InterProcess Communications 1

Outline IPC fundamentals UNIX sockets Remote procedural call Shared memory and large memory spaces IPC in Windows NT Shared memory IPC Shared memory and large address spaces Windows NT IPC Mechanisms 2

IPC Fundamentals What is IPC? Mechanisms to transfer data between processes Why is it needed? Not all important procedures can be easily built in a single process 3

Why do processes communicate? To share resources Client/server paradigms Inherently distributed applications Reusable software components Other good software engineering reasons 4

The Basic Concept of IPC A process needs to send data to a receiving process Sender wants to avoid details of receiver’s condition Receiver wants to get the data in an organized way 5

IPC from the OS Point of View OS address space Process A Process B Private address space 6

Fundamental IPC Problem for the OS Each process has a private address space Normally, no process can write to another process’s space How to get important data from process A to process B? 7

OS Solutions to IPC Problem Fundamentally, two options 1. Support some form of shared address space Shared memory 2. Use OS mechanisms to transport data from one address space to another Files, messages, pipes, RPC 8

Fundamental Differences in OS Treatment of IPC Solutions Shared memory OS has job of setting it up And perhaps synchronizing But not transporting data Messages, etc OS involved in every IPC OS transports data 9

Desirable IPC Characteristics Fast Easy to use Well defined synchronization model Versatile Easy to implement Works Remotely 10

IPC and Synchronization Synchronization is a major concern for IPC Allowing sender to indicate when data is transmitted Allowing receiver to know when data is ready Allowing both to know when more IPC is possible 11

IPC and Connections IPC mechanisms can be connection less or require connection Connectionless IPC mechanisms require no preliminary setup Connection IPC mechanisms require negotiation and setup before data flows 12

Connectionless IPC Data simply flows Typically, no permanent data structures shared in OS by sender and receiver + Good for quick, short communication + Less long-term OS overhead - Less efficient for large, frequent communications - Each communication takes more OS resources per byte 13

Connection-Oriented IPC Sender and receiver pre-arrange IPC delivery details OS typically saves IPC-related information for them Advantages/disadvantages pretty much the opposites of connectionless IPC 14

Basic IPC Mechanisms File system IPC Message-based IPC Procedure call IPC Shared memory IPC 15

IPC Through the File System Sender writes to a file Receiver reads from it But when does the receiver do the read? Often synchronized with file locking or lock files Special types of files can make file-based IPC easier 16

File IPC Diagram Process A Process B Data 17

Message-Based IPC Sender formats data into a formal message With some form of address for receiver OS delivers message to receiver’s message input queue (might signal too) Receiver (when ready) reads a message from the queue Sender might or might not block 18

Message-Based IPC Diagram Process A Process B Data sent from A to B OS B’s message queue 19

File IPC Diagram Process A Process B Data as parameters main () {. call();. }. server();. Data as return values 20

Shared Memory IPC Different processes share a common piece of memory Either physically or virtually Communications via normal reads/writes May need semaphores or locks In or associated with the shared memory 21

Shared Memory IPC Diagram Process A Process B write variable x main () {. x = 10. }. print(x);. read variable x x: 10 22

Synchronizing in IPC How do sending and receiving process synchronize their communications? Many possibilities Based on which process block when Examples that follow in message context, but more generally applicable 23

Blocking Send, Blocking Receive Both sender and receiver block Sender blocks till receiver receives Receiver blocks until sender sends Often called message rendezvous 24

Non-Blocking Send, Blocking Receive Sender issues send, can proceed without waiting to discover fate of message Receiver waits for message arrival before proceeding Essentially, receiver is message-driven 25

Non-Blocking Send, Non-Blocking Receive Neither party blocks Sender proceeds after sending message Receiver works until message arrives Either receiver periodically checks in non- blocking fashion Or some form of interrupt delivered 26

Addressing in IPC How does the sender specify where the data goes? In some cases, the mechanism makes it explicit (e.g., shared memory and RPC) In others, there are options 27

Direct Addressing Sender specifies name of the receiving process Using some form of unique process name Receiver can either specify name of expected sender Or take stuff from anyone 28

Indirect Addressing Data is sent to queues, mailboxes, or some other form of shared data structure Receiver performs some form of read operations on that structure Much more flexible than direct addressing 29

Duality in IPC Mechanisms Many aspects of IPC mechanisms are duals of each other Which implies that these mechanisms have the same power First recognized in context of messages vs. procedure calls At least, IPC mechanisms can be simulated by each other 30

So which IPC mechanism to build/choose/use? Depends on model of computation And on philosophy of user In particular cases, hardware or existing software may make one perform better 31

Typical UNIX IPC Mechanisms Different versions of UNIX introduced different IPC mechanisms Pipes Message queues Semaphores Shared memory Sockets RPC 32

Pipes Only IPC mechanism in early UNIX systems (other than files) Uni-directional Unformatted Un interpreted Inter process byte streams Accessed in file-like way 33

Pipe Details One process feeds bytes into pipe A second process reads the bytes from it Potentially blocking communication mechanism Requires close cooperation between processes to set up Named pipes allow more flexibility 34

Pipes and Blocking Writing more bytes than pipe capacity blocks the sender Until the receiver reads some of them Reading bytes when none are available blocks the receiver Until the sender writes some Single pipe can’t cause deadlock 35

UNIX Message Queues Introduced in System V Release 3 UNIX Like pipes, but data organized into messages Message component include: Type identifier Length Data 36

Semaphores Also introduced in System V Release 3 UNIX Mostly for synchronization only Since they only communicate one bit of information Often used in conjunction with shared memory 37

UNIX Shared Memory Also introduced in System V Release 3 Allows two or more processes to share some memory segments With some control over read/write permissions Often used to implement threads packages for UNIX 38

Sockets Introduced in 4.3 BSD A socket is an IPC channel with generated endpoints Great flexibility in it characteristics Intended as building block for communication Endpoints established by the source and destination processes 39

UNIX Remote Procedure Calls Procedure calls from one address space to another On the same or different machines Requires cooperation from both processes In UNIX, often built on sockets Often used in client/server computing 40

More on Sockets Created using the socket() system call Specifying domain, type, and protocol Sockets can be connected or connectionless Each side responsible for proper setup/access 41

Socket Domains the socket domain describes a protocol family used by the socket Generally related to the address family Domains can be: Internal protocols Internet protocols IMP link layer protocols 42

Socket Types The socket type describes what the socket does Several types are defined SOCK_STREAM SOCK_DGRAM SOCK_SEQPACKET SOCK_RAW SOCK_RDM 43

Socket Protocols This parameter specifies a particular protocol to be used by the socket Must match other parameters Not all protocols usable with all domains and types Generally, only one protocol per socket type available 44

Some Examples of Sockets Socket streams Socket sequential packets Socket datagrams 45

Socket Streams Of type SOCK_STREAM Full-duplex reliable byte streams Like 2-way pipes Requires other side to connect 46

Socket Sequential Packets Similar to streams But for fixed-sized packets So reads always return a fixed number of bytes Allow easy use of buffers 47

Socket Datagrams Like sequential packets But non-reliable delivery Which implies certain simplifications And lower overhead send(), rather than write(), used to send data 48

Socket Options Connection or connectionless Blocking or non-blocking Out-of-band information Broadcast Buffer sizes (input and output) Routing options And others 49

Binding Sockets Binding is the process of preparing a socket for use by a process Sockets are typically bound to local names For IPC on a single machine Often accessed using descriptors Requires clean-up when done Binding can be to IP addresses, as well 50

Connecting to Sockets Method for setting up the receiving end of a socket In local domain, similar to opening file Multiple clients can connect to a socket Program establishing socket can limit connections 51

Remote Procedural Call Method of calling procedures in other address spaces Either on the same machine Or other machines Attempts to provide interface just like local procedure call Request/reply communications model 52

Semantics of RPC Similar to regular procedure call 1. Calling procedure blocks 2. Set of parameters transferred to called procedure 3. Called procedure computes till it returns 4. Return value delivered to calling procedure 5. Calling procedure continues 53

High-Level RPC Mechanics Hide details from applications Clients pass requests to stub programs Client-end stub sends request to server stub Server-end stub calls user-level server Results travel in reverse direction Network transport or OS actually moves data 54

Diagram of RPC in Action Server (callee)Client (caller) OS or Network Server stub Client stub callreplycallreply 55

What do the stubs do? Stubs handle complex details like: Marshaling arguments Message construction Data format translation Finding the server process 56

Setting Up RPC Caller must have a way to find the called procedure But it can’t be found at link time Unlike local procedure Potential servers must make their presence known 57

Registering Servers Either register the server at a “well-known” port Or register it with a name server Which in turn is at a “well-known” port Calling procedure “addresses” RPC to that port 58

Binding to a Service A client process binds to the service it wants Two major sub-problem: Naming Location 59

Binding: The Naming Problem How does a caller name the server program? Depends on the RPC mechanism And perhaps the protocol or type within it Do you name the server explicitly? Or do you name the service and let some other authority choose which server? 60

Binding: The Location Problem Where is the remote server? Some naming schemes make it explicit Some don’t If it’s not explicit, system must convert symbolic names to physical locations 61

Binding in Cedar RPC Client applications bind to a symbolic name Composed of: Type Instance Type is which kind of service Instance is which particular implementer 62

Locating Cedar RPC Service Names do not contain physical location So Cedar consults a database of services Services register with database Binding call automatically looks up location in database 63

Binding in UNIX RPC bind() system call used by servers Allows servers to bind naming/location information to particular socket connect() system call used by clients Using information similar to that bound by the server Automatic code generation hides details 64

UNIX Binding Information Like most socket operations, it’s flexible Fill in all or part of a socket address data structure Create an appropriate socket Then call bind to link socket to address information connect works similarly 65

UNIX Binding Example On server side, struct sockaddr_un sin; int sd; strcpy(sin.sun_path,”./socket”); sd = socket(AF_UNIX, SOCK_STREAM, 0); bind(sd, &sin, sizeof(sin)); 66

UNIX Binding Example, Con’t For client side, struct sockaddr_un sin; int sd; strcpy(sin.sun_path, “./socket”); sd = socket(AF_UNIX, SOCK_STREAM, 0); connect(sd, &sin, sizeof(sin)); 67

Locating Remove Services in UNIX RPC Similar to Cedar methods Register services with the portmapper The portmapper is typically called through automatically generated code the portmapper runs on each machine Another service (e.g., NIS) deals with intermachine requests 68

Using RPC Once it’s bound, how does the client use RPC? Just call the routines As if they were local And the results come back 69

What’s happening under the covers? When a client calls a remote routine, he really calls a local stub program Stub program packages up request to remote server And sends it to local transport code When reply arrives, local transport code returns results to stub Which returns to client program 70

What happens at the server side? A request comes in to the RPC transport code It routes it to the appropriate server stub Which converts it into a local procedure call Which is made within the context of the server 71

Conceptual Diagram of RPC call_procedure(); Message from client stub to server stub call_procedure(); Message received by server stub Network transport client program continues Return value formatted as message to client stub Network transportClient stub converts message to return value return(…); 72

Transport for RPC In Cedar, special-purpose RPC transport protocol In UNIX RPC, can use either UDP or TCP for transport Typically protocol is chosen automatically by stub generator program 73

Other RPC Issues Mostly related to intermachine RPC Data format conversions Security and authentication Locating remote services Multipacket RPC 74

Shared Memory A simple and powerful form of IPC Most multiprogramming OS’s use some form of shared memory E.g., sharing executables Not all OS’s make shared memory available to applications 75

Shared Memory Diagram Process A Process B x: 10 y: 20 z: __ a: __b: __ 76

Problems with Shared Memory Synchronization Protection Pointers 77

Synchronization Shared memory itself does not provide synchronization of communications Except at the single-word level Typically, some other synchronization mechanism is used E.g., semaphore in UNIX Events, semaphores, or hardware locks in Windows NT 78

Protection Who can access a segment? And in what ways? UNIX allows some read/write controls Windows NT has general security monitoring based on the object-status of shared memory 79

Pointers in Shared Memory Pointers in a shared memory segment can be troublesome For that matter, pointers in any IPC can be troublesome 80

Shared Memory Containing Pointers Process A Process B x: 10 y: 20 z: __ a: __b: __ w: 5 81

A Troublesome Pointer Process A Process B x: 10 y: 20 z: __ a: __b: __ w: 5 82

So, how do you share pointers? Several methods are in use Copy-time translation Reference-time translation Pointer swizzling All involve somehow translating pointers at some point before they are used 83

Copy-Time Pointer Translation When a process sends data containing pointers to another process Locate each pointer within old version of the data Then translate pointers are required Requires both sides to traverse entire structure Not really feasible for shared memory 84

Reference-Time Translation Encode pointers in shared memory segment as pointer surrogates Typically as offsets into some other segment in separate contexts So each sharer can have its own copy of what is pointed to Slow, pointers in two formats 85

Pointer Swizzling Like reference-time, but cache results in the memory location Only first reference is expensive But each sharer must have his own copy Must “unswizzle” pointers to transfer data outside of local context Stale swizzled pointers can cause problems 86

Shared Memory in a Wide Virtual Address Space When virtual memory was created, 16 or 32 bit addresses were available Reasonable size for one process But maybe not for all processes on a machine And certainly not for all processes ever on a machine 87

Wide Address Space Architectures Computer architects can now give us 64-bit virtual addresses A 64-bit address space, consumed at 100 MB/sec, lasts 5000 years Orders of magnitude beyond any process’s needs 40 bits can address a TB 88

Do we care? Should OS designers care about wide address space? Well, what can we do with them? One possible answer: Put all processes in the same address space Maybe all processes for all time? 89

Implications of Single Shared Address Space IPC is trivial Shared memory, RPC Separation of concepts of address space and protection domain Uniform address space 90

Address Space and Protection Domain A process has a protection domain The data that cannot be touched by other processes And an address space The addresses it can generate and access In standard systems, these concepts are merged 91

Separating the Concepts These concepts are potentially orthogonal Just because you can issue an address doesn’t mean you can access it (Though clearly to access an address you must be able to issue it) Existing hardware can support this separation 92

Context-Independent Addressing Addresses mean the same thing in any execution context So, a given address always refers to the same piece of data Key concept of uniform-address systems Allows many OS optimizations/improvements 93

Uniform-Addressing Allows Easy Sharing Any process can issue any address So any data can be shared All that’s required is changing protection to permit desired sharing Suggests programming methods that make wider use of sharing 94

To Opal System New OS using uniform-addressing Developed at University of Washington Not intended as slight alteration to existing UNIX system Most of the rest of material specific to Opal 95

Protection Mechanisms for Uniform- Addressing Protection domains are assigned portions of the address space They can allow other protection domains to access them Read-only Transferable access permissions System-enforced page-level locking 96

Program Execution in Uniform- Access Memory Executing a program creates a new protection domain The new domain is assigned an unused portion of the address space But it may also get access to used portions E.g., a segment containing the required executable image 97

Virtual Segments Global address space is divided into segments Each composed of variable number of contiguous virtual pages Domains can only access segments they attach to Attempting to access unattached segment causes a segment fault 98

Persistent Memory in Opal Persistent segments exist even when attached to no current domain Recoverable segments are permanently stored And can thus survive crashes All Opal segments can be persistent and recoverable Pointers can thus live forever on disk 99

Code Modules in Opal Executable code stored in modules Independent of protection domains Pure modules can be easily shared Because they are essentially static Can get benefit of dynamic loading without run-time linking 100

Address Space Reclamation Trivial in non-uniform-address systems Tricky in uniform-address systems Problem akin to reclaiming i_nodes in the presence of hard links But even if segments improperly reclaimed, only trusting domains can be hurt 101

Windows NT IPC Inter-thread communications Within a single process Local procedure calls Between processes on same machine Shared memory 102

Windows NT and Threads Windows NT support multiple threads of control in a single process address space Threads share address space So communication among them is through memory 103

Windows NT and Client/Server Computing Windows NT strongly supports the client/server model of computing Various OS services are built as servers, rather than part of the kernel Windows NT needs facilities to support client/server operations Which guide users to building client/server solution 104

Client/Server Computing and RPC In client/server computing, clients request services from servers Service can be requested in many ways But RPC is a typical way Windows NT uses a specialized service for single machine RPC 105

Local Procedure Call (LPC) Similar in many ways to RPC But optimized to only work on a single machine Primarily used to communicate with protected subsystems Windows NT also provides a true RPC facility for genuinely distributed computing 106

Basic Flow of Control in Windows NT LPC Application calls routine in an application programming interface Which is usually in a dynamically linked library Which sends a message to the server through a messaging mechanism 107

Windows NT LPC Messaging Mechanisms Messages between port objects Message pointers into shared memory Using dedicated shared memory segments 108

Port Objects Windows NT is generally object-oriented Port objects support communications Two types: Connection ports Communication ports 109

Connection Ports Used to establish connections between clients and servers Named, so they can be located Only used to set up communication ports 110

Communication Ports Used to actually pass data Created in pairs, between given client and given server Private to those two processes Destroyed when communications end 111

Windows NT Port Example Client process Server process Connection port 112

Windows NT Port Example Client process Server process Connection port 113

Communication ports Windows NT Port Example Client process Server process Connection port 114

Communication ports Windows NT Port Example Client process Server process Connection port Send request 115

Message Passing through Port Object Message Queues One of three methods in Windows NT to pass messages 1. Client submits message to OS 2. OS copies to receiver’s queue 3. Receiver copies from queue to its own address space 116

Characteristics of Message Passing via Queues Two message copies required Fixed-sized, fairly short message ~256 bytes Port objects stored in system memory So always accessible to OS Fixed number of entries in message queue 117

Message Passing Through Shared Memory Used for messages larger than 256 bytes Client must create section object Shared memory segment Of arbitrary size Message goes into the section Pointer to message sent to receiver’s queue 118

Setting up Section Objects Pre-arranged through OS calls Using virtual memory to map segment into both sender and receiver’s address space If replies are large, need another segment for the receiver to store responses OS doesn’t format section objects 119

Characteristics of Message Passing via Shared Memory Capable of handling arbitrarily large transfers Sender and receiver can share a single copy of data i.e., data copied only once Requires pre-arrangement for section object 120

Server Handling of Requests Windows NT servers expect requests from multiple clients Typically, they have multiple threads to handle requests Must be sufficiently general to handle many different ports and section objects 121

Message Passing Through Quick LPC Third way to pass messages in Windows NT Used exclusively with Win32 subsystem Like shared memory, but with a key difference Dedicated resources 122

Use of Dedicated Resources in Quick LPC To avoid overhead of copying Notification messages to port queue And thread switching overhead Client sets up dedicated server thread only for its use Also dedicated 64KB section object And event pair object for synchronization 123

Characteristics of Message Passing via Quick LPC Transfers of limited size Very quick Minimal copying of anything Wasteful of OS resources 124

Shared Memory in Windows NT Similar in most ways to other shared memory services Windows NT runs on multiprocessors, which complicates things Done through virtual memory 125

Shared Memory Sections Block of memory shared by two or more processes Created with unique name Can be very, very large 126

Section Views Given process might not want to waste lots of its address space on big sections So a process can have a view into a shared memory section Different processes can have different views of same section Or multiple views for single process 127

Shared Memory View Diagram Process A Process B Physical memory Section view 1view 2 view 3 128