Skip to content

MemoryManagement.md

Latest commit

 

History

History
300 lines (207 loc) · 7.97 KB

MemoryManagement.md

File metadata and controls

300 lines (207 loc) · 7.97 KB

Linux Memory Management

  • Memory Addressing
    • Segmentation
  • Physical and Virtual Memory
  • Page Table Management
  • Data Structures
  • Allocators
    • vmalloc
    • Page Allocation
    • sla/ub
  • Process address space

Memory Addressing

  • Logical Address
    • In machine language instruction, used to determine a operator or instruction address
    • Include segment and offset
  • Linear Address (aka. Virtual Address) => "Express Ability"
    • In 32-bit machine => 4GB i.e. 4294967295. From 0x00000000 to 0xFFFFFFFF
  • Physical Address => "Actually You Have"
Logical Address ===Segmentation Unit==> Linear Address ===Paging Unit==> Physical address
                             \                               /
                                Memory Management Unit (MMU)

memory addressing

IA-32 Memory Management Facilities:

  • Segmentation: from Logical Address to Linear Address
  • Paging (demand paging): from Linear Address to Phsical Address

In Linux, it used these two mechanism at the same time. (because you have to support hardware... but tend to weaker the segmentation)

Segmentation

Segment Selectors

segment selectors

  • Index
    • which is an index into the GDT and thus points to an entry of the GDT called a segment descriptor
  • Table Indicator
    • which indicates whether the selector points to a GDT or an LDT
  • Requestor Privilege Level (RPL)
    • which is the privilege level of the CPU when the selector is loaded

Privilege Level in Linux only 0 and 3. One for operating system (kernel) level, another for user space level.

Privilege Level

  • RPL (Request Privilege Level)

  • CPL (Current Privilege Level)

  • DPL (Descriptor Privilege Level)

  • Access data, gate descripter: max(CPL, RPL) <= DPL

  • ...

Segment Register

  • Visible part: Segment Selector
  • Hidden part: Base Address, Limit, Access Information

Linux uses segmentation only when required by the 80x86 architecture!

Segment Model

Basic Flat Model - Linux

Continuous, non-segment address space

basic flat model

Protected Flat Model

Multi-Segment Model

Linux Physical Address

Node

Linux .....

NUMA (Non Uniform Memory Access)

In x86 used UMA

Zone

Each node split into many zones

Because of hardware limitations, the kernel cannot treat all pages as identical

  • ZONE_DMA
  • ZONE_NORMAL
  • ZONE_HIGHMEM (in 32-bit system, there may have some physical address that cannot be expressed by virtual address thus need this zone.)

In X86

  • ZONE_DMA => 16MB of memory
  • ZONE_NORMAL => 16MB ~ 896MB
  • ZONE_HIGHMEM => 896MB ~ End (in 64-bit system it is empty)

Linux Linear Address Space

...TBD

1GB Kernel Space + 3GB User Space

  • All the user share the 1GB Kernel Space
  • But for a single thread vision. It can see a 4GB memory space.

Kernel Address Space

Kernel Address Space

  • vmalloc address space
    • Noncontiguous physical memory allocation
  • kmap address space
    • Allocation of memory from ZONE_HITHMEM
  • Fixed mapping
    • Compile-time virtual memory allocation

64-bit machine

Used only 48-bit. => 128TB for user and another half for kernel

Translating Kernel Virtual Address

  • Memory in ZONE_DMA and ZONE_NORMAL is direct-mapped and all page frames are described by mem_map array

Kernel virtual address => physical address

Kernel virtual address => struct page

arch/x86/include/asm/io.h - line 131

/**
 *	virt_to_phys	-	map virtual addresses to physical
 *	@address: address to remap
 *
 *	The returned physical address is the physical (CPU) mapping for
 *	the memory address given. It is only valid to use this function on
 *	addresses directly mapped or allocated via kmalloc.
 *
 *	This function does not give bus mappings for DMA transfers. In
 *	almost all conceivable cases a device driver should not be using
 *	this function
 */

static inline phys_addr_t virt_to_phys(volatile void *address)
{
	return __pa(address);
}
#define virt_to_phys virt_to_phys

arch/x86/include/asm/page.h - line 42 and 69

#define __pa(x)		__phys_addr((unsigned long)(x))
#define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)

Page Table Management

Page and Page Frame

  • Page: a block of data, which may be stored in any page frame or on disk! (not necessary be in memory)
  • Page Frame: physical addresses in main memory

Paging in Linux

4 level paging model

  • PGD Page Global Directory (頁全局目錄)
  • PUD Page Upper Directory (頁上級目錄)
  • PMD Page Middle Directory (頁中間目錄)
  • PT Page Table (頁表)

Multi-level is good when data is sparse

TBD

High Memory Mappings

Physical Address Extension (PAE)

PDPT Page Directory Pointer Table

Data Structure

Node Stucture

include/linux/mmzone.h

mm/memory.c

include/linux/mm_types.h

include/linux/page-flags.h

Relationship Between Nodes, Zones and Pages

Macro

TBD

#define PAGE_SHIFT	12
#ifdef __ASSEMBLY__
#define PAGE_SIZE	(1 << PAGE_SHIFT)
#else
#define PAGE_SIZE	(1UL << PAGE_SHIFT)
#endif
#define PAGE_MASK	(~(PAGE_SIZE-1))
typedef struct {
	unsigned long pte;
} pte_t;
typedef struct {
	unsigned long pmd[16];
} pmd_t;
typedef struct {
	unsigned long pgd;
} pgd_t;
typedef struct {
	unsigned long pgprot;
} pgprot_t;
typedef struct page *pgtable_t;
#define pte_val(x)	((x).pte)
#define pmd_val(x)	((&x)->pmd[0])
#define pgd_val(x)	((x).pgd)
#define pgprot_val(x)	((x).pgprot)

#define __pte(x)	((pte_t) { (x) } )
#define __pmd(x)	((pmd_t) { (x) } )
#define __pgd(x)	((pgd_t) { (x) } )
#define __pgprot(x)	((pgprot_t) { (x) } )

Allocator

  • Boot Memory Allocator
    • alloc_bootmem()
    • free_bootmem()
  • Physical Page Allocator (buddy system)
    • alloc_pages()
    • __get_free_pages()
  • Slab Allocation
    • kmalloc()
    • kfree()
  • Virtual Memory Allocator
    • vmalloc()
    • vfree()

Noncontiguous ....

Linear Address of Noncontiguous Memory Areas

vmalloc.c

Physical Page Allocation

Not yet

Resources