A Whirlwind Tour through Concurrency

1
A Whirlwind Tour through Concurrency

• Kedar Namjoshi
• Bell Labs
• Columbia University, Jan. 2008

2
Outline

• Concurrency
• Common Concurrency Patterns
• The Plan9 Thread Model
• Reasoning about Concurrency
• The Session Data Type Model

3
Concurrency

• Processes which overlap in time and exchange
  information
• The world is concurrent. Purely sequential
  behavior is often artificial (like a machine)
• However, writing concurrent programs is viewed as
  a hard problem. Why?

4
Concurrency in programming

• For a long time, concurrency has been a concern
  mainly for designers and implementers of
• operating systems
• supercomputers
• processor chips
• database systems
• and phone networks
• This is about to change!
• Faster Chips Are Leaving Programmers In Their
  Dust
• (The New York Times, December 17, 2007)

5
Multi-Core Processors
(from http//www.amd.com)
Dual-core already common Quad-core on its way
expect more!
6
Shared-Memory Concurrency

• initially, account x holds 0
• P1 x x50 // deposit 50 to x
• P2 x x100 // deposit 100 to x
• What is the amount in x after P1P2?
• It should be 150 --- but it might be 50 !
• How could this happen?

7
Shared-Memory II

• Lets take a closer look
• P1 t1 x t1 t150 x t1
• P2 t2 x t2 t2100 x t2
• For the schedule P2P1P2P1P2P1, the final
  value of x is 50!
• Mistake implicit assumption that the deposit
  action is atomic.

8
Shared-Memory III

• Atomicity must be enforced.
• P1 lock x deposit 50 unlock x
• P2 lock x deposit 100 unlock x
• Locks are difficult to get right.
• programmers forget to place them (data races)
• or forget to release locks, or do so in the wrong
  order (deadlock)
• or put too many locks (minimal concurrency)
• Locking could result in priority inversion (a
  higher priority process gets locked out) --- the
  Mars Pathfinder failure was traced back to
  priority inversion
• Locks often dont match up well to the structure
  of a problem.

9
Shared Memory IV

• Compilers can get in the way, too!
• P1 P2
• x 0 y 0
• y 1 x 1
• On termination, we may expect anything but
  x0,y0. However, the compiler (or the processor)
  is free to reorder instructions!
• P1 P2
• y 1 x 1
• x 0 y 0
• Now x0,y0 is a possibility!
• This leads to considerations of a weak
  consistency model.

10
CSP Hoare 1978

• CSP Communicating Sequential Processes
• Three key concepts
• No shared variables
• Processes communicate through synchronized
  send-receive actions
• A process may respond to one of a set of
  communication actions
• Simple. Clean. Elegant.
• Hoare received the Turing Award in 1980 for
• this and many other contributions to computing

11
Communication Patterns

• Often, the most natural description of a
  concurrent program is through its communication
  pattern.
• Well see a number of very common patterns.

12
1. The knock on the door

• also known as a real-time interrupt
• C Unix signals Java
  Thread.interrupt()
• As captain of the crew I had had extensive
  experience (dating back to 1958) in making basic
  software dealing with real time interrupts and I
  knew by bitter experience that as a result of the
  irreproducibility of the interrupt moments, a
  program error could present itself misleadingly
  like an occasional machine malfunctioning. As a
  result I was terribly afraid. Having fears
  regarding the possibility of debugging we decided
  to be as careful as possible and prevention is
  better than cure! to try to prevent bugs from
  entering the construction.
• -- Edsger W. Dijkstra, The Structure of the
  THEmultiprogramming System (1968) EWD 196
• Dijkstra received the Turing Award in 1972
• for his work on concurrency and structured
• programming

13
How does one prevent bugs?

• The non-determinism introduced by the real-time
  interrupt implies that a program execution may be
  stopped at any point, in any intermediate state.
• The key is to write down the minimal set of facts
  that should be invariant (i.e., unchanged)
  across an interrupt, for the main program to work
  correctly.
• Then write the interrupt handling routine to
  respect the invariant, or disable interrupts over
  a short execution sequence.

14
2. Copy

• process copy(channel in, channel out)
• var x
• in?x ? out!x
• ? is the input action
• ! is the output action
• means repeat forever

15
3. Pipeline

• A chain of N (N 1) copy processes
• copy(c0,c1) copy(c1,c2)
  copy(c(N-1),c(N))
• If the ith process is altered to
• in?x ? out!f(i)(x)
• the pipeline computes the function
• f(N) f(N-1) f(2) f(1)

16
4. Delegation Taxi Rank

• All taxis follow the same protocol.
• process Taxii
• taxi-rank! i dispatchi?p ? drive(p)
  taxi-rank! i
• process Dispatcher
• queue?p ? (taxi-rank?i ? dispatchi!p)
• Also known as a thread pool

17
5. Delegation Futures

• process A
• toB ! (x2,x10)
• . stuff .
• fromB ? y
• Process B
• toB? (expr, binding) ? fromB!
  eval(expr,binding)
• Introduced in Scheme (1975?) as delay and force

18
6. Co-authoring a paper

• process Kedar
• toDennis! token
• fromDennis? token ? write toDennis! token
• process Dennis
• toDennis? token ? write fromDennis! token
• Also known as a co-routine .

19
7. Callback

• K
• C! let me know if you cant pick up the kids
  today
• work interrupt (C? cant do it ? pick up
  kids)
• Best understood as an expected interrupt.
• All warnings regarding interrupt processing
  apply.

20
8. Multiplexing

• Merge messages from two channels into a single
  channel
• process mux
• channel1? x ? out! x
• channel2? y ? out! y
• The operator defines alternatives. If both
  alternatives are available, the choice is made
  non-deterministically.
• The result may not be a fair merge!

21
9. De-Multiplexing

• Copy input from one channel into two
• process demux
• input? x ? out1! x out2! x
• Or, perhaps
• process demux
• input? X ? if (x0) out1!x else out2!x

22
10. On Friday

• K
• C! let me know if you cant pick up the kids
  today
• work interrupt
• (
• C? cant do it ? pick up kids
• Al? would you like to teach a
  class? ? chat
• Dennis? lunch? ? go to lunch
• )
• All warnings regarding interrupt processing
  apply.

23
The Plan9 Thread Model

• Plan9 is an operating system developed at Bell
  Labs from about 1990 onwards.
• Lots of interesting features.
• In particular, the OS thread model influenced by
  CSP.
• CSP also influenced two languages at Bell Labs
• newsqueak (Pike, Cardelli) Rob Pikes talk
• Alef (Winterbottom) Russ Coxs CSP-Plan9 page
• and, in the wider world
• The design of the Ada programming language
• The Occam programming language
• The Transputer chip
• The BPEL web services language
• The CSP book is available.

24
What is the Plan9 thread model?

• Two kinds of entities processes and threads.
• Threads are co-routines within a process (i.e.,
  scheduling is cooperative)
• Processes are pre-emptively scheduled
• Channels are used for communication and
  synchronization. They can be synchronous (as in
  CSP) or buffered (with finite capacity,
  essentially a pipeline)
• Further reading these lecture notes (Sape
  Mullender) and a description of the structure of
  the Plan9 window system (Pike)

25
How is it used?

• Typically, shared data is held in one process,
  and accessed through commands sent through a
  channel (e.g., read, write)
• Hence, no locks. Synchronization is enforced
  through channels. Hence, no race conditions
• Clean fit to the design
• Communication deadlock is possible, but it can
  often be detected early by analyzing the
  communication skeleton

26
Programming Languages Shared-variable concurrency

• High-Performance Fortran (HPF), OpenMP, X10, C
  designed for high-performance, scientific
  computing. forall construct, data layout
  specification
• Java Threads, synchronized methods, external
  locks. Also, lock-free implementations of basic
  data structures in java.util.concurrent
• C/C concurrency provided by external thread
  libraries (e.g., pthreads). C will include
  language constructs for concurrency in its next
  version
• OCaml/F OCaml and F use external thread
  libraries. F has a concurrent garbage collector

27
Programming Languages Message-passing concurrency

• Ada processes (called tasks) communicating
  through rendezvous operations (like CSP ! and
  ?)
• Erlang processes communicating though unbounded
  message queues. Used by Ericsson for programming
  telephony switches
• MPI Message-passing interface. Used for
  wide-area and supercomputer applications

28
Session Data Types Simple Management of the
Life-cycle of Session-oriented Services

• Al Aho, Dennis Dams, Rick Hull, John Letourneau,
  Kedar Namjoshi, Frans Panken, Mans van Tellingen
• Alcatel-Lucent and Columbia University

29
Session Data Types

• A new project at Bell Labs
• a simple model and a (fairly simple) API for
  programming real-time communications
• Targets applications that mix web-based setup and
  control with real-time communications via
  telephony, video, and IM
• Model based around the key concept of a session.

30
The impedance mismatch
Device OSs and platforms
Devices
Service Developer/ Blender
Eclipse-based SCE
SIP, SIP Servlets, SB Steplets,
SIP App Server (including SB)
Web Services App Server
Web Services, SOAP, WSDL,
Other?
. . .

• Programming of converged services programmers
  still think in multiple paradigms
• Web HTTP, SOAP, WSDL, XML
• All-IP Telecom SIP, IMS, Parlay/ParlayX, AIN,
  PacketTV, XMPP,

31
SDT Model

• Three basic building blocks
• Party e.g., kedar_at_cellphone, dennis_at_IMserver,
  aho_at_laptop
• Bubble a group of parties in conversation. The
  group could be empty, or have a single person in
  it (usually a temporary situation)
• Session a group of bubbles, related in some way
• And three key concepts
• Application a control program
• State Machine Each party is limited to behavior
  specified by a state machine over events such as
  pickup, drop, mute
• Notify-Respond the application is notified for
  events it registers for. It can trigger changes
  to a state machine or to a session in response

32
SDT Model
Session Manager
Session 1
Bubble 2
Bubble 1
Global Address Book / Social Network
Bob
people, devices, friends
Alice
Deborah
Charles
33
Hello World in SDT

• // register as a client for a session manger
• ISessionManager sm SessionManager.register(conf
  ig.properties, this)
• // start a session with a predefined audio bubble
• AudioSDT s new AudioSDT(sm)
• AudioBubble b s.getBubble()
• // add yourself and a friend
• b.addParty(new PartyID(Kedar, 19085821891),
  Role.ReadWrite)
• b.addParty(new PartyID(Dennis, 19085828859),
  Role.ReadWrite)
• // talk
• System.out.println(Hello World!)
• Thread.sleep(whatever)
• // exit sm shuts down sessions and bubbles

34
A bit more

• // set up a handler for pickup actions
• public class pickupHandler implements IHandler
• public Response run(TriggerBinding b)
• num_pickup
• return Response.Continue
• // link the handler to the pickup action
• PartyVar who new PartyVar(who)
• Expr e b.mkEvent(AudioBubble.SM.Events.pickup(),
  who)
• b.addTrigger(e, new pickupHandler())

35
And thats it!

• The session/bubble model is simple, yet very
  flexible, thanks to the notify-respond mechanism.
  
• Typically, the application program sets up
  sessions and bubbles and waits for events to
  happen
• The response to an event may create (or delete)
  sessions and bubbles, add (or drop) parties, or
  create new event triggers, all at run-time
• This results in an incredible amount of
  flexibility, out of a very few basic primitives.

36
What we have done

• We have an implementation of this model, running
  on an externally available server
• We have written sample applications, including
• A customizable, network-hosted, speed-conference
  Through a web interface, set your phone buttons
  to perform SDT actions. As the service is
  network-hosted, these bindings work from
  anywhere. subject to some fine print
• A conference system with the ability to create
  whisper sub-sessions

37
What you might do

• Design a scripting language that hides some of
  the Java intricacies
• Design a graphical, web-based language to
  implement your own view of speed dial
• Create a conferencing application which you can
  distribute through Facebook
• or something far more interesting! ?
• We are available throughout the semester for
  help with the API and the system.