Lect13
0 likes333 views
The document discusses different techniques for managing memory in operating systems, including fixed and dynamic partitioning, paging, segmentation, and swapping, which allows processes to be swapped temporarily to disk to improve memory utilization and enable more processes to run simultaneously than can fit in physical memory. Memory management aims to allocate memory efficiently while supporting protection between processes and relocation of processes in memory.
Lect13
- 1. Memory Management 1
- 2. Process • One or more threads of execution • Resources required for execution – Memory (RAM) • Program code (“text”) • Data (initialised, uninitialised, stack) • Buffers held in the kernel on behalf of the process – Others • CPU time • Files, disk space, printers, etc. 2
- 3. Some Goals of an Operating System • Maximise memory utilisation • Maximise CPU utilization • Minimise response time • Prioritise “important” processes • Note: Conflicting goals ⇒ tradeoffs – E.g. maximising CPU utilisation (by running many processes) increases (degrades) system response time. 3
- 4. Memory Management • Keeps track of what memory is in use and what memory is free • Allocates free memory to process when needed – And deallocates it when they don’t • Manages the transfer of memory between RAM and disk. 4
- 5. Memory Hierarchy • Ideally, programmers want memory that is – Fast – Large – Nonvolatile • Not possible • Memory manager coordinates how memory hierarchy is used. – Focus usually on RAM ⇔ Disk 5
- 6. Memory Management • Two broad classes of memory management systems – Those that transfer processes to and from disk during execution. • Called swapping or paging – Those that don’t • Simple • Might find this scheme in an embedded device, phone, smartcard, or PDA. 6
- 7. Basic Memory Management Monoprogramming without Swapping or Paging Three simple ways of organizing memory - an operating system with one user process7
- 8. Monoprogramming • Okay if – Only have one thing to do – Memory available approximately equates to memory required • Otherwise, – Poor CPU utilisation in the presence of I/O waiting – Poor memory utilisation with a varied job mix 8
- 9. Idea • Subdivide memory and run more than one process at once!!!! – Multiprogramming, Multitasking 9
- 10. Modeling Multiprogramming Degree of multiprogramming CPU utilization as a function of number of processes in10 memory
- 11. Problem: How to divide memory • One approach Process A – divide memory into fixed equal-sized partitions – Any process <= partition Process B size can be loaded into any partition Process C Process D 11
- 12. Simple MM: Fixed, equal-sized partitions • Any unused space in the Process A partition is wasted – Called internal fragmentation Process B • Processes smaller than main memory, but larger than a partition cannot Process C run. Process D 12
- 13. Simple MM: Fixed, variable-sized partitions • Multiple Queues: – Place process in queue for smallest partition that it fits in. 13
- 14. • Issue – Some partitions may be idle • Small jobs available, but only large partition free 14
- 15. • Single queue, search for any jobs that fits • Small jobs in large partition if necessary – Increases internal memory fragmentation 15
- 16. Fixed Partition Summary • Simple • Easy to implement • Can result in poor memory utilisation – Due to internal fragmentation • Used on OS/360 operating system (OS/MFT) – Old mainframe batch system • Still applicable for simple embedded systems 16
- 17. Dynamic Partitioning • Partitions are of variable length • Process is allocated exactly what it needs – Assume a process knows what it needs 17
- 20. Dynamic Partitioning • In previous diagram – We have 16 meg free in total, but it can’t be used to run any more processes requiring > 6 meg as it is fragmented – Called external fragmentation • We end up with unusable holes • Reduce external fragmentation by compaction – Shuffle memory contents to place all free memory together in one large block. – Compaction is possible only if relocation is dynamic, and is done at execution time. 20
- 21. Recap: Fragmentation • External Fragmentation: – The space wasted external to the allocated memory regions. – Memory space exists to satisfy a request, but it is unusable as it is not contiguous. • Internal Fragmentation: – The space wasted internal to the allocated memory regions. – allocated memory may be slightly larger than requested memory; this size difference is wasted memory internal to a partition. 21
- 22. Dynamic Partition Allocation Algorithms • Basic Requirements – Quickly locate a free partition satisfying the request – Minimise external fragmentation – Efficiently support merging two adjacent free partitions into a larger partition 22
- 23. Classic Approach • Represent available memory as a linked list of available “holes”. – Base, size – Kept in order of increasing address • Simplifies merging of adjacent holes into larger holes. Address Address Address Address Size Size Size Size Link Link Link Link 23
- 24. Coalescing Free Partitions with Linked Lists Four neighbor combinations for the terminating process X 24
- 25. Dynamic Partitioning Placement Algorithm • First-fit algorithm – Scan the list for the first entry that fits • If greater in size, break it into an allocated and free part • Intent: Minimise amount of searching performed – Aims to find a match quickly – Generally can result in holes at the front end of memory that must be searched over when trying to find a free block. – May have lots of unusable holes at the beginning. • External fragmentation – Tends to preserve larger blocks at the end of memory Address Address Address Address Size Size Size Size Link Link Link Link 25
- 26. Dynamic Partitioning Placement Algorithm • Next-fit – Like first-fit, except it begins its search from the point in list where the last request succeeded instead of at the beginning. • Spread allocation more uniformly over entire memory – More often allocates a block of memory at the end of memory where the largest block is found • The largest block of memory is broken up into smaller blocks – May not be able to service larger request as well as first fit. Address Address Address Address Size Size Size Size Link Link Link Link 26
- 27. Dynamic Partitioning Placement Algorithm • Best-fit algorithm – Chooses the block that is closest in size to the request – Poor performer • Has to search complete list – does more work than first- or next-fit • Since smallest block is chosen for a process, the smallest amount of external fragmentation is left – Create lots of unusable holes Address Address Address Address Size Size Size Size Link Link Link Link 27
- 28. Dynamic Partitioning Placement Algorithm • Worst-fit algorithm – Chooses the block that is largest in size (worst-fit) • (whimsical) idea is to leave a usable fragment left over – Poor performer • Has to do more work (like best fit) to search complete list • Does not result in significantly less fragmentation Address Address Address Address Size Size Size Size Link Link Link Link 28
- 30. Dynamic Partition Allocation Algorithm • Summary – First-fit and next-fit are generally better than the others and easiest to implement • Note: Used rarely these days – Typical in-kernel allocators used are lazy buddy, and slab allocators • Might go through these later in extended • You should be aware of them – useful as a simple allocator for simple systems 30
- 31. Compaction • We can reduce Process A external fragmentation by compaction – Only if we can relocate running programs – Generally requires Process A Process B hardware support Process B Process C Process C Process D Process D 31
- 32. Some Remaining Issues with Dynamic Partitioning • We have ignored Process A – Relocation • How does a process run in different locations in memory? – Protection Process B • How do we prevent processes interfering with each other Process C Process D 32
- 33. Example Logical Address-Space Layout 0x0000 • Logical addresses refer to specific locations within the program • Once running, these address must refer to real physical memory • When are logical 0xFFFF addresses bound to physical? 33
- 34. When are memory addresses bound? • Compile/link time – Compiler/Linker binds the addresses – Must know “run” location at compile time – Recompile if location changes • Load time – Compiler generates relocatable code – Loader binds the addresses at load time • Run time – Logical compile-time addresses translated to physical addresses by special hardware. 34
- 35. Hardware Support for Runtime Binding and Protection 0xFFFF • For process B to run using logical Process A addresses – Need to add an appropriate offset to its logical addresses • Achieve relocation • Protect memory “lower” than B Process B limit base – Must limit the maximum logical address B can generate • Protect memory “higher” than B Process C Process D 35 0x0000
- 36. Hardware Support for Relocation and Limit Registers 36
- 37. Base and Limit Registers 0xFFFF base=0x8000 • Also called Process A limit = 0x2000 – Base and bound registers – Relocation and limit registers • Base and limit registers – Restrict and relocate the currently 0x9FFF limit 0x1FFF Process B base 0x0000 active process 0x8000 – Base and limit registers must be changed at • Load time Process C • Relocation (compaction time) • On a context switch Process D 37 0x0000
- 38. Base and Limit Registers 0xFFFF base=0x4000 • Also called Process A limit = 0x3000 – Base and bound registers – Relocation and limit registers • Base and limit registers – Restrict and relocate the currently Process B active process – Base and limit registers must be 0x6FFF changed at 0x2FFF limit Process C • Load time 0x0000 base 0x4000 • Relocation (compaction time) • On a context switch Process D 38 0x0000
- 39. Base and Limit Registers • Cons – Physical memory allocation must still be contiguous – The entire process must be in memory – Do not support partial sharing of address spaces 39
- 40. Timesharing 0xFFFF • Thus far, we have a system Process A suitable for a batch system – Limited number of dynamically allocated processes • Enough to keep CPU utilised – Relocated at runtime Process B – Protected from each other • But what about timesharing? – We need more than just a small Process C number of processes running at once Process D 40 0x0000
- 41. Swapping • A process can be swapped temporarily out of memory to a backing store, and then brought back into memory for continued execution. • Backing store – fast disk large enough to accommodate copies of all memory images for all users; must provide direct access to these memory images. • Can prioritize – lower-priority process is swapped out so higher-priority process can be loaded and executed. • Major part of swap time is transfer time; total transfer time is directly proportional to the amount of memory swapped. – slow 41
- 42. Schematic View of Swapping 42
- 43. So far we have assumed a process is smaller than memory • What can we do if a process is larger than main memory? 43
- 44. Overlays • Keep in memory only those instructions and data that are needed at any given time. • Implemented by user, no special support needed from operating system • Programming design of overlay structure is complex 44
- 45. Overlays for a Two-Pass Assembler 45
- 46. Virtual Memory • Developed to address the issues identified with the simple schemes covered thus far. • Two classic variants – Paging – Segmentation • Paging is now the dominant one of the two • Some architectures support hybrids of the two schemes 46
- 47. Virtual Memory - Paging • Partition physical memory into small equal sized chunks – Called frames • Divide each process’s virtual (logical) address space into same size chunks – Called pages – Virtual memory addresses consist of a page number and offset within the page • OS maintains a page table – contains the frame location for each page – Used by to translate each virtual address to physical address – The relation between virtual addresses and physical memory addresses is given by page table • Process’s physical memory does not have to be contiguous 47
- 50. Paging • No external fragmentation • Small internal fragmentation (in last page) • Allows sharing by mapping several pages to the same frame • Abstracts physical organisation – Programmer only deal with virtual addresses • Minimal support for logical organisation – Each unit is one or more pages 50
- 51. Memory Management Unit (also called Translation Look-aside Buffer – TLB) The position and function of the MMU 51
- 52. MMU Operation Assume for now that the page table is contained wholly in registers within the MMU – in practice it is not Internal operation of simplified MMU with 16 4 KB pages 52
- 53. Virtual Memory - Segmentation • Memory-management scheme that supports user’s view of memory. • A program is a collection of segments. A segment is a logical unit such as: – main program, procedure, function, method, object, local variables, global variables, common block, stack, symbol table, arrays 53
- 54. Logical View of Segmentation 1 4 1 2 3 2 4 3 user space physical memory space 54
- 55. Segmentation Architecture • Logical address consists of a two tuple: <segment- number, offset>, – Addresses identify segment and address with segment • Segment table – each table entry has: – base – contains the starting physical address where the segments reside in memory. – limit – specifies the length of the segment. • Segment-table base register (STBR) points to the segment table’s location in memory. • Segment-table length register (STLR) indicates number of segments used by a program; segment number s is legal if s < STLR. 55
- 56. Segmentation Hardware 56
- 57. Example of Segmentation 57
- 58. Segmentation Architecture • Protection. With each entry in segment table associate: – validation bit = 0 ⇒ illegal segment – read/write/execute privileges • Protection bits associated with segments; code sharing occurs at segment level. • Since segments vary in length, memory allocation is a dynamic partition-allocation problem. • A segmentation example is shown in the following diagram 58
- 59. Sharing of Segments 59
- 60. Segmentation Architecture • Relocation. – dynamic ⇒ by segment table • Sharing. – shared segments ⇒ same physical backing multiple segments ⇒ ideally, same segment number • Allocation. – First/next/best fit ⇒ external fragmentation 60
- 61. Comparison Comparison of paging and segmentation 61