Virtual Memory

Dr. N.Nalini
SCOPE
VIT
 Background
• Code needs to be in memory to execute, but entire program rarely used
• Error code, unusual routines, large data structures
• Entire program code not needed at same time
• Consider ability to execute partially-loaded program
• Program no longer constrained by limits of physical memory
• Each program takes less memory while running -> more programs run at the
same time
• Increased CPU utilization and throughput with no increase in response
time or turnaround time
• Less I/O needed to load or swap programs into memory -> each user
program runs faster
 Virtual memory
• Virtual memory – separation of user logical memory from physical memory
• Only part of the program needs to be in memory for execution
• Logical address space can therefore be much larger than physical address
space
• Allows address spaces to be shared by several processes
• Allows for more efficient process creation
• More programs running concurrently
• Less I/O needed to load or swap processes
 Virtual memory (Cont.)
• Virtual address space – logical view of how process is stored in
memory
• Usually start at address 0, contiguous addresses until end of
space
• Meanwhile, physical memory organized in page frames
• MMU must map logical to physical
• Virtual memory can be implemented via:
• Demand paging
• Demand segmentation
Virtual Memory That is Larger Than Physical Memory
 Virtual-address Space
• Usually design logical address space for stack to start
at Max logical address and grow “down” while heap
grows “up”
• Maximizes address space use
• Unused address space between the two is hole
• No physical memory needed until heap or
stack grows to a given new page
• Enables sparse address spaces with holes left for
growth, dynamically linked libraries, etc.
• System libraries shared via mapping into virtual
address space
• Shared memory by mapping pages read-write into
virtual address space
• Pages can be shared during fork(), speeding
process creation
Shared Library Using Virtual Memory
 Demand Paging
• Could bring entire process into
memory at load time
• Or bring a page into memory only
when it is needed
– Less I/O needed, no
unnecessary I/O
– Less memory needed
– Faster response
– More users
• Similar to paging system with
swapping
 Basic Concepts
• With swapping, pager guesses which pages will be used before
swapping out again
• Instead, pager brings in only those pages into memory
• How to determine that set of pages?
• Need new MMU functionality to implement demand paging
• If pages needed are already memory resident
• No difference from non demand-paging
• If page needed and not memory resident
• Need to detect and load the page into memory from storage
• Without changing program behavior
• Without programmer needing to change code
 Valid-Invalid Bit
• With each page table entry a valid–invalid bit is associated
(v  in-memory – memory resident, i  not-in-memory)
• Initially valid–invalid bit is set to i on all entries
• Example of a page table snapshot:

• During MMU address translation, if valid–invalid bit in page table entry is i 

page fault
Page Table When Some Pages Are Not
in Main Memory
 Steps in Handling Page Fault
1. If there is a reference to a page, first reference to that page will trap
to operating system
• Page fault
2. Operating system looks at another table to decide:
• Invalid reference  abort
• Just not in memory
3. Find free frame
4. Swap page into frame via scheduled disk operation
5. Reset tables to indicate page now in memory
Set validation bit = v
6. Restart the instruction that caused the page fault
Steps in Handling a Page Fault (Cont.)
 Aspects of Demand Paging
• Extreme case – start process with no pages in memory
• OS sets instruction pointer to first instruction of process, non-memory-resident -> page
fault
• And for every other process pages on first access
• Pure demand paging
• Actually, a given instruction could access multiple pages -> multiple page faults
• Consider fetch and decode of instruction which adds 2 numbers from memory and stores
result back to memory
• Pain decreased because of locality of reference
• Hardware support needed for demand paging
• Page table with valid / invalid bit
• Secondary memory (swap device with swap space)
• Instruction restart
 Instruction Restart
• Consider an instruction that could access several different locations
– Block move

– Auto increment/decrement location

– Restart the whole operation?
• What if source and destination overlap?
 Free-Frame List
• When a page fault occurs, the operating system must bring the desired page
from secondary storage into main memory.
• Most operating systems maintain a free-frame list -- a pool of free frames for
satisfying such requests.

• Operating system typically allocate free frames using a technique known as

zero-fill-on-demand -- the content of the frames zeroed-out before being
allocated.
• When a system starts up, all available memory is placed on the free-frame list.
 Stages in Demand Paging – Worse Case
1. Trap to the operating system
2. Save the user registers and process state
3. Determine that the interrupt was a page fault
4. Check that the page reference was legal and determine the location of the page
on the disk
5. Issue a read from the disk to a free frame:
a) Wait in a queue for this device until the read request is serviced
b) Wait for the device seek and/or latency time
c) Begin the transfer of the page to a free frame
 Stages in Demand Paging (Cont.)
6. While waiting, allocate the CPU to some other user
7. Receive an interrupt from the disk I/O subsystem (I/O completed)
8. Save the registers and process state for the other user
9. Determine that the interrupt was from the disk
10. Correct the page table and other tables to show page is now in
memory
11. Wait for the CPU to be allocated to this process again
12. Restore the user registers, process state, and new page table, and then
resume the interrupted instruction
 Performance of Demand Paging
• Three major activities
– Service the interrupt – careful coding means just several hundred
instructions needed
– Read the page – lots of time
– Restart the process – again just a small amount of time
• Page Fault Rate 0  p  1
– if p = 0 no page faults
– if p = 1, every reference is a fault
• Effective Access Time (EAT)
EAT = (1 – p) x memory access
+ p (page fault overhead
+ swap page out
+ swap page in )
 Demand Paging Example
• Memory access time = 200 nanoseconds
• Average page-fault service time = 8 milliseconds
• EAT = (1 – p) x 200 + p (8 milliseconds)
= (1 – p x 200 + p x 8,000,000
= 200 + p x 7,999,800
• If one access out of 1,000 causes a page fault, then
EAT = 8.2 microseconds.
This is a slowdown by a factor of 40!!
• If want performance degradation < 10 percent
– 220 > 200 + 7,999,800 x p
20 > 7,999,800 x p
– p < .0000025
– < one page fault in every 400,000 memory accesses
 Demand Paging Optimizations
• Swap space I/O faster than file system I/O even if on the same device
• Swap allocated in larger chunks, less management needed than file system
• Copy entire process image to swap space at process load time
• Then page in and out of swap space
• Used in older BSD Unix
• Demand page in from program binary on disk, but discard rather than paging out when freeing
frame
• Used in Solaris and current BSD
• Still need to write to swap space
• Pages not associated with a file (like stack and heap) – anonymous memory
• Pages modified in memory but not yet written back to the file system
• Mobile systems
• Typically don’t support swapping
• Instead, demand page from file system and reclaim read-only pages (such as code)
 Copy-on-Write
• Copy-on-Write (COW) allows both parent and child processes to initially share the same pages in
memory
• If either process modifies a shared page, only then is the page copied
• COW allows more efficient process creation as only modified pages are copied
• In general, free pages are allocated from a pool of zero-fill-on-demand pages
• Pool should always have free frames for fast demand page execution
• Don’t want to have to free a frame as well as other processing on page fault
• Why zero-out a page before allocating it?
• vfork() variation on fork() system call has parent suspend and child using copy-on-write
address space of parent
• Designed to have child call exec()
• Very efficient
Before Process 1 Modifies Page C
After Process 1 Modifies Page C
 What Happens if There is no Free Frame?
• Used up by process pages
• Also in demand from the kernel, I/O buffers, etc
• How much to allocate to each?
• Page replacement – find some page in memory, but not
really in use, page it out
– Algorithm – terminate? swap out? replace the page?
– Performance – want an algorithm which will result in
minimum number of page faults
• Same page may be brought into memory several times
 Page Replacement
• Prevent over-allocation of memory by modifying page-fault service
routine to include page replacement
• Use modify (dirty) bit to reduce overhead of page transfers – only
modified pages are written to disk
• Page replacement completes separation between logical memory
and physical memory – large virtual memory can be provided on a
smaller physical memory
Need For Page Replacement
 Basic Page Replacement
1. Find the location of the desired page on disk
2. Find a free frame:
- If there is a free frame, use it
- If there is no free frame, use a page replacement algorithm to select a
victim frame
- Write victim frame to disk if dirty
3. Bring the desired page into the (newly) free frame; update the page and frame
tables
4. Continue the process by restarting the instruction that caused the trap

Note now potentially 2 page transfers for page fault – increasing EAT
Page Replacement
 Page and Frame Replacement Algorithms

• Frame-allocation algorithm determines

• How many frames to give each process
• Which frames to replace
• Page-replacement algorithm
• Want lowest page-fault rate on both first access and re-access
• Evaluate algorithm by running it on a particular string of memory references
(reference string) and computing the number of page faults on that string
• String is just page numbers, not full addresses
• Repeated access to the same page does not cause a page fault
• Results depend on number of frames available
• In all our examples, the reference string of referenced page numbers is
7,0,1,2,0,3,0,4,2,3,0,3,0,3,2,1,2,0,1,7,0,1
Graph of Page Faults Versus the Number of Frames
 First-In-First-Out (FIFO) Algorithm
• Reference string: 7,0,1,2,0,3,0,4,2,3,0,3,0,3,2,1,2,0,1,7,0,1
• 3 frames (3 pages can be in memory at a time per process)

15 page faults

• Giving more memory to a process would improve its performance (Increasing no.of frames)
– But, this assumption was not always true.
• Can vary by reference string: consider 1,2,3,4,1,2,5,1,2,3,4,5 find page fault if #frame=3 and 4 ?
• Adding more frames can cause more page faults!
• Belady’s Anomaly
• How to track ages of pages?
• Just use a FIFO queue
 FIFO Illustrating Belady’s Anomaly

1, 2, 3, 4, 1, 2, 5, 1, 2, 3, 4, 5 #frames=3, Page fault= 9

#frames=4, Page fault= 10
 Optimal Algorithm
• Replace page that will not be used for longest period of time
– 9 is optimal for the example
• How do you know this?
– Can’t read the future
• Used for measuring how well your algorithm performs
 Least Recently Used (LRU) Algorithm

• Use past knowledge rather than future

• Replace page that has not been used in the most amount of time
• Associate time of last use with each page

• 12 faults – better than FIFO but worse than OPT

• Generally good algorithm and frequently used
• But how to implement?
 LRU Algorithm (Cont.)
• Counter implementation
– Every page entry has a counter; every time page is referenced through this
entry, copy the clock into the counter
– When a page needs to be changed, look at the counters to find smallest
value
• Search through table needed
• Stack implementation
– Keep a stack of page numbers in a double link form:
– Page referenced:
• move it to the top
• requires 6 pointers to be changed
– But each update more expensive
– No search for replacement
 LRU Algorithm (Cont.)
• LRU and OPT are cases of stack algorithms that don’t have Belady’s
Anomaly
• Use Of A Stack to Record Most Recent Page References
 LRU Approximation Algorithms
• LRU needs special hardware and still slow
• Reference bit
• With each page associate a bit, initially = 0
• When page is referenced bit set to 1
• Replace any with reference bit = 0 (if one exists)
• We do not know the order, however
 LRU Approximation Algorithms (cont.)

• Second-chance algorithm
• Generally FIFO, plus hardware-provided reference
bit
• Clock replacement
• If page to be replaced has
• Reference bit = 0 -> replace it
• reference bit = 1 then:
– set reference bit 0, leave page in memory
– replace next page, subject to same rules
Second-chance Algorithm
Second-chance Algorithm
Enhanced Second-Chance Algorithm
 Counting Algorithms
• Keep a counter of the number of references that have been
made to each page
• Not common
• Lease Frequently Used (LFU) Algorithm:
• Replaces page with smallest count
• Most Frequently Used (MFU) Algorithm:
• Based on the argument that the page with the smallest count
was probably just brought in and has yet to be used
 Page-Buffering Algorithms
• Keep a pool of free frames, always
– Then frame available when needed, not found at fault time
– Read page into free frame and select victim to evict and add to free pool
– When convenient, evict victim
• Possibly, keep list of modified pages
– When backing store otherwise idle, write pages there and set to non-dirty
• Possibly, keep free frame contents intact and note what is in them
– If referenced again before reused, no need to load contents again from disk
– Generally useful to reduce penalty if wrong victim frame selected
 Applications and Page Replacement
• All of these algorithms have OS guessing about future page access
• Some applications have better knowledge – i.e. databases
• Memory intensive applications can cause double buffering
• OS keeps copy of page in memory as I/O buffer
• Application keeps page in memory for its own work
• Operating system can given direct access to the disk, getting out of
the way of the applications
• Raw disk mode
• Bypasses buffering, locking, etc.
 Allocation of Frames
• Each process needs minimum number of frames
• Example: IBM 370 – 6 pages to handle SS MOVE instruction:
– instruction is 6 bytes, might span 2 pages
– 2 pages to handle from
– 2 pages to handle to
• Maximum of course is total frames in the system
• Two major allocation schemes
– fixed allocation
– priority allocation
• Many variations
 Fixed Allocation
• Equal allocation – if there are 93 frames and 5 processes, each
process will get 18 frames. The 3 leftover frames can be used as a
free-frame buffer pool.
• Proportional allocation – Allocate according to the size of process
– Dynamic as degree of multiprogramming, process sizes change
m  64
si  size of process pi s1  10( size of process1)
S   si s2  127( size of process 2)
m  total number of frames a1 
10
 62  4
s 137
ai  allocation for pi  i  m 127
S a2   62  57
137
 Global vs. Local Allocation
• Global replacement – process selects a replacement frame from the
set of all frames; one process can take a frame from another
(replacement frames with another process/outside frames)
• But then process execution time can vary greatly
• But greater throughput so more common
• Local replacement – each process selects from only its own set of
allocated frames (replacement frames with its own process frames)
• More consistent per-process performance
• But possibly underutilized memory
 Reclaiming Pages
• A strategy to implement global page-replacement policy
• All memory requests are satisfied from the free-frame list,
rather than waiting for the list to drop to zero before we
begin selecting pages for replacement,
• Page replacement is triggered when the list falls below a
certain threshold.
• This strategy attempts to ensure there is always sufficient
free memory to satisfy new requests.
Reclaiming Pages Example
 Non-Uniform Memory Access
• So far, we assumed that all memory accessed equally
• Many systems are NUMA – speed of access to memory varies
• Consider system boards containing CPUs and memory, interconnected
over a system bus
• NUMA multiprocessing architecture
 Non-Uniform Memory Access (Cont.)
• Optimal performance comes from allocating memory “close to” the CPU on
which the thread is scheduled
• And modifying the scheduler to schedule the thread on the same system
board when possible
• Solved by Solaris by creating lgroups
• Structure to track CPU / Memory low latency groups
• Used my schedule and pager
• When possible schedule all threads of a process and allocate all
memory for that process within the lgroup
 Thrashing
• If a process does not have “enough” pages, the page-fault rate is
very high
– Page fault to get page
– Replace existing frame
– But quickly need replaced frame back
– This leads to:
• Low CPU utilization
• Operating system thinking that it needs to increase the
degree of multiprogramming
• Another process added to the system
 Thrashing (Cont.)
• Thrashing. A process is busy swapping pages in and out.
A process is thrashing if it is spending more time paging
than executing.
 Demand Paging and Thrashing
• Why does demand paging work?

Locality model
• Process migrates from one locality to another
• Localities may overlap
• Why does thrashing occur?

 size of locality > total memory size

• Limit effects by using local or priority page replacement

Locality In A Memory-Reference Pattern
 Working-Set Model
•   working-set window  a fixed number of page references
Example: 10,000 instructions
• WSSi (working set of Process Pi) = total number of pages
referenced in the most recent  (varies in time)
– if  too small will not encompass entire locality
– if  too large will encompass several localities
– if  =   will encompass entire program
• D =  WSSi  total demand frames
– Approximation of locality
 Working-Set Model (Cont.)
• if D > m  Thrashing
• Policy if D > m, then suspend or swap out
one of the processes
 Keeping Track of the Working Set
• Approximate with interval timer + a reference bit
• Example:  = 10,000
– Timer interrupts after every 5000 time units
– Keep in memory 2 bits for each page
– Whenever a timer interrupts copy and sets the values of all
reference bits to 0
– If one of the bits in memory = 1  page in working set
• Why is this not completely accurate?
• Improvement = 10 bits and interrupt every 1000 time units
 Page-Fault Frequency
• More direct approach than WSS
• Establish “acceptable” page-fault frequency (PFF) rate and use
local replacement policy
• If actual rate too low, process loses frame
• If actual rate too high, process gains frame
 Working Sets and Page Fault Rates
 Direct relationship between working set of a process and its page-fault rate
 Working set changes over time
 Peaks and valleys over time
 Allocating Kernel Memory
• Treated differently from user memory
• Often allocated from a free-memory pool
– Kernel requests memory for structures of varying
sizes
– Some kernel memory needs to be contiguous
• i.e., for device I/O
 Buddy System
• Allocates memory from fixed-size segment consisting of physically-contiguous pages
• Memory allocated using power-of-2 allocator
• Satisfies requests in units sized as power of 2
• Request rounded up to next highest power of 2
• When smaller allocation needed than is available, current chunk split into two buddies of
next-lower power of 2
• Continue until appropriate sized chunk available
• For example, assume 256KB chunk available, kernel requests 21KB
• Split into AL and AR of 128KB each
• One further divided into BL and BR of 64KB
– One further into CL and CR of 32KB each – one used to satisfy request
• Advantage – quickly coalesce unused chunks into larger chunk
• Disadvantage - fragmentation
Buddy System Allocator
 Slab Allocator
• Alternate strategy
• Slab is one or more physically contiguous pages
• Cache consists of one or more slabs
• Single cache for each unique kernel data structure
• Each cache filled with objects – instantiations of the data
structure
• When cache created, filled with objects marked as free
• When structures stored, objects marked as used
• If slab is full of used objects, next object allocated from empty slab
• If no empty slabs, new slab allocated
• Benefits include no fragmentation, fast memory request satisfaction
Slab Allocation
 Slab Allocator in Linux
• For example process descriptor is of type struct task_struct
• Approx 1.7KB of memory
• New task -> allocate new struct from cache
– Will use existing free struct task_struct
• Slab can be in three possible states
1. Full – all used
2. Empty – all free
3. Partial – mix of free and used
• Upon request, slab allocator
1. Uses free struct in partial slab
2. If none, takes one from empty slab
3. If no empty slab, create new empty
 Slab Allocator in Linux (Cont.)
• Slab started in Solaris, now wide-spread for both kernel mode and
user memory in various OSes
• Linux 2.2 had SLAB, now has both SLOB and SLUB allocators
– SLOB for systems with limited memory
• Simple List of Blocks – maintains 3 list objects for small,
medium, large objects
– SLUB is performance-optimized SLAB removes per-CPU queues,
metadata stored in page structure
 Other Considerations
• Prepaging
• Page size
• TLB reach
• Inverted page table
• Program structure
• I/O interlock and page locking
 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 was wasted
• Assume s pages are prepaged and α of the pages is used
– Is cost of s * α save pages faults > or < than the cost of prepaging
s * (1- α) unnecessary pages?
– α near zero  prepaging loses
 Page Size
• Sometimes OS designers have a choice
• Especially if running on custom-built CPU
• Page size selection must take into consideration:
• Fragmentation
• Page table size
• Resolution
• I/O overhead
• Number of page faults
• Locality
• TLB size and effectiveness
• Always power of 2, usually in the range 212 (4,096 bytes) to 222 (4,194,304 bytes)
• On average, growing over time
 TLB Reach
• TLB Reach - The amount of memory accessible from the TLB
• TLB Reach = (TLB Size) X (Page Size)
• Ideally, the working set of each process is stored in the TLB
– Otherwise there is a high degree of page faults
• Increase the Page Size
– This may lead to an increase in fragmentation as not all applications require
a large page size
• Provide Multiple Page Sizes
– This allows applications that require larger page sizes the opportunity to use
them without an increase in fragmentation
 Program Structure
• Program structure
– int[128,128] data;
– Each row is stored in one page
– Program 1
for (j = 0; j <128; j++)
for (i = 0; i < 128; i++)
data[i,j] = 0;

128 x 128 = 16,384 page faults

– Program 2
for (i = 0; i < 128; i++)
for (j = 0; j < 128; j++)
data[i,j] = 0;

128 page faults

 I/O interlock
• I/O Interlock – Pages must
sometimes be locked into
memory
• 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
• Pinning of pages to lock into
memory
 Operating System Examples
• Windows

• Solaris
 Windows
• Uses demand paging with clustering. Clustering brings in pages surrounding
the faulting page
• Processes are assigned working set minimum and working set maximum
• Working set minimum is the minimum number of pages the process is
guaranteed to have in memory
• A process may be assigned as many pages up to its working set maximum
• When the amount of free memory in the system falls below a threshold,
automatic working set trimming is performed to restore the amount of
free memory
• Working set trimming removes pages from processes that have pages in
excess of their working set minimum
 Solaris
• Maintains a list of free pages to assign faulting processes
• Lotsfree – threshold parameter (amount of free memory) to begin paging
• Desfree – threshold parameter to increasing paging
• Minfree – threshold parameter to being swapping
• Paging is performed by pageout process
• Pageout scans pages using modified clock algorithm
• Scanrate is the rate at which pages are scanned. This ranges from slowscan
to fastscan
• Pageout is called more frequently depending upon the amount of free
memory available
• Priority paging gives priority to process code pages
Solaris 2 Page Scanner