Operating Systems: Internals and Design Principles, 6/E William Stallings - Chapter 8: Virtual Memory

Chapter 8Virtual MemoryOperating Systems:Internals and Design Principles, 6/EWilliam StallingsDave BremerOtago Polytechnic, N.Z.©2008, Prentice HallRoadmapHardware and Control StructuresOperating System SoftwareUNIX and Solaris Memory ManagementLinux Memory ManagementWindows Memory ManagementTerminologyKey points inMemory Management1) Memory references are logical addresses dynamically translated into physical addresses at run timeA process may be swapped in and out of main memory occupying different regions at different times during execution2) A process may be broken up into pieces that do not need to located contiguously in main memoryBreakthrough in Memory ManagementIf both of those two characteristics are present, then it is not necessary that all of the pages or all of the segments of a process be in main memory during execution.If the next instruction, and the next data location are in memory then execution can proceed at least for a timeExecution of a ProcessOperating system brings into main memory a few pieces of the programResident set - portion of process that is in main memoryAn interrupt is generated when an address is needed that is not in main memoryOperating system places the process in a blocking stateExecution of a ProcessPiece of process that contains the logical address is brought into main memoryOperating system issues a disk I/O Read requestAnother process is dispatched to run while the disk I/O takes placeAn interrupt is issued when disk I/O complete which causes the operating system to place the affected process in the Ready stateImplications of this new strategyMore processes may be maintained in main memoryOnly load in some of the pieces of each processWith so many processes in main memory, it is very likely a process will be in the Ready state at any particular timeA process may be larger than all of main memoryReal and Virtual MemoryReal memoryMain memory, the actual RAMVirtual memoryMemory on diskAllows for effective multiprogramming and relieves the user of tight constraints of main memoryThrashingA state in which the system spends most of its time swapping pieces rather than executing instructions.To avoid this, the operating system tries to guess which pieces are least likely to be used in the near future.The guess is based on recent historyPrinciple of LocalityProgram and data references within a process tend to clusterOnly a few pieces of a process will be needed over a short period of timeTherefore it is possible to make intelligent guesses about which pieces will be needed in the futureThis suggests that virtual memory may work efficientlyA Processes Performance in VM EnvironmentNote that during the lifetime of the process, references are confined to a subset of pages.Support Needed for Virtual MemoryHardware must support paging and segmentation Operating system must be able to manage the movement of pages and/or segments between secondary memory and main memoryPagingEach process has its own page tableEach page table entry contains the frame number of the corresponding page in main memoryTwo extra bits are needed to indicate:whether the page is in main memory or notWhether the contents of the page has been altered since it was last loaded(see next slide)Paging TableAddress TranslationPage TablesPage tables are also stored in virtual memoryWhen a process is running, part of its page table is in main memoryTwo-Level Hierarchical Page TableAddress Translation for Hierarchical page tablePage tables grow proportionallyA drawback of the type of page tables just discussed is that their size is proportional to that of the virtual address space.An alternative is Inverted Page TablesInverted Page TableUsed on PowerPC, UltraSPARC, and IA-64 architecturePage number portion of a virtual address is mapped into a hash valueHash value points to inverted page tableFixed proportion of real memory is required for the tables regardless of the number of processesInverted Page TableEach entry in the page table includes:Page numberProcess identifierThe process that owns this page. Control bits includes flags, such as valid, referenced, etcChain pointer the index value of the next entry in the chain.Inverted Page TableTranslation LookasideBufferEach virtual memory reference can cause two physical memory accessesOne to fetch the page tableOne to fetch the dataTo overcome this problem a high-speed cache is set up for page table entriesCalled a Translation Lookaside Buffer (TLB)Contains page table entries that have been most recently usedTLB OperationGiven a virtual address, processor examines the TLBIf page table entry is present (TLB hit), the frame number is retrieved and the real address is formedIf page table entry is not found in the TLB (TLB miss), the page number is used to index the process page tableLooking into the Process Page TableFirst checks if page is already in main memory If not in main memory a page fault is issuedThe TLB is updated to include the new page entryTranslation Lookaside BufferTLB operationAssociative MappingAs the TLB only contains some of the page table entries we cannot simply index into the TLB based on the page numberEach TLB entry must include the page number as well as the complete page table entryThe process is able to simultaneously query numerous TLB entries to determine if there is a page number matchTranslation Lookaside BufferTLB and Cache OperationPage SizeSmaller page size, less amount of internal fragmentationBut Smaller page size, more pages required per processMore pages per process means larger page tablesLarger page tables means large portion of page tables in virtual memoryPage SizeSecondary memory is designed to efficiently transfer large blocks of data so a large page size is betterFurther complications to Page SizeSmall page size, large number of pages will be found in main memoryAs 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.Page SizeExample Page SizeSegmentationSegmentation allows the programmer to view memory as consisting of multiple address spaces or segments.May be unequal, dynamic sizeSimplifies handling of growing data structuresAllows programs to be altered and recompiled independentlyLends itself to sharing data among processesLends itself to protectionSegment OrganizationStarting address corresponding segment in main memoryEach entry contains the length of the segmentA bit is needed to determine if segment is already in main memoryAnother bit is needed to determine if the segment has been modified since it was loaded in main memorySegment Table EntriesAddress Translation in SegmentationCombined Paging and SegmentationPaging is transparent to the programmerSegmentation is visible to the programmerEach segment is broken into fixed-size pagesCombined Paging and SegmentationAddress TranslationProtection and sharingSegmentation lends itself to the implementation of protection and sharing policies.As each entry has a base address and length, inadvertent memory access can be controlledSharing can be achieved by segments referencing multiple processesProtection RelationshipsRoadmapHardware and Control StructuresOperating System SoftwareUNIX and Solaris Memory ManagementLinux Memory ManagementWindows Memory ManagementMemory Management DecisionsWhether or not to use virtual memory techniquesThe use of paging or segmentation or bothThe algorithms employed for various aspects of memory managementKey Design ElementsKey aim: Minimise page faultsNo definitive best policyFetch PolicyDetermines when a page should be brought into memoryTwo main types:Demand Paging PrepagingDemand Pagingand PrepagingDemand pagingonly brings pages into main memory when a reference is made to a location on the pageMany page faults when process first startedPrepaging brings in more pages than neededMore efficient to bring in pages that reside contiguously on the diskDon’t confuse with “swapping”Placement PolicyDetermines where in real memory a process piece is to resideImportant in a segmentation systemPaging or combined paging with segmentation hardware performs address translationReplacement PolicyWhen all of the frames in main memory are occupied and it is necessary to bring in a new page, the replacement policy determines which page currently in memory is to be replaced.ButWhich page is replaced?Page removed should be the page least likely to be referenced in the near future How is that determined?Principal of locality againMost policies predict the future behavior on the basis of past behaviorReplacement Policy: Frame LockingFrame LockingIf frame is locked, it may not be replacedKernel of the operating systemKey control structuresI/O buffersAssociate a lock bit with each frameBasic Replacement AlgorithmsThere are certain basic algorithms that are used for the selection of a page to replace, they includeOptimalLeast recently used (LRU)First-in-first-out (FIFO)ClockExamples ExamplesAn example of the implementation of these policies will use a page address stream formed by executing the program is2 3 2 1 5 2 4 5 3 2 5 2Which means that the first page referenced is 2, the second page referenced is 3, And so on. Optimal policySelects for replacement that page for which the time to the next reference is the longestBut Impossible to have perfect knowledge of future eventsOptimal Policy Example The optimal policy produces three page faults after the frame allocation has been filled.Least Recently Used (LRU)Replaces the page that has not been referenced for the longest timeBy the principle of locality, this should be the page least likely to be referenced in the near futureDifficult to implementOne approach is to tag each page with the time of last reference. This requires a great deal of overhead.LRU ExampleThe LRU policy does nearly as well as the optimal policy.In this example, there are four page faultsFirst-in, first-out (FIFO)Treats page frames allocated to a process as a circular bufferPages are removed in round-robin styleSimplest replacement policy to implementPage that has been in memory the longest is replacedBut, these pages may be needed again very soon if it hasn’t truly fallen out of useFIFO ExampleThe FIFO policy results in six page faults.Note that LRU recognizes that pages 2 and 5 are referenced more frequently than other pages, whereas FIFO does not.Clock PolicyUses and additional bit called a “use bit”When a page is first loaded in memory or referenced, the use bit is set to 1When it is time to replace a page, the OS scans the set flipping all 1’s to 0The first frame encountered with the use bit already set to 0 is replaced.Clock Policy ExampleNote that the clock policy is adept at protecting frames 2 and 5 from replacement.Clock PolicyClock PolicyClock PolicyCombined ExamplesComparisonPage BufferingLRU and Clock policies both involve complexity and overhead Also, replacing a modified page is more costly than unmodified as needs written to secondary memorySolution: Replaced page is added to one of two listsFree page list if page has not been modifiedModified page listReplacement Policy and Cache SizeMain memory size is getting larger and the locality of applications is decreasing. So, cache sizes have been increasingWith large caches, replacement of pages can have a performance impactimprove performance by supplementing the page replacement policy with a with a policy for page placement in the page bufferResident Set ManagementThe OS must decide how many pages to bring into main memoryThe smaller the amount of memory allocated to each process, the more processes that can reside in memory.Small number of pages loaded increases page faults.Beyond a certain size, further allocations of pages will not affect the page fault rate.Resident Set SizeFixed-allocationGives a process a fixed number of pages within which to executeWhen a page fault occurs, one of the pages of that process must be replacedVariable-allocationNumber of pages allocated to a process varies over the lifetime of the processReplacement ScopeThe scope of a replacement strategy can be categorized as global or local.Both types are activated by a page fault when there are no free page frames.A local replacement policy chooses only among the resident pages of the process that generated the page faultA global replacement policy considers all unlocked pages in main memory Fixed Allocation, Local ScopeDecide ahead of time the amount of allocation to give a processIf allocation is too small, there will be a high page fault rateIf allocation is too large there will be too few programs in main memoryIncreased processor idle time orIncreased swapping.Variable Allocation, Global ScopeEasiest to implementAdopted by many operating systemsOperating system keeps list of free framesFree frame is added to resident set of process when a page fault occursIf no free frame, replaces one from another processTherein lies the difficulty which to replace.Variable Allocation, Local ScopeWhen new process added, allocate number of page frames based on application type, program request, or other criteriaWhen page fault occurs, select page from among the resident set of the process that suffers the faultReevaluate allocation from time to timeResident Set Management SummaryCleaning PolicyA cleaning policy is concerned with determining when a modified page should be written out to secondary memory.Demand cleaningA page is written out only when it has been selected for replacementPrecleaningPages are written out in batchesCleaning PolicyBest approach uses page bufferingReplaced pages are placed in two listsModified and unmodifiedPages in the modified list are periodically written out in batchesPages in the unmodified list are either reclaimed if referenced again or lost when its frame is assigned to another pageLoad ControlDetermines the number of processes that will be resident in main memoryThe multiprogramming levelToo few processes, many occasions when all processes will be blocked and much time will be spent in swappingToo many processes will lead to thrashingMultiprogrammingProcess SuspensionIf the degree of multiprogramming is to be reduced, one or more of the currently resident processes must be suspended (swapped out).Six possibilities existSuspension policiesLowest priority processFaulting processThis process does not have its working set in main memory so it will be blocked anywayLast process activatedThis process is least likely to have its working set residentSuspension policies cont.Process with smallest resident setThis process requires the least future effort to reloadLargest processObtains the most free frames Process with the largest remaining execution windowRoadmapHardware and Control StructuresOperating System SoftwareUNIX and Solaris Memory ManagementLinux Memory ManagementWindows Memory ManagementUnixIntended to be machine independent so implementations varyEarly Unix: variable partitioning with no virtual memory to pagedRecent Unix (SVR4 & Solaris) using paged virtual memorySVR4 uses two separate schemes:Paging system and a kernel memory allocator.Paging System and Kernel Memory AllocatorPaging system provides a virtual memory capability that allocates page frames in main memory to processes Also allocates page frames to disk block buffers.Kernel Memory Allocator allocates memory for the kernelThe paging system is less suited for this taskPaged VM Data StructuresPage Table Entry FieldsDisk Block Descriptor FieldsPage Frame and Swap Use fieldsPage ReplacementThe page frame data table is used for page replacementPointers used to create several lists within the tableFree frame listWhen the number of free frames drops below a threshold, the kernel will steal a number of frames to compensate.“Two Handed” Clock Page ReplacementParameters for Two Handed ClockScanrate: The rate at which the two hands scan through the page list, in pages per secondHandspread: The gap between fronthand and backhandBoth have defaults set at boot time based on physical memoryKernel Memory AllocatorThe kernel generates and destroys small tables and buffers frequently during the course of execution, each of which requires dynamic memory allocation.Most of these blocks significantly smaller than typical pages, Therefore normal paging would be inefficientVariation of “buddy system” is usedLazy BuddyUNIX often exhibits steady-state behavior in kernel memory demand; i.e. the amount of demand for blocks of a particular size varies slowly in time.To avoid unnecessary joining and splitting of blocks, the lazy buddy system defers coalescing until it seems likely that it is needed, and then coalesces as many blocks as possible.Lazy Buddy System ParametersNi = current number of blocks of size 2iAi = current number of blocks of size 2i that are allocated (occupied).Gi = current number of blocks of size 2i that are globally free.Li = current number of blocks of size 2i that are locally freeLazy Buddy System AllocatorLinux Memory ManagementShares many characteristics with UnixBut is quite complexTwo main aspectsProcess virtual memory, and Kernel memory allocation.Linux Memory ManagementPage directoryPage middle directoryPage tableLinux Virtual MemoryThree level page table structureEach table is the size of one pagePage directoryEach process has one page directory1 page in size, must be in main memoryPage middle directory:May be multiple pages, each entry points to one page in the page tableLinux Memory contPage table May also span multiple pages. Each page table entry refers to one virtual page of the process.Address TranslationPage ReplacementBased on the clock algorithmThe “use bit” is replace with an 8-bit age variableIncremented with each page accessPeriodically decrements the age bitsAny page with an age of 0 is “old” and is a candidate for replacementA form of Least Frequently Used policyWindows Memory ManagementThe Windows virtual memory manager controls how memory is allocated and how paging is performed. Designed to operate over a variety of platformsuses page sizes ranging from 4 Kbytes to 64 Kbytes. Windows Virtual Address MapOn 32 bit platforms each user process sees a separate 32 bit address spaceAllowing 4G per processSome reserved for the OS, Typically each user process has 32G of available virtual address spaceWith all processes sharing the same 2G system space32 bit Windows Address SpaceWindows PagingOn creation, a process can make use of the entire user space of almost 2 Gbytes. This space is divided into fixed-size pages managed in contiguous regions allocated on 64Kbyte boundariesRegions may be in one of three statesAvailableReservedCommittedResident Set Management SystemWindows uses “variable allocation, local scope”When activated a process is assigned data structures to manage its working setWorking sets of active processes are adjusted depending on the availability of main memory