>PAGE_SHIFT*/ unsignedlongzone_start_pfn;//当前区域的起始物理页号 /* *spanned_pagesisthetotalpagesspannedbythezone,including *holes,whichiscalculatedas: *spanned_pages=zone_end_pfn-zone_start_pfn; * *present_pagesisphysicalpagesexistingwithinthezone,which *iscalculatedas: *present_pages=spanned_pages-absent_pages(pagesinholes); * *managed_pagesispresentpagesmanagedbythebuddysystem,which *iscalculatedas(reserved_pagesincludespagesallocatedbythe *bootmemallocator): *managed_pages=present_pages-reserved_pages; * *Sopresent_pagesmaybeusedbymemoryhotplugormemorypower *managementlogictofigureoutunmanagedpagesbychecking *(present_pages-managed_pages).Andmanaged_pagesshouldbeused *bypageallocatorandvmscannertocalculateallkindsofwatermarks *andthresholds. * *Lockingrules: * *zone_start_pfnandspanned_pagesareprotectedbyspan_seqlock. *Itisaseqlockbecauseithastobereadoutsideofzone->lock, *anditisdoneinthemainallocatorpath.But,itiswritten *quiteinfrequently. * *Thespan_seqlockisdeclaredalongwithzone->lockbecauseitis *frequentlyreadinproximitytozone->lock.It'sgoodto *givethemachanceofbeinginthesamecacheline. * *Writeaccesstopresent_pagesatruntimeshouldbeprotectedby *mem_hotplug_begin/end().Anyreaderwhocan'ttolerantdriftof *present_pagesshouldget_online_mems()togetastablevalue. * *Readaccesstomanaged_pagesshouldbesafebecauseit'sunsigned *long.Writeaccesstozone->managed_pagesandtotalram_pagesare *protectedbymanaged_page_count_lockatruntime.Idealyonly *adjust_managed_page_count()shouldbeusedinsteadofdirectly *touchingzone->managed_pagesandtotalram_pages. */ unsignedlongmanaged_pages;//伙伴分配器管理的物理页的数量 unsignedlongspanned_pages;//当前区域跨越的总页数,包括空洞 unsignedlongpresent_pages;//当前区域存在的物理页的数量,不包括空洞 constchar*name;//区域名称 #ifdefCONFIG_MEMORY_ISOLATION /* *Numberofisolatedpageblock.Itisusedtosolveincorrect *freepagecountingproblemduetoracyretrievingmigratetype *ofpageblock.Protectedbyzone->lock. */ unsignedlongnr_isolate_pageblock; #endif #ifdefCONFIG_MEMORY_HOTPLUG /*seespanned/present_pagesformoredescription*/ seqlock_tspan_seqlock; #endif /* *wait_table--thearrayholdingthehashtable *wait_table_hash_nr_entries--thesizeofthehashtablearray *wait_table_bits--wait_table_size==(1<< wait_table_bits) * * The purpose of all these is to keep track of the people * waiting for a page to become available and make them * runnable again when possible. The trouble is that this * consumes a lot of space, especially when so few things * wait on pages at a given time. So instead of using * per-page waitqueues, we use a waitqueue hash table. * * The bucket discipline is to sleep on the same queue when * colliding and wake all in that wait queue when removing. * When something wakes, it must check to be sure its page is * truly available, a la thundering herd. The cost of a * collision is great, but given the expected load of the * table, they should be so rare as to be outweighed by the * benefits from the saved space. * * __wait_on_page_locked() and unlock_page() in mm/filemap.c, are the * primary users of these fields, and in mm/page_alloc.c * free_area_init_core() performs the initialization of them. */ wait_queue_head_t *wait_table; unsigned long wait_table_hash_nr_entries; unsigned long wait_table_bits; ZONE_PADDING(_pad1_) /* free areas of different sizes */ struct free_area free_area[MAX_ORDER]; // 不同长度的空间区域 /* zone flags, see below */ unsigned long flags; /* Write-intensive fields used from the page allocator */ spinlock_t lock; ZONE_PADDING(_pad2_) /* Write-intensive fields used by page reclaim */ /* Fields commonly accessed by the page reclaim scanner */ spinlock_t lru_lock; struct lruvec lruvec; /* Evictions & activations on the inactive file list */ atomic_long_t inactive_age; /* * When free pages are below this point, additional steps are taken * when reading the number of free pages to avoid per-cpu counter * drift allowing watermarks to be breached */ unsigned long percpu_drift_mark; #if defined CONFIG_COMPACTION || defined CONFIG_CMA /* pfn where compaction free scanner should start */ unsigned long compact_cached_free_pfn; /* pfn where async and sync compaction migration scanner should start */ unsigned long compact_cached_migrate_pfn[2]; #endif #ifdef CONFIG_COMPACTION /* * On compaction failure, 1<
3.3 物理页
页是内存管理当中最小单位,页面中的内存其物理地址是连续的,每个物理页由struct page描述。为了节省内存,struct page是个联合体。
页,又称为页帧,在内核当中,内存管理单元MMU(负责虚拟地址和物理地址转换的硬件)是把物理页page作为内存管理的基本单位。体系结构不同,支持的页大小也不同。(32位体系结构支持4KB的页、64位体系结构支持8KB的页、MIPS64架构体系支持16KB的页)
审核编辑:汤梓红
-
ARM
+关注
关注
134文章
9045浏览量
366793 -
内核
+关注
关注
3文章
1362浏览量
40226 -
Linux
+关注
关注
87文章
11225浏览量
208909 -
物理内存
+关注
关注
0文章
11浏览量
8449 -
虚拟内存
+关注
关注
0文章
70浏览量
8052
原文标题:Linux内核 | 物理内存组织结构
文章出处:【微信号:嵌入式开发AIoT,微信公众号:嵌入式开发AIoT】欢迎添加关注!文章转载请注明出处。
发布评论请先 登录
相关推荐
评论