repairing sequential consistency in c c 11
play

Repairing Sequential Consistency in C/C++11 Ori Lahav 1 Viktor - PowerPoint PPT Presentation

Nov 12, 2023 •193 likes •696 views

Repairing Sequential Consistency in C/C++11 Ori Lahav 1 Viktor Vafeiadis 1 Jeehoon Kang 2 Chung-Kil Hur 2 Derek Dreyer 1 1 Max Planck Institute for Software Systems (MPI-SWS) 2 Seoul National University PLDI 2017 C11s spectrum of consistency

  1. Repairing Sequential Consistency in C/C++11 Ori Lahav 1 Viktor Vafeiadis 1 Jeehoon Kang 2 Chung-Kil Hur 2 Derek Dreyer 1 1 Max Planck Institute for Software Systems (MPI-SWS) 2 Seoul National University PLDI 2017

  2. C11’s spectrum of consistency Access modes non- release/ relaxed sc � � � atomic acquire

  3. C11’s spectrum of consistency Access modes non- release/ relaxed sc � � � atomic acquire Message passing x := sc 1; a := y sc ; / / 1 x := rlx 1; a := y acq ; / / 1 ∼ = y := sc 1; b := x sc ; / / 0 y := rel 1; b := x rlx ; / / 0 Store buffer x := sc 1; y := sc 1; x := rel 1; y := sc 1; �∼ = a := y sc ; / 0 b := x sc ; / 0 a := y sc ; / / 0 b := x sc ; / / 0 / /

  4. C11’s spectrum of consistency Access modes non- release/ relaxed sc � � � atomic acquire Message passing x := sc 1; a := y sc ; / / 1 x := rlx 1; a := y acq ; / / 1 ∼ = y := sc 1; b := x sc ; / / 0 y := rel 1; b := x rlx ; / / 0 Store buffer x := sc 1; y := sc 1; x := rel 1; y := sc 1; �∼ = a := y sc ; / 0 b := x sc ; / 0 a := y sc ; / / 0 b := x sc ; / / 0 / /

  5. C11’s spectrum of consistency Access modes non- release/ relaxed sc � � � atomic acquire Message passing 1. SC semantics is too strong (new correctness problem!) x := sc 1; a := y sc ; / / 1 x := rlx 1; a := y acq ; / / 1 ∼ = y := sc 1; b := x sc ; / / 0 y := rel 1; b := x rlx ; / / 0 2. SC semantics is too weak (SC-fences) Store buffer x := sc 1; y := sc 1; x := rel 1; y := sc 1; �∼ = a := y sc ; 3. Out-of-thin-air reads (relaxed accesses) / 0 b := x sc ; / 0 a := y sc ; / / 0 b := x sc ; / / 0 / /

  6. C11’s spectrum of consistency Access modes non- release/ relaxed sc � � � atomic acquire Message passing 1. SC semantics is too strong (new correctness problem!) x := sc 1; a := y sc ; / / 1 x := rlx 1; a := y acq ; / / 1 ∼ = y := sc 1; b := x sc ; / / 0 y := rel 1; b := x rlx ; / / 0 We show how to get SC semantics just right! 2. SC semantics is too weak (SC-fences) Store buffer x := sc 1; y := sc 1; x := rel 1; y := sc 1; �∼ = a := y sc ; 3. Out-of-thin-air reads (relaxed accesses) / 0 b := x sc ; / 0 a := y sc ; / / 0 b := x sc ; / / 0 / /

  7. C11’s spectrum of consistency Access modes non- release/ relaxed sc � � � atomic acquire Message passing 1. SC semantics is too strong (new correctness problem!) x := sc 1; a := y sc ; / / 1 x := rlx 1; a := y acq ; / / 1 ∼ = y := sc 1; b := x sc ; / / 0 y := rel 1; b := x rlx ; / / 0 We show how to get SC semantics just right! 2. SC semantics is too weak (SC-fences) Store buffer x := sc 1; y := sc 1; x := rel 1; y := sc 1; �∼ = a := y sc ; 3. Out-of-thin-air reads (relaxed accesses) / 0 b := x sc ; / 0 a := y sc ; / / 0 b := x sc ; / / 0 / /

  8. Semantics of SC-atomics is too strong! Example due to Yatin Manerkar et al. [CoRR abs/1611.01507] a := x acq ; / / 1 c := y acq ; / / 1 x := sc 1; y := sc 1; b := y sc ; / / 0 d := x sc ; / / 0 C/C++11: behavior disallowed

  9. Semantics of SC-atomics is too strong! Example due to Yatin Manerkar et al. [CoRR abs/1611.01507] a := x acq ; / / 1 c := y acq ; / / 1 x := sc 1; y := sc 1; b := y sc ; / / 0 d := x sc ; / / 0 C/C++11: behavior disallowed Compilation of C/C++11 to Power R rlx �→ W rlx �→ ld st R acq �→ W rel �→ lwsync;st ld;lwsync R sc �→ sync;ld;lwsync W sc �→ Leading sync : sync;st R sc �→ W sc �→ lwsync;st;sync Trailing sync : ld; sync

  10. Semantics of SC-atomics is too strong! Example due to Yatin Manerkar et al. [CoRR abs/1611.01507] a := x acq ; / / 1 c := y acq ; / / 1 x := sc 1; y := sc 1; b := y sc ; / / 0 d := x sc ; / / 0 C/C++11: behavior disallowed Compilation of C/C++11 to Power R rlx �→ W rlx �→ ld st R acq �→ W rel �→ lwsync;st ld;lwsync R sc �→ sync;ld;lwsync W sc �→ Leading sync : sync;st R sc �→ W sc �→ lwsync;st;sync Trailing sync : ld; sync Compilation result with “trailing sync ” convention: a := x ; / / 1 c := y ; / / 1 lwsync ; x := 1; y := 1; lwsync ; b := y ; / / 0 sync ; sync ; d := x ; / / 0 sync ; sync ; Power: behavior allowed

  11. Semantics of SC-atomics is too strong! Other examples show unsoundness of: ◮ Leading sync compilation (implemented in GCC and LLVM) ◮ Placing sync both before and after SC-accesses In order to recover the correctness of existing compilers, we suggest to weaken the standard.

  12. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / /

  13. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0)

  14. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 0: choose reads-from Every read reads from a corresponding write.

  15. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 0: choose reads-from Every read reads from a corresponding write.

  16. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 1: calculate happens-before po rf hb hb a : W ⊒ rel b : R ⊒ acq a b a c b b hb hb hb a a b b a c

  17. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 1: calculate happens-before po rf hb hb a : W ⊒ rel b : R ⊒ acq a b a c b b hb hb hb a a b b a c

  18. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 2: “SC-per-location” W na W na x (0) y (0) R acq R acq W sc W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0)

  19. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 2: “SC-per-location” W na W na x (0) y (0) R acq R acq W sc W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0)

  20. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 2: “SC-per-location” W na W na x (0) y (0) R acq R acq W sc W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0)

  21. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) R sc R sc y (0) x (0) Stage 2: “SC-per-location” W na W na x (0) ① y (0) R acq R acq W sc W sc x (1) ④ x (1) ③ y (1) y (1) R sc R sc y (0) x (0) ②

  22. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) sc-per-loc R sc R sc y (0) x (0) Stage 2: “SC-per-location” W na W na x (0) ① y (0) R acq R acq W sc W sc x (1) ④ x (1) ③ y (1) y (1) R sc R sc y (0) x (0) ②

  23. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) sc-per-loc R sc R sc y (0) x (0) Stage 3: global restrictions on SC-accesses Order all SC-accesses while respecting: sc-per-loc hb a : * sc b : W sc a : * sc b : * sc x x sc-order sc-order a a b b

  24. C11’s declarative semantics 101 Batty et al. [POPL’16] a := x acq ; / 1 c := y acq ; / 1 / / x := sc 1; y := sc 1; b := y sc ; / 0 d := x sc ; / 0 / / program order W na W na y (0) x (0) reads from happens-before R acq W sc R acq W sc x (1) x (1) y (1) y (1) sc-per-loc R sc R sc y (0) x (0) sc-order Stage 3: global restrictions on SC-accesses Order all SC-accesses while respecting: sc-per-loc hb a : * sc b : W sc a : * sc b : * sc x x sc-order sc-order a a b b

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Sequential consistency considered harmful Viktor Vafeiadis Max Planck Institute for Software

Sequential consistency considered harmful Viktor Vafeiadis Max Planck Institute for Software

Sequential consistency considered harmful Viktor Vafeiadis Max Planck Institute for Software Systems (MPI-SWS) 7 November 2017 The standard WMC talk . . . Sequential consistency (SC) CPU CPU Interleaving semantics read write

514 views • 15 slides
An introduction to weak memory consistency and the out-of-thin-air problem Viktor Vafeiadis Max

An introduction to weak memory consistency and the out-of-thin-air problem Viktor Vafeiadis Max

An introduction to weak memory consistency and the out-of-thin-air problem Viktor Vafeiadis Max Planck Institute for Software Systems (MPI-SWS) CONCUR, 7 September 2017 Sequential consistency Sequential consistency (SC) The standard

774 views • 73 slides
Separation Logic for Weak Consistency PLMW15 Viktor Vafeiadis Max Planck Institute for

Separation Logic for Weak Consistency PLMW15 Viktor Vafeiadis Max Planck Institute for

Separation Logic for Weak Consistency PLMW15 Viktor Vafeiadis Max Planck Institute for Software Systems (MPI-SWS) 14 January 2015 Sequential consistency Sequential consistency (SC): The standard model for concurrency. Interleave

375 views • 21 slides
C++ 11 Memory Consistency Model Sebastian Gerstenberg NUMA Seminar 07.01.2015 Agenda 1.

C++ 11 Memory Consistency Model Sebastian Gerstenberg NUMA Seminar 07.01.2015 Agenda 1.

C++ 11 Memory Consistency Model Sebastian Gerstenberg NUMA Seminar 07.01.2015 Agenda 1. Sequential Consistency 2. Violation of Sequential Consistency Non-Atomic Operations Instruction Reordering 3. C++ 11 Memory Consistency

554 views • 27 slides
Sequential Consistency Versus Linearizability Hagit A(ya, and

Sequential Consistency Versus Linearizability Hagit A(ya, and

Sequential Consistency Versus Linearizability Hagit A(ya, and Jennifer L. Welch (ACM Transac<ons on Computer Systems 1994) Presented by: Haonan Lu 0

299 views • 8 slides
Vulcan: Hardware Support for Detecting Sequential Consistency Violations Dynamically Abdullah

Vulcan: Hardware Support for Detecting Sequential Consistency Violations Dynamically Abdullah

Vulcan: Hardware Support for Detecting Sequential Consistency Violations Dynamically Abdullah Muzahid , Shanxiang Qi , and Josep Torrellas University of Illinois at Urbana-Champaign http://iacoma.cs.uiuc.edu MICRO December 2012 Sequential

261 views • 24 slides
Multiple- -Writer Distributed Memory Writer Distributed Memory Multiple The Sequential

Multiple- -Writer Distributed Memory Writer Distributed Memory Multiple The Sequential

Multiple- -Writer Distributed Memory Writer Distributed Memory Multiple The Sequential Consistency Memory Model The Sequential Consistency Memory Model sequential processors P3 P1 P2 issue memory ops in program order switch randomly

209 views • 20 slides
Web Cache Consistency Web Cache Consistency Web Cache Consistency Web Cache Consistency

Web Cache Consistency Web Cache Consistency Web Cache Consistency Web Cache Consistency

Web Cache Consistency Web Cache Consistency Web Cache Consistency Web Cache Consistency Requirements of performance, availability, and disconnected operation require us to relax the goal of semantic transparency. - HTTP 1.1 specification

598 views • 13 slides
Consistency - Chapter 5 Introduce several notions of Local Consistency: arc consistency,

Consistency - Chapter 5 Introduce several notions of Local Consistency: arc consistency,

Constraints Course - Justin Pearson - Consistency 1 Consistency - Chapter 5 Introduce several notions of Local Consistency: arc consistency, hyper-arc consistency, k-consistency and strong k-consistency. Local

728 views • 37 slides
Constraint Programming - An overview Node-consistency Arc-consistency Path-consistency

Constraint Programming - An overview Node-consistency Arc-consistency Path-consistency

Constraint Programming - An overview Node-consistency Arc-consistency Path-consistency I-consistency Generalised Arc-consistency Bounds-Consistency Propagators: The COMET example 7 November 2011 Constraint Propagation

615 views • 60 slides
Weak Memory Models: A Tutorial Jade Alglave University College London February 3rd, 2014

Weak Memory Models: A Tutorial Jade Alglave University College London February 3rd, 2014

Weak Memory Models: A Tutorial Jade Alglave University College London February 3rd, 2014 Sequential Consistency A comfortable model for concurrent programming would be Sequential Consistency (SC), as defined by Leslie Lamport in 1979: The

2.4k views • 51 slides
Robustness against Parallel Snapshot Isolation Giovanni Bernardi jointly with Andrea Cerone

Robustness against Parallel Snapshot Isolation Giovanni Bernardi jointly with Andrea Cerone

Robustness against Parallel Snapshot Isolation Giovanni Bernardi jointly with Andrea Cerone Alexey Gotsman 1 / 22 strong consistency weak consistency Eventual Consistency, Sequential DB Causal Consistency, Parallel Snapshot Isolation

361 views • 22 slides
Causal Consistency CS 240: Computing Systems and Concurrency Lecture 16 Marco Canini Credits:

Causal Consistency CS 240: Computing Systems and Concurrency Lecture 16 Marco Canini Credits:

Causal Consistency CS 240: Computing Systems and Concurrency Lecture 16 Marco Canini Credits: Michael Freedman and Kyle Jamieson developed much of the original material. Consistency models Causal Linearizability Eventual Sequential 2

622 views • 39 slides
On Verifying Causal Consistency Ahmed Bouajjani, Constantin Enea, Rachid Guerraoui, Jad Hamza

On Verifying Causal Consistency Ahmed Bouajjani, Constantin Enea, Rachid Guerraoui, Jad Hamza

On Verifying Causal Consistency Ahmed Bouajjani, Constantin Enea, Rachid Guerraoui, Jad Hamza IRIF, Universit e Paris Diderot May 2017 Geo-Replicated Data Structures Strong (sequential) consistency 1 S. Gilbert and N. A. Lynch. Brewers

1.6k views • 95 slides
Shared Memory Consistency Models: A Tutorial By Sarita Adve, Kourosh Gharachorloo WRL

Shared Memory Consistency Models: A Tutorial By Sarita Adve, Kourosh Gharachorloo WRL

Shared Memory Consistency Models: A Tutorial By Sarita Adve, Kourosh Gharachorloo WRL Research Report, 1995 Presentation: Vince Schuster Contents Overview Uniprocessor Review Sequential Consistency Relaxed Memory Models

598 views • 25 slides
Distributed Shared Memory Distributed Shared Memory Systems Page based

Distributed Shared Memory Distributed Shared Memory Systems Page based

Operating Systems DSM Distributed Shared Memory Distributed Shared Memory Systems Page based Shared-variable based Consistency Models Strict consistency Sequential consistency Release consistency

320 views • 11 slides
Problem Definition Machine learning problems, such as image classification or voice recognition,

Problem Definition Machine learning problems, such as image classification or voice recognition,

A Simple Proximal Stochastic Gradient Method for Nonsmooth Nonconvex Optimization Zhize Li , Jian Li IIIS, Tsinghua University https://zhizeli.github.io/ Dec 6th, NeurIPS 2018 Problem Definition Machine learning problems, such as image

615 views • 12 slides
Multilinguism and Linked Open Data in the EU Open Data Portal and other projects of the

Multilinguism and Linked Open Data in the EU Open Data Portal and other projects of the

Multilinguism and Linked Open Data in the EU Open Data Portal and other projects of the Publications Office Audience: W3C Workshop: The Multilingual Web Linked Open Data W3C Workshop: The Multilingual Web Linked Open Data -LT

276 views • 12 slides
UT Participatory Day 2019 Yes, if versus No, unless 15:30 till 16:10 ish 25

UT Participatory Day 2019 Yes, if versus No, unless 15:30 till 16:10 ish 25

https://www.utwente.nl/en/ces/celt/programme-committee/ UT Participatory Day 2019 Yes, if versus No, unless 15:30 till 16:10 ish 25 September 2019 Dr. Hans van den Berg, DipM Centre of Expertise in Learning & Teaching

623 views • 16 slides
K3: Language Design for Building Multi-Platform Domain-Specific Runtimes P.C. Shyamshankar with

K3: Language Design for Building Multi-Platform Domain-Specific Runtimes P.C. Shyamshankar with

DSRs Language Design Annotations Closing K3: Language Design for Building Multi-Platform Domain-Specific Runtimes P.C. Shyamshankar with Zachary Palmer and Yanif Ahmad Department of Computer Science, The Johns Hopkins University First

499 views • 25 slides
Natural Language Processing with Deep Learning CS224N/Ling284 Christopher Manning Lecture 12:

Natural Language Processing with Deep Learning CS224N/Ling284 Christopher Manning Lecture 12:

Natural Language Processing with Deep Learning CS224N/Ling284 Christopher Manning Lecture 12: Information from parts of words: Subword Models Announcements (Changes!!!) Assignment 5 written questions Will be updated tomorrow

687 views • 58 slides
Outline I. Flags what is a flag? who cares? II. (commutative) Generic matrices

Outline I. Flags what is a flag? who cares? II. (commutative) Generic matrices

PO-set paths and q -commuting minors Aaron Lauve LaCIM - UQAM S eminaire de combinatoire et dinformatique math ematique le 7 avril 2006 http://www.lacim.uqam.ca/ lauve (Research) lauve@lacim.uqam.ca Outline I. Flags

819 views • 39 slides
Prelim 1 tonight 7:30 PM NetID ends in 0-6: Olin Hall 155 NetID ends in 7-9: Olin Hall

Prelim 1 tonight 7:30 PM NetID ends in 0-6: Olin Hall 155 NetID ends in 7-9: Olin Hall

Try to leave every 3 rd row (mostly) empty Wall Wall usable seats empty seats Prelim 1 tonight 7:30 PM NetID ends in 0-6: Olin Hall 155 NetID ends in 7-9: Olin Hall 165 Previous Lecture: Introduction to 2-d array matrix

831 views • 42 slides
Purchase Requisitions and Receiving Non Vendor Catalog Revised 12/06/2016 Ground Rules

Purchase Requisitions and Receiving Non Vendor Catalog Revised 12/06/2016 Ground Rules

Purchase Requisitions and Receiving Non Vendor Catalog Revised 12/06/2016 Ground Rules Please put cell phones on vibrate, dont answer in class. If urgent, please step outside the classroom. Please do not perform work including email

474 views • 22 slides

More recommend