• For each process, let r = T/t; where T is the wall clock time this process has been in system and t is the running time of the process to date.
• If r=5, that means the job has been running 1/5 of the time it has been in the system.
• We call r the penalty ratio and run the process having the highest r value.
• HPRN is normally defined to be non-preemptive (i.e., the system only checks r when a burst ends), but there is an preemptive analogue
  • Do not worry about a process that just enters the system (its ratio is undefined)
  • When putting process into the run state compute the time at which it will no longer have the highest ratio and set a timer.
  • When a process is moved into the ready state, compute its ratio and preempt if needed.
• HRN stands for highest response ratio next and means the same thing.
• This policy is yet another example of priority scheduling

Multilevel Queues (**, **, MLQ, **)

Put different classes of processs in different queues

• Processs do not move from one queue to another.
• Can have different policies on the different queues.
  For example, might have a background (batch) queue that is FCFS and one or more foreground queues that are RR.
• Must also have a policy among the queues.
  For example, might have two queues, foreground and background, and give the first absolute priority over the second
  • Might apply aging to prevent background starvation.
  • But might not, i.e., no guarantee of service for background processes. View a background process as a ``cycle soaker''.
  • Might have 3 queues, foreground, background, cycle soaker

Multilevel Feedback Queues (FB, MFQ, MLFBQ, MQ)

Many queues and processs move from queue to queue in an attempt to dynamically separate ``batch-like'' from interactive processs so that we can favor the latter.

• Run processs from the highest priority nonempty queue in a RR manner.
• When a process uses its full quanta (looks a like batch process), move it to a lower priority queue.
• When a process doesn't use a full quanta (looks like an interactive process), move it to a higher priority queue.
• A long process with frequent (perhaps spurious) I/O will remain in the upper queues.
• Might have the bottom queue FCFS.
• Many variants
  For example, might let process stay in top queue 1 quantum, next queue 2 quanta, next queue 4 quanta (i.e. return a process to the rear of the same queue it was in if the quantum expires).

Theoretical Issues

Considerable theory has been developed.

• NP completeness results abound.
• Much work in queuing theory to predict performance.
• Not covered in this course.

Medium-Term Scheduling

In addition to the short-term scheduling we have discussed, we add medium-term scheduling in which decisions are made at a coarser time scale.

• Called memory scheduling by Tanenbaum (part of three level scheduling).
• Suspend (swap out) some process if memory is over-committed.
• Criteria for choosing a victim.
  • How long since previously suspended.
  • How much CPU time used recently.
  • How much memory does it use.
  • External priority (pay more, get swapped out less).
• We will discuss medium term scheduling again when we study memory management.

Long Term Scheduling

• ``Job scheduling''. Decide when to start jobs, i.e., do not necessarily start them when submitted.
• Force user to log out and/or block logins if over-committed.
  • CTSS (an early time sharing system at MIT) did this to insure decent interactive response time.
  • Unix does this if out of processes (i.e., out of PTEs).
  • ``LEM jobs during the day'' (Grumman).
• Called admission scheduling by Tanenbaum (part of three level scheduling).
• Many supercomputer sites.

2.5.4: Scheduling in Real Time Systems

Skipped

2.5.5: Policy versus Mechanism

Skipped.

2.5.6: Thread Scheduling

Skipped.

Research on Processes and Threads

Skipped.

Notes on lab (scheduling)

1. If several processes are waiting on I/O, you may assume noninterference. For example, assume that on cycle 100 process A flips a coin and decides its wait is 6 units (i.e., during cycles 101-106 A will be blocked. Assume B begins running at cycle 101 for a burst of 1 cycle. So during 101 process B flips a coin and decides its wait is 3 units. You do NOT alter process A. That is, Process A will become ready after cycle 106 (100+6) so enters the ready list cycle 107 and process B becomes ready after cycle 104 (101+3) and enters ready list cycle 105.
   
   
2. For processor sharing (PS), which is part of the extra credit:
   PS (processor sharing). Every cycle you see how many jobs are ready (or running). Say there are 7. Then during this cycle (an exception will be described below) each process gets 1/7 of a cycle.
   EXCEPTION: Assume there are exactly 2 ready jobs, one needs 1/3 cycle and one needs 1/2 cycle. The process needing only 1/3 gets only 1/3, i.e. it is finished after 2/3 cycle. So the other process gets 1/3 cycle during the first 2/3 cycle and then starts to get all the CPU. Hence it finishes after 2/3 + 1/6 = 5/6 cycle. The last 1/6 cycle is not used by any process.

Chapter 3: Deadlocks

A deadlock occurs when every member of a set of processes is waiting for an event that can only be caused by a member of the set.

Often the event waited for is the release of a resource.

In the automotive world deadlocks are called gridlocks.

• The processes are the cars.
• The resources are the spaces occupied by the cars

Reward: One point extra credit on the final exam for anyone who brings a real (e.g., newspaper) picture of an automotive deadlock. You must bring the clipping to the final and it must be in good condition. Hand it in with your exam paper. Note that it must really be a gridlock, i.e., motion is not possible without breaking the traffic rules. A huge traffic jam is not sufficient.

For a computer science example consider two processes A and B that each want to print a file currently on tape.

1. A has obtained ownership of the printer and will release it after printing one file.
2. B has obtained ownership of the tape drive and will release it after reading one file.
3. A tries to get ownership of the tape drive, but is told to wait for B to release it.
4. B tries to get ownership of the printer, but is told to wait for A to release the printer.

Bingo: deadlock!

3.1: Resources:

The resource is the object granted to a process.

3.1.1: Preemptable and Nonpreemptable Resourses

• Resources come in two types
  1. Preemptable, meaning that the resource can be taken away from its current owner (and given back later). An example is memory.
  2. Non-preemptable, meaning that the resource cannot be taken away. An example is a printer.
• The interesting issues arise with non-preemptable resources so those are the ones we study.
• Life history of a resource is a sequence of
  1. Request
  2. Allocate
  3. Use
  4. Release
• Processes make requests, use the resourse, and release the resourse. The allocate decisions are made by the system and we will study policies used to make these decisions.

3.1.2: Resourse Acquisition

Simple example of the trouble you can get into.

• Two resources and two processes.
• Each process wants both resources.
• Use a semaphore for each. Call them S and T.
• If both processes execute P(S); P(T); --- V(T); V(S)
  all is well.
• But if one executes instead P(T); P(S); -- V(S); V(T)
  disaster! This was the printer/tape example just above.

Recall from the semaphore/critical-section treatment last chapter, that it is easy to cause trouble if a process dies or stays forever inside its critical section. Similarly, we assume that no process maintains a resource forever. It may obtain the resource an unbounded number of times (i.e. it can have a loop forever with a resource request inside), but each time it gets the resource, it must release it eventually.

3.2: Introduction to Deadlocks

To repeat: A deadlock occurs when a every member of a set of processes is waiting for an event that can only be caused by a member of the set.

Often the event waited for is the release of a resource.

3.2.1: (Necessary) Conditions for Deadlock

The following four conditions (Coffman; Havender) are necessary but not sufficient for deadlock. Repeat: They are not sufficient.

1. Mutual exclusion: A resource can be assigned to at most one process at a time (no sharing).
2. Hold and wait: A processing holding a resource is permitted to request another.
3. No preemption: A process must release its resources; they cannot be taken away.
4. Circular wait: There must be a chain of processes such that each member of the chain is waiting for a resource held by the next member of the chain.

The first three are characteristics of the system and resources. For a given system fixed set of resource they are either true or false, i.e., they don't change with time. The truth or falsehood of the last condition does indeed change with time as the resources are requested/allocated/released.

3.2.2: Deadlock Modeling

On the right is the Resource Allocation Graph, also called the Reusable Resource Graph.

• The processes are circles.
• The resources are squares.
• An arc (directed line) from a process P to a resource R signifies that process P has requested (but not yet been allocated) resource R.
• An arc from a resource R to a process P indicates that process P has been allocated resource R.

Homework: 5.

Consider two concurrent processes P1 and P2 whose programs are.

P1: request R1       P2: request R2
    request R2           request R1
    release R2           release R1
    release R1           release R2

On the board draw the resource allocation graph for various possible executions of the processes, indicating when deadlock occurs and when deadlock is no longer avoidable.

There are four strategies used for dealing with deadlocks.

1. Ignore the problem
2. Detect deadlocks and recover from them
3. Avoid deadlocks by carefully deciding when to allocate resources.
4. Prevent deadlocks by violating one of the 4 necessary conditions.

3.3: Ignoring the problem--The Ostrich Algorithm

The ``put your head in the sand approach''.

• If the likelihood of a deadlock is sufficiently small and the cost of avoiding a deadlock is sufficiently high it might be better to ignore the problem. For example if each PC deadlocks once per 100 years, the one reboot may be less painful that the restrictions needed to prevent it.
• Clearly not a good philosophy for nuclear missile launchers.
• For embedded systems (e.g., missile launchers) the programs run are fixed in advance so many of the questions Tanenbaum raises (such as many processes wanting to fork at the same time) don't occur.

3.4: Detecting Deadlocks and Recovering From Them

3.4.1: Detecting Deadlocks with Single Unit Resources

Consider the case in which there is only one instance of each resource.

• So a request can be satisfied by only one specific resource.
• In this case the 4 necessary conditions for deadlock are also sufficient.
• Remember we are making an assumption (single unit resources) that is often invalid. For example, many systems have several printers and a request is given for ``a printer'' not a specific printer. Similarly, one can have many tape drives.
• So the problem comes down to finding a directed cycle in the resource allocation graph. Why?
  Answer: Because the other three conditions are either satisfied by the system we are studying or are not in which case deadlock is not a question. That is, conditions 1,2,3 are conditions on the system in general not on what is happening right now.

To find a directed cycle in a directed graph is not hard. The algorithm is in the book. The idea is simple.

1. For each node in the graph do a depth first traversal (hoping the graph is a DAG (directed acyclic graph), building a list as you go down the DAG.
2. If you ever find the same node twice on your list, you have found a directed cycle and the graph is not a DAG and deadlock exists among the processes in your current list.
3. If you never find the same node twice, the graph is a DAG and no deadlock occurs.
4. The searches are finite since the list size is bounded by the number of nodes.

3.4.2: Detecting Deadlocks with Multiple Unit Resources

This is more difficult.

• The figure on the right shows a resource allocation graph with multiple unit resources.
• Each unit is represented by a dot in the box.
• Request edges are drawn to the box since they represent a request for any dot in the box.
• Allocation edges are drawn from the dot to represent that this unit of the resource has been assigned (but all units of a resource are equivalent and the choice of which one to assign is arbitrary).
• Note that there is a directed cycle in black, but there is no deadlock. Indeed the middle process might finish, erasing the magenta arc and permitting the blue dot to satisfy the rightmost process.
• The book gives an algorithm for detecting deadlocks in this more general setting. The idea is as follows.
  1. look for a process that might be able to terminate (i.e., all its request arcs can be satisfied).
  2. If one is found pretend that it does terminate (erase all its arcs), and repeat step 1.
  3. If any processes remain, they are deadlocked.
• We will soon do in detail an algorithm (the Banker's algorithm) that has some of this flavor.

3.4.3: Recovery from deadlock

Preemption

Perhaps you can temporarily preempt a resource from a process. Not likely.

Rollback

Database (and other) systems take periodic checkpoints. If the system does take checkpoints, one can roll back to a checkpoint whenever a deadlock is detected. Somehow must guarantee forward progress.

Kill processes

Can always be done but might be painful. For example some processes have had effects that can't be simply undone. Print, launch a missile, etc.

Remark: We are doing 3.6 before 3.5 since 3.6 is easier.

3.6: Deadlock Prevention

Attack one of the coffman/havender conditions

3.6.1: Attacking Mutual Exclusion

Idea is to use spooling instead of mutual exclusion. Not possible for many kinds of resources

3.6.2: Attacking Hold and Wait

Require each processes to request all resources at the beginning of the run. This is often called One Shot.

3.6.3: Attacking No Preempt

Normally not possible.

3.6.4: Attacking Circular Wait

Establish a fixed ordering of the resources and require that they be requested in this order. So if a process holds resources #34 and #54, it can request only resources #55 and higher.

It is easy to see that a cycle is no longer possible.

Homework: 7.

3.5: Deadlock Avoidance

Let's see if we can tiptoe through the tulips and avoid deadlock states even though our system does permit all four of the necessary conditions for deadlock.

An optimistic resource manager is one that grants every request as soon as it can. To avoid deadlocks with all four conditions present, the manager must be smart not optimistic.

3.5.1 Resource Trajectories

We plot progress of each process along an axis. In the example we show, there are two processes, hence two axes, i.e., planar. This procedure assumes that we know the entire request and release pattern of the processes in advance so it is not a practical solution. I present it as it is some motivation for the practical solution that follows, the Banker's Algorithm.

• We have two processes H (horizontal) and V.
• The origin represents them both starting.
• Their combined state is a point on the graph.
• The parts where the printer and plotter are needed by each process are indicated.
• The dark green is where both processes have the plotter and hence execution cannot reach this point.
• Light green represents both having the printer; also impossible.
• Pink is both having both printer and plotter; impossible.
• Gold is possible (H has plotter, V has printer), but you can't get there.
• The upper right corner is the goal; both processes have finished.
• The red dot is ... (cymbals) deadlock. We don't want to go there.
• The cyan is safe. From anywhere in the cyan we have horizontal and vertical moves to the finish point (the upper right corner) without hitting any impossible area.
• The magenta interior is very interesting. It is
  • Possible: each processor has a different resource
  • Not deadlocked: each processor can move within the magenta
  • Deadly: deadlock is unavoidable. You will hit a magenta-green boundary and then will no choice but to turn and go to the red dot.
• The cyan-magenta border is the danger zone.
• The dashed line represents a possible execution pattern.
• With a uniprocessor no diagonals are possible. We either move to the right meaning H is executing or move up indicating V is executing.
• The trajectory shown represents.
  1. H excuting a little.
  2. V excuting a little.
  3. H executes; requests the printer; gets it; executes some more.
  4. V executes; requests the plotter.
• The crisis is at hand!
• If the resource manager gives V the plotter, the magenta has been entered and all is lost. ``Abandon all hope ye who enter here'' --Dante.
• The right thing to do is to deny the request, let H execute moving horizontally under the magenta and dark green. At the end of the dark green, no danger remains, both processes will complete successfully. Victory!
• This procedure is not practical for a general purpose OS since it requires knowing the programs in advance. That is, the resource manager, knows in advance what requests each process will make and in what order.

Homework: 10, 11, 12.

3.5.2: Safe States

Avoiding deadlocks given some extra knowledge.

• Not surprisingly, the resource manager knows how many units of each resource it had to begin with.
• Also it knows how many units of each resource it has given to each process.
• It would be great to see all the programs in advance and thus know all future requests, but that is asking for too much.
• Instead, when each process when it starts, it gives its maximum usage. That is each process at startup states, for each resource, the maximum number of units it can possibly ask for. This is called the claim of the process.
  • If during the run the process asks for more than its claim, abort it.
  • If it claims more than it needs, the result is that the resource manager will be more conservative than need be and there will be more waiting.

Give an example of all four possibilities. A state that is

1. Safe and deadlocked--not possible
2. Safe and not deadlocked
3. Not safe and deadlocked
4. Not safe and not deadlocked--interesting

A manager can determine if a state is safe.

• Since the manager know all the claims, it can determine the maximum amount of additional resources each process can request.
• The manager knows how many units of each resource it has left.

The manager then follows the following procedure, which is part of Banker's Algorithms discovered by Dijkstra, to determine if the state is safe.

1. If there are no processes remaining, the state is safe.
   
   
2. Seek a process P whose max additional requests is less than what remains (for each resource type).
   • If no such process can be found, then the state is not safe.
   • The banker (manager) knows that if it refuses all requests excepts those from P, then it will be able to satisfy all of P's requests. Why?
     Ans: Look at how P was chosen.
   
   
3. The banker now pretends that P has terminated (since the banker knows that it can guarantee this will happen). Hence the banker pretends that all of P's currently held resources are returned. This makes the banker richer and hence perhaps a process that was not eligible to be chosen as P previously, can now be chosen.
   
   
4. Repeat these steps.

Example 1

• One resource type R with 22 unit
• Three processes X, Y, and Z with claims 3, 11, and 19 respectively.
• Currently the processes have 1, 5, and 10 units respectively.
• Hence the manager currently has 6 units left.
• Also note that the max additional needs for processes are 2, 6, 9
• So the manager cannot assure (with its current remaining supply of 6 units) that Z can terminate. But that is not the question.
• This state is safe
  1. Use 2 units to satisfy X; now the manager has 7 units.
  2. Use 6 units to satisfy Y; now the manager has 12 units.
  3. Use 9 units to satisfy Z; done!

Example 2

Start with example 1 and assume that Z now requests 2 units and we grant them.

• Currently the processes have 1, 5, and 12 units respectively.
• The manager has 4 units.
• The max additional needs are 2, 6, and 7.
• This state is unsafe
  1. Use 2 unit to satisfy X; now the manager has 5 units.
  2. Y needs 6 and Z needs 7 so we can't guarantee satisfying either
• Note that we were able to find a process that can terminate (X) but then we were stuck. So it is not enough to find one process. Must find a sequence of all the processes.

3.5.3: The Banker's Algorithm (Dijkstra) for a Single Resource

The algorithm is simple: Stay in safe states. Initially, we assume all the processes are present before execution begins and that all claims are given before execution begins. We will relax these assumptions very soon.

• Before execution begins, check that the system is safe. That is, check that no process claims more than the manager has). If not, then the offending process is trying to claim more than the system has so cannot be guaranteed to complete even if run by itself. You might say that it can become deadlocked all by itself.
• When the manager receives a request, it pretends to grant it and checks if the resulting state is safe. If it is safe the request is granted, if not the process is blocked.
• When a resource is returned, the manager (politely thanks the process and then) checks to see if any of the pending requests can be granted (i.e., if the result would now be safe). If so the request is granted and the manager checks to see if another pending request can be granted, etc..

Homework: 13.

3.5.4: The Banker's Algorithm for Multiple Resources

At a high level the algorithm is identical: Stay in safe states.

• What is a safe state?
• The same definition (if processes are run in a certain order they will all terminate).
• Checking for safety is the same idea as above. The difference is that to tell if there enough free resources for a processes to terminate, the manager must check that for all resources, the number of free units is at least equal to the max additional need of the process.

Limitations of the banker's algorithm

• Often users don't know the maximum requests a process will make. They can estimate conservatively (i.e., use big numbers for the claim) but then the manager becomes very conservative.
• New processes arriving cause a problem (but not so bad as Tanenbaum suggests).
  • The process's claim must be less than the total number of units of the resource in the system. If not, the process is not accepted by the manager.
  • Since the state without the new process is safe, so is the state with the new process! Just use the order you had originally and put the new process at the end.
  • Insuring fairness (starvation freedom) needs a little more work, but isn't too hard either (once an hour stop taking new processes until all current processes finish).
• A resource becoming unavailable (e.g., a tape drive breaking), can result in an unsafe state.

Homework: 21, 27, and 20. There is an interesting typo in 20 (2nd edition of book): A has claimed 3 units of resource 5, but there are only 2 units in the entire system. Change A's claim to 2.

3.7: Other Issues

3.7.1: Two-phase locking

This is covered (MUCH better) in a database text. We will skip it.

3.7.2: Non-resource deadlocks

You can get deadlock from semaphores as well as resources. This is trivial. Semaphores can be considered resources. P(S) is request S and V(S) is release S. The manager is the module implementing P and V. When the manager returns from P(S), it has granted the resource S.

3.7.3: Starvation

As usual FCFS is a good cure. Often this is done by priority aging and picking the highest priority process to get the resource. Also can periodically stop accepting new processes until all old ones get their resources.

3.8: Research on Deadlocks

Skipped.

3.9: Summary

Read.

Chapter 4: Memory Management

Also called storage management or space management.

Memory management must deal with the storage hierarchy present in modern machines.

• Registers, cache, central memory, disk, tape (backup)
• Move data from level to level of the hierarchy.
• How should we decide when to move data up to a higher level?
  • Fetch on demand (e.g. demand paging, which is dominant now).
  • Prefetch
    • Read-ahead for file I/O.
    • Large cache lines and pages.
    • Extreme example. Entire job present whenever running.

We will see in the next few lectures that there are three independent decision:

1. Segmentation (or no segmentation)
2. Paging (or no paging)
3. Fetch on demand (or no fetching on demand)

Memory management implements address translation.

• Convert virtual addresses to physical addresses
  • Also called logical to real address translation.
  • A virtual address is the address expressed in the program.
  • A physical address is the address understood by the computer hardware.
• The translation from virtual to physical addresses is performed by the Memory Management Unit or (MMU).
• Another example of address translation is the conversion of relative addresses to absolute addresses by the linker.
• The translation might be trivial (e.g., the identity) but not in a modern general purpose OS.
• The translation might be difficult (i.e., slow).
  • Often includes addition/shifts/mask--not too bad.
  • Often includes memory references.
    • VERY serious.
    • Solution is to cache translations in a Translation Lookaside Buffer (TLB). Sometimes called a translation buffer (TB).

Homework: 6.

When is address translation performed?

1. At compile time
   • Compiler generates physical addresses.
   • Requires knowledge of where the compilation unit will be loaded.
   • No linker.
   • Loader is trivial.
   • Primitive.
   • Rarely used (MSDOS .COM files).
   
   
2. At link-edit time (the ``linker lab'')
   • Compiler
     • Generates relocatable addresses for each compilation unit.
     • References external addresses.
   • Linkage editor
     • Converts the relocatable addr to absolute.
     • Resolves external references.
     • Misnamed ld by unix.
     • Also converts virtual to physical addresses by knowing where the linked program will be loaded. Linker lab ``does'' this, but it is trivial since we assume the linked program will be loaded at 0.
   • Loader is still trivial.
   • Hardware requirements are small.
   • A program can be loaded only where specified and cannot move once loaded.
   • Not used much any more.
   
   
3. At load time
   • Similar to at link-edit time, but do not fix the starting address.
   • Program can be loaded anywhere.
   • Program can move but cannot be split.
   • Need modest hardware: base/limit registers.
   • Loader sets the base/limit registers.
   
   
4. At execution time
   • Addresses translated dynamically during execution.
   • Hardware needed to perform the virtual to physical address translation quickly.
   • Currently dominates.
   • Much more information later.

Extensions

• Dynamic Loading
  • When executing a call, check if module is loaded.
  • If not loaded, call linking loader to load it and update tables.
  • Slows down calls (indirection) unless you rewrite code dynamically.
  • Not used much.
• Dynamic Linking
  • The traditional linking described above is today often called static linking.
  • With dynamic linking, frequently used routines are not linked into the program. Instead, just a stub is linked.
  • When the routine is called, the stub checks to see if the real routine is loaded (it may have been loaded by another program).
    • If not loaded, load it.
    • If already loaded, share it. This needs some OS help so that different jobs sharing the library don't overwrite each other's private memory.
  • Advantages of dynamic linking.
    • Saves space: Routine only in memory once even when used many times.
    • Bug fix to dynamically linked library fixes all applications that use that library, without having to relink the application.
  • Disadvantages of dynamic linking.
    • New bugs in dynamically linked library infect all applications.
    • Applications ``change'' even when they haven't changed.

4.1: Basic Memory Management (Without Swapping or Paging)

Entire process remains in memory from start to finish and does not move.

The sum of the memory requirements of all jobs in the system cannot exceed the size of physical memory.

** 4.1.1: Monoprogramming without swapping or paging (Single User)

The ``good old days'' when everything was easy.

• No address translation done by the OS (i.e., address translation is not performed dynamically during execution).
• Either reload the OS for each job (or don't have an OS, which is almost the same), or protect the OS from the job.
  • One way to protect (part of) the OS is to have it in ROM.
  • Of course, must have the OS (read-write) data in ram.
  • Can have a separate OS address space only accessible in supervisor mode.
  • Might just put some drivers in ROM (BIOS).
• The user employs overlays if the memory needed by a job exceeds the size of physical memory.
  • Programmer breaks program into pieces.
  • A ``root'' piece is always memory resident.
  • The root contains calls to load and unload various pieces.
  • Programmer's responsibility to ensure that a piece is already loaded when it is called.
  • No longer used, but we couldn't have gotten to the moon in the 60s without it (I think).
  • Overlays have been replaced by dynamic address translation and other features (e.g., demand paging) that have the system support logical address sizes greater than physical address sizes.
  • Fred Brooks (leader of IBM's OS/360 project and author of ``The mythical man month'') remarked that the OS/360 linkage editor was terrific, especially in its support for overlays, but by the time it came out, overlays were no longer used.

**4.1.2: Multiprogramming with fixed partitions

Two goals of multiprogramming are to improve CPU utilization, by overlapping CPU and I/O and to permit short jobs to finish quickly.

• This scheme was used by IBM for system 360 OS/MFT (multiprogramming with a fixed number of tasks).
• Can have a single input queue instead of one for each partition.
  • So that if there are no big jobs can use big partition for little jobs.
  • But I don't think IBM did this.
  • Can think of the input queue(s) as the ready list(s) with a scheduling policy of FCFS in each partition.
• The partition boundaries are not movable (must reboot to move a job).
  • MFT can have large internal fragmentation, i.e., wasted space inside a region
• Each process has a single ``segment'' (we will discuss segments later)
  • No sharing between process.
  • No dynamic address translation.
  • At load time must ``establish addressability''.
    • i.e. must set a base register to the location at which the process was loaded (the bottom of the partition).
    • The base register is part of the programmer visible register set.
    • This is an example of address translation during load time.
    • Also called relocation.
• Storage keys are adequate for protection (IBM method).
• Alternative protection method is base/limit registers.
• An advantage of base/limit is that it is easier to move a job.
• But MFT didn't move jobs so this disadvantage of storage keys is moot.
• Tanenbaum says a job was ``run until it terminates. This must be wrong as that would mean monoprogramming.
• He probably means that jobs not swapped out and each queue is FCFS without preemption.

4.1.3: Modeling Multiprogramming

• Consider a job that is unable to compute (i.e., it is waiting for I/O) a fraction p of the time.
• Then, with monoprogramming, the CPU utilization is 1-p.
• Note that p is often > .5 so CPU utilization is poor.
• But, if the probability that a job is waiting for I/O is p and n jobs are in memory, then the probability that all n are waiting for I/O is approximately p^n.
• So, with a multiprogramming level (MPL) of n, the CPU utilization is approximately 1-p^n.
• If p=.5 and n=4, then 1-p^n = 15/16, which is much better than 1/2, which would occur for monoprogramming (n=1).
• This is a crude model, but it is correct that increasing MPL does increase CPU utilization up to a point.
• The limitation is memory, which is why we discuss it here instead of process management. That is, we must have many jobs loaded at once, which means we must have enough memory for them. There are other issues as well and we will discuss them.
• Some of the CPU utilization is time spent in the OS executing context switches so the gains are not a great as the crude model predicts.

Homework: 1, 2 (typo in book; figure 4.21 seems irrelevant).

4.1.4: Analysis of Multiprogramming System Performance

Skipped

4.1.5: Relocation and Protection

Relocation was discussed as part of linker lab and at the beginning of this chapter. When done dynamically, a simple method is to have a base register whose value is added to every address by the hardware.

Similarly a limit register is checked by the hardware to be sure that the address (before the base register is added) is not bigger than the size of the program.

The base and limit register are set by the OS when the job starts.

4.2: Swapping

Moving entire processes between disk and memory is called swapping.

Multiprogramming with Variable Partitions

• IBM OS/MVT (multiprogramming with a varying number of tasks).
• Also early PDP-10 OS.
• Job still has only one segment (as with MFT) but now can be of any size up to the size of the machine and can change with time.
• A single ready list.
• A job can move (might be swapped back in a different place).
• This is dynamic address translation (during run time).
• Must perform an addition on every memory reference (i.e. on every address translation) to add the start address of the partition.
• Called a DAT (dynamic address translation) box by IBM.
• Eliminates internal fragmentation.
  • Find a region the exact right size (leave a hole for the remainder).
  • Not quite true, can't get a piece with 10A755 bytes. Would get say 10A760. But internal fragmentation is much reduced compared to MFT. Indeed, we say that internal fragmentation has been eliminated.
• Introduces external fragmentation, i.e., holes outside any region.
• What do you do if no hole is big enough for the request?
  • Can compactify
    • Transition from bar 3 to bar 4 in diagram below.
    • This is expensive.
    • Not suitable for real time (MIT ping pong).
  • Can swap out one process to bring in another, e.g., bars 5-6 and 6-7 in the diagram.
• There are more processes than holes. Why?
  • Because next to a process there might be a process or a hole but next to a hole there must be a process
  • So can have ``runs'' of processes but not of holes
  • If after a process equally likely to have a process or a hole, you get about twice as many processes as holes.
• Base and limit registers are used.
  • Storage keys not good since compactifying or moving would require changing many keys.
  • Storage keys might need a fine granularity to permit the boundaries move by small amounts. Hence many keys would need to be changed

Homework: 3

MVT Introduces the ``Placement Question''

That is, which hole (partition) should one choose?

• Best fit, worst fit, first fit, circular first fit, quick fit, Buddy
  • Best fit doesn't waste big holes, but does leave slivers and is expensive to run.
  • Worst fit avoids slivers, but eliminates all big holes so a big job will require compaction. Even more expensive than best fit (best fit stops if it finds a perfect fit).
  • Quick fit keeps lists of some common sizes (but has other problems, see Tanenbaum).
  • Buddy system
    • Round request to next highest power of two (causes internal fragmentation).
    • Look in list of blocks this size (as with quick fit).
    • If list empty, go higher and split into buddies.
    • When returning coalesce with buddy.
    • Do splitting and coalescing recursively, i.e. keep coalescing until can't and keep splitting until successful.
    • See Tanenbaum for more details (or an algorithms book).
• A current favorite is circular first fit, also know as next fit.
  • Use the first hole that is big enough (first fit) but start looking where you left off last time.
  • Doesn't waste time constantly trying to use small holes that have failed before, but does tend to use many of the big holes, which can be a problem.
• Buddy comes with its own implementation. How about the others?

4.2.1: Memory Management with Bitmaps

Divide memory into blocks and associate a bit with each block, used to indicate if the corresponding block is free or allocated. To find a chunk of size N blocks need to find N consecutive bits indicating a free block.

The only design question is how much memory does one bit represent.

• Big: Serious internal fragmentation.
• Small: Many bits to store and process.

4.2.2: Memory Management with Linked Lists

• Each item on list gives the length and starting location of the corresponding region of memory and says whether it is a Hole or Process.
• The items on the list are not taken from the memory to be used by processes.
• Keep in order of starting address.
• Merge adjacent holes
• Double linked

Memory Management using Boundary Tags

• Use the same memory for list items as for processes
• Don't need an entry in linked list for blocks in use, just the avail blocks are linked.
• For the blocks currently in use, just need a hole/process bit at each end and the length. Keep this in the block itself.
• See Knuth, The Art of Computer Programming vol 1.

Homework: 5.

MVT also introduces the ``Replacement Question''

That is, which victim should we swap out? Note that this is an example of the suspend arc mentioned in process scheduling.

We will study this question more when we discuss demand paging in which case we swap out part of a process.

Considerations in choosing a victim

• Cannot replace a job that is pinned, i.e. whose memory is tied down. For example, if Direct Memory Access (DMA) I/O is scheduled for this process, the job is pinned until the DMA is complete.
• Victim selection is a medium term scheduling decision
  • A job that has been in a wait state for a long time is a good candidate.
  • Often choose as a victim a job that has been in memory for a long time.
  • Another question is how long should it stay swapped out.
• For demand paging, where swaping out a page is not as drastic as swapping out a job, choosing the victim is an important memory management decision and we shall study several policies,

1. So far the schemes presented so far have had two properties:
   1. Each job is stored contiguously in memory. That is, the job is contiguous in physical addresses.
   2. Each job cannot use more memory than exists in the system. That is, the virtual addresses space cannot exceed the physical address space.
   
   
2. Tanenbaum now attacks the second item. I wish to do both and start with the first.
   
   
3. Tanenbaum (and most of the world) uses the term ``paging'' to mean what I call demand paging. This is unfortunate as it mixes together two concepts.
   1. Paging (dicing the address space) to solve the placement problem and essentially eliminate external fragmentation.
   2. Demand fetching, to permit the total memory requirements of all loaded jobs to exceed the size of physical memory.
   
   
4. Tanenbaum (and most of the world) uses the term virtual memory as a synonym for demand paging. Again I consider this unfortunate.
   1. Demand paging is a fine term and is quite descriptive.
   2. Virtual memory ``should'' be used in contrast with physical memory to describe any virtual to physical address translation.

** (non-demand) Paging

Simplest scheme to remove the requirement of contiguous physical memory.

• Chop the program into fixed size pieces called pages (invisible to the programmer).
• Chop the real memory into fixed size pieces called page frames or simply frames.
• Size of a page (the page size) = size of a frame (the frame size).
• Sprinkle the pages into the frames.
• Keep a table (called the page table) having an entry for each page. The page table entry or PTE for page p contains the number of the frame f that contains page p.

Properties of (non-demand) paging.

• Entire job must be memory resident to run.
• No holes, i.e. no external fragmentation.
• If there are 50 frames available and the page size is 4KB than a job requiring <= 200KB will fit, even if the available frames are scattered over memory.
• Hence (non-demand) paging is useful.
• Introduces internal fragmentation approximately equal to 1/2 the page size for every process (really every segment).
• Can have a job unable to run due to insufficient memory and have some (but not enough) memory available. This is not called external fragmentation since it is not due to memory being fragmented.
• Eliminates the placement question. All pages are equally good since don't have external fragmentation.
• The replacement question remains.
• Since page boundaries occur at ``random'' points and can change from run to run (the page size can change with no effect on the program--other than performance), pages are not appropriate units of memory to use for protection and sharing. This is discussed further when we introduce segmentation.

Homework: 16.

Address translation

• Each memory reference turns into 2 memory references
  1. Reference the page table
  2. Reference central memory
• This would be a disaster!
• Hence the MMU caches page#-->frame# translations. This cache is kept near the processor and can be accessed rapidly.
• This cache is called a translation lookaside buffer (TLB) or translation buffer (TB).
• For the above example, after referencing virtual address 3372, there would be an entry in the TLB containing the mapping 3-->459.
• Hence a subsequent access to virtual address 3881 would be translated to physical address 459881 without a memory reference.

Choice of page size is discuss below.

Homework: 8.

4.3: Virtual Memory (meaning fetch on demand)

Idea is that a program can execute even if only the active portion of its address space is memory resident. That is, swap in and swap out portions of a program. In a crude sense this can be called ``automatic overlays''.

Advantages

• Can run a program larger than the total physical memory.
• Can increase the multiprogramming level since the total size of the active, i.e. loaded, programs (running + ready + blocked) can exceed the size of the physical memory.
• Since some portions of a program are rarely if ever used, it is an inefficient use of memory to have them loaded all the time. Fetch on demand will not load them if not used and will unload them during replacement if they are not used for a long time (hopefully).

4.3.1: Paging (meaning demand paging)

Fetch pages from disk to memory when they are referenced, with a hope of getting the most actively used pages in memory.

• Very common: dominates modern operating systems.
• Started by the Atlas system at Manchester University in the 60s (Fortheringham).
• Each PTE continues to have the frame number if the page is loaded.
• But what if the page is not loaded (exists only on disk)?
  • The PTE has a flag indicating if it is loaded (can think of the X in the diagram on the right as indicating that this flag is not set).
  • If not loaded, the location on disk could be kept in the PTE (not shown in the diagram). But normally it is not (discussed below).
  • When a reference is made to a non-loaded page (sometimes called a non-existent page, but that is a bad name), the system has a lot of work to do. We give more details below.
    1. Choose a free frame, if one exists.
    2. If not
       1. Choose a victim frame.
          • More later on how to choose a victim.
          • Called the replacement question
       2. Write victim back to disk if dirty,
       3. Update the victim PTE to show that it is not loaded and, perhaps, where on disk it has been put.
    3. Copy the referenced page from disk to the free frame.
    4. Update the PTE of the referenced page to show that it is loaded and give the frame number.
    5. Do the standard paging address translation (p#,off)-->(f#,off).
• Really not done quite this way
  • There is ``always'' a free frame because ...
  • There is a deamon active that checks the number of free frames and if this is too low, chooses victims and ``pages them out'' (writing them back to disk if dirty).
• Choice of page size is discussed below.

Homework: 12.

4.3.2: Page tables

A discussion of page tables is also appropriate for (non-demand) paging, but the issues are more acute with demand paging since the tables can be much larger. Why?

1. The total size of the active processes is no longer limited to the size of physical memory. Since the total size of the processes is greater, the total size of the page tables is greater and hence concerns over the size of the page table are more acute.
2. With demand paging an important question is the choice of a victim page to page out. Data in the page table can be useful in this choice.

We must be able access to the page table very quickly since it is needed for every memory access.

Unfortunate laws of hardware.

• Big and fast are essentially incompatible.
• Big and fast and low cost is hopeless.

So we can't just say, put the page table in fast processor registers, and let it be huge, and sell the system for $1500.

For now, put the (one-level) page table in main memory.

• Seems too slow since all memory references require two reference.
• TLB very helpful to reduce the average delay as mentioned above (discussed later in more detail).
• The page table might be too big.
  • Currently we are considering contiguous virtual addresses ranges (i.e. the virtual addresses have no holes).
  • Typically put the stack at one end of virtual address and the global (or static) data at the other end and let them grow towards each other.
  • The memory in between is unused.
  • This unused memory can be huge (in address range) and hence the page table will mostly contain unneeded PTEs.
• Works fine if the maximum virtual address size is small, which was once true (e.g., the PDP-11 of the 1970s) but is no longer the case.

Contents of a PTE

Each page has a corresponding page table entry (PTE). The information in a PTE is for use by the hardware. Information set by and used by the OS is normally kept in other OS tables. The page table format is determined by the hardware so access routines are not portable. The following fields are often present.

1. The valid bit. This tells if the page is currently loaded (i.e., is in a frame). If set, the frame pointer is valid. It is also called the presence or presence/absence bit. If a page is accessed with the valid bit zero, a page fault is generated by the hardware.
2. The frame number. This is the main reason for the table. It is needed for virtual to physical address translation.
3. The Modified bit. Indicates that some part of the page has been written since it was loaded. This is needed when the page is evicted so the OS can know that the page must be written back to disk.
4. The referenced bit. Indicates that some word in the page has been referenced. Used to select a victim: unreferenced pages make good victims by the locality property (discussed below).
5. Protection bits. For example one can mark text pages as execute only. This requires that boundaries between regions with different protection are on page boundaries. Normally many consecutive (in logical address) pages have the same protection so many page protection bits are redundant. Protection is more naturally done with segmentation.

Multilevel page tables (Not on 202 final exam)

Recall the previous diagram. Most of the virtual memory is the unused space between the data and stack regions. However, with demand paging this space does not waste real memory. But the single large page table does waste real memory.

The idea of multi-level page tables (a similar idea is used in Unix i-node-based file systems) is to add a level of indirection and have a page table containing pointers to page tables.

• Imagine one big page table.
• Call it the second level page table and cut it into pieces each the size of a page.
  Note that you can get many PTEs in one page so you will have far fewer of these pages than PTEs
• Now construct a first level page table containing PTEs that point to these pages.
• This first level PT is small enough to store in memory.
• But since we still have the 2nd level PT, we have made the world bigger not smaller!
• Don't store in memory those 2nd level page tables all of whose PTEs refer to unused memory. That is use demand paging on the (second level) page table

• For a two level page table the virtual address is divided into three pieces

  +-----+-----+-------+
  | P#1 | P#2 | Offset|
  +-----+-----+-------+
  

• P#1 gives the index into the first level page table.
• Follow the pointer in the corresponding PTE to reach the frame containing the relevant 2nd level page table.
• P#2 gives the index into this 2nd level page table
• Follow the pointer in the corresponding PTE to reach the frame containing the (originally) requested frame.
• Offset gives the offset in this frame where the requested word is located.

Do an example on the board

The VAX used a 2-level page table structure, but with some wrinkles (see Tanenbaum for details).

Naturally, there is no need to stop at 2 levels. In fact the SPARC has 3 levels and the Motorola 68030 has 4 (and the number of bits of Virtual Address used for P#1, P#2, P#3, and P#4 can be varied).

4.3.3: TLBs--Translation Lookaside Buffers (and General Associative Memory)

An associative memory is a content addressable memory. That is you access the memory by giving the value of some field and the hardware searches all the records and returns the record whose field contains the requested value.

For example

Name  | Animal | Mood     | Color
======+========+==========+======
Moris | Cat    | Finicky  | Grey
Fido  | Dog    | Friendly | Black
Izzy  | Iguana | Quiet    | Brown
Bud   | Frog   | Smashed  | Green

Izzy  | Iguana | Quiet    | Brown

A Translation Lookaside Buffer or TLB is an associate memory where the index field is the page number. The other fields include the frame number, dirty bit, valid bit, and others.

• A TLB is small and expensive but at least it is fast. When the page number is in the TLB, the frame number is returned very quickly.
• On a miss, the page number is looked up in the page table. The record found is placed in the TLB and a victim is discarded. There is no placement question since all entries are accessed at the same time. But there is a replacement question.

Homework: 17.

4.3.4: Inverted page tables

Keep a table indexed by frame number with the entry f containing the number of the page currently loaded in frame f.

• Since modern machine have a smaller physical address space than virtual address space, the table is smaller
• But on a TLB miss, must search the inverted page table.
• Would be hopelessly slow except that some tricks are employed.
• The book mentions some but not all of the tricks, we are skipping this topic.

4.4: Page Replacement Algorithms (PRAs)

These are solutions to the replacement question.

Good solutions take advantage of locality.

• Temporal locality: If a word is referenced now, it is likely to be referenced in the near future.
  • This argues for caching referenced words, i.e. keeping the referenced word near the processor for a while.
• Spatial locality: If a word is referenced now, nearby words are likely to be referenced in the near future.
  • This argues for prefetching words around the currently referenced word.
• These are lumped together into locality: If any word in a page is referenced, each word in the page is ``likely'' to be referenced.
  • So it is good to bring in the entire page on a miss and to keep the page in memory for a while.
• When programs begin there is no history so nothing to base locality on. At this point the paging system is said to be undergoing a ``cold start''.
• Programs exhibit ``phase changes'', when the set of pages referenced changes abruptly (similar to a cold start). At the point of a phase change, many page faults occur because locality is poor.

Pages belonging to processes that have terminated are of course perfect choices for victims.

Pages belonging to processes that have been blocked for a long time are good choices as well.

Random PRA

A lower bound on performance. Any decent scheme should do better.

4.4.1: The optimal page replacement algorithm (opt PRA) (aka Belady's min PRA)

Replace the page whose next reference will be furthest in the future.

• Also called Belady's min algorithm.
• Provably optimal. That is, generates the fewest number of page faults.
• Unimplementable: Requires predicting the future.
• Good upper bound on performance.

4.4.2: The not recently used (NRU) PRA

Divide the frames into four classes and make a random selection from the lowest nonempty class.

0. Not referenced, not modified
1. Not referenced, modified
2. Referenced, not modified
3. Referenced, modified

Assumes that in each PTE there are two extra flags R (sometimes called U, for used) and M (often called D, for dirty).

Also assumes that a page in a lower priority class is cheaper to evict.

• If not referenced, probably will not referenced again soon and hence is a good candidate for eviction.
• If not modified, do not have to write it out so the cost of the eviction is lower.
• When a page is brought in, OS resets R and M (i.e. R=M=0)
• On a read, hardware sets R.
• On a write, hardware sets R and M.

We again have the prisoner problem, we do a good job of making little ones out of big ones, but not the reverse. Need more resets.

Every k clock ticks, reset all R bits

• Why not reset M?
  Answer: Must have M accurate to know if victim needs to be written back
• Could have two M bits one accurate and one reset, but I don't know of any system (or proposal) that does so.

What if the hardware doesn't set these bits?

• OS can use tricks
• When the bits are reset, make the PTE indicate the page is not resident (i.e. lie). On the page fault, set the appropriate bit(s).

4.4.3: FIFO PRA

Simple but poor since usage of the page is ignored.

Belady's Anomaly: Can have more frames yet generate more faults. Example given later.

4.4.4: Second chance PRA

Similar to the FIFO PRA but when time choosing a victim, if the page at the head of the queue has been referenced (R bit set), don't evict it. Instead reset R and move the page to the rear of the queue (so it looks new). The page is being a second chance.

What if all frames have been referenced?
Becomes the same as fifo (but takes longer).

Might want to turn off the R bit more often (say every k clock ticks).

4.4.5: Clock PRA

Same algorithm as 2nd chance, but a better (and I would say obvious) implementation: Use a circular list.

Do an example.

LIFO PRA

This is terrible! Why?
Ans: All but the last frame are frozen once loaded so you can replace only one frame. This is especially bad after a phase shift in the program when it is using all new pages.

4.4.6:Least Recently Used (LRU) PRA

When a page fault occurs, choose as victim that page that has been unused for the longest time, i.e. that has been least recently used.

LRU is definitely

• Implementable: The past is knowable.
• Good: Simulation studies have shown this.
• Difficult. Essentially need to either:
  1. Keep a time stamp in each PTE, updated on each reference and scan all the PTEs when choosing a victim to find the PTE with the oldest timestamp.
  2. Keep the PTEs in a linked list in usage order, which means on each reference moving the PTE to the end of the list

Homework: 29, 23

A hardware cutsie in Tanenbaum

• For n pages, keep an nxn bit matrix.
• On a reference to page i, set row i to all 1s and col i to all 0s
• At any time the 1 bits in the rows are ordered by inclusion. I.e. one row's 1s are a subset of another row's 1s, which is a subset of a third. (Tanenbaum forgets to mention this.)
• So the row with the fewest 1s is a subset of all the others and is hence least recently used
• Cute, but still impractical.

4.4.7: Simulating (Approximating) LRU in Software

The Not Frequently Used (NFU) PRA

• Include a counter in each PTE (and have R in each PTE).
• Set counter to zero when page is brought into memory.
• For each PTE, every k clock ticks.
  1. Add R to counter.
  2. Clear R.
• Choose as victim the PTE with lowest count.

The Aging PRA

NFU doesn't distinguish between old references and recent ones. The following modification does distinguish.

• Include a counter in each PTE (and have R in each PTE).
• Set counter to zero when page is brought into memory.
• For each PTE, every k clock ticks.
  1. Shift counter right one bit.
  2. Insert R as new high order bit (HOB).
  3. Clear R.
• Choose as victim the PTE with lowest count.

Homework: 25, 34

4.4.8: The Working Set Page Replacement Problem (Peter Denning)

The working set policy (Peter Denning)

The goal is to specify which pages a given process needs to have memory resident in order for the give process to run without too many page faults.

• But this is impossible since it requires predicting the future.
• So we make the assumption that the immediate future is well approximated by the immediate past.
• Measure time in units of memory references, so t=1045 means the time when the 1045th memory reference is issued.
• In fact we measure time separately for each process, so t=1045 really means the time when this process made its 1045th memory reference.
• W(t,&omega) is the set of pages referenced (by the given process) from time t-ω to time t.
• That is, W(t,ω) is the set pages referenced during the window of size ω ending at time t.
• That is, W(t,ω) is the set of pages referenced by the last ω memory references ending at reference t.
• W(t,ω) is called the working set at time t (with window ω).
• Does this Netscape support the ω notation to give the Greek letter?
• w(t,ω) is the size of the set W(t,ω), i.e. is the number of pages referenced in the window.

The idea of the working set policy is to ensure that each process keeps its working set in memory.

• Allocate w(t,ω) frames to each process. This number differs for each process and changes with time.
• On a fault, one replaces a page not in the working set. But it is not easy to find such a page quickly.
• Indeed determining W(t,ω) is difficult.
• We will see that the working set algorithm is essentially a ``global policy'' (defined below). I would actually prefer covering the working set policy after defining local and global policies but decided to follow Tannenbaum.
• If a process is suspended, it is often swapped out; the working set then can be used to say which pages should be brought back when the process is resumed.

Interesting questions include:

• What value should be used for ω?
  Experiments have been done and ω is surprisingly robust (i.e., for a given system a fixed value works reasonably for a wide variety of job mixes)
• How should we calculate W(t,ω)?
  Hard so do exactly so ...

... Various approximations to the working set, have been devised. We will study three: using virtual time instead of memory references (immediately below), WSClock (section 4.4.9), and Page Fault Frequency (section 4.6).

Using virtual time

Approximate the working set as those pages referenced during the last m milliseconds. Then clear the reference bit every m milliseconds and set it on every reference. Note that the time is measured only while this process is running. That is why it is called virtual time. So now to choose a victim, we need to find a page with the R bit clear. Similar to NRU.

4.4.9: The WSClock Page Replacement Algorithm

• Use the aging algorithm above to maintain a counter for each PTE and declare a page whose counter is above a certain threshold to be part of the working set.
• Apply the clock algorithm globally (i.e. to all pages) but refuse to page out any page in a working set. The resulting algorithm is called wsclock.
• What if we find there are no pages we can page out?
  Simple answer: Pick some page (almost at random).
  Another answer: Reduce the multiprogramming level (explained in 4.6 below).

4.4.10: Summary of Page Replacement Algorithms

4.5: Modeling Paging Algorithms

4.5.1: Belady's anomaly

Consider a system that has no pages loaded and that uses the FIFO PRU.
Consider the following ``reference string'' (sequences of pages referenced).

 0 1 2 3 0 1 4 0 1 2 3 4

If we have 3 frames this generates 9 page faults (do it).

If we have 4 frames this generates 10 page faults (do it).

Theory has been developed and certain PRA (so called ``stack algorithms'') cannot suffer this anomaly for any reference string. FIFO is clearly not a stack algorithm. LRU is. Tannenbaum has a few details, but we are skipping it.

Repeat the above calculations for LRU.

4.6: Design issues for (demand) Paging Systems

4.6.1: Local vs Global Allocation Policies

A local PRA is one is which a victim page is chosen among the pages of the same process that requires a new page. That is the number of pages for each process is fixed. So LRU means the page least recently used by this process.

• Of course we can't have a purely local policy, why?
  Answer: A new process has no pages and even if we didn't apply this for the first page loaded, the process would remain with only one page.
• Perhaps wait until a process has been running a while or give the process an initial allocation based on the size of the executable.
• A global policy is one in which the choice of victim is made among all pages of all processes.

If we apply global LRU indiscriminately with some sort of RR processor scheduling policy, and memory is somewhat over-committed, then by the time we get around to a process, all the others have run and have probably paged out this process.

If this happens each process will need to page fault at a high rate; this is called thrashing.

It is therefore important to get a good idea of how many pages a process needs, so that we can balance the local and global desires. The working set W(t,ω) is good for this.

An approximation to the working set policy that is useful for determining how many frames a process needs (but not which pages) is the Page Fault Frequency (PFF) algorithm.

• For each process keep track of the page fault frequency, which is the number of faults divided by the number of references.
• Actually, must use a window or a weighted calculation since you are really interested in the recent page fault frequency.
• If the PFF is too high, allocate more frames to this process. Either
  1. Raise its number of frames and use a local policy; or
  2. Bar its frames from eviction (for a while) and use a global policy.
• What if there are not enough frames?
  Answer: Reduce the MPL (see next section).

As mentioned above a question arises what to do if the sum of the working set sizes exceeds the amount of physical memory available. This question is similar to the final point about PFF and brings us to consider controlling the load (or memory pressure).

4.6.2: Load Control

To reduce the overall memory pressure, we must reduce the multiprogramming level (or install more memory while the system is running, which is hardly practical). That is, we have a connection between memory management and process management. This is the suspend/resume arcs we saw way back when.

4.6.3: Page size

• Page size ``must'' be a multiple of the disk block size. Why?
  Answer: When copying out a page if you have a partial disk block, you must do a read/modify/write (i.e., 2 I/Os).
• Important property of I/O that we will learn later this term is that eight I/Os each 1KB takes considerably longer than one 8KB I/O
• Characteristics of a large page size.
  • Good for user I/O.
    • If I/O done using physical addresses, then I/O crossing a page boundary is not contiguous and hence requires multiple I/Os
    • If I/O uses virtual addresses, then page size doesn't effect this aspect of I/O. That is the addresses are contiguous in virtual address and hence one I/O is done.
  • Good for demand paging I/O.
    • Better to swap in/out one big page than several small pages.
    • But if page is too big you will be swapping in data that is really not local and hence might well not be used.
  • Large internal fragmentation (1/2 page size).
  • Small page table.
  • A very large page size leads to very few pages. Process will have many faults if using demand paging and the process frequently references more regions than frames.
• A small page size has the opposite characteristics.

4.6.4: Separate Instruction and Data (I and D) Spaces

Skipped.

4.6.5: Shared pages

Permit several processes to each have a page loaded in the same frame. Of course this can only be done if the processes are using the same program and/or data.

• Really should share segments.
• Must keep reference counts or something so that when a process terminates, pages (even dirty pages) it shares with another process are not automatically discarded.
• Similarly, a reference count would make a widely shared page (correctly) look like a poor choice for a victim.
• A good place to store the reference count would be in a structure pointed to by both PTEs. If stored in the PTEs, must keep them consistent between processes.

4.6.6: Cleaning Policy (Paging Daemons)

Done earlier

4.6.7: Virtual Memory Interface

Skipped.

4.7: Implementation Issues

4.7.1: Operating System Involvement with Paging

4.7.2: Page Fault Handling

1. The hardware detects the fault and traps to the kernel (switches to supervisor mode and saves state).
   
   
2. Some assembly language code save more state, establishes the C-language (or another programming language) environment, and ``calls'' the OS.
   
   
3. The OS determines that a page fault occurred and which page was referenced.
   
   
4. If the virtual address is invalid, process A is killed. If the virtual address is valid, the OS must find a free frame. If there is no free frames, the OS selects a victim frame. Call the process owning the victim frame, process B. (If the page replacement algorithm is local, the victim is process A.)
   
   
5. The PTE of the victim page is updated to show that the page is no longer resident.
   
   
6. If the victim page is dirty, the OS schedules an I/O write to copy the frame to disk and blocks A waiting for this I/O to occur.
   
   
7. Since process A is blocked, the scheduler is invoked to perform a context switch.
   
   
   • Tanenbaum ``forgot'' some here.
   • The process selected by the scheduler (say process C) runs.
   • Perhaps C is preempted for D or perhaps C blocks and D runs and then perhaps D is blocked and E runs, etc.
   • When the I/O to write the victim frame completes, a Disk interrupt occurs. Assume processes C is running at the time.
   • Hardware trap / assembly code / OS determines I/O done.
   • The scheduler picks a process to run, maybe A, maybe B, maybe C, maybe another processes.
   • At some point the scheduler does pick process A to run. Recall that at this point A is still executing OS code.
   
   
8. Now the O/S has a clean frame (this may be much later in wall clock time if a victim frame had to be written). The O/S schedules an I/O to read the desired page into this clean frame. Process A is blocked (perhaps for the second time) and hence the process scheduler is invoked to perform a context switch.
   
   
9. A Disk interrupt occurs when the I/O completes (trap / asm / OS determines I/O done). The PTE in process A is updated to indicate that the page is in memory.
   
   
10. The O/S may need to fix up process A (e.g. reset the program counter to re-execute the instruction that caused the page fault).
    
    
11. Process A is placed on the ready list and eventually is chosen by the scheduler to run. Recall that process A is executing O/S code.
    
    
12. The OS returns to the first assembly language routine.
    
    
13. The assembly language routine restores registers, etc. and ``returns'' to user mode.

The user's program running as process A is unaware that all this happened (except for the time delay).

4.7.3: Instruction Backup

A cute horror story. The 68000 was so bad in this regard that early demand paging systems for the 68000, used two processors one running one instruction behind. If the first got a page fault, there wasn't always enough information to figure out what to do so the system switched to the second processor after it did the page fault. Don't worry about instruction backup. Very machine dependent and modern implementations tend to get it right. The next generation machine, 68010, provided extra information on the stack so the horrible 2-processor kludge was no longer necessary.

4.7.4: Locking (Pinning) Pages in Memory

We discussed pinning jobs already. The same (mostly I/O) considerations apply to pages.

4.7.5: Backing Store

The issue is where on disk do we put pages.

• For program text, which is presumably read only, a good choice is the file itself.
  
  
• What if we decide to keep the data and stack each contiguous on the backing store. Data and stack grow so we must be prepared to grow the space on disk, which leads to the same issues and problems as we saw with MVT.
  
  
• If those issues/problems are painful, we can scatter the pages on the disk.
  • That is we employ paging!
  • This is NOT demand paging.
  • Need a table to say where the backing space for each page is located.
    • This corresponds to the page table used to tell where in real memory a page is located.
    • The format of the ``memory page table'' is determined by the hardware since the hardware modifies/accesses it.
    • The format of the ``disk page table'' is decided by the OS designers and is machine independent.
    • If the format of the memory page table was flexible, then we might well keep the disk information in it as well.
  
  
• What if we felt disk space was too expensive and wanted to put some of these disk pages on say tape?
  Ans: We have demand paging of the disk blocks!

4.7.6: Separation of Policy and Mechanism

Skipped.

4.8: Segmentation

Up to now, the virtual address space has been contiguous.

• Among other issues this makes memory management difficult when there are more that two dynamically growing regions.
• With two regions you start them on opposite sides of the virtual space as we did before.
• Better is to have many virtual address spaces each starting at zero.
• This split up is user visible.
• Without segmentation (equivalently said with just one segment) all procedures are packed together so if one changes in size all the virtual addresses following are changed and the program must be re-linked.
• Eases flexible protection and sharing (share a segment). For example, can have a shared library.

Homework: 37.

** Two Segments

Late PDP-10s and TOPS-10

• One shared text segment, that can also contain shared (normally read only) data.
• One (private) writable data segment.
• Permission bits on each segment.
• Which kind of segment is better to evict?
  • Swap out shared segment hurts many tasks.
  • The shared segment is read only (probably) so no writeback is needed.
• ``One segment'' is OS/MVT done above.

** Three Segments

Traditional (early) Unix shown at right.

• Shared text marked execute only.
• Data segment (global and static variables).
• Stack segment (automatic variables).
• In reality, since the text doesn't grow, this was sometimes treated as 2 segments by combining text and data into one segment

** Four Segments

Just kidding.

** General (not necessarily demand) Segmentation

• Permits fine grained sharing and protection. For a simple example can share the text segment in early unix.
• Visible division of program.
• Variable size segments.
• Virtual Address = (seg#, offset).
• Does not mandate how stored in memory.
  • One possibility is that the entire program must be in memory in order to run it. Use whole process swapping. Very early versions of Unix did this.
  • Can also implement demand segmentation.
  • Can combine with demand paging (done below).
• Requires a segment table with a base and limit value for each segment. Similar to a page table. Why is there no limit value in a page table?
  Ans: All pages are the same size so the limit is obvious.
• Entries are called STEs, Segment Table Entries.
• (seg#, offset) --> if (offset<limit) base+offset else error.
• Segmentation exhibits external fragmentation, just as whole program swapping. Since segments are smaller than programs (several segments make up one program), the external fragmentation is not as bad.

** Demand Segmentation

Same idea as demand paging applied to segments.

• If a segment is loaded, base and limit are stored in the STE and the valid bit is set in the PTE.
• The PTE is accessed for each memory reference (not really, TLB).
• If the segment is not loaded, the valid bit is unset. The base and limit as well as the disk address of the segment is stored in the an OS table.
• A reference to a non-loaded segment generate a segment fault (analogous to page fault).
• To load a segment, we must solve both the placement question and the replacement question (for demand paging, there is no placement question).
• I believe demand segmentation was once implemented by Burroughs, but am not sure. It is not used in modern systems.

The following table mostly from Tanenbaum compares demand paging with demand segmentation.

** 4.8.2 and 4.8.3: Segmentation With Paging

(Tanenbaum gives two sections to explain the differences between Multics and the Intel Pentium. These notes cover what is common to all segmentation).

Combines both segmentation and paging to get advantages of both at a cost in complexity. This is very common now.

• A virtual address becomes a triple: (seg#, page#, offset).
• Each segment table entry (STE) points to the page table for that segment. Compare this with a multilevel page table.
  
  
• The size of each segment is a multiple of the page size (since the segment consists of pages). Perhaps not. Can keep the exact size in the STE (limit value) and shoot the process if it referenced beyond the limit. In this case the last page of each segment is partially valid.
  
  
• The page# field in the address gives the entry in the chosen page table and the offset gives the offset in the page.
  
  
• From the limit field, one can easily compute the size of the segment in pages (which equals the size of the corresponding page table in PTEs). Implementations may require the size of a segment to be a multiple of the page size in which case the STE would store the number of pages in the segment.
  
  
• A straightforward implementation of segmentation with paging would requires 3 memory references (STE, PTE, referenced word) so a TLB is crucial.
  
  
• Some books carelessly say that segments are of fixed size. This is wrong. They are of variable size with a fixed maximum and possibly with the requirement that the size of a segment is a multiple of the page size.
• The first example of segmentation with paging was Multics.
  
  
• Keep protection and sharing information on segments. This works well for a number of reasons.
  1. A segment is variable size.
  2. Segments and their boundaries are user (i.e., linker) visible.
  3. Segments are shared by sharing their page tables. This eliminates the problem mentioned above with shared pages.
  
  
• Do replacement on pages so there is no placement question and no external fragmentation.
  
  
• Do fetch-on-demand with pages (i.e., do demand paging).
  
  
• In general, segmentation with demand paging works well and is widely used. The only problems are the complexity and the resulting 3 memory references for each user memory reference. The complexity is real, but can be managed. The three memory references would be fatal were it not for TLBs, which considerably ameliorate the problem. TLBs have high hit rates and for a TLB hit there is essentially no penalty.

Homework: 38.

4.9: Research on Memory Management

Skipped

4.10: Summary

Read

Some Last Words on Memory Management

• Segmentation / Paging / Demand Loading (fetch-on-demand)
  • Each is a yes or no alternative.
  • Gives 8 possibilities.
• Placement and Replacement.
• Internal and External Fragmentation.
• Page Size and locality of reference.
• Multiprogramming level and medium term scheduling.

Chapter 5: Input/Output

5.1: Principles of I/O Hardware

5.1.1: I/O Devices

• Not much to say. Devices are varied.
• Block versus character devices:
  • Devices, such as disks and CDROMs, with addressable chunks (sectors in this case) are called block devices,
    These devices support seeking.
  • Devices, such as Ethernet and modem connections, that are a stream of characters are called character devices.
    These devices do not support seeking.
  • Some cases, like tapes, are not so clear.

5.1.2: Device Controllers

These are the ``devices'' as far as the OS is concerned. That is, the OS code is written with the controller spec in hand not with the device spec.

• Also called adaptors.
  
  
• The controller abstracts away some of the low level features of the device.
  
  
• For disks, the controller does error checking and buffering.
  
  
• (Unofficial) In the old days it handled interleaving of sectors. (Sectors are interleaved if the controller or CPU cannot handle the data rate and would otherwise have to wait a full revolution. This is not a concern with modern systems since the electronics have increased in speed faster than the devices.)
  
  
• For analog monitors (CRTs) the controller does a great deal. Analog video is far from a bunch of ones and zeros.
  
  
• Controllers are also called adaptors.

5.1.3: Memory-Mapped I/O

Think of a disk controller and a read request. The goal is to copy data from the disk to some portion of the central memory. How do we do this?

• The controller contains a microprocessor and memory and is connected to the disk (by a cable).
  
  
• When the controller asks the disk to read a sector, the contents come to the controller via the cable and are stored by the controller in its memory.
  
  
• The question is how does the OS, which is running on another processor, let the controller know that a disk read is desired and how is the data eventually moved from the controller's memory to the general system memory.
  
  
• Typically the interface the OS sees consists of some device registers located on the controller.
  • These are memory locations into which the OS writes information such as sector to access, read vs. write, length, where in system memory to put the data (for a read) or from where to take the data (for a write).
  • There is also typically a device register that acts as a ``go button''.
  • There are also devices registers that the OS reads, such as status of the controller, errors found, etc.
  
  
• So now the question is how does the OS read and write the device register
  • With Memory-mapped I/O the device registers appear as normal memory. All that is needed is to know at which address each device regester appears. Then the OS uses normal load and store instructions to write the registers.
  • Some systems instead have a special ``I/O space'' into which the registers are mapped and require the use of special I/O space instructions to accomplish the load and store. From a conceptual point of view there is no difference between the two models.

5.1.4: Direct Memory Access (DMA)

• With or without DMA, the disk controller pulls the desired data from the disk to its buffer (and pushes data from the buffer to the disk).
  
  
• Without DMA, i.e., with programmed I/O (PIO), the cpu then does loads and stores (or I/O instructions) to copy the data from the buffer to the desired memory location.
  
  
• With a DMA controller, the controller writes the memory without intervention of the CPU.
  
  
• Clearly DMA saves CPU work. But this might not be important if the CPU is limited by the memory or by system buses.
  
  
• Very important is that there is less data movement so the buses are used less and the entire operation takes less time.
  
  
• Since PIO is pure software it is easier to change, which is an advantage.
  
  
• DMA does need a number of bus transfers from the CPU to the controller to specify the DMA. So DMA is most effective for large transfers where the setup is amortized.
  
  
• Why have the buffer? Why not just go from the disk straight to the memory.
  Answer: Speed matching. The disk supplies data at a fixed rate, which might exceed the rate the memory can accept it. In particular the memory might be busy servicing a request from the processor or from another DMA controller.

Homework: 12

5.1.5: Interrupts Revisited

Skipped.

5.2: Principles of I/O Software

As with any large software system, good design and layering is important.

5.2.1: Goals of the I/O Software

Device independence

We want to have most of the OS, unaware of the characteristics of the specific devices attached to the system. Indeed we also want the OS to be largely unaware of the CPU type itself.

Due to this device independence, programs are written to read and write generic devices and then at run time specific devices are assigned. Writing to a disk has differences from writing to a terminal, but Unix cp and DOS copy do not see these differences. Indeed, most of the OS, including the file system code, is unaware of whether the device is a floppy or hard disk.

Homework: 5.9

Uniform naming

Recall that we discussed the value of the name space implemented by file systems. There is no dependence between the name of the file and the device on which it is stored. So a file called IAmStoredOnAHardDisk might well be stored on a floppy disk.

Error handling

1. Detection should be done as close to where the error occurred as possible before more damage is done (fault containment). This is not trivial.
   
   
2. Correction is sometimes easy, for example ECC memory does this automatically (but the OS wants to know about the error so that it can schedule replacement of the faulty chips before unrecoverable double errors occur).

   Other easy cases include successful retries for failed ethernet transmissions. In this example, while logging is appropriate, it is quite reasonable for no action to be taken.
   

3. Error reporting tends to be awful. The trouble is that the error occurs at a low level but by the time it is reported the context is lost. Unix/Linux in particular is horrible in this area.

Creating the illusion of synchronous I/O

• I/O must be asynchronous for good performance. That is the OS cannot simply wait for an I/O to complete. Instead, it proceeds with other activities and responds to the notification when the I/O has finished.
  
  
• Users (mostly) want no part of this. The code sequence

  Read X
  Y <-- X+1
  Print Y
  

  should print a value one greater than that read. But if the assignment is performed before the read completes, the wrong value is assigned.
  
  
• Performance junkies sometimes do want the asynchrony so that they can have another portion of their program executed while the I/O is underway. That is they implement a mini-scheduler in their application code.

Buffering

• Often needed to hold data for examination prior to sending it to its desired destination.
• But this involves copying and takes time.

Sharable vs dedicated devices

For devices like printers and tape drives, only one user at a time is permitted. These are called serially reusable devices, and were studied in the deadlocks chapter. Devices like disks and Ethernet ports can be shared by processes running concurrently.

5.2.2: Programmed I/O

• As mentioned just above, with programmed I/O the processor moves the data between memory and the device.
• How does the process know when the device is ready to accept or supply new data?
• In the simplest implementation, the processor loops continually asking the device. This is called polling or busy waiting.
• If the device is slow, polling is clearly wasteful, which leads us to.

5.2.3: Interrupt-Driven I/O

• The device interrupts the processor when it is ready.
• An interrupt service routine then initiates transfer of the next datum
• Normally better than polling, but not always. Interrupts are expensive on modern machines.
• To minimize interrupts, better controllers often employ ...

I/O Using DMA

• We discussed DMA above.
• An additional advantage of dma, not mentioned above, is that the processor is interrupted only at the end of a command not after each datum is transferred.
• Many devices receive a character at a time, but with a dma controller, an interrupt occurs only after a buffer has been transferred.

5.3: I/O Software Layers

Layers of abstraction as usual prove to be effective. Most systems are believed to use the following layers (but for many systems, the OS code is not available for inspection).

1. User level I/O routines.
2. Device independent I/O software.
3. Device drivers.
4. Interrupt handlers.

We will give a bottom up explanation.

5.3.1: Interrupt Handlers

We discussed an interrupt handler before when studying page faults. Then it was called ``assembly language code''.

In the present case, we have a process blocked on I/O and the I/O event has just completed. So the goal is to make the process ready. Possible methods are.

• Releasing a semaphore on which the process is waiting.
• Sending a message to the process.
• Inserting the process table entry onto the ready list.

Once the process is ready, it is up to the scheduler to decide when it should run.

5.3.2: Device Drivers

The portion of the OS that ``knows'' the characteristics of the controller.

The driver has two ``parts'' corresponding to its two access points. Recall the following figure from the beginning of the course.

1. Accessed by the main line OS via the envelope in response to an I/O system call. The portion of the driver accessed in this way is sometimes call the ``top'' part.
2. Accessed by the interrupt handler when the I/O completes (this completion is signaled by an interrupt). The portion of the driver accessed in this way is sometimes call the ``bottom'' part.

Tanenbaum describes the actions of the driver assuming it is implemented as a process (which he recommends). I give both that view point and the self-service paradigm in which the driver is invoked by the OS acting in behalf of a user process (more precisely the process shifts into kernel mode).

Driver in a self-service paradigm

1. The user (A) issues an I/O system call.
   
   
2. The main line, machine independent, OS prepares a generic request for the driver and calls (the top part of) the driver.
   1. If the driver was idle (i.e., the controller was idle), the driver writes device registers on the controller ending with a command for the controller to begin the actual I/O.
   2. If the controller was busy (doing work the driver gave it previously), the driver simply queues the current request (the driver dequeues this request below).
   
   
3. The driver jumps to the scheduler indicating that the current process should be blocked.
   
   
4. The scheduler blocks A and runs (say) B.
   
   
5. B starts running.
   
   
6. An interrupt arrives (i.e., an I/O has been completed).
   
   
7. The interrupt handler invokes (the bottom part of) the driver.
   1. The driver informs the main line perhaps passing data and surely passing status (error, OK).
   2. The top part is called to start another I/O if the queue is nonempty. We know the controller is free. Why?
      Answer: We just received an interrupt saying so.
   
   
8. The driver jumps to the scheduler indicating that process A should be made ready.
   
   
9. The scheduler picks a ready process to run. Assume it picks A.
   
   
10. A resumes in the driver, which returns to the main line, which returns to the user code.

Driver as a process (Tanenbaum) (less detailed than above)

• The user issues an I/O request. The main line OS prepares a generic request (e.g. read, not read using Buslogic BT-958 SCSI controller) for the driver and the driver is awakened (perhaps a message is sent to the driver to do both jobs).
  1. The driver wakes up.
     1. If the driver was idle (i.e., the controller is idle), the driver writes device registers on the controller ending with a command for the controller to begin the actual I/O.
     2. If the controller is busy (doing work the driver gave it), the driver simply queues the current request (the driver dequeues this below).
  2. The driver blocks waiting for an interrupt or for more requests.
• An interrupt arrives (i.e., an I/O has been completed).
  1. The driver wakes up.
  2. The driver informs the main line perhaps passing data and surely passing status (error, OK).
  3. The driver finds the next work item or blocks.
     1. If the queue of requests is non-empty, dequeue one and proceed as if just received a request from the main line.
     2. If queue is empty, the driver blocks waiting for an interrupt or a request from the main line.

5.3.3: Device-Independent I/O Software

The device-independent code does most of the functionality, but not necessarily most of the code since there can be many drivers all doing essentially the same thing in slightly different ways due to slightly different controllers.

• Naming. Again an important O/S functionality. Must offer a consistent interface to the device drivers.
  • In Unix this is done by associating each device with a (special) file in the /dev directory.
  • The i-nodes for these files contain an indication that these are special files and also contain so called major and minor device numbers.
  • The major device number gives the number of the driver. (These numbers are rather ad hoc, they correspond to the position of the function pointer to the driver in a table of function pointers.)
  • The minor number indicates for which device (e.g., which scsi cdrom drive) the request is intended
  
  
• Protection. A wide range of possibilities are actually done in real systems. Including both extreme examples of everything is permitted and nothing is permitted (directly).
  • In ms-dos any process can write to any file. Presumably, our offensive nuclear missile launchers do not run dos.
  • In IBM and other mainframe OS's, normal processors do not access devices. Indeed the main CPU doesn't issue the I/O requests. Instead an I/O channel is used and the mainline constructs a channel program and tells the channel to invoke it.
  • Unix uses normal rwx bits on files in /dev (I don't believe x is used).
  
  
• Buffering is necessary since requests come in a size specified by the user and data is delivered in a size specified by the device.
  
  
• Enforce exclusive access for non-shared devices like tapes.

5.3.4: User-Space Software

A good deal of I/O code is actually executed in user space. Some is in library routines linked into user programs and some is in daemon processes.

• Some library routines are trivial and just move their arguments into the correct place (e.g., a specific register) and then issue a trap to the correct system call to do the real work.
  
  
• Some, notably standard I/O (stdio) in Unix, are definitely not trivial. For example consider the formatting of floating point numbers done in printf and the reverse operation done in scanf.
  
  
• Printing to a local printer is often performed in part by a regular program (lpr in Unix) and part by a daemon (lpd in Unix). The daemon might be started when the system boots or might be started on demand. I guess it is called a daemon because it is not under the control of any user. A student has suggested that the usage comes from the meaning of daemon in Greek mythology, namely a guardian spirit. Does anyone know the reason.
  
  
• Printing uses spooling, i.e., the file to be printed is copied somewhere by lpr and then the daemon works with this copy. Mail uses a similar technique (but generally it is called queuing, not spooling).

Homework: 10, 13.

5.4: Disks

The ideal storage device is

1. Fast
2. Big (in capacity)
3. Cheap
4. Impossible

Disks are big and cheap, but slow.

5.4.1: Disk Hardware

Show a real disk opened up and illustrate the components

• Platter
• Surface
• Head
• Track
• Sector
• Cylinder
• Seek time
• Rotational latency
• Transfer rate

Overlapping I/O operations is important. Many controllers can do overlapped seeks, i.e. issue a seek to one disk while another is already seeking.

Modern disks cheat and do not have the same number of sectors on outer cylinders as on inner one. However, the disks have electronics and software (firmware) that hides the cheat and gives the illusion of the same number of sectors on all cylinders.

(Unofficial) Despite what tanenbaum says later, it is not true that when one head is reading from cylinder C, all the heads can read from cylinder C with no penalty. It is, however, true that the penalty is very small.

Choice of block size

• We discussed this before when studying page size.
• Current commodity disk characteristics (not for laptops) result in about 15ms to transfer the first byte and 10K bytes per ms for subsequent bytes (if contiguous).
  • Rotation rate is 5400, 7600, or 10,000 RPM (15K just now available).
  • Recall that 6000 RPM is 100 rev/sec or one rev per 10ms. So half a rev (the average time for to rotate to a given point) is 5ms.
  • Transfer rates around 10MB/sec = 10KB/ms.
  • Seek time around 10ms.
• This favors large blocks, 100KB or more.
• But the internal fragmentation would be severe since many files are small.
• Multiple block sizes have been tried as have techniques to try to have consecutive blocks of a given file near each other.
• Typical block sizes are 4KB-8KB.

Homework: Consider a disk with an average seek time of 10ms, an average rotational latency of 5ms, and a transfer rate of 10MB/sec.

1. If the block size is 1KB, how long would it take to read a block?
2. If the block size is 100KB, how long would it take to read a block?
3. If the goal is to read 1K, a 1KB block size is better as the remaining 99KB are wasted. If the goal is to read 100KB, the 100KB block size is better since the 1KB block size needs 100 seeks and 100 rotational latencies. What is the minimum size request for which a disk with a 100KB block size would complete faster than one with a 1KB block size?

RAID (Redundant Array of Inexpensive Disks) (Skipped)

• The name RAID is from Berkeley.
  
  
• IBM changed the name to Redundant Array of Independent Disks. I wonder why?
  
  
• A simple form is mirroring, where two disks contain the same data.
  
  
• Another simple form is striping (interleaving) where consecutive blocks are spread across multiple disks. This helps bandwidth, but is not redundant. Thus it shouldn't be called RAID, but it sometimes is.
  
  
• One of the normal RAID methods is to have N (say 4) data disks and one parity disk. Data is striped across the data disks and the bitwise parity of these sectors is written in the corresponding sector of the parity disk.
  
  
• On a read if the block is bad (e.g., if the entire disk is bad or even missing), the system automatically reads the other blocks in the stripe and the parity block in the stripe. Then the missing block is just the bitwise exclusive or of all these blocks.
  
  
• For reads this is very good. The failure free case has no penalty (beyond the space overhead of the parity disk). The error case requires N+1 (say 5) reads.
  
  
• A serious concern is the small write problem. Writing a sector requires 4 I/O. Read the old data sector, compute the change, read the parity, compute the new parity, write the new parity and the new data sector. Hence one sector I/O became 4, which is a 300% penalty.
  
  
• Writing a full stripe is not bad. Compute the parity of the N (say 4) data sectors to be written and then write the data sectors and the parity sector. Thus 4 sector I/Os become 5, which is only a 25% penalty and is smaller for larger N, i.e., larger stripes.
  
  
• A variation is to rotate the parity. That is, for some stripes disk 1 has the parity, for others disk 2, etc. The purpose is to not have a single parity disk since that disk is needed for all small writes and could become a point of contention.
  
  

5.4.2: Disk Formatting

Skipped.

5.4.3: Disk Arm Scheduling Algorithms

There are three components to disk response time: seek, rotational latency, and transfer time. Disk arm scheduling is concerned with minimizing seek time by reordering the requests.

These algorithms are relevant only if there are several I/O requests pending. For many PCs this is not the case. For most commercial applications, I/O is crucial and there are often many requests pending.

1. FCFS (First Come First Served): Simple but has long delays.
   
   
2. Pick: Same as FCFS but pick up requests for cylinders that are passed on the way to the next FCFS request.
   
   
3. SSTF or SSF (Shortest Seek (Time) First): Greedy algorithm. Can starve requests for outer cylinders and almost always favors middle requests.
   
   
4. Scan (Look, Elevator): The method used by an old fashioned jukebox (remember ``Happy Days'') and by elevators. The disk arm proceeds in one direction picking up all requests until there are no more requests in this direction at which point it goes back the other direction. This favors requests in the middle, but can't starve any requests.
   
   
5. C-Scan (C-look, Circular Scan/Look): Similar to Scan but only service requests when moving in one direction. When going in the other direction, go directly to the furthest away request. This doesn't favor any spot on the disk. Indeed, it treats the cylinders as though they were a clock, i.e. after the highest numbered cylinder comes cylinder 0.
   
   
6. N-step Scan: This is what the natural implementation of Scan gives.
   • While the disk is servicing a Scan direction, the controller gathers up new requests and sorts them.
   • At the end of the current sweep, the new list becomes the next sweep.

Minimizing Rotational Latency

Use Scan based on sector numbers not cylinder number. For rotational latency Scan which is the same as C-Scan. Why?
Ans: Because the disk only rotates in one direction.

Homework: 24, 25

5.4.4: Error Handling

Disks error rates have dropped in recent years. Moreover, bad block forwarding is normally done by the controller (or disk electronic) so this topic is no longer as important for OS.

5.5: Clocks

Also called timers.

5.5.1: Clock Hardware

• Generates an interrupt when timer goes to zero
• Counter reload can be automatic or under software (OS) control.
• If done automatically, the interrupt occurs periodically and thus is perfect for generating a clock interrupt at a fixed period.

5.5.2: Clock Software

1. TOD: Bump a counter each tick (clock interupt). If counter is only 32 bits must worry about overflow so keep two counters: low order and high order.
   
   
2. Time quantum for RR: Decrement a counter at each tick. The quantum expires when counter is zero. Load this counter when the scheduler runs a process.
   
   
3. Accounting: At each tick, bump a counter in the process table entry for the currently running process.
   
   
4. Alarm system call and system alarms:
   • Users can request an alarm at some future time.
   • The system also on occasion needs to schedule some of its own activities to occur at specific times in the future (e.g. turn off the floppy motor).
   • The conceptually simplest solution is to have one timer for each event.
   • Instead, we simulate many timers with just one.
   • The data structure on the right works well.
   • The time in each list entry is the time after the preceding entry that this entry's alarm is to ring.
   • For example, if the time is zero, this event occurs at the same time as the previous event.
   • The other entry is a pointer to the action to perform.
   • At each tick, decrement next-signal.
   • When next-signal goes to zero, process the first entry on the list and any others following immediately after with a time of zero (which means they are to be simultaneous with this alarm). Then set next-signal to the value in the next alarm.
5. Profiling
   • Want a histogram giving how much time was spent in each 1KB (say) block of code.
   • At each tick check the PC and bump the appropriate counter.
   • A user-mode program can determine the software module associated with each 1K block.
   • If we use finer granularity (say 10B instead of 1KB), we get increased accuracy but more memory overhead.

Homework: 27

5.6: Character-Oriented Terminals

5.6.1: RS-232 Terminal Hardware

Quite dated. It is true that modern systems can communicate to a hardwired ascii terminal, but most don't. Serial ports are used, but they are normally connected to modems and then some protocol (SLIP, PPP) is used not just a stream of ascii characters. So skip this section.

Memory-Mapped Terminals

Not as dated as the previous section but it still discusses the character not graphics interface.

• Today, software writes into video memory the bits that are to be put on the screen and then the graphics controller converts these bits to analog signals for the monitor (actually laptop displays and some modern monitors are digital).
• But it is much more complicated than this. The graphics controllers can do a great deal of video themselves (like filling).
• This is a subject that would take many lectures to do well.
• I believe some of this is covered in 201.

Keyboards

Tanenbaum description of keyboards is correct.

• At each key press and key release a code is written into the keyboard controller and the computer is interrupted.
• By remembering which keys have been depressed and not released the software can determine Cntl-A, Shift-B, etc.

5.6.2: Input Software

• We are just looking at keyboard input. Once again graphics is too involved to be treated here.
• There are two fundamental modes of input, sometimes called raw and cooked.
• In raw mode the application sees every ``character'' the user types. Indeed, raw mode is character oriented.
  • All the OS does is convert the keyboard ``scan codes'' to ``characters'' and and pass these characters to the application.
  • Some examples
    1. down-cntl down-x up-x up-cntl is converted to cntl-x
    2. down-cntl up-cntl down-x up-x is converted to x
    3. down-cntl down-x up-cntl up-x is converted to cntl-x (I just tried it to be sure).
    4. down-x down-cntl up-x up-cntl is converted to x
  • Full screen editors use this mode.
• Cooked mode is line oriented. The OS delivers lines to the application program.
  • Special characters are interpreted as editing characters (erase-previous-character, erase-previous-word, kill-line, etc).
  • Erased characters are not seen by the application but are erased by the keyboard driver.
  • Need an escape character so that the editing characters can be passed to the application if desired.
  • The cooked characters must be echoed (what should one do if the application is also generating output at this time?)
• The (possibly cooked) characters must be buffered until the application issues a read (and an end-of-line EOL has been received for cooked mode).

5.6.3: Output Software

Again too dated and the truth is too complicated to deal with in a few minutes.

5.7: Graphical User Interfaces (GUIs)

Skipped.

5.8: Network Terminals

Skipped.

5.9: Power Management

Skipped.

5.10: Research on Input/Output

Skipped.

5.11: Summary

Read.

Chapter 6: File Systems

Requirements

1. Size: Store very large amounts of data.
2. Persistence: Data survives the creating process.
3. Access: Multiple processes can access the data concurrently.

Solution: Store data in files that together form a file system.

6.1: Files

6.1.1: File Naming

Very important. A major function of the file system.

• Does each file have a unique name?
  Answer: Often no. We will discuss this below when we study links.
  
  
• Extensions, e.g. the ``html'' in ``class-notes.html''.
  
  
  1. Conventions just for humans: letter.teq (my convention).
     
     
  2. Conventions giving default behavior for some programs.
     • The emacs editor thinks .html files should be edited in html mode but
       can edit them in any mode and can edit any file in html mode.
     • Netscape thinks .html means an html file, but
       <html> ... </html> works as well
     • Gzip thinks .gz means a compressed file but accepts a --suffix flag
     
     
  3. Default behavior for Operating system or window manager or desktop environment.
     • Click on .xls file in windows and excel is started.
     • Click on .xls file in nautilus under linux and gnumeric is started.
  4. Required extensions for programs
     • The gnu C compiler (and probably others) requires C programs be named *.c and assembler programs be named *.s
     
     
  5. Required extensions by operating systems
     • MS-DOS treats .com files specially
     • Windows 95 requires (as far as I can tell) shortcuts to end in .lnk.
  
  
• Case sensitive?
  Unix: yes. Windows: no.

6.1.2: File structure

A file is a

1. Byte stream
   • Unix, dos, windows (I think).
   • Maximum flexibility.
   • Minimum structure.
   
   
2. (fixed size) Record stream: Out of date
   • 80-character records for card images.
   • 133-character records for line printer files. Column 1 was for control (e.g., new page) Remaining 132 characters were printed.
   
   
3. Varied and complicated beast.
   • Indexed sequential.
   • B-trees.
   • Supports rapidly finding a record with a specific key.
   • Supports retrieving (varying size) records in key order.
   • Treated in depth in database courses.

Remark: In spring 2002, due to my attending a conference, showing all the parts of a disk and doing a problem involving I/O time are done here.

6.1.3: File types

Examples

1. (Regular) files.
   
   
2. Directories: studied below.
   
   
3. Special files (for devices). Uses the naming power of files to unify many actions.

       dir             # prints on screen
       dir > file      # result put in a file
       dir > /dev/tape # results written to tape
       

4. ``Symbolic'' Links (similar to ``shortcuts''): Also studied below.

``Magic number'': Identifies an executable file.

• There can be several different magic numbers for different types of executables.
• unix: #!/usr/bin/perl

Strongly typed files:

• The type of the file determines what you can do with the file.
• This make the easy and (hopefully) common case easier and, more importantly safer.
• It tends to make the unusual case harder. For example, you have a program that turns out data (.dat) files. But you want to use it to turn out a java file but the type of the output is data and cannot be easily converted to type java.

6.1.4: File access

There are basically two possibilities, sequential access and random access (a.k.a. direct access). Previously, files were declared to be sequential or random. Modern systems do not do this. Instead all files are random and optimizations are applied when the system dynamically determines that a file is (probably) being accessed sequentially.

1. With Sequential access the bytes (or records) are accessed in order (i.e., n-1, n, n+1, ...). Sequential access is the most common and gives the highest performance. For some devices (e.g. tapes) access ``must'' be sequential.
2. With random access, the bytes are accessed in any order. Thus each access must specify which bytes are desired.

6.1.5: File attributes

A laundry list of properties that can be specified for a file For example:

• hidden
• do not dump
• owner
• key length (for keyed files)

6.1.6: File operations

• Create: Essential if a system is to add files. Need not be a separate system call (can be merged with open).
• Delete: Essential if a system is to delete files.
• Open: Not essential. An optimization in which the translation from file name to disk locations is perform only once per file rather than once per access.
• Close: Not essential. Free resources.
• Read: Essential. Must specify filename, file location, number of bytes, and a buffer into which the data is to be placed. Several of these parameters can be set by other system calls and in many OS's they are.
• Write: Essential if updates are to be supported. See read for parameters.
• Seek: Not essential (could be in read/write). Specify the offset of the next (read or write) access to this file.
• Get attributes: Essential if attributes are to be used.
• Set attributes: Essential if attributes are to be user settable.
• Rename: Tanenbaum has strange words. Copy and delete is not acceptable for big files. Moreover copy-delete is not atomic. Indeed link-delete is not atomic so even if link (discussed below) is provided, renaming a file adds functionality.

Homework: 6, 7.

6.1.7: An Example Program Using File System Calls

Homework: Read and understand ``copyfile''.

Notes on copyfile

• Normally in unix one wouldn't call read and write directly.
• Indeed, for copyfile, getchar() and putchar() would be nice since they take care of the buffering (standard I/O, stdio).
• If you compare copyfile from the 1st to 2nd edition, you can see the addition of error checks.

6.1.8: Memory mapped files (Unofficial)

Conceptually simple and elegant. Associate a segment with each file and then normal memory operations take the place of I/O.

Thus copyfile does not have fgetc/fputc (or read/write). Instead it is just like memcopy

while ( *(dest++) = *(src++) );

The implementation is via segmentation with demand paging but the backing store for the pages is the file itself. This all sounds great but ...

1. How do you tell the length of a newly created file? You know which pages were written but not what words in those pages. So a file with one byte or 10, looks like a page.
2. What if same file is accessed by both I/O and memory mapping.
   
3. What if the file is bigger than the size of virtual memory (will not be a problem for systems built 3 years from now as all will have enormous virtual memory sizes).

6.2: Directories

Unit of organization.

6.2.1-6.2.3: Single-level, Two-level, and Hierarchical directory systems

Possibilities

• One directory in the system (Single-level)
• One per user and a root above these (Two-level)
• One tree
• One tree per user
• One forest
• One forest per user

These are not as wildly different as they sound.

• If the system only has one directory, but allows the character / in a file name. Then one could fake a tree by having a file named
  /allan/gottlieb/courses/arch/class-notes.html
  rather than a directory allan, a subdirectory gottlieb, ..., a file class-notes.html.
• Dos (windows) is a forest, unix a tree. In dos there is no common parent of a:\ and c:\.
• But windows explorer makes the dos forest look quite a bit like a tree. Indeed, the original gnome file manager for linux, looks A LOT like windows explorer.
• You can get an effect similar to (but not the same as) one X per user by having just one X in the system and having permissions that permits each user to visit only a subset. Of course if the system doesn't have permissions, this is not possible.
• Today's systems have a tree per system or a forest per system.

6.2.4: Path Names

You can specify the location of a file in the file hierarchy by using either an absolute or a Relative path to the file

• An absolute path starts at the (or ``one of the'', if we have a forest) root(s).
• A relative path starts at the current (a.k.a working) directory.
• The special directories . and .. represent the current directory and the parent of the current directory respectively.

Homework: 1, 9.

6.2.5: Directory operations

1. Create: Produces an ``empty'' directory. Normally the directory created actually contains . and .., so is not really empty
   
   
2. Delete: Requires the directory to be empty (i.e., to just contain . and ..). Commands are normally written that will first empty the directory (except for . and ..) and then delete it. These commands make use of file and directory delete system calls.
   
   
3. Opendir: Same as for files (creates a ``handle'')
   
   
4. Closedir: Same as for files
   
   
5. Readdir: In the old days (of unix) one could read directories as files so there was no special readdir (or opendir/closedir). It was believed that the uniform treatment would make programming (or at least system understanding) easier as there was less to learn.

   However, experience has taught that this was not a good idea since the structure of directories then becomes exposed. Early unix had a simple structure (and there was only one). Modern systems have more sophisticated structures and more importantly they are not fixed across implementations.
   

6. Rename: As with files
   
   
7. Link: Add a second name for a file; discussed below.
   
   
8. Unlink: Remove a directory entry. This is how a file is deleted. But if there are many links and just one is unlinked, the file remains. Discussed in more detail below.

6.3: File System Implementation

6.3.1: File System Layout

• One disk starts with a Master Boot Record (MBR).
• Each disk has a partition table.
• Each partition holds one file system.
• Each partition typically contains some parameters (e.g., size), free blocks, and blocks in use. The details vary.
• In unix some of the in use blocks contains I-nodes each of which describes a file or directory and is described below.
• During boot the MBR is read and executed. It transfers control to the boot block of the active partition.

6.3.2: Implementing Files

• A disk cannot read or write a single word. Instead it can read or write a sector, which is often 512 bytes.
• Disks are written in blocks whose size is a multiple of the sector size.

Contiguous allocation

• This is like OS/MVT.
• The entire file is stored as one piece.
• Simple and fast for access, but ...
• Problem with growing files
  • Must either evict the file itself or the file it is bumping into.
  • Same problem with an OS/MVT kind of system if jobs grow.
• Problem with external fragmentation.
• No longer used for general purpose rewritable file systems.
• Ideal for file systems where files do not change size.
• Used for CD-ROM file systems.

Homework: 12.

Linked allocation

• The directory entry contains a pointer to the first block of the file.
• Each block contains a pointer to the next.
• Horrible for random access.
• Not used.

FAT (file allocation table)

• Used by dos and windows (but not windows/NT).
• Directory entry points to first block (i.e. specifies the block number).
• A FAT is maintained in memory having one (word) entry for each disk block. The entry for block N contains the block number of the next block in the same file as N.
• This is linked but the links are stored separately.
• Time to access a random block is still is linear in size of file but now all the references are to this one table which is in memory. So it is bad but not horrible for random access.
• Size of table is one word per disk block. If blocks are of size 4K and the FAT uses 4-byte words, the table is one megabyte for each disk gigabyte. Large but perhaps not prohibitive.
• If blocks are of size 512 bytes (the sector size of most disks) then the table is 8 megs per gig, which is probably prohibitive.

I-Nodes

• Used by unix/linux.
• Directory entry points to i-node (index-node).
• I-Node points to first few data blocks, often called direct blocks.
• I-Node also points to an indirect block, which points to disk blocks.
• I-Node also points to a double indirect, which points an indirect ...
• For some implementations there are triple indirect as well.
• The i-node is in memory for open files. So references to direct blocks take just one I/O.
• For big files most references require two I/Os (indirect + data).
• For huge files most references require three I/Os (double indirect, indirect, and data).

6.3.3: Implementing Directories

Recall that a directory is a mapping that converts file (or subdirectory) names to the files (or subdirectories) themselves.

Trivial File System (CP/M)

• Only one directory in the system.
• Directory entry contains pointers to disk blocks.
• If need more blocks, get another directory entry.

MS-DOS and Windows (FAT)

• Subdirectories supported.
• Directory entry contains metatdata such as date and size as well as pointer to first block.
• The FAT has the pointers to the remaining blocks.

Unix/linux

• Each entry contains a name and a pointer to the corresponding i-node.
• Metadata is in the i-node.
• Early unix had limit of 14 character names.
• Name field now is varying length.
• To go down a level in directory takes two steps: get i-node, get file (or subdirectory).
• Do on the blackboard the steps for /allan/gottlieb/courses/os/class-notes.html

Homework: 27

6.3.4: Shared files (links)

• ``Shared'' files is Tanenbaum's terminology.
• More descriptive would be ``multinamed files''.
• If a file exists, one can create another name for it (quite possibly in another directory).
• This is often called creating a (or another) link to the file.
• Unix has two flavor of links, hard links and symbolic links or symlinks.
• Dos/windows has symlinks, but I don't believe it has hard links.
• These links often cause confusion, but I really believe that the diagrams I created make it all clear.

Hard Links

• Symmetric multinamed files.
• When a hard like is created another name is created for the same file.
• The two names have equal status.
• It is not, I repeat NOT, true that one name is the ``real name'' and the other one is ``just a link''.

Start with an empty file system (i.e., just the root directory) and then execute:

cd /
mkdir /A; mkdir /B
touch /A/X; touch /B/Y

We have the situation shown on the right.

• Circles represent ordinary files.
• Squares represent directories.
• One name for the left circle is /A/X.
• I have written the names on the edges.
  • This is not customary, normally they are written in the circles or squares.
  • When there are no multi-named files, it doesn't matter if they are written in the node or edge.
  • We will see that when files can have multiple names it is much better to write the name on the edge.

ln /B/Y /A/New

At this point there are two equally valid name for the right hand yellow file, /B/Y and /A/New. The fact that /B/Y was created first is NOT detectable.

• Both point to the same i-node.
• Only one owner (the one who created the file initially).
• One date, one set of permissions, one ... .

Assume Bob created /B and /B/Y and Alice created /A, /A/X, and /A/New. Later Bob tires of /B/Y and removes it by executing

rm /B/Y

The file /A/New is still fine (see third diagram on the right). But it is owned by Bob, who can't find it! If the system enforces quotas bob will likely be charged (as the owner), but he can neither find nor delete the file (since bob cannot unlink, i.e. remove, files from /A)

Since hard links are only permitted to files (not directories) the resulting file system is a dag (directed acyclic graph). That is, there are no directed cycles. We will now proceed to give away this useful property by studying symlinks, which can point to directories.

Symlinks

• Asymmetric multinamed files.
• When a symlink is created another file is created. The contents of the new file is the name of the original file.
• A hard link in contrast points to the original file.
• he examples will make this clear.

Again start with an empty file system and this time execute

cd /
mkdir /A; mkdir /B
touch /A/X; touch /B/Y
ln -s /B/Y /A/New

We now have an additional file /A/New, which is a symlink to /B/Y.

• The file named /A/New has the name /B/Y as its data (not metadata).
• The system notices that A/New is a diamond (symlink) so reading /A/New will return the contents of /B/Y (assuming the reader has read permission for /B/Y).
• If /B/Y is removed /A/New becomes invalid.
• If a new /B/Y is created, A/New is once again valid.
• Removing /A/New has no effect of /B/Y.
• If a user has write permission for /B/Y, then writing /A/New is possible and writes /B/Y.

The bottom line is that, with a hard link, a new name is created for the file. This new name has equal status with the original name. This can cause some surprises (e.g., you create a link but I own the file). With a symbolic link a new file is created (owned by the creator naturally) that contains the name of the original file. We often say the new file points to the original file.

Question: Consider the hard link setup above. If Bob removes /B/Y and then creates another /B/Y, what happens to /A/New?
Answer: Nothing. /A/New is still a file with the same contents as the original /B/Y.

Question: What about with a symlink?
Answer: /A/New becomes invalid and then valid again, this time pointing to the new /B/Y. (It can't point to the old /B/Y as that is completely gone.)

Note:

Shortcuts in windows contain more that symlinks in unix. In addition to the file name of the original file, they can contain arguments to pass to the file if it is executable. So a shortcut to

netscape.exe

netscape.exe //allan.ultra.nyu.edu/~gottlieb/courses/os/class-notes.html

What about symlinking a directory?

cd /
mkdir /A; mkdir /B
touch /A/X; touch /B/Y
ln -s /B /A/New

Is there a file named /A/New/Y ?
Yes.

What happens if you execute cd /A/New/.. ?

• Answer: Not clear!
• Clearly you are changing directory to the parent directory of /A/New. But is that /A or /?
• The command interpreter I use offers both possibilities.
  • cd -L /A/New/.. takes you to A (L for logical).
  • cd -P /A/New/.. takes you to / (P for physical).
  • cd /A/New/.. takes you to A (logical is the default).

What did I mean when I said the pictures made it all clear?
Answer: From the file system perspective it is clear. It is not always so clear what programs will do.

6.3.5: Disk space management

All general purpose systems use a (non-demand) paging algorithm for file storage. Files are broken into fixed size pieces, called blocks that can be scattered over the disk. Note that although this is paging, it is never called paging.

The file is completely stored on the disk, i.e., it is not demand paging.

Actually, it is more complicated

1. Various optimizations are performed to try to have consecutive blocks of a single file stored consecutively on the disk. Discussed below
   .
   
2. One can imagine systems that store only parts of the file on disk with the rest on tertiary storage (some kind of tape).
   
   
3. This would be just like demand paging.
   
   
4. Perhaps NASA does this with their huge datasets.
   
   
5. Caching (as done for example in microprocessors) is also the same as demand paging.
   
   
6. We unify these concepts in the computer architecture course.

Choice of block size

• We discussed this before when studying page size.
• Current commodity disk characteristics (not for laptops) result in about 15ms to transfer the first byte and 10K bytes per ms for subsequent bytes (if contiguous).
  • This was explained in the I/O chapter.
  • Rotation rate is 5400, 7600, or 10,000 RPM (15K is now available).
  • Recall that 6000 RPM is 100 rev/sec or one rev per 10ms. So half a rev (the average time for to rotate to a given point) is 5ms.
  • Transfer rates around 10MB/sec = 10KB/ms.
  • Seek time is around 10ms.
• This favors large blocks, 100KB or more.
• But the internal fragmentation would be severe since many files are small.
• Typical block sizes are 4KB and 8KB.
• Unofficial
  • Systems that contain multiple block sizes have been tried (i.e., the system uses blocks of size A for some files and blocks of size B for other files).
  • Some systems use techniques to try to have consecutive blocks of a given file near each other as well as blocks of ``related'' files (e.g., files in the same directory).

Storing free blocks

There are basically two possibilities

1. An in-memory bit map.
   • One bit per block
   • If block size is 4KB = 32K bits, 1 bit per 32K bits
   • So 32GB disk (potentially all free) needs 1MB ram.
   • Variation is to demand page the bit map. This saves space (RAM) at the cost of I/O.
   
   
2. Linked list with each free block pointing to next.
   • Thus you must do a read for each request.
   • But reading a free block is a wasted I/O.
   • Instead some free blocks contain pointers to other free blocks. This has much less wasted I/O, but is more complicated.
   • When read a block of pointers store them in memory.
   • See diagram on right.

6.3.6: File System reliability

Bad blocks on disks

Not so much of a problem now. Disks are more reliable and, more importantly, disks take care of the bad blocks themselves. That is, there is no OS support needed to map out bad blocks. But if a block goes bad, the data is lost (not always).

Backups

All modern systems support full and incremental dumps.

• A level 0 dump is a called a full dump (i.e., dumps everything).
• A level n dump (n>0) is called an incremental dump and the standard unix utility dumps all files that have changed since the previous level n-1 dump.
• Other dump utilities dump all files that have changed since the last level n dump.
• Keep on the disk the dates of the most recent level i dumps for all i. In Unix this is traditionally in /etc/dumpdates.
• What about the nodump attribute?
  • Default policy (for Linux at least) is to dump such files anyway when doing a full dump, but not dump them for incremental dumps.
  • Another way to say this is the nodump attribute is honored for level n dumps if n>1.
  • The dump command has an option to override the default policy (can specify k so that nodump is honored for level n dumps if n>k).

Consistency

• Fsck (file system check) and chkdsk (check disk)
  • If the system crashed, it is possible that not all metadata was written to disk. As a result the file system may be inconsistent. These programs check, and often correct, inconsistencies.
  • Scan all i-nodes (or fat) to check that each block is in exactly one file, or on the free list, but not both.
  • Also check that the number of links to each file (part of the metadata in the file's i-node) is correct (by looking at all directories).
  • Other checks as well.
  • Offers to ``fix'' the errors found (for most errors).
• ``Journaling'' file systems
  • An idea from database theory (transaction logs).
  • Vastly reduces the need for fsck.
  • NTFS has had journaling from day 1.
  • Many Unix systems have it. IBM's AIX converted to journaling in the early 90s.
  • Linux distributions now have journaling (2001-2002).
  • FAT does not have journaling.

6.3.7 File System Performance

Buffer cache or block cache

An in-memory cache of disk blocks.

• Demand paging again!
• Clearly good for reads as it is much faster to read memory than to read a disk.
• What about writes?
  • Must update the buffer cache (otherwise subsequent reads will return the old value).
  • The major question is whether the system should also update the disk block.
  • The simplest alternative is write through in which each write is performed at the disk before it declared complete.
    • Since floppy disk drivers adopt a write through policy, one can remove a floppy as soon as an operation is complete.
    • Write through results in heavy I/O write traffic.
      • If a block is written many times all the writes are sent the disk. Only the last one was ``needed''.
      • If a temporary file is created, written, read, and deleted, all the disk writes were wasted.
    • DOS uses write-through
  • The other alternative is write back in which the disk is not updated until the in-memory copy is evicted (i.e., at replacement time).
    • Much less write traffic than write through.
    • Trouble if a crash occurs.
    • Used by Unix and others for hard disks.
    • Can write dirty blocks periodically, say every minute. This limits the possible damage, but also the possible gain.
    • Ordered writes. Do not write a block containing pointers until the block pointed to has been written. Especially if the block pointed to contains pointers since the version of these pointers on disk may be wrong and you are giving a file pointers to some random blocks.--unofficial for 202

Homework: 29.

Block Read Ahead

When the access pattern ``looks'' sequential read ahead is employed. This means that after completing a read() request for block n of a file. The system guesses that a read() request for block n+1 will shortly be issued so it automatically fetches block n+1.

• How do you decide that the access pattern looks sequential?
  
  • If a seek system call is issued, the access pattern is not sequential.
  • If a process issues consecutive read() system calls for block n-1 and then n, the access patters is guessed to be sequential.
  
  
• What if block n+1 is already in the block cache?
  Ans: Don't issue the read ahead.
  
  
• Would it be reasonable to read ahead two or three blocks?
  Ans: Yes.
  
  
• Would it be reasonable to read ahead the entire file?
  Ans: No, it could easily pollute the cache evicting needed blocks and could waste considerable disk bandwidth.
  
  

Reducing Disk Arm Motion

Try to place near each other blocks that are going to be read in succession.

1. If the system uses a bitmap for the free list, it can allocate a new block for a file close to the previous block (guessing that the file will be accessed sequentially).
   
   
2. The system can perform allocations in ``super-blocks'', consisting of several contiguous blocks.
   • Block cache and I/O requests are still in blocks.
   • If the file is accessed sequentially, consecutive blocks of a super-block will be accessed in sequence and these are contiguous on the disk.
   
   
3. For a unix-like file system, the i-nodes can be placed in the middle of the disk, instead of at one end, to reduce the seek time to access an i-node followed by a block of the file.
4. Can divide the disk into cylinder groups, each of which is a consecutive group of cylinders.
   • Each cylinder group has its own free list and, for a unix-like file system, its own space for i-nodes.
   • If possible, the blocks for a file are allocated in the same cylinder group as is the i-node.
   • This reduces seek time if consecutive accesses are for the same file.

6.3.8: Log-Structured File Systems (unofficial)

• Assumption is that large block caches will eliminate most disk reads so we need to improve writes.
• Buffer writes until have (say) 1MB to write.
• When the buffer is full, write it to the end of the disk (treating the disk as a log).
• Thus writes are sequential and hence fast
• The 1MB units on the disk are called (unfortunately) segments. I will refer to the buffer as the segment buffer.
• A segment can have i-nodes, direct blocks, indirect blocks, blocks forming part of a file, blocks forming part of a directory. In short a segment contains the most recently modified (or created) 1MB of blocks.
• Note that modified blocks are not reclaimed!
• The system keeps a map of where the most recent version of each i-node is located. The map is on disk (but the heavily accessed parts will be in the block cache.
• So the (most up to date) i-node of a file can be found and from that the entire file can be found.
• But the disk will fill with garbage since modified blocks are not reclaimed.
• A ``cleaner'' process runs in the background and examines segments starting from the beginning. It removes overwritten blocks and then adds the remaining blocks to the segment buffer. (This is not trivial.)
• Thus the disk is compacted and is treated like a circular array of segments.

6.4: Example File Systems (watch this space)

6.4.1: CD-ROM File Systems (skipped)

6.4.2: The CP/M File System

This was done above.

6.4.3: The MS-DOS File System

This was done above.

6.4.4: The windows 98 File System

Two changes were made: Long file names were supported and the allocation table was switched from FAT-16 to FAT-32.

1. The only hard part was to keep compatibility with the old 8.3 naming rule. This is called ``backwards compatibility''. A file has two name a long one and an 8.3. If the long name fits the 8.3 format, only one name is kept. If the long name does not fit the 8+3, an 8+3 version is produce via an algorithm, that works but the names produced are not lovely.
   
   
2. FAT-32 used 32 bit words for the block numbers so the fat table could be huge. Windows 98 kept only a portion of the FAT-32 table in memory at a time. (I do not know the replacement policy, number of blocks kept in memory, etc).

6.4.5: The Unix V7 File System

This was done above.

Draw the steps to find /allan/gottlieb/linker.java

6.5: Research on File Systems (skipped)

6.6 Summary (read)

The End: Good luck on the final