Unit III

MEMORY MANAGEMENT
1. MAIN MEMORY

 In general, to rum a program, it must be brought into memory. 

 Input queue – collection of processes on the disk that are waiting to be brought
into memory to run the program. 
 User programs go through several steps before being run 
 Address binding: Mapping of instructions and data from one address to another
address in memory. 

Three different stages of binding:

1. Compile time: Must generate absolute code if memory location is known in prior.
2. Load time: Must generate relocatable code if memory location is not known at
compile time
3. Execution time: Need hardware support for address maps (e.g., base and limit
registers)
Multistep Processing of a User Program

Logical vs. Physical Address Space

 Logical address – generated by the CPU; also referred to as “virtual address“ 

 Physical address – address seen by the memory unit. 
 Logical and physical addresses are the same in ―compile-time and load-time
 address-binding schemes‖ 
 Logical (virtual) and physical addresses differ in ―execution-time address-
binding scheme‖ 

Memory-Management Unit (MMU)

 It is a hardware device that maps virtual / Logical address to physical address 

 In this scheme, the relocation register‘s value is added to Logical address
generated by a user process. 
 The user program deals with logical addresses; it never sees the real physical
addresses 
 Logical address range: 0 to max 
 Physical address range: R+0 to R+max, where R—value in relocation register

Dynamic relocation using relocation register

Dynamic Loading

 Through this, the routine is not loaded until it is called.

o Better memory-space utilization; unused routine is never loaded
o Useful when large amounts of code are needed to handle infrequently
occurring cases
o No special support from the operating system is required implemented
through program design

Dynamic Linking
 Linking postponed until execution time & is particularly useful for libraries 
 Small piece of code called stub, used to locate the appropriate memory-resident
 library routine or function. 

 Stub replaces itself with the address of the routine, and executes the routine 
 Operating system needed to check if routine is in processes‘ memory address 
 Shared libraries: Programs linked before the new library was installed will
continue using the older library 
Overlays:
 Enable a process larger than the amount of memory allocated to it. 
 At a given time, the needed instructions & data are to be kept within a memory. 

Swapping
 A process can be swapped temporarily out of memory to a backing store (SWAP
OUT)and then brought back into memory for continued execution (SWAP IN). 
 Backing store – fast disk large enough to accommodate copies of all memory
images for all users & it must provide direct access to these memory images 
 Roll out, roll in – swapping variant used for priority-based scheduling algorithms;
lower-priority process is swapped out so higher-priority process can be loaded and
executed 
 Transfer time : 

Major part of swap time is transfer time 

Total transfer time is directly proportional to the amount of memory swapped. 

Example: Let us assume the user process is of size 1MB & the backing store
is a standard hard disk with a transfer rate of 5MBPS. 

Transfer time = 1000KB/5000KB per second

= 1/5 sec = 200ms

2. CONTIGUOUS MEMORY ALLCOATION
(i) Memory Protection:
o It should consider;
a) Protecting the OS from user process.
b) Protecting user processes from one another.
o The above protection is done by “Relocation-register & Limit-register scheme
o Relocation register contains value of smallest physical address i.e base value.
o Limit register contains range of logical addresses – each logical address must be
less than the limit register

A base and a limit register define a logical address space

HW address protection with base and limit registers

ii) Memory Allocation

 Each process is contained in a single contiguous section of memory. 

 There are two methods namely : 

Fixed – Partition Method 

Variable – Partition Method 
 Fixed – Partition Method : 
o Divide memory into fixed size partitions, where each partition has exactly
one process.
o The drawback is memory space unused within a partition is
wasted.(eg.when process size < partition size)
 Variable-partition method:
o Divide memory into variable size partitions, depending upon the size of the
incoming process.
o When a process terminates, the partition becomes available for another
process.
o As processes complete and leave they create holes in the main memory.
o Hole – block of available memory; holes of various size are scattered
throughout memory.

 Dynamic Storage-Allocation Problem: 


How to satisfy a request of size ‗n‘ from a list of free holes? 
Solution:
o First-fit: Allocate the first hole that is big enough.
o Best-fit: Allocate the smallest hole that is big enough; must search entire list,
unless ordered by size. Produces the smallest leftover hole.
o Worst-fit: Allocate the largest hole; must also search entire list. Produces the
largest leftover hole.
 iii)Fragmentation

o External Fragmentation – This takes place when enough total memory

space exists to satisfy a request, but it is not contiguous i.e, storage is
fragmented into a large number of small holes scattered throughout the main
memory.
o Internal Fragmentation – Allocated memory may be slightly larger than
requested memory.
Example: hole = 184 bytes
Process size = 182 bytes.
We are left with a hole of 2 bytes.
o Solutions:
1. Coalescing : Merge the adjacent holes together.
2. Compaction: Move all processes towards one end of memory, hole towards
other end of memory, producing one large hole of available memory. This
scheme is expensive as it can be done if relocation is dynamic and done at
execution time.
3. Permit the logical address space of a process to be non-contiguous. This is
achieved through two memory management schemes namely paging and
segmentation.

3. PAGING
 It is a memory management scheme that permits the physical address space of a
process to be noncontiguous. 
 It avoids the considerable problem of fitting the varying size memory chunks on to
the backing store. 

3.1 Basic Method:
o Divide logical memory into blocks of same size called “pages”. o
Divide physical memory into fixed-sized blocks called “frames” o
Page size is a power of 2, between 512 bytes and 16MB.

Address Translation Scheme

o Address generated by CPU(logical address) is divided into:

Page number (p) – used as an index into a page table which contains base
address of each page in physical memory 

Page offset (d) – combined with base address to define the physical address
i.e., 
Physical address = base address + offset 
 Paging Hardware

Paging model of logical and physical memory

Paging example for a 32-byte memory with 4-byte pages

Page size = 4 bytes

Physical memory size = 32 bytes i.e ( 4 X 8 = 32 so, 8 pages)

Logical address ‗0‘ maps to physical address 20 i.e ( (5 X 4) +0)

Where Frame no = 5, Page size = 4, Offset =0

Allocation
o When a process arrives into the system, its size (expressed in pages) is
examined.
 o Each page of process needs one frame. Thus if the process requires ‗n‘ pages, at
least ‗n‘ frames must be available in memory.
o If ‗n‘ frames are available, they are allocated to this arriving process.
o The 1st page of the process is loaded into one of the allocated frames & the
frame number is put into the page table.
o Repeat the above step for the next pages & so on.

(a) Before Allocation (b) After Allocation

Frame table: It is used to determine which frames are allocated, which frames are
available, how many total frames are there, and so on.(ie) It contains all the information
about the frames in the physical memory.

3.2 Hardware implementation of Page Table

o This can be done in several ways :
1. Using PTBR
2. TLB
o The simplest case is Page-table base register (PTBR), is an index to point the
page table.
o TLB (Translation Look-aside Buffer)

It is a fast lookup hardware cache. 

It contains the recently or frequently used page table entries. 

It has two parts: Key (tag) & Value. 

More expensive. 
Paging Hardware with TLB

o When a logical address is generated by CPU, its page number is presented to TLB.
o TLB hit: If the page number is found, its frame number is immediately
available & is used to access memory
o TLB miss: If the page number is not in the TLB, a memory reference to the page
table must be made.
o Hit ratio: Percentage of times that a particular page is found in the TLB.

For example hit ratio is 80% means that the desired page
number in the TLB is 80% of the time. 
o Effective Access Time:

Assume hit ratio is 80%. 

If it takes 20ns to search TLB & 100ns to access memory, then the
memory access takes 120ns(TLB hit) 

If we fail to find page no. in TLB (20ns), then we must 1st access
memory for page table (100ns) & then access the desired byte in
memory (100ns). 
Therefore Total = 20 + 100 + 100 
= 220 ns(TLB miss).
Then Effective Access Time (EAT) = 0.80 X (120 + 0.20) X 220.
= 140 ns.
(iii) Protectio
o Memory protection implemented by associating protection bit with each
frame
o Valid-invalid bit attached to each entry in the page table:

“valid (v)” indicates that the associated page is in the process‘ logical
address space, and is thus a legal page 



“invalid (i)” indicates that the page is not in the process‘ logical address
space 

3.3 Structures of the Page Table

a) Hierarchical Paging
b) Hashed Page Tables
c) Inverted Page Tables

a) Hierarchical Paging
o Break up the Page table into smaller pieces. Because if the page table is too
large then it is quit difficult to search the page number.
Example: “Two-Level Paging “
Address-Translation Scheme/32 PAGING ARCHITECTURE

Address-translation scheme for a two-level 32-bit paging architecture

It requires more number of memory accesses, when the number of levels is increased.

(b) Hashed Page Tables

o Each entry in hash table contains a linked list of elements that hash to the
same location.
o Each entry consists of;

(a) Virtual page numbers

(b) Value of mapped page frame.
(c) Pointer to the next element in the linked list.
o Working Procedure:

The virtual page number in the virtual address is hashed into the hash
table. 

Virtual page number is compared to field (a) in the 1st element in the
linked list. 

If there is a match, the corresponding page frame (field (b)) is used to form
the desired physical address. 

If there is no match, subsequent entries in the linked list are searched for a
matching virtual page number. 


Clustered page table: It is a variation of hashed page table & is similar to hashed page
table except that each entry in the hash table refers to several pages rather than a single
page.

(c)Inverted Page Table

o It has one entry for each real page (frame) of memory & each entry consists of
the virtual address of the page stored in that real memory location, with
information about the process that owns that page. So, only one page table is in
the system.
 o When a memory reference occurs, part of the virtual address ,consisting of
<Process-id, Page-no> is presented to the memory sub-system.
o Then the inverted page table is searched for match:
(i) If a match is found, then the physical address is generated.
(ii) If no match is found, then an illegal address access has been attempted.
o Merit: Reduce the amount of memory needed.
o Demerit: Improve the amount of time needed to search the table when a page
reference oocurs.

(v) Shared Pages

o One advantage of paging is the possibility of sharing common code.
o Shared code

One copy of read-only (reentrant) code shared among processes (i.e., text
editors, compilers, window systems). 

Shared code must appear in same location in the logical address space of all
processes 
o Reentrant code (Pure code): Non-self modifying code. If the code is
reentrant, then it never changes during execution. Thus two or more
processes can execute the same code at the same time.
o Private code and data

Each process keeps a separate copy of the code and data 

The pages for the private code and data can appear anywhere in the
logical address space 
EXAMPLE:
Drawback of Paging – Internal fragmentation

o In the worst case a process would need n pages plus one byte.It would be allocated
n+1 frames resulting in an internal fragmentation of almost an entire frame.
Example:
Page size = 2048 bytes
Process size= 72766 bytes
Process needs 35 pages plus 1086 bytes.
It is allocated 36 frames resulting in an internal fragmentation of 962 bytes.

5. SEGMENTATION
o Memory-management scheme that supports user view of memory
o 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

User’s View of a Program

Logical View of Segmentation

5.1 Segmentation Hardware

o Logical address consists of a two tuple :

<Segment-number, offset>
o Segment table – maps two-dimensional physical addresses; each table entry has:

Base – contains the starting physical address where the segments reside in
memory 

Limit – specifies the length of the segment 
o Segment-table base register (STBR) points to the segment table‘s location in
memory
o Segment-table length register (STLR) indicates number of segments used by a
program;
Segment number‗s‘ is legal, if s < STLR
o Relocation.

dynamic 

by segment table 
o Sharing. 

shared segments 

same segment number 
o Allocation. 

first fit/best fit 

external fragmentation 
o Protection: With each entry in segment table associate:

validation bit = 0  illegal segment 

read/write/execute privileges 
o Protection bits associated with segments; code sharing occurs at segment level
o Since segments vary in length, memory allocation is a dynamic storage-allocation
problem
o A segmentation example is shown in the following diagram

Address Translation scheme

EXAMPLE:

Sharing of Segments

o Another advantage of segmentation involves the sharing of code or data.

o Each process has a segment table associated with it, which the dispatcher uses to
define the hardware segment table when this process is given the CPU.
o Segments are shared when entries in the segment tables of two different processes
point to the same physical location.

5.2 Segmentation with paging

o The IBM OS/ 2.32 bit version is an operating system running on top of the Intel 386
architecture. The 386 uses segmentation with paging for memory management. The
maximum number of segments per process is 16 KB, and each segment can be as
large as 4 gigabytes.
o The local-address space of a process is divided into two partitions.

The first partition consists of up to 8 KB segments that are private to that
process. 

The second partition consists of up to 8KB segments that are shared
among all the processes. 
o Information about the first partition is kept in the local descriptor table (LDT),
information about the second partition is kept in the global descriptor table
(GDT).

o Each entry in the LDT and GDT consist of 8 bytes, with detailed information about
a particular segment including the base location and length of the segment.
The logical address is a pair (selector, offset) where the selector is a16-bit
number:
s g p

13 1 2

Where s designates the segment number, g indicates whether the segment

is in the GDT or LDT, and p deals with protection. The offset is a 32-bit number
specifying the location of the byte within the segment in question.
o The base and limit information about the segment in question are used to generate a
linear-address.
o First, the limit is used to check for address validity. If the address is not valid, a
memory fault is generated, resulting in a trap to the operating system. If it is valid,
then the value of the offset is added to the value of the base, resulting in a 32-bit
linear address. This address is then translated into a physical address.
o The linear address is divided into a page number consisting of 20 bits, and a page
offset consisting of 12 bits. Since we page the page table, the page number is further
divided into a 10-bit page directory pointer and a 10-bit page table pointer. The
logical address is as follows.
 p1 p2 d

10 10 12

o To improve the efficiency of physical memory use. Intel 386 page tables can be
swapped to disk. In this case, an invalid bit is used in the page directory entry to
indicate whether the table to which the entry is pointing is in memory or on disk.
o If the table is on disk, the operating system can use the other 31 bits to specify the
disk location of the table; the table then can be brought into memory on demand.

6. VIRTUAL MEMORY
o It is a technique that allows the execution of processes that may not be completely
in main memory.
o Advantages:

Allows the program that can be larger than the physical memory. 

Separation of user logical memory from physical memory 

Allows processes to easily share files & address space. 

Allows for more efficient process creation. 
o Virtual memory can be implemented using

Demand paging 

Demand segmentation 
Virtual Memory That is Larger than Physical Memory
6.1 Demand Paging
o It is similar to a paging system with swapping.
o Demand Paging - Bring a page into memory only when it is needed
o To execute a process, swap that entire process into memory. Rather than swapping
the entire process into memory however, we use ―Lazy Swapper‖
o Lazy Swapper - Never swaps a page into memory unless that page will be
needed.
o Advantages

Less I/O needed 

Less memory needed 

Faster response 

More users 

Transfer of a paged memory to contiguous disk space

Basic Concepts:
o Instead of swapping in the whole processes, the pager brings only those necessary
pages into memory. Thus,
1. It avoids reading into memory pages that will not be used anyway.
2. Reduce the swap time.
3. Reduce the amount of physical memory needed.
o To differentiate between those pages that are in memory & those that are on the
disk we use the Valid-Invalid bit

Valid-Invalid bit
o A valid – invalid bit is associated with each page table entry. o
Valid  associated page is in memory.
In-Valid 

invalid page 

valid page but is currently on the disk 
Page Fault
o Access to a page marked invalid causes a page fault trap.

Steps in Handling a Page Fault

1. Determine whether the reference is a valid or invalid memory access

2. a) If the reference is invalid then terminate the process.
b) If the reference is valid then the page has not been yet brought into main
memory.
3. Find a free frame.
4. Read the desired page into the newly allocated frame.
 5. Reset the page table to indicate that the page is now in memory.
6. Restart the instruction that was interrupted .

Pure demand paging

o Never bring a page into memory until it is required.
o We could start a process with no pages in memory.
o When the OS sets the instruction pointer to the 1st instruction of the process, which
is on the non-memory resident page, then the process immediately faults for the
page.
o After this page is bought into the memory, the process continue to execute,
faulting as necessary until every page that it needs is in memory.

Performance of demand paging

o Let p be the probability of a page fault 0  p  1
o Effective Access Time (EAT)
EAT = (1 – p) x ma + p x page fault time.
Where ma  memory access, p  Probability of page fault (0≤ p ≤ 1)
o The memory access time denoted ma is in the range 10 to 200 ns. o
If there are no page faults then EAT = ma.
o To compute effective access time, we must know how much time is needed to
service a page fault.
o A page fault causes the following sequence to occur:
1. Trap to the OS
2. Save the user registers and process state.
3. Determine that the interrupt was a page fault.
4. Check whether the reference was legal and find the location of page on
disk.
5. Read the page from disk to free frame.
a. Wait in a queue until read request is serviced.
b. Wait for seek time and latency time.
c. Transfer the page from disk to free frame.
6. While waiting ,allocate CPU to some other user.
7. Interrupt from disk.
8. Save registers and process state for other users.
9. Determine that the interrupt was from disk.
7. Reset the page table to indicate that the page is now in memory.
8. Wait for CPU to be allocated to this process again.
9. Restart the instruction that was interrupted .

Process Creation
o Virtual memory enhances the performance of creating and running processes:
 - Copy-on-Write
- Memory-Mapped Files

a) Copy-on-Write

o fork() creates a child process as a duplicate of the parent process & it worked by
creating copy of the parent address space for child, duplicating the pages belonging to
the parent.
o Copy-on-Write (COW) allows both parent and child processes to initially share the
same pages in memory. These shared pages are marked as Copy-on-Write pages,
meaning that if either process modifies a shared page, a copy of the shared page is
created.
o vfork():

With this the parent process is suspended & the child process uses the address
space of the parent.

Because vfork() does not use Copy-on-Write, if the child process changes any
pages of the parent‘s address space, the altered pages will be visible to the parent
once it resumes. 

Therefore, vfork() must be used with caution, ensuring that the child process does
not modify the address space of the parent. 

(b)Memory – mapped files:

o Sequential read of a file on disk uses open() , read() and write()
o Every time a file is accessed it requires a system call and disk access.
o Alternative method: “Memory – mapped files”
 Allowing a part of virtual address space to be logically associated with file 
 Mapping a disk block to a page in memory. 

6.2 Page Replacement

o If no frames are free, we could find one that is not currently being used & free it.
o We can free a frame by writing its contents to swap space & changing the page
table to indicate that the page is no longer in memory.
o Then we can use that freed frame to hold the page for which the process faulted.

Basic Page Replacement

1. Find the location of the desired page on disk

2. Find a free frame
- If there is a free frame , then use it.
- If there is no free frame, use a page replacement algorithm to select a victim
 frame
- Write the victim page to the disk, change the page & frame tables accordingly.
3. Read the desired page into the (new) free frame. Update the page and frame tables.
4. Restart the process

If no frames are free, two page transfers are required & this situation effectively
doubles the page- fault service time.

Modify (dirty) bit:

o It indicates that any word or byte in the page is modified.
o When we select a page for replacement, we examine its modify bit.

If the bit is set, we know that the page has been modified & in this case
we must write that page to the disk. 

If the bit is not set, then if the copy of the page on the disk has not been
overwritten, then we can avoid writing the memory page on the disk as it
is already there. 
Page Replacement Algorithms

1. FIFO Page Replacement

2. Optimal Page Replacement
3. LRU Page Replacement
4. LRU Approximation Page Replacement
5. Counting-Based Page Replacement

o We evaluate an algorithm by running it on a particular string of memory references

& computing the number of page faults. The string of memory reference is called a
―reference string.
p The algorithm that provides less number of page faults is termed to be a good one.
o As the number of available frames increases , the number of page faults decreases.
 This is shown in the following graph:

(a) FIFO page replacement algorithm

o Replace the oldest page.

o This algorithm associates with each page ,the time when that page was brought in.

Example:
Reference string: 7,0,1,2,0,3,0,4,2,3,0,3,2,1,2,0,1,7,0,1
No.of available frames = 3 (3 pages can be in memory at a time per process)

No. of page faults = 15

Drawback:
o FIFO page replacement algorithm ‗s performance is not always good.
o To illustrate this, consider the following example:
Reference string: 1, 2, 3, 4, 1, 2, 5, 1, 2, 3, 4, 5
o If No.of available frames -= 3 then the no.of page faults =9
o If No.of available frames =4 then the no.of page faults =10
o Here the no. of page faults increases when the no.of frames increases .This is
called as Belady’s Anomaly.
(b) Optimal page replacement algorithm
o Replace the page that will not be used for the longest period of time.
Example:
Reference string: 7,0,1,2,0,3,0,4,2,3,0,3,2,1,2,0,1,7,0,1
No.of available frames = 3

No. of page faults = 9

Drawback:
o It is difficult to implement as it requires future knowledge of the reference string.

(c) LRU(Least Recently Used) page replacement algorithm

o Replace the page that has not been used for the longest period of time.
Example:
Reference string: 7,0,1,2,0,3,0,4,2,3,0,3,2,1,2,0,1,7,0,1
No.of available frames = 3

No. of page faults = 12

o LRU page replacement can be implemented using

1. Counters

Every page table entry has a time-of-use field and a clock or counter is
associated with the CPU. 
 
The counter or clock is incremented for every memory reference. 

Each time a page is referenced , copy the counter into the time-of-use
field. 

When a page needs to be replaced, replace the page with the smallest
counter value. 
2. Stack

Keep a stack of page numbers 

Whenever a page is referenced, remove the page from the stack and put
it on top of the stack. 

When a page needs to be replaced, replace the page that is at the bottom
of the stack.(LRU page)

(d) LRU Approximation Page Replacement

o Reference bit

With each page associate a reference bit, initially set to 0 

When page is referenced, the bit is set to 1 
o When a page needs to be replaced, replace the page whose reference bit is 0
o The order of use is not known , but we know which pages were used and which
were not used.

(i) Additional Reference Bits Algorithm

o Keep an 8-bit byte for each page in a table in memory.
o At regular intervals , a timer interrupt transfers control to OS.
o The OS shifts reference bit for each page into higher- order bit shifting the
other bits right 1 bit and discarding the lower-order bit.
Example:
o If reference bit is 00000000 then the page has not been used for 8 time
periods.
o If reference bit is 11111111 then the page has been used atleast once each
time period.
o If the reference bit of page 1 is 11000100 and page 2 is 01110111 then page
2 is the LRU page.
(ii) Second Chance Algorithm
o Basic algorithm is FIFO
o When a page has been selected , check its reference bit.

If 0 proceed to replace the page 

If 1 give the page a second chance and move on to the next FIFO
page.

When a page gets a second chance, its reference bit is cleared and
arrival time is reset to current time. 

Hence a second chance page will not be replaced until all other
 pages are replaced. 

(iii) Enhanced Second Chance Algorithm

o Consider both reference bit and modify bit
o There are four possible classes
1. (0,0) – neither recently used nor modified  Best page to replace

2. (0,1) – not recently used but modified  page has to be written out before
replacement.
3. (1,0) - recently used but not modified  page may be used again
4. (1,1) – recently used and modified  page may be used again and page has
to be written to disk

(e) Counting-Based Page Replacement

o Keep a counter of the number of references that have been made to each page
1. Least Frequently Used (LFU )Algorithm: replaces page with smallest
count
2. Most Frequently Used (MFU )Algorithm: replaces page with largest
count

It is based on the argument that the page with the smallest count was
probably just brought in and has yet to be used 

Page Buffering Algorithm

o These are used along with page replacement algorithms to improve their
performance
Technique 1:
o A pool of free frames is kept.
o When a page fault occurs, choose a victim frame as before.
o Read the desired page into a free frame from the pool
 o The victim frame is written onto the disk and then returned to the pool of free
frames.
Technique 2:
o Maintain a list of modified pages.
o Whenever the paging device is idles, a modified is selected and written to disk
and its modify bit is reset.
Technique 3:
o A pool of free frames is kept.
o Remember which page was in each frame.
o If frame contents are not modified then the old page can be reused directly from
the free frame pool when needed

6.3 Allocation of Frames

o There are two major allocation schemes

Equal Allocation 

Proportional Allocation
o Equal allocation 

If there are n processes and m frames then allocate m/n frames to each
process. 



Example: If there are 5 processes and 100 frames, give each process 20
frames. 

o Proportional allocation

Allocate according to the size of process 
Let si be the size of process i.
Let m be the total no. of frames
Then S = ∑ si
a i = si / S * m
where ai is the no.of frames allocated to process i.
Global vs. Local Replacement
o Global replacement – each process selects a replacement frame from the set of all
frames; one process can take a frame from another.
o Local replacement – each process selects from only its own set of allocated
frames.

6.4 Thrashing
o High paging activity is called thrashing.
o If a process does not have ―enough‖ pages, the page-fault rate is very high. This
leads to:

low CPU utilization 
 
operating system thinks that it needs to increase the degree of
multiprogramming 

another process is added to the system 
o When the CPU utilization is low, the OS increases the degree of
multiprogramming.
o If global replacement is used then as processes enter the main memory they tend to
steal frames belonging to other processes.
o Eventually all processes will not have enough frames and hence the page fault rate
becomes very high.
o Thus swapping in and swapping out of pages only takes place. o
This is the cause of thrashing.

o To limit thrashing, we can use a local replacement algorithm.

o To prevent thrashing, there are two methods namely ,

Working Set Strategy

Page Fault Frequency 


1. Working-Set Strategy
o It is based on the assumption of the model of locality.
o Locality is defined as the set of pages actively used together.
o Working set is the set of pages in the most recent  page references
o  is the working set window.


if  too small , it will not encompass entire locality 

if  too large ,it will encompass several localities 

if  =  it will encompass entire program 
o D =  WSSi

Where WSSi is the working set size for process i. 

D is the total demand of frames 
o if D > m then Thrashing will occur.
2. Page-Fault Frequency Scheme

o If actual rate too low, process loses frame

o If actual rate too high, process gains frame

Other Issues
o Prepaging

To reduce the large number of page faults that occurs at process startup 

Prepage all or some of the pages a process will need, before they are
referenced 

But if prepaged pages are unused, I/O and memory are wasted 
o Page Size
Page size selection must take into consideration:
o fragmentation
o table size
o I/O overhead
o locality
o TLB Reach
o
TLB Reach - The amount of memory accessible from the TLB 
o
TLB Reach = (TLB Size) X (Page Size) 
o
Ideally, the working set of each process is stored in the TLB. Otherwise there
is a high degree of page faults. 
o
Increase the Page Size. This may lead to an increase in fragmentation as not
all applications require a large page size 
o
Provide Multiple Page Sizes. This allows applications that require larger page
sizes the opportunity to use them without an increase in fragmentation. 
 o I/O interlock
o
Pages must sometimes be locked into memory 
o
Consider I/O. Pages that are used for copying a file from a device must be

locked from being selected for eviction by a page replacement

algorithm.


7. ALLOCATING KERNEL MEMORY

 The kernel is limited to about 1GB of virtual and physical memory.

 The kernel's memory is not pageable.
 The kernel usually wants physically contiguous memory.
 Often, the kernel must allocate the memory without sleeping.
 Mistakes in the kernel have a much higher price than they do elsewhere.

Although easy access to an abundance of memory certainly is not a luxury to the kernel, a
little understanding of the issues can go a long way toward making the process relatively
painless.

A General-Purpose Allocator

The general interface for allocating memory inside of the kernel is kmalloc():

#include <linux/slab.h>
void * kmalloc(size_t size, int flags);

example:

struct falcon *p;

p = kmalloc(sizeof (struct falcon), GFP_KERNEL);
if (!p)
/* the allocation failed - handle appropriately */

Flags
 The flags field controls the behavior of memory allocation.
 We can divide flags into three groups: action modifiers, zone modifiers and types.
 Action modifiers tell the kernel how to allocate memory.
 They specify, for example, whether the kernel can sleep (that is, whether the call
to kmalloc() can block) in order to satisfy the allocation. Zone modifiers, on the
other hand, tell the kernel from where the request should be satisfied.
 For example, some requests may need to be satisfied from memory that hardware
can access through direct memory access (DMA).
 Finally, type flags specify a type of allocation. They group together relevant action
and zone modifiers into a single mnemonic.
 In general, instead of specifying multiple action and zone modifiers, you specify a
single type flag.
 8. OS Examples

Windows XP
 Windows XP implements virtual memory using demand paging with clustering.
 Clustering handles page faults by bringing in not only the faulting page but also
several pages following the faulting page.
 When a process is first created, it is assigned a working-set minimum and
maximum.
 The working-set minimum is the minimum number of pages the process is
guaranteed to have in memory.
 If sufficient memory is available, a process may be assigned as many pages as its
working-set maximum.
 For most applications, the value of working-set minimum and working-set
maximum is 50 and 345 pages, respectively
 The virtual memory manager maintains a list of free page frames. Associated with
this list is a threshold value that is used to indicate whether sufficient free memory
is available.
 If a page fault occurs for a process that is below its working-set maximum, the
virtual memory manager allocates a page from this list of free pages.
 If a process is at its working-set maximum and it incurs a page fault, it must select
a page for replacement using a local page-replacement policy.
 When the amount of free memory falls below the threshold, the virtual memory
manager uses a tactic known as automatic working-set trimming to restore the
value above the threshold. Automatic working-set trimming works by evaluating
the number of pages allocated to processes.
 If a process has been allocated more pages than its working-set minimum, the
virtual memory manager removes pages until the process reaches its working-set
minimum.
 A process that is at its working-set minimum may be allocated pages from the
free-page frame list once sufficient free memory is available.
 The algorithm used to determine which page to remove from a working set
depends on the type of processor.
 On single-processor 80x86 systems, Windows XP uses a variation of the clock
algorithm discussed in Section 9.4.5.2
 On Alpha and, multiprocessor x86 systems, clearing the reference bit may require
invalidating the entry in the translation look-aside buffer on other processors.
Solaris
 In Solaris, when a thread incurs a page fault, the kernel assigns a page to the
faulting thread from the list of free pages it maintains.
 Therefore, it is imperative that the kernel keep a sufficient amount of free memory
available.
 Associated with this list of free pages is a parameter—lotsfree—that represents a
threshold to begin paging.
 The lotsfree parameter is typically set to 1/64 the size of the physical memory.
Four times per second, the kernel checks whether the amount of free memory is
less than lotsfree.
 If the number of free pages falls below lotsfree, a process known as the pageout
starts up.
 The pageout process works as follows: The front hand of the clock scans all pages
in memory, setting the reference bit to 0.
 Later, the back hand of the clock examines the reference bit for the pages in
memory, appending those pages whose bit is still set to 0 to the free list and
writing to disk their contents if modified.
 Solaris maintains a cache list of pages that have been "freed" but have not yet been
overwritten. The free list contains frames that have invalid contents.
 Pages can be reclaimed from the cache list if they are accessed before being
moved to the free list.