1. 1 Virtual Memory Chapter 8

2. 2 Virtual memory concept Simple paging/ segmentation  pages or segments are loaded into frames that may not necessarily be contiguous.  address references in the loaded pages or segments is dynamically calculated. Virtual memory  Similar to the above two things except that we do not need to load the whole program into memory.  Only a small portion of the whole program is first loaded into memory (the resident portion)  The rest loaded when needed.  Swapped out to disk when space is needed for other processes

3. 3 Hardware and Control Structures Memory references are dynamically translated into physical addresses at run time  A process may be swapped in and out of main memory such that it occupies different regions A process may be broken up into pieces that do not need to located contiguously in main memory All pieces of a process do not need to be loaded in main memory during execution

4. 4 Execution of a Program Operating system brings into main memory a few pieces of the program Resident set - portion of process that is in main memory An interrupt is generated when an address is needed that is not in main memory Operating system places the process in a blocking state Piece of process that contains the logical address is brought into main memory  Operating system issues a disk I/O Read request  Another process is dispatched to run while the disk I/O takes place  An interrupt is issued when disk I/O complete which causes the operating system to place the affected process in the Ready state

5. 5 Advantages of breaking up a Process More processes may be maintained in main memory  Only load in some of the pieces of each process  With so many processes in main memory, it is very likely a process will be in the Ready state at any particular time A process may be larger than all of main memory

6. 6 Types of Memory Real memory  Main memory where programs are actually loaded. Virtual memory  Memory view that is unlimited - that includes the space on disk.  A programmer’s view of memory - Allows for effective multiprogramming and relieves the user of tight constraints of main memory

7. 7 Concept of Thrashing Sometimes, when portion of a program is loaded into memory to run and the loaded portion immediately need another module which is not in memory. So, it have to be loaded from disk. Sometimes, a piece of a process is swapped out just before that piece is needed, So, it have to be reloaded from disk. There will be a time when there are many processes loaded in memory that needs to do the swap in and the processor spends more time swapping in and out as compared to doing real processing. This is called thrashing.

8. 8 To avoid Thrashing - Principle of Locality To avoid thrashing, OS will guess which part of a process is not needed in the near future and swap it out for new process to be loaded in. Only process that will be used in a short time will remain in memory. The selection is based on principle of locality (POL).

9. 9 Principle of Locality Principle of locality states that : program and data references within a process tend to cluster. Only a few pieces of a process will be needed over a short period of time ie: within a short period of time the same set of instructions and data will be repeatedly used. Possible to make intelligent guesses about which pieces will be needed in the future This suggests that virtual memory may work efficiently

10. 10 Support Needed for Virtual Memory Hardware support for paging and segmentation Operating system must be able to management the movement of pages and/or segments between secondary memory (disk) and main memory.

11. 11 Implementation of Virtual memory Virtual memory can be implemented for system using  paging  segmentation  combination of paging and segmentation.

12. 12 Virtual memory using Paging Each process has its own page table Each page table entry contains :  the frame number of the corresponding page in main memory  Present bit - to indicate whether the page is in main memory or not.  Modify bit - to indicate if the page has been altered since it was last loaded into main memory. If no change has been made, the page does not have to be written to the disk when it needs to be swapped out Address refered in the program (the logical address) will have two parts :  Page number  Offset

13. 13

14. 14 Two-Level Scheme for 32-bit Address

15. 15 2-level Page Table The entire page table may take up too much main memory Page table can itself reside in virtual memory for a system with large number of frames per processes and many processes can be running. In a two level page table scheme – at first level, a logical address points to a page table directory of page table. From page table directory, there will be another pointer to the actual page table. When a process is running, part of its page table is in main memory

16. 16 2-level Page Table scheme

17. 17 Inverted Page Table Page number portion of a virtual address is mapped into a hash value Hash value points to inverted page table Fixed proportion of real memory is required for the tables regardless of the number of processes Contents of page table: Page number Process identifier Control bits Chain pointer Used on PowerPC, UltraSPARC, and IA-64 architecture

18. 18

19. 19 Translation Lookaside Buffer Each virtual memory reference can cause two physical memory accesses  One to fetch the page table  One to fetch the data This will increase memory access time. To overcome this, we can cache (store temporarily) a group of page table entries that have been most recently used in a buffer called Translation Lookaside Buffer (TLB). TLB normally resides in main memory.

20. 20 Translation Lookaside Buffer Given a virtual address, processor examines the TLB If page table entry is present (TLB hit), the frame number is retrieved and the real address is formed If page table entry is not found in the TLB (TLB miss), the page number is used to index the process page table

21. 21 Translation Lookaside Buffer First checks if page is already in main memory  If not in main memory a page fault is issued The TLB is updated to include the new page entry

22. 22

23. 23

24. 24

25. 25

26. 26 Issue in Paging - Page Size Smaller page size means :  less amount of internal fragmentation  But more pages required per process More pages per process means  larger page tables  this means large portion of page tables in virtual memory Secondary memory is designed to efficiently transfer large blocks of data so a large page size is better

27. 27 Page Size Small page size, large number of pages will be found in main memory As time goes on during execution, the pages in memory will all contain portions of the process near recent references. Page faults low. Increased page size causes pages to contain locations further from any recent reference. Page faults rise.

28. 28

29. 29 Example Page Sizes

30. 30 Segmentation May be unequal, dynamic size Simplifies handling of growing data structures Allows programs to be altered and recompiled independently Lends itself to sharing data among processes Lends itself to protection

31. 31 Segment Tables Corresponding segment in main memory Each entry contains the length of the segment A bit is needed to determine if segment is already in main memory Another bit is needed to determine if the segment has been modified since it was loaded in main memory

32. 32 Segment Table Entries

33. 33

34. 34 Combined Paging and Segmentation Paging is transparent to the programmer Segmentation is visible to the programmer Each segment is broken into fixed-size pages

35. 35 Combined Segmentation and Paging

36. 36

37. 37

38. 38 OS Support for Virtual Memory Virtual memory using Paging and segmentation require hardware support. OS software support the various algorithm involved in managing the policies for :  page fetching  page placement  page replacement after swap out  Management of resident set of pages  Cleaning of pages  Load control.

39. 39 All the above policies rely on :  Main memory size  Relative speed of main and secondary memory  Size of processes  Number of processes  Execution behaviour of individual program The nature of the application. The programming language it is written in. The compiler used. The style of the programmerwho wrote it. The behaviour of the user (in the case of interactive program).

40. 40 The OS designer must therefore choose a set of policies depending on the target users of the OS, based on the current knowledge. Most modern OS allow system administrators to configure the system or to tune the OS for the maximum performance.

41. 41 Fetch Policy  Determines when a page should be brought into memory  Demand paging only brings pages into main memory when a reference is made to a location on the page Many page faults when process first started  Prepaging brings in more pages than needed More efficient to bring in pages that reside contiguously on the disk

42. 42 Placement Policy Determines where in real memory a process piece is to reside Important in a segmentation system Paging or combined paging with segmentation hardware performs address translation

43. 43 Replacement Policy  Which page is replaced?  Page removed should be the page least likely to be referenced in the near future  Most policies predict the future behavior on the basis of past behavior

44. 44 Replacement Policy Frame Locking  If frame is locked, it may not be replaced  Kernel of the operating system  Control structures  I/O buffers  Associate a lock bit with each frame

45. 45 Basic Replacement Algorithms Optimal policy Least Recently Used (LRU) First-in, first-out (FIFO) Clock Policy

46. 46 Basic Replacement Algorithms Optimal policy  Selects for replacement that page for which the time to the next reference is the longest  Impossible to have perfect knowledge of future events

47. 47 Basic Replacement Algorithms Least Recently Used (LRU)  Replaces the page that has not been referenced for the longest time  By the principle of locality, this should be the page least likely to be referenced in the near future  Each page could be tagged with the time of last reference. This would require a great deal of overhead.

48. 48 Basic Replacement Algorithms First-in, first-out (FIFO)  Treats page frames allocated to a process as a circular buffer  Pages are removed in round-robin style  Simplest replacement policy to implement  Page that has been in memory the longest is replaced  These pages may be needed again very soon

49. 49 Basic Replacement Algorithms Clock Policy  Additional bit called a use bit  When a page is first loaded in memory, the use bit is set to 1  When the page is referenced, the use bit is set to 1  When it is time to replace a page, the first frame encountered with the use bit set to 0 is replaced.  During the search for replacement, each use bit set to 1 is changed to 0

50. 50

51. 51

52. 52 Comparison of Replacement Algorithms

53. 53

54. 54 To improve paging performance A strategy that can improve paging performance is by using a simple replacement policy but on top of that employ page buffering.

55. 55 Page Buffering In page buffering, a replaced page is put into either one of the two lists :  Free page list - page that have not been modified.  Modified page list - page that have been modified. This page have to be written back to hard disk. When a page is to be read in,  the page frame ahead in the free page list is overwritten.  If the page is to be replaced is an unmodified page, the page frame is moved to the tail of the free page list.  If the page is to be replaced is a modified page, the page frame is moved to the tail of the modified page list.  The important fact here is that the replaced pages still remains in memory.  The modified pages can later be written back to disk.

56. 56 Resident set management Resident set management deals with the followings issues :  How many page frames are to be allocated to each active process  Whether to limit the pages to replace to just the pages that cause the page fault or the whole pages that belong to the process.

57. 57 Factors in Resident set management The following factors comes into play :  Less amount of memory allocated to process means more process can be loaded, thus more ready process available.  Rate of page fault is high if less memory frames allocated to process.  Beyond certain size, even though more memory is allocated to process, the will be no noticable impact.

58. 58 Policies and Scope in Resident set management Two different policies can be employed in resident set management :  Fixed-allocation - fixing number of pages to allocate for a process at load time.  Variable-allocation - let the number of pages allocated to be varied over the lifetime of the process. OS will have to assess the behaviour of each process. This require software overhead. The scope of replacement strategy can be  Global - choose pages to replace among all unlocked pages in main memory from any processes at all.  Local - choose pages to replace only among the resident pages of the process that generate page fault.

59. 59 Relationship between policies and scope The relationship between resident set size allocation and the scope can be one of the followings :  Fixed allocation, local scope  Variable allocation, local scope  Variable allocation, global scope

60. 60 Fixed Allocation, Local Scope Decide ahead of time the amount of allocation to give a process If allocation is too small, there will be a high page fault rate If allocation is too large there will be too few programs in main memory

61. 61 Variable Allocation, Global Scope Easiest to implement Adopted by many operating systems Operating system keeps list of free frames Free frame is added to resident set of process when a page fault occurs If no free frame, replaces one from another process

62. 62 Variable Allocation, Local Scope When new process added, allocate number of page frames based on application type, program request, or other criteria When page fault occurs, select page from among the resident set of the process that suffers the fault Reevaluate allocation from time to time

63. 63 Relationship between policies and scope

64. 64 Cleaning Policy Cleaning policy determines when a modified page should be written to disk. 2 main alternatives are :  Demand cleaning A page is written out only when it has been selected for replacement  Precleaning Pages are written out in batches

65. 65 Cleaning Policy Best approach uses page buffering  Replaced pages are placed in two lists Modified and unmodified  Pages in the modified list are periodically written out in batches  Pages in the unmodified list are either reclaimed if referenced again or lost when its frame is assigned to another page

66. 66 Load Control Load control is concerned with determining the number of processes that will be resident in main memory. Too few processes, many occasions when all processes will be blocked and much time will be spent in swapping Too many processes will lead to thrashing

67. 67 Multiprogramming Load control is Also refered to as multiprogramming level. The objective is to avoid thrashing.

68. 68 Process Suspension If multiprogramming level is to be reduced, one or more of the currently resident process needs to be suspended (swapped out). There are 6 possibilities :  Lowest priority process  Faulting process This process does not have its working set in main memory so it will be blocked anyway  Last process activated This process is least likely to have its working set resident  Process with smallest resident set This process requires the least future effort to reload  Largest process Obtains the most free frames  Process with the largest remaining execution window

69. 69 UNIX and Solaris Memory Management Paging System  Page table  Disk block descriptor  Page frame data table  Swap-use table

70. 70

71. 71

72. 72

73. 73 UNIX and Solaris Memory Management Page Replacement  Refinement of the clock policy

74. 74 Kernel Memory Allocator Lazy buddy system

75. 75 Linux Memory Management Page directory Page middle directory Page table

76. 76

77. 77

78. 78 Windows Memory Management Paging  Available  Reserved  Committed