- Timestamp:
- Jun 3, 2020, 2:35:13 PM (4 years ago)
- Branches:
- ADT, arm-eh, ast-experimental, enum, forall-pointer-decay, jacob/cs343-translation, master, new-ast, new-ast-unique-expr, pthread-emulation, qualifiedEnum
- Children:
- fe9cf9e
- Parents:
- 4e7c0fc0
- Location:
- doc/papers
- Files:
-
- 1 added
- 3 edited
Legend:
- Unmodified
- Added
- Removed
-
doc/papers/AMA/AMA-stix/ama/WileyNJD-v2.cls
r4e7c0fc0 r04b4a71 2444 2444 \@afterheading} 2445 2445 2446 \renewcommand\section{\@startsection{section}{1}{\z@}{-2 5pt \@plus -2pt \@minus -2pt}{12\p@}{\sectionfont}}%2447 \renewcommand\subsection{\@startsection{subsection}{2}{\z@}{- 22pt \@plus -2pt \@minus -2pt}{5\p@}{\subsectionfont}}%2446 \renewcommand\section{\@startsection{section}{1}{\z@}{-20pt \@plus -2pt \@minus -2pt}{8\p@}{\sectionfont}}% 2447 \renewcommand\subsection{\@startsection{subsection}{2}{\z@}{-18pt \@plus -2pt \@minus -2pt}{5\p@}{\subsectionfont}}% 2448 2448 \renewcommand\subsubsection{\@startsection{subsubsection}{3}{\z@}{-20pt \@plus -2pt \@minus -2pt}{2\p@}{\subsubsectionfont}}% 2449 2449 % -
doc/papers/concurrency/Paper.tex
r4e7c0fc0 r04b4a71 99 99 \newcommand{\CRT}{\global\columnposn=\gcolumnposn} 100 100 101 % Denote newterms in particular font and index them without particular font and in lowercase, e.g.,\newterm{abc}.102 % The option parameter provides an index term different from the new term, e.g.,\newterm[\texttt{abc}]{abc}101 % Denote newterms in particular font and index them without particular font and in lowercase, \eg \newterm{abc}. 102 % The option parameter provides an index term different from the new term, \eg \newterm[\texttt{abc}]{abc} 103 103 % The star version does not lowercase the index information, e.g., \newterm*{IBM}. 104 104 \newcommand{\newtermFontInline}{\emph} … … 235 235 {\lstset{language=python,moredelim=**[is][\protect\color{red}]{`}{`},#1}\lstset{#1}} 236 236 {} 237 \lstnewenvironment{java}[1][] 238 {\lstset{language=java,moredelim=**[is][\protect\color{red}]{`}{`},#1}\lstset{#1}} 239 {} 237 240 238 241 % inline code @...@ … … 272 275 % Library extension for executors, futures, and actors are built on these basic mechanisms. 273 276 The runtime provides significant programmer simplification and safety by eliminating spurious wakeup and monitor barging. 274 The runtime also ensures multiple monitors can be safely acquired \emph{simultaneously} (deadlock free), and this feature is fully integrated with all monitor synchronization mechanisms.277 The runtime also ensures multiple monitors can be safely acquired in a deadlock-free way, and this feature is fully integrated with all monitor synchronization mechanisms. 275 278 All control-flow features integrate with the \CFA polymorphic type-system and exception handling, while respecting the expectations and style of C programmers. 276 279 Experimental results show comparable performance of the new features with similar mechanisms in other concurrent programming languages. … … 293 296 % "Trait inheritance" works for me. "Interface inheritance" might also be a good choice, and distinguish clearly from implementation inheritance. 294 297 % You'll want to be a little bit careful with terms like "structural" and "nominal" inheritance as well. CFA has structural inheritance (I think Go as well) -- it's inferred based on the structure of the code. Java, Rust, and Haskell (not sure about Swift) have nominal inheritance, where there needs to be a specific statement that "this type inherits from this type". 295 However, functions \emph{cannot} be nested in structures, so there is no lexical binding between a structure and set of functions (member/method)implemented by an implicit \lstinline@this@ (receiver) parameter.},298 However, functions \emph{cannot} be nested in structures, so there is no lexical binding between a structure and set of functions implemented by an implicit \lstinline@this@ (receiver) parameter.}, 296 299 backwards-compatible extension of the C programming language. 297 In many ways, \CFA is to C as Scala~\cite{Scala} is to Java, providing a \emph{research vehicle}for new typing and control-flow capabilities on top of a highly popular programming language\footnote{298 The TIOBE index~\cite{TIOBE} for December 2019 ranks the top five \emph{popular} programming languages as Java 17\%, C 16\%, Python 10\%, and \CC 6\%, \Csharp 5\% = 54\%, and over the past 30 years, C has always ranked either first or second in popularity.}300 In many ways, \CFA is to C as Scala~\cite{Scala} is to Java, providing a vehicle for new typing and control-flow capabilities on top of a highly popular programming language\footnote{ 301 The TIOBE index~\cite{TIOBE} for May 2020 ranks the top five \emph{popular} programming languages as C 17\%, Java 16\%, Python 9\%, \CC 6\%, and \Csharp 4\% = 52\%, and over the past 30 years, C has always ranked either first or second in popularity.} 299 302 allowing immediate dissemination. 300 303 This paper discusses the design philosophy and implementation of advanced language-level control-flow and concurrent/parallel features in \CFA and its runtime, which is written entirely in \CFA. … … 304 307 % the concurrency extensions allow high-level management of threads. 305 308 306 Call/return control-flow with argument /parameter passing appeared in the first programming languages.309 Call/return control-flow with argument and parameter passing appeared in the first programming languages. 307 310 Over the past 50 years, call/return has been augmented with features like static/dynamic call, exceptions (multi-level return) and generators/coroutines (retain state between calls). 308 While \CFA has mechanisms for dynamic call (algebraic effects ) and exceptions\footnote{311 While \CFA has mechanisms for dynamic call (algebraic effects~\cite{Zhang19}) and exceptions\footnote{ 309 312 \CFA exception handling will be presented in a separate paper. 310 The key feature that dovetails with this paper is nonlocal exceptions allowing exceptions to be raised across stacks, with synchronous exceptions raised among coroutines and asynchronous exceptions raised among threads, similar to that in \uC~\cite[\S~5]{uC++}}, this work only discusses retaining state between calls via generators /coroutines.313 The key feature that dovetails with this paper is nonlocal exceptions allowing exceptions to be raised across stacks, with synchronous exceptions raised among coroutines and asynchronous exceptions raised among threads, similar to that in \uC~\cite[\S~5]{uC++}}, this work only discusses retaining state between calls via generators and coroutines. 311 314 \newterm{Coroutining} was introduced by Conway~\cite{Conway63} (1963), discussed by Knuth~\cite[\S~1.4.2]{Knuth73V1}, implemented in Simula67~\cite{Simula67}, formalized by Marlin~\cite{Marlin80}, and is now popular and appears in old and new programming languages: CLU~\cite{CLU}, \Csharp~\cite{Csharp}, Ruby~\cite{Ruby}, Python~\cite{Python}, JavaScript~\cite{JavaScript}, Lua~\cite{Lua}, \CCtwenty~\cite{C++20Coroutine19}. 312 315 Coroutining is sequential execution requiring direct handoff among coroutines, \ie only the programmer is controlling execution order. … … 314 317 Coroutines are only a stepping stone towards concurrency where the commonality is that coroutines and threads retain state between calls. 315 318 316 \Celeven/\CCeleven define concurrency~\cite[\S~7.26]{C11}, but it is largely wrappers for a subset of the pthreads library~\cite{Pthreads}.\footnote{Pthreads concurrency is based on simple thread fork /join in a function and mutex/condition locks, which is low-level and error-prone}317 Interestingly, almost a decade after the \Celeven standard, neither gcc-9, clang-9 nor msvc-19 (most recent versions)support the \Celeven include @threads.h@, indicating no interest in the C11 concurrency approach (possibly because of the recent effort to add concurrency to \CC).319 \Celeven/\CCeleven define concurrency~\cite[\S~7.26]{C11}, but it is largely wrappers for a subset of the pthreads library~\cite{Pthreads}.\footnote{Pthreads concurrency is based on simple thread fork and join in a function and mutex or condition locks, which is low-level and error-prone} 320 Interestingly, almost a decade after the \Celeven standard, the most recent versions of gcc, clang, and msvc do not support the \Celeven include @threads.h@, indicating no interest in the C11 concurrency approach (possibly because of the recent effort to add concurrency to \CC). 318 321 While the \Celeven standard does not state a threading model, the historical association with pthreads suggests implementations would adopt kernel-level threading (1:1)~\cite{ThreadModel}, as for \CC. 319 322 In contrast, there has been a renewed interest during the past decade in user-level (M:N, green) threading in old and new programming languages. … … 323 326 As a result, many current languages implementations adopt the 1:1 kernel-threading model, like Java (Scala), Objective-C~\cite{obj-c-book}, \CCeleven~\cite{C11}, C\#~\cite{Csharp} and Rust~\cite{Rust}, with a variety of presentation mechanisms. 324 327 From 2000 onwards, several language implementations have championed the M:N user-threading model, like Go~\cite{Go}, Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, D~\cite{D}, and \uC~\cite{uC++,uC++book}, including putting green threads back into Java~\cite{Quasar}, and many user-threading libraries have appeared~\cite{Qthreads,MPC,Marcel}. 325 The main argument for user-level threading is that it is lighter weight than kernel threading (locking and context switching do not cross the kernel boundary), so there is less restriction on programming styles that encourages large numbers of threads performing medium-sized work to facilitate load balancing by the runtime~\cite{Verch12}.328 The main argument for user-level threading is that it is lighter weight than kernel threading because locking and context switching do not cross the kernel boundary, so there is less restriction on programming styles that encourages large numbers of threads performing medium-sized work to facilitate load balancing by the runtime~\cite{Verch12}. 326 329 As well, user-threading facilitates a simpler concurrency approach using thread objects that leverage sequential patterns versus events with call-backs~\cite{Adya02,vonBehren03}. 327 Finally, performant user-threading implementations (both time and space)meet or exceed direct kernel-threading implementations, while achieving the programming advantages of high concurrency levels and safety.330 Finally, performant user-threading implementations, both time and space, meet or exceed direct kernel-threading implementations, while achieving the programming advantages of high concurrency levels and safety. 328 331 329 332 A further effort over the past two decades is the development of language memory models to deal with the conflict between language features and compiler/hardware optimizations, \eg some language features are unsafe in the presence of aggressive sequential optimizations~\cite{Buhr95a,Boehm05}. 330 The consequence is that a language must provide sufficient tools to program around safety issues, as inline and library code is all sequential to the compiler.331 One solution is low-level qualifiers and functions (\eg @volatile@ and atomics) allowing \emph{programmers} to explicitly write safe (race-free~\cite{Boehm12})programs.332 A safer solution is high-level language constructs so the \emph{compiler} knows the concurrency boundaries (where mutual exclusion and synchronization are acquired/released)and provide implicit safety at and across these boundaries.333 The consequence is that a language must provide sufficient tools to program around safety issues, as inline and library code is compiled as sequential without any explicit concurrent directive. 334 One solution is low-level qualifiers and functions, \eg @volatile@ and atomics, allowing \emph{programmers} to explicitly write safe, race-free~\cite{Boehm12} programs. 335 A safer solution is high-level language constructs so the \emph{compiler} knows the concurrency boundaries, \ie where mutual exclusion and synchronization are acquired and released, and provide implicit safety at and across these boundaries. 333 336 While the optimization problem is best known with respect to concurrency, it applies to other complex control-flow, like exceptions and coroutines. 334 337 As well, language solutions allow matching the language paradigm with the approach, \eg matching the functional paradigm with data-flow programming or the imperative paradigm with thread programming. 335 338 336 Finally, it is important for a language to provide safety over performance \emph{as the default}, allowing careful reduction of safety (unsafe code)for performance when necessary.337 Two concurrency violations of this philosophy are \emph{spurious wakeup} (random wakeup~\cite[\S~9]{Buhr05a}) and \emph{barging}\footnote{339 Finally, it is important for a language to provide safety over performance \emph{as the default}, allowing careful reduction of safety for performance when necessary. 340 Two concurrency violations of this philosophy are \emph{spurious or random wakeup}~\cite[\S~9]{Buhr05a}) and \emph{barging}\footnote{ 338 341 Barging is competitive succession instead of direct handoff, \ie after a lock is released both arriving and preexisting waiter threads compete to acquire the lock. 339 342 Hence, an arriving thread can temporally \emph{barge} ahead of threads already waiting for an event, which can repeat indefinitely leading to starvation of waiter threads. 340 } (signals-as-hints~\cite[\S~8]{Buhr05a}), where one is a consequence of the other, \ie once there is spurious wakeup, signals-as-hints follow.343 } or signals-as-hints~\cite[\S~8]{Buhr05a}, where one is a consequence of the other, \ie once there is spurious wakeup, barging follows. 341 344 (Author experience teaching concurrency is that students are confused by these semantics.) 342 345 However, spurious wakeup is \emph{not} a foundational concurrency property~\cite[\S~9]{Buhr05a}; … … 356 359 357 360 \item 358 monitor synchronization without barging, and the ability to safely acquiring multiple monitors \emph{simultaneously} (deadlock free), while seamlessly integrating these capabilities with all monitor synchronization mechanisms,361 monitor synchronization without barging, and the ability to safely acquiring multiple monitors in a deadlock-free way, while seamlessly integrating these capabilities with all monitor synchronization mechanisms, 359 362 360 363 \item … … 368 371 369 372 \item 370 a dynamic partitioning mechanism to segregate groups of executing user and kernel threads performing specialized work (\eg web-server or compute engine) or requiring different scheduling (\eg NUMA or real-time).373 a dynamic partitioning mechanism to segregate groups of executing user and kernel threads performing specialized work, \eg web-server or compute engine, or requiring different scheduling, \eg NUMA or real-time. 371 374 372 375 % \item … … 380 383 Section~\ref{s:StatefulFunction} begins advanced control by introducing sequential functions that retain data and execution state between calls producing constructs @generator@ and @coroutine@. 381 384 Section~\ref{s:Concurrency} begins concurrency, or how to create (fork) and destroy (join) a thread producing the @thread@ construct. 382 Section~\ref{s:MutualExclusionSynchronization} discusses the two mechanisms to restricted nondeterminism when controlling shared access to resources (mutual exclusion) and timing relationships among threads (synchronization).385 Section~\ref{s:MutualExclusionSynchronization} discusses the two mechanisms to restricted nondeterminism when controlling shared access to resources, called mutual exclusion, and timing relationships among threads, called synchronization. 383 386 Section~\ref{s:Monitor} shows how both mutual exclusion and synchronization are safely embedded in the @monitor@ and @thread@ constructs. 384 Section~\ref{s:CFARuntimeStructure} describes the large-scale mechanism to structure (cluster)threads and virtual processors (kernel threads).387 Section~\ref{s:CFARuntimeStructure} describes the large-scale mechanism to structure threads and virtual processors (kernel threads). 385 388 Section~\ref{s:Performance} uses a series of microbenchmarks to compare \CFA threading with pthreads, Java 11.0.6, Go 1.12.6, Rust 1.37.0, Python 3.7.6, Node.js 12.14.1, and \uC 7.0.0. 386 389 … … 392 395 To this end, the control-flow features created for \CFA are based on the fundamental properties of any language with function-stack control-flow (see also \uC~\cite[pp.~140-142]{uC++}). 393 396 The fundamental properties are execution state, thread, and mutual-exclusion/synchronization (MES). 394 These independent properties can be used alone, in pairs, or in triplets to compose different language features, forming a compositional hierarchy where the most advanced feature has all the properties (state/thread/MES).395 While it is possible for a language to only support the most advanced feature~\cite{Hermes90}, this unnecessarily complicates and makes inefficient solutions to certain classes of problems.396 As is shown, each of the (non-rejected) composedfeatures solves a particular set of problems, and hence, has a defensible position in a programming language.397 If a compositional feature is missing, a programmer has too few /manyfundamental properties resulting in a complex and/or is inefficient solution.397 These independent properties can be used to compose different language features, forming a compositional hierarchy, where the combination of all three is the most advanced feature, called a thread/task/process. 398 While it is possible for a language to only provide threads for composing programs~\cite{Hermes90}, this unnecessarily complicates and makes inefficient solutions to certain classes of problems. 399 As is shown, each of the non-rejected composed language features solves a particular set of problems, and hence, has a defensible position in a programming language. 400 If a compositional feature is missing, a programmer has too few fundamental properties resulting in a complex and/or is inefficient solution. 398 401 399 402 In detail, the fundamental properties are: 400 403 \begin{description}[leftmargin=\parindent,topsep=3pt,parsep=0pt] 401 404 \item[\newterm{execution state}:] 402 is the state information needed by a control-flow feature to initialize, manage compute data and execution location(s), and de-initialize .403 State is retained in fixed-sized aggregate structures and dynamic-sized stack(s), often allocated in the heap(s) managed by the runtime system.405 is the state information needed by a control-flow feature to initialize, manage compute data and execution location(s), and de-initialize, \eg calling a function initializes a stack frame including contained objects with constructors, manages local data in blocks and return locations during calls, and de-initializes the frame by running any object destructors and management operations. 406 State is retained in fixed-sized aggregate structures (objects) and dynamic-sized stack(s), often allocated in the heap(s) managed by the runtime system. 404 407 The lifetime of the state varies with the control-flow feature, where longer life-time and dynamic size provide greater power but also increase usage complexity and cost. 405 Control-flow transfers among execution states occursin multiple ways, such as function call, context switch, asynchronous await, etc.408 Control-flow transfers among execution states in multiple ways, such as function call, context switch, asynchronous await, etc. 406 409 Because the programming language determines what constitutes an execution state, implicitly manages this state, and defines movement mechanisms among states, execution state is an elementary property of the semantics of a programming language. 407 410 % An execution-state is related to the notion of a process continuation \cite{Hieb90}. … … 410 413 is execution of code that occurs independently of other execution, \ie the execution resulting from a thread is sequential. 411 414 Multiple threads provide \emph{concurrent execution}; 412 concurrent execution becomes parallel when run on multiple processing units (hyper-threading, cores, sockets).413 There must be language mechanisms to create, block /unblock, and join with a thread.415 concurrent execution becomes parallel when run on multiple processing units, \eg hyper-threading, cores, or sockets. 416 There must be language mechanisms to create, block and unblock, and join with a thread. 414 417 415 418 \item[\newterm{MES}:] 416 419 is the concurrency mechanisms to perform an action without interruption and establish timing relationships among multiple threads. 417 These two properties are independent, \ie mutual exclusion cannot provide synchronization and vice versa without introducing additional threads~\cite[\S~4]{Buhr05a}.420 We contented these two properties are independent, \ie mutual exclusion cannot provide synchronization and vice versa without introducing additional threads~\cite[\S~4]{Buhr05a}. 418 421 Limiting MES, \eg no access to shared data, results in contrived solutions and inefficiency on multi-core von Neumann computers where shared memory is a foundational aspect of its design. 419 422 \end{description} … … 421 424 422 425 423 \subsection{Execution Properties} 424 425 Table~\ref{t:ExecutionPropertyComposition} shows how the three fundamental execution properties: state, thread, and mutual exclusion compose a hierarchy of control-flow features needed in a programming language. 426 (When doing case analysis, not all combinations are meaningful.) 427 Note, basic von Neumann execution requires at least one thread and an execution state providing some form of call stack. 426 \subsection{Structuring Execution Properties} 427 428 Programming languages seldom present the fundamental execution properties directly to programmers. 429 Instead, the properties are packaged into higher-level constructs that encapsulate details and provide safety to these low-level mechanisms. 430 Interestingly, language designers often pick and choose among these execution properties proving a varying subset of constructs. 431 432 Table~\ref{t:ExecutionPropertyComposition} shows all combinations of the three fundamental execution properties available to language designers. 433 (When doing combination case-analysis, not all combinations are meaningful.) 434 The combinations of state, thread, and mutual exclusion compose a hierarchy of control-flow features all of which have appeared in prior programming languages, where each of these languages have found the feature useful. 435 To understand the table, it is important to review the basic von Neumann execution requirement of at least one thread and execution state providing some form of call stack. 428 436 For table entries missing these minimal components, the property is borrowed from the invoker (caller). 429 430 Case 1 is a function that borrows storage for its state (stack frame/activation) and a thread from its invoker and retains this state across \emph{callees}, \ie function local-variables are retained on the stack across calls. 431 Case 2 is case 1 with access to shared state so callers are restricted during update (mutual exclusion) and scheduling for other threads (synchronization). 432 Case 3 is a stateful function supporting resume/suspend along with call/return to retain state across \emph{callers}, but has some restrictions because the function's state is stackless. 433 Note, stackless functions still borrow the caller's stack and thread, where the stack is used to preserve state across its callees. 434 Case 4 is cases 2 and 3 with protection to shared state for stackless functions. 435 Cases 5 and 6 are the same as 3 and 4 but only the thread is borrowed as the function state is stackful, so resume/suspend is a context switch from the caller's to the function's stack. 436 Cases 7 and 8 are rejected because a function that is given a new thread must have its own stack where the thread begins and stack frames are stored for calls, \ie there is no stack to borrow. 437 Cases 9 and 10 are rejected because a thread with a fixed state (no stack) cannot accept calls, make calls, block, or be preempted, all of which require an unknown amount of additional dynamic state. 438 Hence, once started, this kind of thread must execute to completion, \ie computation only, which severely restricts runtime management. 439 Cases 11 and 12 have a stackful thread with and without safe access to shared state. 440 Execution properties increase the cost of creation and execution along with complexity of usage. 437 Each entry in the table, numbered \textbf{1}--\textbf{12}, is discussed with respect to how the execution properties combine to generate a high-level language construct. 441 438 442 439 \begin{table} … … 452 449 \hline 453 450 \hline 454 No & No & \textbf{1}\ \ \ function & \textbf{2}\ \ \ @monitor@ function\\451 No & No & \textbf{1}\ \ \ @struct@ & \textbf{2}\ \ \ @mutex@ @struct@ \\ 455 452 \hline 456 Yes (stackless) & No & \textbf{3}\ \ \ @generator@ & \textbf{4}\ \ \ @m onitor@ @generator@ \\453 Yes (stackless) & No & \textbf{3}\ \ \ @generator@ & \textbf{4}\ \ \ @mutex@ @generator@ \\ 457 454 \hline 458 Yes (stackful) & No & \textbf{5}\ \ \ @coroutine@ & \textbf{6}\ \ \ @m onitor@ @coroutine@ \\455 Yes (stackful) & No & \textbf{5}\ \ \ @coroutine@ & \textbf{6}\ \ \ @mutex@ @coroutine@ \\ 459 456 \hline 460 457 No & Yes & \textbf{7}\ \ \ {\color{red}rejected} & \textbf{8}\ \ \ {\color{red}rejected} \\ … … 462 459 Yes (stackless) & Yes & \textbf{9}\ \ \ {\color{red}rejected} & \textbf{10}\ \ \ {\color{red}rejected} \\ 463 460 \hline 464 Yes (stackful) & Yes & \textbf{11}\ \ \ @thread@ & \textbf{12}\ \ @m onitor@ @thread@ \\461 Yes (stackful) & Yes & \textbf{11}\ \ \ @thread@ & \textbf{12}\ \ @mutex@ @thread@ \\ 465 462 \end{tabular} 466 463 \end{table} 467 464 468 Given the execution-properties taxonomy, programmers can now answer three basic questions: is state necessary across calls and how much, is a separate thread necessary, is access to shared state necessary. 469 The answers define the optimal language feature need for implementing a programming problem. 470 The next sections discusses how \CFA fills in the table with language features, while other programming languages may only provide a subset of the table. 465 Case 1 is a structure where access functions borrow local state (stack frame/activation) and thread from the invoker and retain this state across \emph{callees}, \ie function local-variables are retained on the borrowed stack during calls. 466 Structures are a foundational mechanism for data organization, and access functions provide interface abstraction and code sharing in all programming languages. 467 Case 2 is case 1 with thread safety to a structure's state where access functions provide serialization (mutual exclusion) and scheduling among calling threads (synchronization). 468 A @mutex@ structure, often called a \newterm{monitor}, provides a high-level interface for race-free access of shared data in concurrent programming-languages. 469 Case 3 is case 1 where the structure can implicitly retain execution state and access functions use this execution state to resume/suspend across \emph{callers}, but resume/suspend does not retain a function's local state. 470 A stackless structure, often called a \newterm{generator} or \emph{iterator}, is \newterm{stackless} because it still borrow the caller's stack and thread, where the stack is used only to preserve state across its callees not callers. 471 Generators provide the first step toward directly solving problems like finite-state machines that retain data and execution state between calls, whereas normal functions restart on each call. 472 Case 4 is cases 2 and 3 with thread safety during execution of the generator's access functions. 473 A @mutex@ generator extends generators into the concurrent domain. 474 Cases 5 and 6 are like cases 3 and 4 where the structure is extended with an implicit separate stack, so only the thread is borrowed by access functions. 475 A stackful generator, often called a \newterm{coroutine}, is \newterm{stackful} because resume/suspend now context switch to/from the caller's and coroutine's stack. 476 A coroutine extends the state retained between calls beyond the generator's structure to arbitrary call depth in the access functions. 477 Cases 7 and 8 are rejected because a new thread must have its own stack, where the thread begins and stack frames are stored for calls, \ie it is unrealistic for a thread to borrow a stack. 478 Cases 9 and 10 are rejected because a thread needs a growable stack to accept calls, make calls, block, or be preempted, all of which compound to require an unknown amount of execution state. 479 If this kind of thread exists, it must execute to completion, \ie computation only, which severely restricts runtime management. 480 Cases 11 and 12 are a stackful thread with and without safe access to shared state. 481 A thread is the language mechanism to start another thread of control in a program with growable execution state for call/return execution. 482 In general, more execution properties increase the cost of creation and execution along with complexity of usage. 483 484 Given the execution-properties taxonomy, programmers now ask three basic questions: is state necessary across callers and how much, is a separate thread necessary, is thread safety necessary. 485 Table~\ref{t:ExecutionPropertyComposition} then suggests the optimal language feature needed for implementing a programming problem. 486 The following sections describe how \CFA fills in \emph{all} the non-rejected table entries with language features, while other programming languages may only provide a subset of the table. 471 487 472 488 … … 481 497 \item 482 498 Direct interaction among language features must be possible allowing any feature to be selected without restricting comm\-unication. 483 For example, many concurrent languages do not provide direct communication (calls)among threads, \ie threads only communicate indirectly through monitors, channels, messages, and/or futures.484 Indirect communication increases the number of objects, consuming more resources, and require additional synchronization and possibly data transfer.499 For example, many concurrent languages do not provide direct communication calls among threads, \ie threads only communicate indirectly through monitors, channels, messages, and/or futures. 500 Indirect communication increases the number of objects, consuming more resources, and requires additional synchronization and possibly data transfer. 485 501 486 502 \item … … 493 509 494 510 \item 495 MES must be available implicitly in language constructs as well as explicitly for specialized requirements, because requiring programmers to build MES using low-level locks often leads to incorrect programs.511 MES must be available implicitly in language constructs, \eg Java built-in monitors, as well as explicitly for specialized requirements, \eg @java.util.concurrent@, because requiring programmers to build MES using low-level locks often leads to incorrect programs. 496 512 Furthermore, reducing synchronization scope by encapsulating it within language constructs further reduces errors in concurrent programs. 497 513 … … 502 518 \item 503 519 Synchronization must be able to control the service order of requests including prioritizing selection from different kinds of outstanding requests, and postponing a request for an unspecified time while continuing to accept new requests. 504 Otherwise, certain concurrency problems are difficult, e.g.\web server, disk scheduling, and the amount of concurrency is inhibited~\cite{Gentleman81}.520 Otherwise, certain concurrency problems are difficult, \eg web server, disk scheduling, and the amount of concurrency is inhibited~\cite{Gentleman81}. 505 521 \end{itemize} 506 522 We have satisfied these requirements in \CFA while maintaining backwards compatibility with the huge body of legacy C programs. … … 511 527 512 528 Asynchronous await/call is a caller mechanism for structuring programs and/or increasing concurrency, where the caller (client) postpones an action into the future, which is subsequently executed by a callee (server). 513 The caller detects the action's completion through a \newterm{future} /\newterm{promise}.529 The caller detects the action's completion through a \newterm{future} or \newterm{promise}. 514 530 The benefit is asynchronous caller execution with respect to the callee until future resolution. 515 531 For single-threaded languages like JavaScript, an asynchronous call passes a callee action, which is queued in the event-engine, and continues execution with a promise. … … 517 533 A promise-completion call-back can be part of the callee action or the caller is rescheduled; 518 534 in either case, the call back is executed after the promise is fulfilled. 519 While asynchronous calls generate new callee (server) events, we conten t this mechanism is insufficient for advanced control-flow mechanisms like generators or coroutines (which are discussed next).535 While asynchronous calls generate new callee (server) events, we contend this mechanism is insufficient for advanced control-flow mechanisms like generators or coroutines, which are discussed next. 520 536 Specifically, control between caller and callee occurs indirectly through the event-engine precluding direct handoff and cycling among events, and requires complex resolution of a control promise and data. 521 537 Note, @async-await@ is just syntactic-sugar over the event engine so it does not solve these deficiencies. 522 538 For multi-threaded languages like Java, the asynchronous call queues a callee action with an executor (server), which subsequently executes the work by a thread in the executor thread-pool. 523 The problem is when concurrent work-units need to interact and/or block as this effects the executor , \eg stopsthreads.539 The problem is when concurrent work-units need to interact and/or block as this effects the executor by stopping threads. 524 540 While it is possible to extend this approach to support the necessary mechanisms, \eg message passing in Actors, we show monitors and threads provide an equally competitive approach that does not deviate from normal call communication and can be used to build asynchronous call, as is done in Java. 525 541 … … 540 556 There are two styles of activating a stateful function, \emph{asymmetric} or \emph{symmetric}, identified by resume/suspend (no cycles) and resume/resume (cycles). 541 557 These styles \emph{do not} cause incremental stack growth, \eg a million resume/suspend or resume/resume cycles do not remember each cycle just the last resumer for each cycle. 542 Selecting between stackless/stackful semantics and asymmetric/symmetric style is a tradeoff between programming requirements, performance, and design, where stackless is faster and smaller (modified call/return between closures), stackful is more general but slower and larger (context switching between distinct stacks), and asymmetric is simpler control-flow than symmetric.543 Additionally, storage management for the closure/stack (especially in unmanaged languages, \ie no garbage collection) must be factored into design and performance.558 Selecting between stackless/stackful semantics and asymmetric/symmetric style is a tradeoff between programming requirements, performance, and design, where stackless is faster and smaller using modified call/return between closures, stackful is more general but slower and larger using context switching between distinct stacks, and asymmetric is simpler control-flow than symmetric. 559 Additionally, storage management for the closure/stack must be factored into design and performance, especially in unmanaged languages without garbage collection. 544 560 Note, creation cost (closure/stack) is amortized across usage, so activation cost (resume/suspend) is usually the dominant factor. 545 561 … … 690 706 \hspace{3pt} 691 707 \subfloat[C generated code for \CFA version]{\label{f:CFibonacciSim}\usebox\myboxC} 692 \caption{Fibonacci (output)asymmetric generator}708 \caption{Fibonacci output asymmetric generator} 693 709 \label{f:FibonacciAsymmetricGenerator} 694 710 … … 765 781 \subfloat[C generated code for \CFA version]{\label{f:CFormatGenImpl}\usebox\myboxB} 766 782 \hspace{3pt} 767 \caption{Formatter (input)asymmetric generator}783 \caption{Formatter input asymmetric generator} 768 784 \label{f:FormatterAsymmetricGenerator} 769 785 \end{figure} 770 786 771 Figure~\ref{f:FibonacciAsymmetricGenerator} shows an unbounded asymmetric generator for an infinite sequence of Fibonacci numbers written (left to right)in C, \CFA, and showing the underlying C implementation for the \CFA version.787 Figure~\ref{f:FibonacciAsymmetricGenerator} shows an unbounded asymmetric generator for an infinite sequence of Fibonacci numbers written left to right in C, \CFA, and showing the underlying C implementation for the \CFA version. 772 788 This generator is an \emph{output generator}, producing a new result on each resumption. 773 789 To compute Fibonacci, the previous two values in the sequence are retained to generate the next value, \ie @fn1@ and @fn@, plus the execution location where control restarts when the generator is resumed, \ie top or middle. 774 An additional requirement is the ability to create an arbitrary number of generators (of any kind), \ie retaining one state in global variables is insufficient;790 An additional requirement is the ability to create an arbitrary number of generators of any kind, \ie retaining one state in global variables is insufficient; 775 791 hence, state is retained in a closure between calls. 776 792 Figure~\ref{f:CFibonacci} shows the C approach of manually creating the closure in structure @Fib@, and multiple instances of this closure provide multiple Fibonacci generators. … … 794 810 Figure~\ref{f:CFibonacciSim} shows the C implementation of the \CFA asymmetric generator. 795 811 Only one execution-state field, @restart@, is needed to subscript the suspension points in the generator. 796 At the start of the generator main, the @static@ declaration, @states@, is initialized to the N suspend points in the generator (where operator @&&@ dereferences/references a label~\cite{gccValueLabels}).812 At the start of the generator main, the @static@ declaration, @states@, is initialized to the N suspend points in the generator, where operator @&&@ dereferences or references a label~\cite{gccValueLabels}. 797 813 Next, the computed @goto@ selects the last suspend point and branches to it. 798 The cost of setting @restart@ and branching via the computed @goto@ adds very little cost to the suspend /resume calls.814 The cost of setting @restart@ and branching via the computed @goto@ adds very little cost to the suspend and resume calls. 799 815 800 816 An advantage of the \CFA explicit generator type is the ability to allow multiple type-safe interface functions taking and returning arbitrary types. … … 917 933 When a message byte arrives, it triggers an interrupt, and the operating system services the interrupt by calling the device driver with the byte read from a hardware register. 918 934 The device driver returns a status code of its current state, and when a complete message is obtained, the operating system read the message accumulated in the supplied buffer. 919 Hence, the device driver is an input/output generator, where the cost of resuming the device-driver generator is the same as call /return, so performance in an operating-system kernel is excellent.935 Hence, the device driver is an input/output generator, where the cost of resuming the device-driver generator is the same as call and return, so performance in an operating-system kernel is excellent. 920 936 The key benefits of using a generator are correctness, safety, and maintenance because the execution states are transcribed directly into the programming language rather than table lookup or stack ripping. 921 937 The conclusion is that FSMs are complex and occur in important domains, so direct generator support is important in a system programming language. … … 982 998 Constructing the cycle must deal with definition-before-use to close the cycle, \ie, the first generator must know about the last generator, which is not within scope. 983 999 (This issue occurs for any cyclic data structure.) 984 The example creates the generators, @ping@ /@pong@, and then assigns the partners that form the cycle.1000 The example creates the generators, @ping@ and @pong@, and then assigns the partners that form the cycle. 985 1001 % (Alternatively, the constructor can assign the partners as they are declared, except the first, and the first-generator partner is set after the last generator declaration to close the cycle.) 986 1002 Once the cycle is formed, the program main resumes one of the generators, @ping@, and the generators can then traverse an arbitrary cycle using @resume@ to activate partner generator(s). … … 1066 1082 While the semantics of call forward is a tail-call optimization, which compilers perform, the generator state is different on each call rather a common state for a tail-recursive function (i.e., the parameter to the function never changes during the forward calls. 1067 1083 However, this assembler code depends on what entry code is generated, specifically if there are local variables and the level of optimization. 1068 Hence, internal compiler support is necessary for any forward call (or backwards return), \eg LLVM has various coroutine support~\cite{CoroutineTS}, and \CFA can leverage this support should it eventually fork @clang@.1069 For this reason, \CFA does not support general symmetric generators at this time, but, it is possible to hand generate any symmetric generators (as in Figure~\ref{f:CPingPongSim})for proof of concept and performance testing.1084 Hence, internal compiler support is necessary for any forward call or backwards return, \eg LLVM has various coroutine support~\cite{CoroutineTS}, and \CFA can leverage this support should it eventually fork @clang@. 1085 For this reason, \CFA does not support general symmetric generators at this time, but, it is possible to hand generate any symmetric generators, as in Figure~\ref{f:CPingPongSim}, for proof of concept and performance testing. 1070 1086 1071 1087 Finally, part of this generator work was inspired by the recent \CCtwenty coroutine proposal~\cite{C++20Coroutine19}, which uses the general term coroutine to mean generator. … … 1126 1142 \begin{cfa} 1127 1143 int Crc() { 1128 `suspend;` 1129 short int crc = byte << 8; 1130 `suspend;` 1131 status = (crc | byte) == sum ? MSG : ECRC; 1144 `suspend;` short int crc = byte << 8; 1145 `suspend;` status = (crc | byte) == sum ? MSG : ECRC; 1132 1146 return crc; 1133 1147 } … … 1139 1153 \begin{comment} 1140 1154 Figure~\ref{f:Coroutine3States} creates a @coroutine@ type, @`coroutine` Fib { int fn; }@, which provides communication, @fn@, for the \newterm{coroutine main}, @main@, which runs on the coroutine stack, and possibly multiple interface functions, \eg @restart@. 1141 Like the structure in Figure~\ref{f:ExternalState}, the coroutine type allows multiple instances, where instances of this type are passed to the (overloaded)coroutine main.1155 Like the structure in Figure~\ref{f:ExternalState}, the coroutine type allows multiple instances, where instances of this type are passed to the overloaded coroutine main. 1142 1156 The coroutine main's stack holds the state for the next generation, @f1@ and @f2@, and the code represents the three states in the Fibonacci formula via the three suspend points, to context switch back to the caller's @resume@. 1143 1157 The interface function @restart@, takes a Fibonacci instance and context switches to it using @resume@; … … 1373 1387 1374 1388 Figure~\ref{f:ProdCons} shows the ping-pong example in Figure~\ref{f:CFAPingPongGen} extended into a producer/consumer symmetric-coroutine performing bidirectional communication. 1375 This example is illustrative because both producer /consumer have two interface functions with @resume@s that suspend execution in these interface (helper)functions.1389 This example is illustrative because both producer and consumer have two interface functions with @resume@s that suspend execution in these interface functions. 1376 1390 The program main creates the producer coroutine, passes it to the consumer coroutine in its initialization, and closes the cycle at the call to @start@ along with the number of items to be produced. 1377 1391 The call to @start@ is the first @resume@ of @prod@, which remembers the program main as the starter and creates @prod@'s stack with a frame for @prod@'s coroutine main at the top, and context switches to it. … … 1401 1415 The question now is where does control continue? 1402 1416 1403 The na\"{i}ve semantics for coroutine-cycle termination is to context switch to the last resumer, like executing a @suspend@ /@return@ in a generator.1417 The na\"{i}ve semantics for coroutine-cycle termination is to context switch to the last resumer, like executing a @suspend@ or @return@ in a generator. 1404 1418 However, for coroutines, the last resumer is \emph{not} implicitly below the current stack frame, as for generators, because each coroutine's stack is independent. 1405 1419 Unfortunately, it is impossible to determine statically if a coroutine is in a cycle and unrealistic to check dynamically (graph-cycle problem). … … 1410 1424 For asymmetric coroutines, it is common for the first resumer (starter) coroutine to be the only resumer; 1411 1425 for symmetric coroutines, it is common for the cycle creator to persist for the lifetime of the cycle. 1412 For other scenarios, it is always possible to devise a solution with additional programming effort, such as forcing the cycle forward (backward)to a safe point before starting termination.1426 For other scenarios, it is always possible to devise a solution with additional programming effort, such as forcing the cycle forward or backward to a safe point before starting termination. 1413 1427 1414 1428 Note, the producer/consumer example does not illustrate the full power of the starter semantics because @cons@ always ends first. 1415 1429 Assume generator @PingPong@ in Figure~\ref{f:PingPongSymmetricGenerator} is converted to a coroutine. 1416 1430 Unlike generators, coroutines have a starter structure with multiple levels, where the program main starts @ping@ and @ping@ starts @pong@. 1417 By adjusting $N$ for either @ping@ /@pong@, it is possible to have either finish first.1431 By adjusting $N$ for either @ping@ or @pong@, it is possible to have either finish first. 1418 1432 If @pong@ ends first, it resumes its starter @ping@ in its coroutine main, then @ping@ ends and resumes its starter the program main on return; 1419 1433 if @ping@ ends first, it resumes its starter the program main on return. … … 1425 1439 \subsection{Generator / Coroutine Implementation} 1426 1440 1427 A significant implementation challenge for generators /coroutines (and threads in Section~\ref{s:threads}) is adding extra fields to the custom types and related functions, \eg inserting code after/before the coroutine constructor/destructor and @main@ to create/initialize/de-initialize/destroy any extra fields, \egstack.1441 A significant implementation challenge for generators and coroutines (and threads in Section~\ref{s:threads}) is adding extra fields to the custom types and related functions, \eg inserting code after/before the coroutine constructor/destructor and @main@ to create/initialize/de-initialize/destroy any extra fields, \eg the coroutine stack. 1428 1442 There are several solutions to these problem, which follow from the object-oriented flavour of adopting custom types. 1429 1443 … … 1433 1447 \end{cfa} 1434 1448 % The problem is that the programming language and its tool chain, \eg debugger, @valgrind@, need to understand @baseCoroutine@ because it infers special property, so type @baseCoroutine@ becomes a de facto keyword and all types inheriting from it are implicitly custom types. 1435 The problem is that some special properties are not handled by existing language semantics, \eg the execution of constructors /destructors is in the wrong order to implicitly start threads because the thread must start \emph{after} all constructors as it relies on a completely initialized object, but the inherited constructor runs \emph{before} the derived.1449 The problem is that some special properties are not handled by existing language semantics, \eg the execution of constructors and destructors is in the wrong order to implicitly start threads because the thread must start \emph{after} all constructors as it relies on a completely initialized object, but the inherited constructor runs \emph{before} the derived. 1436 1450 Alternatives, such as explicitly starting threads as in Java, are repetitive and forgetting to call start is a common source of errors. 1437 1451 An alternative is composition: … … 1461 1475 forall( `dtype` T | is_coroutine(T) ) void $suspend$( T & ), resume( T & ); 1462 1476 \end{cfa} 1463 Note, copying generators /coroutines/threads is undefined because muliple objects cannot execute on a shared stack and stack copying does not work in unmanaged languages (no garbage collection), like C, because the stack may contain pointers to objects within it that require updating for the copy.1464 The \CFA @dtype@ property provides no \emph{implicit} copying operations and the @is_coroutine@ trait provides no \emph{explicit} copying operations, so all coroutines must be passed by reference (pointer).1465 The function definitions ensure there is a statically typed @main@ function that is the starting point (first stack frame) of a coroutine, and a mechanism to get (read)the coroutine descriptor from its handle.1466 The @main@ function has no return value or additional parameters because the coroutine type allows an arbitrary number of interface functions with corresponding arbitrary typed input /output values versus fixed ones.1477 Note, copying generators, coroutines, and threads is undefined because muliple objects cannot execute on a shared stack and stack copying does not work in unmanaged languages (no garbage collection), like C, because the stack may contain pointers to objects within it that require updating for the copy. 1478 The \CFA @dtype@ property provides no \emph{implicit} copying operations and the @is_coroutine@ trait provides no \emph{explicit} copying operations, so all coroutines must be passed by reference or pointer. 1479 The function definitions ensure there is a statically typed @main@ function that is the starting point (first stack frame) of a coroutine, and a mechanism to read the coroutine descriptor from its handle. 1480 The @main@ function has no return value or additional parameters because the coroutine type allows an arbitrary number of interface functions with corresponding arbitrary typed input and output values versus fixed ones. 1467 1481 The advantage of this approach is that users can easily create different types of coroutines, \eg changing the memory layout of a coroutine is trivial when implementing the @get_coroutine@ function, and possibly redefining \textsf{suspend} and @resume@. 1468 1482 … … 1506 1520 1507 1521 Figure~\ref{f:CoroutineMemoryLayout} shows different memory-layout options for a coroutine (where a thread is similar). 1508 The coroutine handle is the @coroutine@ instance containing programmer specified type global /communication variables across interface functions.1522 The coroutine handle is the @coroutine@ instance containing programmer specified type global and communication variables across interface functions. 1509 1523 The coroutine descriptor contains all implicit declarations needed by the runtime, \eg @suspend@/@resume@, and can be part of the coroutine handle or separate. 1510 1524 The coroutine stack can appear in a number of locations and be fixed or variable sized. … … 1513 1527 Once allocated, a VLS is fixed sized.} 1514 1528 on the allocating stack, provided the allocating stack is large enough. 1515 For a VLS stack allocation /deallocation is an inexpensive adjustment of the stack pointer, modulo any stack constructor costs (\eg initial frame setup).1516 For stack allocation in the heap, allocation /deallocation is an expensive allocation, where the heap can be a shared resource, modulo any stack constructor costs.1517 It is also possible to use a split (segmented)stack calling convention, available with gcc and clang, allowing a variable-sized stack via a set of connected blocks in the heap.1518 Currently, \CFA supports stack /heap allocated descriptors but only fixed-sized heap allocated stacks.1529 For a VLS stack allocation and deallocation is an inexpensive adjustment of the stack pointer, modulo any stack constructor costs to initial frame setup. 1530 For stack allocation in the heap, allocation and deallocation is an expensive allocation, where the heap can be a shared resource, modulo any stack constructor costs. 1531 It is also possible to use a split or segmented stack calling convention, available with gcc and clang, allowing a variable-sized stack via a set of connected blocks in the heap. 1532 Currently, \CFA supports stack and heap allocated descriptors but only fixed-sized heap allocated stacks. 1519 1533 In \CFA debug-mode, the fixed-sized stack is terminated with a write-only page, which catches most stack overflows. 1520 1534 Experience teaching concurrency with \uC~\cite{CS343} shows fixed-sized stacks are rarely an issue for students. … … 1539 1553 The transition to concurrency, even for a single thread with multiple stacks, occurs when coroutines context switch to a \newterm{scheduling coroutine}, introducing non-determinism from the coroutine perspective~\cite[\S~3]{Buhr05a}. 1540 1554 Therefore, a minimal concurrency system requires coroutines \emph{in conjunction with a nondeterministic scheduler}. 1541 The resulting execution system now follows a cooperative threading-model~\cite{Adya02,libdill} because context-switching points to the scheduler (blocking)are known, but the next unblocking point is unknown due to the scheduler.1555 The resulting execution system now follows a cooperative threading-model~\cite{Adya02,libdill} because context-switching points to the scheduler are known, but the next unblocking point is unknown due to the scheduler. 1542 1556 Adding \newterm{preemption} introduces \newterm{non-cooperative} or \newterm{preemptive} scheduling, where context switching points to the scheduler are unknown as they can occur randomly between any two instructions often based on a timer interrupt. 1543 1557 Uncertainty gives the illusion of parallelism on a single processor and provides a mechanism to access and increase performance on multiple processors. 1544 The reason is that the scheduler /runtime have complete knowledge about resources and how to best utilized them.1558 The reason is that the scheduler and runtime have complete knowledge about resources and how to best utilized them. 1545 1559 However, the introduction of unrestricted nondeterminism results in the need for \newterm{mutual exclusion} and \newterm{synchronization}~\cite[\S~4]{Buhr05a}, which restrict nondeterminism for correctness; 1546 1560 otherwise, it is impossible to write meaningful concurrent programs. … … 1589 1603 \vspace{1pt} 1590 1604 \par\noindent 1591 \CFA has a simpler approach using a custom @thread@ type and leveraging declaration semantics (allocation/deallocation), where threads implicitly @fork@ after construction and @join@ before destruction.1605 \CFA has a simpler approach using a custom @thread@ type and leveraging declaration semantics, allocation and deallocation, where threads implicitly @fork@ after construction and @join@ before destruction. 1592 1606 \begin{cfa} 1593 1607 thread MyThread {}; … … 1598 1612 } $\C{// deallocate stack-based threads, implicit joins before destruction}$ 1599 1613 \end{cfa} 1600 This semantic ensures a thread is started and stopped exactly once, eliminating some programming error, and scales to multiple threads for basic (termination)synchronization.1601 For block allocation to arbitrary depth, including recursion, threads are created /destroyed in a lattice structure (tree with top and bottom).1614 This semantic ensures a thread is started and stopped exactly once, eliminating some programming error, and scales to multiple threads for basic termination synchronization. 1615 For block allocation to arbitrary depth, including recursion, threads are created and destroyed in a lattice structure (tree with top and bottom). 1602 1616 Arbitrary topologies are possible using dynamic allocation, allowing threads to outlive their declaration scope, identical to normal dynamic allocation. 1603 1617 \begin{cfa} … … 1670 1684 \end{tabular} 1671 1685 \end{cquote} 1672 Like coroutines, the @dtype@ property prevents \emph{implicit} copy operations and the @is_thread@ trait provides no \emph{explicit} copy operations, so threads must be passed by reference (pointer).1673 Similarly, the function definitions ensure there is a statically typed @main@ function that is the thread starting point (first stack frame), a mechanism to get (read)the thread descriptor from its handle, and a special destructor to prevent deallocation while the thread is executing.1686 Like coroutines, the @dtype@ property prevents \emph{implicit} copy operations and the @is_thread@ trait provides no \emph{explicit} copy operations, so threads must be passed by reference or pointer. 1687 Similarly, the function definitions ensure there is a statically typed @main@ function that is the thread starting point (first stack frame), a mechanism to read the thread descriptor from its handle, and a special destructor to prevent deallocation while the thread is executing. 1674 1688 (The qualifier @mutex@ for the destructor parameter is discussed in Section~\ref{s:Monitor}.) 1675 1689 The difference between the coroutine and thread is that a coroutine borrows a thread from its caller, so the first thread resuming a coroutine creates the coroutine's stack and starts running the coroutine main on the stack; 1676 1690 whereas, a thread is scheduling for execution in @main@ immediately after its constructor is run. 1677 No return value or additional parameters are necessary for this function because the @thread@ type allows an arbitrary number of interface functions with corresponding arbitrary typed input /output values.1691 No return value or additional parameters are necessary for this function because the @thread@ type allows an arbitrary number of interface functions with corresponding arbitrary typed input and output values. 1678 1692 1679 1693 … … 1683 1697 Unrestricted nondeterminism is meaningless as there is no way to know when a result is completed and safe to access. 1684 1698 To produce meaningful execution requires clawing back some determinism using mutual exclusion and synchronization, where mutual exclusion provides access control for threads using shared data, and synchronization is a timing relationship among threads~\cite[\S~4]{Buhr05a}. 1685 The shared data protected by mutual exlusion is called a \newterm{critical section}~\cite{Dijkstra65}, and the protection can be simple (only 1 thread) or complex (only N kinds of threads, \eg group~\cite{Joung00} or readers/writer~\cite{Courtois71}).1686 Without synchronization control in a critical section, an arriving thread can barge ahead of preexisting waiter threads resulting in short/long-term starvation, staleness /freshness problems, and/orincorrect transfer of data.1699 The shared data protected by mutual exlusion is called a \newterm{critical section}~\cite{Dijkstra65}, and the protection can be simple, only 1 thread, or complex, only N kinds of threads, \eg group~\cite{Joung00} or readers/writer~\cite{Courtois71} problems. 1700 Without synchronization control in a critical section, an arriving thread can barge ahead of preexisting waiter threads resulting in short/long-term starvation, staleness and freshness problems, and incorrect transfer of data. 1687 1701 Preventing or detecting barging is a challenge with low-level locks, but made easier through higher-level constructs. 1688 1702 This challenge is often split into two different approaches: barging \emph{avoidance} and \emph{prevention}. … … 1696 1710 Some concurrent systems eliminate mutable shared-state by switching to non-shared communication like message passing~\cite{Thoth,Harmony,V-Kernel,MPI} (Erlang, MPI), channels~\cite{CSP} (CSP,Go), actors~\cite{Akka} (Akka, Scala), or functional techniques (Haskell). 1697 1711 However, these approaches introduce a new communication mechanism for concurrency different from the standard communication using function call/return. 1698 Hence, a programmer must learn and manipulate two sets of design /programming patterns.1712 Hence, a programmer must learn and manipulate two sets of design and programming patterns. 1699 1713 While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account. 1700 In contrast, approaches based on shared-state models more closely resemble the standard call /return programming model, resulting in a single programming paradigm.1714 In contrast, approaches based on shared-state models more closely resemble the standard call and return programming model, resulting in a single programming paradigm. 1701 1715 Finally, a newer approach for restricting non-determinism is transactional memory~\cite{Herlihy93}. 1702 1716 While this approach is pursued in hardware~\cite{Nakaike15} and system languages, like \CC~\cite{Cpp-Transactions}, the performance and feature set is still too restrictive~\cite{Cascaval08,Boehm09} to be the main concurrency paradigm for system languages. … … 1711 1725 For these reasons, \CFA selected monitors as the core high-level concurrency construct, upon which higher-level approaches can be easily constructed. 1712 1726 1713 Specifically, a \textbf{monitor} is a set of functions that ensure mutual exclusion when accessing shared state. 1714 More precisely, a monitor is a programming technique that implicitly binds mutual exclusion to static function scope by call/return, as opposed to locks, where mutual-exclusion is defined by acquire/release calls, independent of lexical context (analogous to block and heap storage allocation). 1715 Restricting acquire/release points eases programming, comprehension, and maintenance, at a slight cost in flexibility and efficiency. 1716 \CFA uses a custom @monitor@ type and leverages declaration semantics (deallocation) to protect active or waiting threads in a monitor. 1717 1718 The following is a \CFA monitor implementation of an atomic counter. 1719 \begin{cfa} 1720 `monitor` Aint { int cnt; }; $\C[4.25in]{// atomic integer counter}$ 1721 int ++?( Aint & `mutex` this ) with( this ) { return ++cnt; } $\C{// increment}$ 1722 int ?=?( Aint & `mutex` lhs, int rhs ) with( lhs ) { cnt = rhs; } $\C{// conversions with int, mutex optional}\CRT$ 1723 int ?=?( int & lhs, Aint & `mutex` rhs ) with( rhs ) { lhs = cnt; } 1724 \end{cfa} 1725 The operators use the parameter-only declaration type-qualifier @mutex@ to mark which parameters require locking during function execution to protect from race conditions. 1726 The assignment operators provide bidirectional conversion between an atomic and normal integer without accessing field @cnt@. 1727 (These operations only need @mutex@, if reading/writing the implementation type is not atomic.) 1728 The atomic counter is used without any explicit mutual-exclusion and provides thread-safe semantics. 1729 \begin{cfa} 1727 Figure~\ref{f:AtomicCounter} compares a \CFA and Java monitor implementing an atomic counter.\footnote{ 1728 Like other concurrent programming languages, \CFA and Java have performant specializations for the basic types using atomic instructions.} 1729 A \newterm{monitor} is a set of functions that ensure mutual exclusion when accessing shared state. 1730 (Note, in \CFA, @monitor@ is short-hand for @mutex struct@.) 1731 More precisely, a monitor is a programming technique that implicitly binds mutual exclusion to static function scope by call and return, as opposed to locks, where mutual exclusion is defined by acquire/release calls, independent of lexical context (analogous to block and heap storage allocation). 1732 Restricting acquire and release points eases programming, comprehension, and maintenance, at a slight cost in flexibility and efficiency. 1733 As for other special types, \CFA has a custom @monitor@ type. 1734 1735 \begin{figure} 1736 \centering 1737 1738 \begin{lrbox}{\myboxA} 1739 \begin{cfa}[aboveskip=0pt,belowskip=0pt] 1740 `monitor` Aint { // atomic integer counter 1741 int cnt; 1742 }; 1743 int ++?( Aint & `mutex` this ) with(this) { return ++cnt; } 1744 int ?=?( Aint & `mutex` lhs, int rhs ) with(lhs) { cnt = rhs; } 1745 int ?=?(int & lhs, Aint & rhs) with(rhs) { lhs = cnt; } 1746 1730 1747 int i = 0, j = 0, k = 5; 1731 Aint x = { 0 }, y = { 0 }, z = { 5 }; $\C{// no mutex required}$ 1732 ++x; ++y; ++z; $\C{// safe increment by multiple threads}$ 1733 x = 2; y = i; z = k; $\C{// conversions}$ 1734 i = x; j = y; k = z; 1735 \end{cfa} 1736 Note, like other concurrent programming languages, \CFA has specializations for the basic types using atomic instructions for performance and a general trait similar to the \CC template @std::atomic@. 1737 1738 \CFA monitors have \newterm{multi-acquire} semantics so the thread in the monitor may acquire it multiple times without deadlock, allowing recursion and calling other interface functions. 1739 \newpage 1740 \begin{cfa} 1741 monitor M { ... } m; 1742 void foo( M & mutex m ) { ... } $\C{// acquire mutual exclusion}$ 1743 void bar( M & mutex m ) { $\C{// acquire mutual exclusion}$ 1744 ... `bar( m );` ... `foo( m );` ... $\C{// reacquire mutual exclusion}$ 1745 } 1746 \end{cfa} 1747 \CFA monitors also ensure the monitor lock is released regardless of how an acquiring function ends (normal or exceptional), and returning a shared variable is safe via copying before the lock is released. 1748 Aint x = { 0 }, y = { 0 }, z = { 5 }; // no mutex 1749 ++x; ++y; ++z; // mutex 1750 x = 2; y = i; z = k; // mutex 1751 i = x; j = y; k = z; // no mutex 1752 \end{cfa} 1753 \end{lrbox} 1754 1755 \begin{lrbox}{\myboxB} 1756 \begin{java}[aboveskip=0pt,belowskip=0pt] 1757 class Aint { 1758 private int cnt; 1759 public Aint( int init ) { cnt = init; } 1760 `synchronized` public int inc() { return ++cnt; } 1761 `synchronized` public void set( int rhs ) {cnt=rhs;} 1762 public int get() { return cnt; } 1763 } 1764 int i = 0, j = 0, k = 5; 1765 Aint x=new Aint(0), y=new Aint(0), z=new Aint(5); 1766 x.inc(); y.inc(); z.inc(); 1767 x.set( 2 ); y.set( i ); z.set( k ); 1768 i = x.get(); j = y.get(); k = z.get(); 1769 \end{java} 1770 \end{lrbox} 1771 1772 \subfloat[\CFA]{\label{f:AtomicCounterCFA}\usebox\myboxA} 1773 \hspace{3pt} 1774 \vrule 1775 \hspace{3pt} 1776 \subfloat[Java]{\label{f:AtomicCounterJava}\usebox\myboxB} 1777 \caption{Atomic counter} 1778 \label{f:AtomicCounter} 1779 \end{figure} 1780 1781 Like Java, \CFA monitors have \newterm{multi-acquire} semantics so the thread in the monitor may acquire it multiple times without deadlock, allowing recursion and calling other interface functions. 1782 % \begin{cfa} 1783 % monitor M { ... } m; 1784 % void foo( M & mutex m ) { ... } $\C{// acquire mutual exclusion}$ 1785 % void bar( M & mutex m ) { $\C{// acquire mutual exclusion}$ 1786 % ... `bar( m );` ... `foo( m );` ... $\C{// reacquire mutual exclusion}$ 1787 % } 1788 % \end{cfa} 1789 \CFA monitors also ensure the monitor lock is released regardless of how an acquiring function ends, normal or exceptional, and returning a shared variable is safe via copying before the lock is released. 1748 1790 Similar safety is offered by \emph{explicit} opt-in disciplines like \CC RAII versus the monitor \emph{implicit} language-enforced safety guarantee ensuring no programmer usage errors. 1749 Furthermore, RAII mechanisms cannot handle complex synchronization within a monitor, where the monitor lock may not be released on function exit because it is passed to an unblocking thread;1791 However, RAII mechanisms cannot handle complex synchronization within a monitor, where the monitor lock may not be released on function exit because it is passed to an unblocking thread; 1750 1792 RAII is purely a mutual-exclusion mechanism (see Section~\ref{s:Scheduling}). 1793 1794 Both Java and \CFA use a keyword @mutex@/\lstinline[language=java]|synchronized| to designate functions that implicitly acquire/release the monitor lock on call/return providing mutual exclusion to the stared data. 1795 Non-designated functions provide no mutual exclusion for read-only access or as an interface to a multi-step protocol requiring several steps of acquiring and releasing the monitor. 1796 Monitor objects can be passed through multiple helper functions without acquiring mutual exclusion, until a designated function associated with the object is called. 1797 \CFA designated functions are marked by an explicitly parameter-only pointer/reference qualifier @mutex@ (discussed further in Section\ref{s:MutexAcquisition}). 1798 Whereas, Java designated members are marked with \lstinline[language=java]|synchronized| that applies to the implicit reference parameter @this@. 1799 In the example, the increment and setter operations need mutual exclusion while the read-only getter operation can be non-mutex if reading the implementation is atomic. 1751 1800 1752 1801 … … 1771 1820 \end{tabular} 1772 1821 \end{cquote} 1773 The @dtype@ property prevents \emph{implicit} copy operations and the @is_monitor@ trait provides no \emph{explicit} copy operations, so monitors must be passed by reference (pointer).1774 Similarly, the function definitions ensures there is a mechanism to get (read) the monitor descriptor from its handle, and a special destructor to prevent deallocation if a threadusing the shared data.1822 The @dtype@ property prevents \emph{implicit} copy operations and the @is_monitor@ trait provides no \emph{explicit} copy operations, so monitors must be passed by reference or pointer. 1823 Similarly, the function definitions ensures there is a mechanism to read the monitor descriptor from its handle, and a special destructor to prevent deallocation if a thread is using the shared data. 1775 1824 The custom monitor type also inserts any locks needed to implement the mutual exclusion semantics. 1825 \CFA relies heavily on traits as an abstraction mechanism, so the @mutex@ qualifier prevents coincidentally matching of a monitor trait with a type that is not a monitor, similar to coincidental inheritance where a shape and playing card can both be drawable. 1776 1826 1777 1827 … … 1779 1829 \label{s:MutexAcquisition} 1780 1830 1781 While the monitor lock provides mutual exclusion for shared data, there are implementation options for when and where the locking/unlocking occurs. 1782 (Much of this discussion also applies to basic locks.) 1783 For example, a monitor may be passed through multiple helper functions before it is necessary to acquire the monitor's mutual exclusion. 1784 1785 \CFA requires programmers to identify the kind of parameter with the @mutex@ keyword and uses no keyword to mean \lstinline[morekeywords=nomutex]@nomutex@, because @mutex@ parameters are rare and no keyword is the \emph{normal} parameter semantics. 1786 Hence, @mutex@ parameters are documentation, at the function and its prototype, to both programmer and compiler, without other redundant keywords. 1787 Furthermore, \CFA relies heavily on traits as an abstraction mechanism, so the @mutex@ qualifier prevents coincidentally matching of a monitor trait with a type that is not a monitor, similar to coincidental inheritance where a shape and playing card can both be drawable. 1788 1789 The next semantic decision is establishing which parameter \emph{types} may be qualified with @mutex@. 1790 The following has monitor parameter types that are composed of multiple objects. 1791 \begin{cfa} 1792 monitor M { ... } 1831 For object-oriented programming languages, the mutex property applies to one object, the implicit pointer/reference to the monitor type. 1832 Because \CFA uses a pointer qualifier, other possibilities exist, \eg: 1833 \begin{cfa} 1834 monitor M { ... }; 1793 1835 int f1( M & mutex m ); $\C{// single parameter object}$ 1794 1836 int f2( M * mutex m ); $\C{// single or multiple parameter object}$ … … 1796 1838 int f4( stack( M * ) & mutex m ); $\C{// multiple parameters object}$ 1797 1839 \end{cfa} 1798 Function @f1@ has a single parameter object, while @f2@'s indirection could be a single or multi-element array, where static array size is often unknown in C. 1799 Function @f3@ has a multiple object matrix, and @f4@ a multiple object data structure. 1800 While shown shortly, multiple object acquisition is possible, but the number of objects must be statically known. 1801 Therefore, \CFA only acquires one monitor per parameter with exactly one level of indirection, and exclude pointer types to unknown sized arrays. 1802 1803 For object-oriented monitors, \eg Java, calling a mutex member \emph{implicitly} acquires mutual exclusion of the receiver object, @`rec`.foo(...)@. 1804 \CFA has no receiver, and hence, the explicit @mutex@ qualifier is used to specify which objects acquire mutual exclusion. 1805 A positive consequence of this design decision is the ability to support multi-monitor functions,\footnote{ 1840 Function @f1@ has a single object parameter, while functions @f2@ to @f4@ can be a single or multi-element parameter with statically unknown size. 1841 Because of the statically unknown size, \CFA only supports a single reference @mutex@ parameter, @f1@. 1842 1843 The \CFA @mutex@ qualifier does allow the ability to support multi-monitor functions,\footnote{ 1806 1844 While object-oriented monitors can be extended with a mutex qualifier for multiple-monitor members, no prior example of this feature could be found.} 1807 called \newterm{bulk acquire}.1845 where the number of acquisitions is statically known, called \newterm{bulk acquire}. 1808 1846 \CFA guarantees bulk acquisition order is consistent across calls to @mutex@ functions using the same monitors as arguments, so acquiring multiple monitors in a bulk acquire is safe from deadlock. 1809 1847 Figure~\ref{f:BankTransfer} shows a trivial solution to the bank transfer problem~\cite{BankTransfer}, where two resources must be locked simultaneously, using \CFA monitors with implicit locking and \CC with explicit locking. … … 1933 1971 % There are many aspects of scheduling in a concurrency system, all related to resource utilization by waiting threads, \ie which thread gets the resource next. 1934 1972 % Different forms of scheduling include access to processors by threads (see Section~\ref{s:RuntimeStructureCluster}), another is access to a shared resource by a lock or monitor. 1935 This section discusses scheduling for waiting threads eligible for monitor entry , \ie which user thread gets the shared resource next. (See Section~\ref{s:RuntimeStructureCluster} for scheduling kernel threads on virtual processors.)1973 This section discusses scheduling for waiting threads eligible for monitor entry~\cite{Buhr95b}, \ie which user thread gets the shared resource next. (See Section~\ref{s:RuntimeStructureCluster} for scheduling kernel threads on virtual processors.) 1936 1974 While monitor mutual-exclusion provides safe access to its shared data, the data may indicate a thread cannot proceed, \eg a bounded buffer may be full/\-empty so produce/consumer threads must block. 1937 1975 Leaving the monitor and retrying (busy waiting) is impractical for high-level programming. … … 1939 1977 Monitors eliminate busy waiting by providing synchronization within the monitor critical-section to schedule threads needing access to the shared data, where threads block versus spin. 1940 1978 Synchronization is generally achieved with internal~\cite{Hoare74} or external~\cite[\S~2.9.2]{uC++} scheduling. 1941 \newterm{Internal} (largely)schedules threads located \emph{inside} the monitor and is accomplished using condition variables with signal and wait.1942 \newterm{External} (largely)schedules threads located \emph{outside} the monitor and is accomplished with the @waitfor@ statement.1943 Note, internal scheduling has a small amount of external scheduling and vice vers us, so the naming denotes where the majority of the block threads reside (inside or outside) for scheduling.1944 For complex scheduling, the approaches can be combined, so there can be an equal number ofthreads waiting inside and outside.1945 1946 \CFA monitors do not allow calling threads to barge ahead of signalled threads (via barging prevention), which simplifies synchronization among threads in the monitor and increases correctness.1979 \newterm{Internal} largely schedules threads located \emph{inside} the monitor and is accomplished using condition variables with signal and wait. 1980 \newterm{External} largely schedules threads located \emph{outside} the monitor and is accomplished with the @waitfor@ statement. 1981 Note, internal scheduling has a small amount of external scheduling and vice versa, so the naming denotes where the majority of the block threads reside (inside or outside) for scheduling. 1982 For complex scheduling, the approaches can be combined, so there are threads waiting inside and outside. 1983 1984 \CFA monitors do not allow calling threads to barge ahead of signalled threads via barging prevention, which simplifies synchronization among threads in the monitor and increases correctness. 1947 1985 A direct consequence of this semantics is that unblocked waiting threads are not required to recheck the waiting condition, \ie waits are not in a starvation-prone busy-loop as required by the signals-as-hints style with barging. 1948 1986 Preventing barging comes directly from Hoare's semantics in the seminal paper on monitors~\cite[p.~550]{Hoare74}. … … 1953 1991 Furthermore, \CFA concurrency has no spurious wakeup~\cite[\S~9]{Buhr05a}, which eliminates an implicit self barging. 1954 1992 1955 Monitor mutual-exclusion means signalling cannot have the signaller and signalled thread in the monitor simultaneously, so only the signaller or signallee can proceed .1956 Figure~\ref{f:MonitorScheduling} shows internal /external scheduling for the bounded-buffer examples in Figure~\ref{f:GenericBoundedBuffer}.1957 For internal scheduling in Figure~\ref{f:BBInt}, the @signal@ moves the signallee (front thread of the specified condition queue) to urgentand the signaller continues (solid line).1993 Monitor mutual-exclusion means signalling cannot have the signaller and signalled thread in the monitor simultaneously, so only the signaller or signallee can proceed and the other waits on an implicit urgent list~\cite[p.~551]{Hoare74}. 1994 Figure~\ref{f:MonitorScheduling} shows internal and external scheduling for the bounded-buffer examples in Figure~\ref{f:GenericBoundedBuffer}. 1995 For internal scheduling in Figure~\ref{f:BBInt}, the @signal@ moves the signallee, front thread of the specified condition queue, to the urgent list (see Figure~\ref{f:MonitorScheduling}) and the signaller continues (solid line). 1958 1996 Multiple signals move multiple signallees to urgent until the condition queue is empty. 1959 When the signaller exits or waits, a thread is implicitly unblocked from urgent (if available)before unblocking a calling thread to prevent barging.1997 When the signaller exits or waits, a thread is implicitly unblocked from urgent, if available, before unblocking a calling thread to prevent barging. 1960 1998 (Java conceptually moves the signalled thread to the calling queue, and hence, allows barging.) 1961 Signal is used when the signaller is providing the cooperation needed by the signallee (\eg creating an empty slot in a buffer for a producer) and the signaller immediately exits the monitor to run concurrently (consume the buffer element)and passes control of the monitor to the signalled thread, which can immediately take advantage of the state change.1999 Signal is used when the signaller is providing the cooperation needed by the signallee, \eg creating an empty slot in a buffer for a producer, and the signaller immediately exits the monitor to run concurrently consuming the buffer element, and passes control of the monitor to the signalled thread, which can immediately take advantage of the state change. 1962 2000 Specifically, the @wait@ function atomically blocks the calling thread and implicitly releases the monitor lock(s) for all monitors in the function's parameter list. 1963 2001 Signalling is unconditional because signalling an empty condition queue does nothing. 1964 2002 It is common to declare condition queues as monitor fields to prevent shared access, hence no locking is required for access as the queues are protected by the monitor lock. 1965 In \CFA, a condition queue can be created /stored independently.2003 In \CFA, a condition queue can be created and stored independently. 1966 2004 1967 2005 \begin{figure} … … 2049 2087 \end{figure} 2050 2088 2051 The @signal_block@ provides the opposite unblocking order, where the signaller is moved to urgent and the signallee continues and a thread is implicitly unblocked from urgent when the signallee exits or waits (dashed line) .2052 Signal block is used when the signallee is providing the cooperation needed by the signaller (\eg if the buffer is removed and a producer hands off an item to a consumer, as in Figure~\ref{f:DatingSignalBlock})so the signaller must wait until the signallee unblocks, provides the cooperation, exits the monitor to run concurrently, and passes control of the monitor to the signaller, which can immediately take advantage of the state change.2089 The @signal_block@ provides the opposite unblocking order, where the signaller is moved to urgent and the signallee continues and a thread is implicitly unblocked from urgent when the signallee exits or waits (dashed line)~\cite[p.~551]{Hoare74}. 2090 Signal block is used when the signallee is providing the cooperation needed by the signaller, \eg if the buffer is removed and a producer hands off an item to a consumer as in Figure~\ref{f:DatingSignalBlock}, so the signaller must wait until the signallee unblocks, provides the cooperation, exits the monitor to run concurrently, and passes control of the monitor to the signaller, which can immediately take advantage of the state change. 2053 2091 Using @signal@ or @signal_block@ can be a dynamic decision based on whether the thread providing the cooperation arrives before or after the thread needing the cooperation. 2054 2092 2055 External scheduling in Figure~\ref{f:BBExt} simplifies internal scheduling byeliminating condition queues and @signal@/@wait@ (cases where it cannot are discussed shortly), and has existed in the programming language Ada for almost 40 years with variants in other languages~\cite{SR,ConcurrentC++,uC++}.2093 For external scheduling in Figure~\ref{f:BBExt}, the internal scheduling is replaced, eliminating condition queues and @signal@/@wait@ (cases where it cannot are discussed shortly), and has existed in the programming language Ada for almost 40 years with variants in other languages~\cite{SR,ConcurrentC++,uC++}. 2056 2094 While prior languages use external scheduling solely for thread interaction, \CFA generalizes it to both monitors and threads. 2057 2095 External scheduling allows waiting for events from other threads while restricting unrelated events, that would otherwise have to wait on condition queues in the monitor. … … 2062 2100 Now when a producer/consumer detects a full/empty buffer, the necessary cooperation for continuation is specified by indicating the next function call that can occur. 2063 2101 For example, a producer detecting a full buffer must have cooperation from a consumer to remove an item so function @remove@ is accepted, which prevents producers from entering the monitor, and after a consumer calls @remove@, the producer waiting on urgent is \emph{implicitly} unblocked because it can now continue its insert operation. 2064 Hence, this mechanism is done in terms of control flow, next call, versus in terms of data, channels, as in Go /Rust @select@.2102 Hence, this mechanism is done in terms of control flow, next call, versus in terms of data, channels, as in Go and Rust @select@. 2065 2103 While both mechanisms have strengths and weaknesses, \CFA uses the control-flow mechanism to be consistent with other language features. 2066 2104 2067 Figure~\ref{f:ReadersWriterLock} shows internal /external scheduling for a readers/writer lock with no barging and threads are serviced in FIFO order to eliminate staleness/freshness among the reader/writer threads.2105 Figure~\ref{f:ReadersWriterLock} shows internal and external scheduling for a readers/writer lock with no barging and threads are serviced in FIFO order to eliminate staleness and freshness among the reader/writer threads. 2068 2106 For internal scheduling in Figure~\ref{f:RWInt}, the readers and writers wait on the same condition queue in FIFO order, making it impossible to tell if a waiting thread is a reader or writer. 2069 2107 To clawback the kind of thread, a \CFA condition can store user data in the node for a blocking thread at the @wait@, \ie whether the thread is a @READER@ or @WRITER@. … … 2229 2267 For signal scheduling, the @exchange@ condition is necessary to block the thread finding the match, while the matcher unblocks to take the opposite number, post its phone number, and unblock the partner. 2230 2268 For signal-block scheduling, the implicit urgent-queue replaces the explicit @exchange@-condition and @signal_block@ puts the finding thread on the urgent stack and unblocks the matcher. 2231 2232 The dating service is an important example of a monitor that cannot be written using external scheduling. 2233 First, because scheduling requires knowledge of calling parameters to make matching decisions, and parameters of calling threads are unavailable within the monitor. 2234 For example, a girl thread within the monitor cannot examine the @ccode@ of boy threads waiting on the calling queue to determine if there is a matching partner. 2235 Second, because a scheduling decision may be delayed when there is no immediate match, which requires a condition queue for waiting, and condition queues imply internal scheduling. 2236 For example, if a girl thread could determine there is no calling boy with the same @ccode@, it must wait until a matching boy arrives. 2237 Finally, barging corrupts the dating service during an exchange because a barger may also match and change the phone numbers, invalidating the previous exchange phone number. 2269 Note, barging corrupts the dating service during an exchange because a barger may also match and change the phone numbers, invalidating the previous exchange phone number. 2238 2270 This situation shows rechecking the waiting condition and waiting again (signals-as-hints) fails, requiring significant restructured to account for barging. 2239 2271 2240 Both internal and external scheduling extend to multiple monitors in a natural way. 2272 Given external and internal scheduling, what guidelines can a programmer use to select between them? 2273 In general, external scheduling is easier to understand and code because only the next logical action (mutex function(s)) is stated, and the monitor implicitly handles all the details. 2274 Therefore, there are no condition variables, and hence, no wait and signal, which reduces coding complexity and synchronization errors. 2275 If external scheduling is simpler than internal, why not use it all the time? 2276 Unfortunately, external scheduling cannot be used if: scheduling depends on parameter value(s) or scheduling must block across an unknown series of calls on a condition variable, \ie internal scheduling. 2277 For example, the dating service cannot be written using external scheduling. 2278 First, scheduling requires knowledge of calling parameters to make matching decisions and parameters of calling threads are unavailable within the monitor. 2279 Specifically, a thread within the monitor cannot examine the @ccode@ of threads waiting on the calling queue to determine if there is a matching partner. 2280 (Similarly, if the bounded buffer or readers/writer are restructured with a single interface function with a parameter denoting producer/consumer or reader/write, they cannot be solved with external scheduling.) 2281 Second, a scheduling decision may be delayed across an unknown number of calls when there is no immediate match so the thread in the monitor must block on a condition. 2282 Specifically, if a thread determines there is no opposite calling thread with the same @ccode@, it must wait an unknown period until a matching thread arrives. 2283 For complex synchronization, both external and internal scheduling can be used to take advantage of best of properties of each. 2284 2285 Finally, both internal and external scheduling extend to multiple monitors in a natural way. 2241 2286 \begin{cquote} 2242 2287 \begin{tabular}{@{}l@{\hspace{2\parindentlnth}}l@{}} … … 2274 2319 Similarly, for @waitfor( rtn )@, the default semantics is to atomically block the acceptor and release all acquired mutex parameters, \ie @waitfor( rtn : m1, m2 )@. 2275 2320 To override the implicit multi-monitor wait, specific mutex parameter(s) can be specified, \eg @waitfor( rtn : m1 )@. 2276 @waitfor@ does statically verify the monitor types passed are the same as the acquired mutex-parameters of the given function or function pointer, hence the function (pointer)prototype must be accessible.2321 @waitfor@ does statically verify the monitor types passed are the same as the acquired mutex-parameters of the given function or function pointer, hence the prototype must be accessible. 2277 2322 % When an overloaded function appears in an @waitfor@ statement, calls to any function with that name are accepted. 2278 % The rationale is that members with the same name should perform a similar function, and therefore, all should be eligible to accept a call.2323 % The rationale is that functions with the same name should perform a similar actions, and therefore, all should be eligible to accept a call. 2279 2324 Overloaded functions can be disambiguated using a cast 2280 2325 \begin{cfa} … … 2285 2330 2286 2331 The ability to release a subset of acquired monitors can result in a \newterm{nested monitor}~\cite{Lister77} deadlock (see Section~\ref{s:MutexAcquisition}). 2287 \newpage2288 2332 \begin{cfa} 2289 2333 void foo( M & mutex m1, M & mutex m2 ) { … … 2300 2344 2301 2345 Figure~\ref{f:ExtendedWaitfor} shows the extended form of the @waitfor@ statement to conditionally accept one of a group of mutex functions, with an optional statement to be performed \emph{after} the mutex function finishes. 2302 For a @waitfor@ clause to be executed, its @when@ must be true and an outstanding call to its corresponding member(s) must exist.2346 For a @waitfor@ clause to be executed, its @when@ must be true and an outstanding call to its corresponding function(s) must exist. 2303 2347 The \emph{conditional-expression} of a @when@ may call a function, but the function must not block or context switch. 2304 If there are multiple acceptable mutex calls, selection occurs top-to-bottom (prioritized)among the @waitfor@ clauses, whereas some programming languages with similar mechanisms accept nondeterministically for this case, \eg Go \lstinline[morekeywords=select]@select@.2305 If some accept guards are true and there are no outstanding calls to these members, the acceptor is blocked until a call to one of these members is made.2348 If there are multiple acceptable mutex calls, selection is prioritized top-to-bottom among the @waitfor@ clauses, whereas some programming languages with similar mechanisms accept nondeterministically for this case, \eg Go \lstinline[morekeywords=select]@select@. 2349 If some accept guards are true and there are no outstanding calls to these functions, the acceptor is blocked until a call to one of these functions is made. 2306 2350 If there is a @timeout@ clause, it provides an upper bound on waiting. 2307 2351 If all the accept guards are false, the statement does nothing, unless there is a terminating @else@ clause with a true guard, which is executed instead. 2308 2352 Hence, the terminating @else@ clause allows a conditional attempt to accept a call without blocking. 2309 2353 If both @timeout@ and @else@ clause are present, the @else@ must be conditional, or the @timeout@ is never triggered. 2310 There is also a traditional future wait queue (not shown) (\eg Microsoft @WaitForMultipleObjects@), to wait for a specified number of future elements in the queue.2354 % There is also a traditional future wait queue (not shown) (\eg Microsoft @WaitForMultipleObjects@), to wait for a specified number of future elements in the queue. 2311 2355 Finally, there is a shorthand for specifying multiple functions using the same set of monitors: @waitfor( f, g, h : m1, m2, m3 )@. 2312 2356 … … 2315 2359 \begin{cfa} 2316 2360 `when` ( $\emph{conditional-expression}$ ) $\C{// optional guard}$ 2317 waitfor( $\emph{mutex- member-name}$ ) $\emph{statement}$ $\C{// action after call}$2361 waitfor( $\emph{mutex-function-name}$ ) $\emph{statement}$ $\C{// action after call}$ 2318 2362 `or` `when` ( $\emph{conditional-expression}$ ) $\C{// any number of functions}$ 2319 waitfor( $\emph{mutex- member-name}$ ) $\emph{statement}$2363 waitfor( $\emph{mutex-function-name}$ ) $\emph{statement}$ 2320 2364 `or` ... 2321 2365 `when` ( $\emph{conditional-expression}$ ) $\C{// optional guard}$ … … 2335 2379 The left example only accepts @mem1@ if @C1@ is true or only @mem2@ if @C2@ is true. 2336 2380 The right example accepts either @mem1@ or @mem2@ if @C1@ and @C2@ are true. 2381 Hence, the @waitfor@ has parallel semantics, accepting any true @when@ clause. 2337 2382 2338 2383 An interesting use of @waitfor@ is accepting the @mutex@ destructor to know when an object is deallocated, \eg assume the bounded buffer is restructured from a monitor to a thread with the following @main@. … … 2432 2477 If W1 waited first, the signaller must retain @m1@ amd @m2@ until completion of the outer mutex statement and then pass both to W1. 2433 2478 % Furthermore, there is an execution sequence where the signaller always finds waiter W2, and hence, waiter W1 starves. 2434 To support this efficient semantics (and prevent barging), the implementation maintains a list of monitors acquired for each blocked thread.2435 When a signaller exits or waits in a m onitor function/statement, the front waiter on urgent is unblocked if all its monitors are released.2479 To support this efficient semantics and prevent barging, the implementation maintains a list of monitors acquired for each blocked thread. 2480 When a signaller exits or waits in a mutex function or statement, the front waiter on urgent is unblocked if all its monitors are released. 2436 2481 Implementing a fast subset check for the necessary released monitors is important and discussed in the following sections. 2437 2482 % The benefit is encapsulating complexity into only two actions: passing monitors to the next owner when they should be released and conditionally waking threads if all conditions are met. … … 2442 2487 2443 2488 In a statically-typed object-oriented programming language, a class has an exhaustive list of members, even when members are added via static inheritance (see Figure~\ref{f:uCinheritance}). 2444 Knowing all members at compilation (even separate compilation) allows uniquely numbered them so the accept-statement implementation can use a fast/compact bit mask with $O(1)$ compare.2489 Knowing all members at compilation, even separate compilation, allows uniquely numbered them so the accept-statement implementation can use a fast and compact bit mask with $O(1)$ compare. 2445 2490 2446 2491 \begin{figure} … … 2493 2538 Hence, function pointers are used to identify the functions listed in the @waitfor@ statement, stored in a variable-sized array. 2494 2539 Then, the same implementation approach used for the urgent stack (see Section~\ref{s:Scheduling}) is used for the calling queue. 2495 Each caller has a list of monitors acquired, and the @waitfor@ statement performs a (short)linear search matching functions in the @waitfor@ list with called functions, and then verifying the associated mutex locks can be transfers.2540 Each caller has a list of monitors acquired, and the @waitfor@ statement performs a short linear search matching functions in the @waitfor@ list with called functions, and then verifying the associated mutex locks can be transfers. 2496 2541 2497 2542 … … 2571 2616 2572 2617 struct Msg { int i, j; }; 2573 m onitorthread GoRtn { int i; float f; Msg m; };2618 mutex thread GoRtn { int i; float f; Msg m; }; 2574 2619 void mem1( GoRtn & mutex gortn, int i ) { gortn.i = i; } 2575 2620 void mem2( GoRtn & mutex gortn, float f ) { gortn.f = f; } … … 2577 2622 void ^?{}( GoRtn & mutex ) {} 2578 2623 2579 void main( GoRtn & gortn ) with( gortn ) {// thread starts2624 void main( GoRtn & mutex gortn ) with(gortn) { // thread starts 2580 2625 2581 2626 for () { … … 2644 2689 2645 2690 \begin{cfa} 2646 m onitorthread DatingService {2691 mutex thread DatingService { 2647 2692 condition Girls[CompCodes], Boys[CompCodes]; 2648 2693 int girlPhoneNo, boyPhoneNo, ccode; … … 2708 2753 % \label{f:pingpong} 2709 2754 % \end{figure} 2710 Note, the ping/pong threads are globally declared, @pi@/@po@, and hence, start (and possibly complete)before the program main starts.2755 Note, the ping/pong threads are globally declared, @pi@/@po@, and hence, start and possibly complete before the program main starts. 2711 2756 \end{comment} 2712 2757 2713 2758 2714 \subsection{\texorpdfstring{\protect\lstinline@m onitor@ Generators / Coroutines / Threads}{monitor Generators / Coroutines / Threads}}2715 2716 \CFA generators, coroutines, and threads can also be monitors(Table~\ref{t:ExecutionPropertyComposition} cases 4, 6, 12) allowing safe \emph{direct communication} with threads, \ie the custom types can have mutex functions that are called by other threads.2759 \subsection{\texorpdfstring{\protect\lstinline@mutex@ Generators / Coroutines / Threads}{monitor Generators / Coroutines / Threads}} 2760 2761 \CFA generators, coroutines, and threads can also be @mutex@ (Table~\ref{t:ExecutionPropertyComposition} cases 4, 6, 12) allowing safe \emph{direct communication} with threads, \ie the custom types can have mutex functions that are called by other threads. 2717 2762 All monitor features are available within these mutex functions. 2718 For example, if the formatter generator (or coroutine equivalent)in Figure~\ref{f:CFAFormatGen} is extended with the monitor property and this interface function is used to communicate with the formatter:2763 For example, if the formatter generator or coroutine equivalent in Figure~\ref{f:CFAFormatGen} is extended with the monitor property and this interface function is used to communicate with the formatter: 2719 2764 \begin{cfa} 2720 2765 void fmt( Fmt & mutex fmt, char ch ) { fmt.ch = ch; resume( fmt ) } … … 2724 2769 Figure~\ref{f:DirectCommunicationComparison} shows a comparison of direct call-communication in \CFA versus indirect channel-communication in Go. 2725 2770 (Ada has a similar mechanism to \CFA direct communication.) 2726 The program thread in \CFA @main@ uses the call/return paradigm to directly communicate with the @GoRtn main@, whereas Go switches to the channel paradigm to indirectly communicate with the goroutine. 2771 % The thread main function is by default @mutex@, so the @mutex@ qualifier for the thread parameter is optional. 2772 % The reason is that the thread logically starts instantaneously in the thread main acquiring its mutual exclusion, so it starts before any calls to prepare for synchronizing these calls. 2773 The \CFA program @main@ uses the call/return paradigm to directly communicate with the @GoRtn main@, whereas Go switches to the unbuffered channel paradigm to indirectly communicate with the goroutine. 2727 2774 Communication by multiple threads is safe for the @gortn@ thread via mutex calls in \CFA or channel assignment in Go. 2775 The different between call and channel send occurs for buffered channels making the send asynchronous. 2776 In \CFA, asynchronous call and multiple buffers is provided using an administrator and worker threads~\cite{Gentleman81} and/or futures (not discussed). 2728 2777 2729 2778 Figure~\ref{f:DirectCommunicationDatingService} shows the dating-service problem in Figure~\ref{f:DatingServiceMonitor} extended from indirect monitor communication to direct thread communication. 2730 When converting a monitor to a thread (server), the coding pattern is to move as much code as possible from the accepted members into the thread main so it does an much work as possible.2779 When converting a monitor to a thread (server), the coding pattern is to move as much code as possible from the accepted functions into the thread main so it does an much work as possible. 2731 2780 Notice, the dating server is postponing requests for an unspecified time while continuing to accept new requests. 2732 For complex servers (web-servers), there can be hundreds of lines of code in the thread main and safe interaction with clients can be complex.2781 For complex servers, \eg web-servers, there can be hundreds of lines of code in the thread main and safe interaction with clients can be complex. 2733 2782 2734 2783 … … 2768 2817 2769 2818 In contrast to direct threading is indirect \newterm{thread pools}, \eg Java @executor@, where small jobs (work units) are inserted into a work pool for execution. 2770 If the jobs are dependent, \ie interact, there is an implicit /explicitdependency graph that ties them together.2819 If the jobs are dependent, \ie interact, there is an implicit dependency graph that ties them together. 2771 2820 While removing direct concurrency, and hence the amount of context switching, thread pools significantly limit the interaction that can occur among jobs. 2772 2821 Indeed, jobs should not block because that also blocks the underlying thread, which effectively means the CPU utilization, and therefore throughput, suffers. … … 2857 2906 \label{s:RuntimeStructureProcessor} 2858 2907 2859 A virtual processor is implemented by a kernel thread (\eg UNIX process), which are scheduled for execution on a hardware processor by the underlying operating system.2908 A virtual processor is implemented by a kernel thread, \eg UNIX process, which are scheduled for execution on a hardware processor by the underlying operating system. 2860 2909 Programs may use more virtual processors than hardware processors. 2861 2910 On a multiprocessor, kernel threads are distributed across the hardware processors resulting in virtual processors executing in parallel. … … 2872 2921 \label{s:Implementation} 2873 2922 2874 A primary implementation challenge is avoiding contention from dynamically allocating memory because of bulk acquire, \eg the internal-scheduling design is (almost)free of allocations.2923 A primary implementation challenge is avoiding contention from dynamically allocating memory because of bulk acquire, \eg the internal-scheduling design is almost free of allocations. 2875 2924 All blocking operations are made by parking threads onto queues, therefore all queues are designed with intrusive nodes, where each node has preallocated link fields for chaining. 2876 2925 Furthermore, several bulk-acquire operations need a variable amount of memory. … … 2918 2967 2919 2968 There are two versions of the \CFA runtime kernel: debug and non-debug. 2920 The debugging version has many runtime checks and internal assertions, \eg stack (non-writable)guard page, and checks for stack overflow whenever context switches occur among coroutines and threads, which catches most stack overflows.2969 The debugging version has many runtime checks and internal assertions, \eg stack non-writable guard page, and checks for stack overflow whenever context switches occur among coroutines and threads, which catches most stack overflows. 2921 2970 After a program is debugged, the non-debugging version can be used to significantly decrease space and increase performance. 2922 2971 … … 2926 2975 2927 2976 To test the performance of the \CFA runtime, a series of microbenchmarks are used to compare \CFA with pthreads, Java 11.0.6, Go 1.12.6, Rust 1.37.0, Python 3.7.6, Node.js 12.14.1, and \uC 7.0.0. 2928 For comparison, the package must be multi-processor (M:N), which excludes libdil l/libmil~\cite{libdill} (M:1)), and use a shared-memory programming model, \eg not message passing.2977 For comparison, the package must be multi-processor (M:N), which excludes libdil and /libmil~\cite{libdill} (M:1)), and use a shared-memory programming model, \eg not message passing. 2929 2978 The benchmark computer is an AMD Opteron\texttrademark\ 6380 NUMA 64-core, 8 socket, 2.5 GHz processor, running Ubuntu 16.04.6 LTS, and pthreads/\CFA/\uC are compiled with gcc 9.2.1. 2930 2979 … … 2940 2989 % tar --exclude-ignore=exclude -cvhf benchmark.tar benchmark 2941 2990 2991 \paragraph{Creation} 2992 2993 Creation is measured by creating and deleting a specific kind of control-flow object. 2994 Figure~\ref{f:creation} shows the code for \CFA with results in Table~\ref{t:creation}. 2995 Note, the call stacks of \CFA coroutines are lazily created on the first resume, therefore the cost of creation with and without a stack are presented. 2996 2997 \begin{multicols}{2} 2998 \lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}} 2999 \begin{cfa} 3000 @coroutine@ MyCoroutine {}; 3001 void ?{}( MyCoroutine & this ) { 3002 #ifdef EAGER 3003 resume( this ); 3004 #endif 3005 } 3006 void main( MyCoroutine & ) {} 3007 int main() { 3008 BENCH( for ( N ) { @MyCoroutine c;@ } ) 3009 sout | result; 3010 } 3011 \end{cfa} 3012 \captionof{figure}{\CFA creation benchmark} 3013 \label{f:creation} 3014 3015 \columnbreak 3016 3017 \vspace*{-16pt} 3018 \captionof{table}{Creation comparison (nanoseconds)} 3019 \label{t:creation} 3020 3021 \begin{tabular}[t]{@{}r*{3}{D{.}{.}{5.2}}@{}} 3022 \multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} & \multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\ 3023 \CFA generator & 0.6 & 0.6 & 0.0 \\ 3024 \CFA coroutine lazy & 13.4 & 13.1 & 0.5 \\ 3025 \CFA coroutine eager & 144.7 & 143.9 & 1.5 \\ 3026 \CFA thread & 466.4 & 468.0 & 11.3 \\ 3027 \uC coroutine & 155.6 & 155.7 & 1.7 \\ 3028 \uC thread & 523.4 & 523.9 & 7.7 \\ 3029 Python generator & 123.2 & 124.3 & 4.1 \\ 3030 Node.js generator & 32.3 & 32.2 & 0.3 \\ 3031 Goroutine thread & 751.0 & 750.5 & 3.1 \\ 3032 Rust tokio thread & 1860.0 & 1881.1 & 37.6 \\ 3033 Rust thread & 53801.0 & 53896.8 & 274.9 \\ 3034 Java thread & 120274.0 & 120722.9 & 2356.7 \\ 3035 Pthreads thread & 31465.5 & 31419.5 & 140.4 3036 \end{tabular} 3037 \end{multicols} 3038 2942 3039 \paragraph{Context Switching} 2943 3040 2944 3041 In procedural programming, the cost of a function call is important as modularization (refactoring) increases. 2945 3042 (In many cases, a compiler inlines function calls to increase the size and number of basic blocks for optimizing.) 2946 Similarly, when modularization extends to coroutines /threads, the time for a context switch becomes a relevant factor.3043 Similarly, when modularization extends to coroutines and threads, the time for a context switch becomes a relevant factor. 2947 3044 The coroutine test is from resumer to suspender and from suspender to resumer, which is two context switches. 2948 3045 %For async-await systems, the test is scheduling and fulfilling @N@ empty promises, where all promises are allocated before versus interleaved with fulfillment to avoid garbage collection. … … 2950 3047 The thread test is using yield to enter and return from the runtime kernel, which is two context switches. 2951 3048 The difference in performance between coroutine and thread context-switch is the cost of scheduling for threads, whereas coroutines are self-scheduling. 2952 Figure~\ref{f:ctx-switch} shows the \CFA code for a coroutine /thread with results in Table~\ref{t:ctx-switch}.3049 Figure~\ref{f:ctx-switch} shows the \CFA code for a coroutine and thread with results in Table~\ref{t:ctx-switch}. 2953 3050 2954 3051 % From: Gregor Richards <gregor.richards@uwaterloo.ca> … … 2996 3093 \uC thread & 97.3 & 97.4 & 1.0 \\ 2997 3094 Python generator & 40.9 & 41.3 & 1.5 \\ 3095 Node.js await & 1852.2 & 1854.7 & 16.4 \\ 2998 3096 Node.js generator & 32.6 & 32.2 & 1.0 \\ 2999 Node.js await & 1852.2 & 1854.7 & 16.4 \\3000 3097 Goroutine thread & 143.0 & 143.3 & 1.1 \\ 3098 Rust async await & 32.0 & 32.0 & 0.0 \\ 3099 Rust tokio thread & 143.0 & 143.0 & 1.7 \\ 3001 3100 Rust thread & 332.0 & 331.4 & 2.4 \\ 3002 3101 Java thread & 405.0 & 415.0 & 17.6 \\ … … 3005 3104 \end{multicols} 3006 3105 3106 \vspace*{-10pt} 3007 3107 \paragraph{Internal Scheduling} 3008 3108 … … 3036 3136 } 3037 3137 \end{cfa} 3138 \vspace*{-8pt} 3038 3139 \captionof{figure}{\CFA Internal-scheduling benchmark} 3039 3140 \label{f:schedint} … … 3104 3205 \paragraph{Mutual-Exclusion} 3105 3206 3106 Uncontented mutual exclusion, which frequently occurs, is measured by entering /leaving a critical section.3107 For monitors, entering and leaving a m onitorfunction is measured, otherwise the language-appropriate mutex-lock is measured.3207 Uncontented mutual exclusion, which frequently occurs, is measured by entering and leaving a critical section. 3208 For monitors, entering and leaving a mutex function is measured, otherwise the language-appropriate mutex-lock is measured. 3108 3209 For comparison, a spinning (versus blocking) test-and-test-set lock is presented. 3109 3210 Figure~\ref{f:mutex} shows the code for \CFA with results in Table~\ref{t:mutex}. … … 3142 3243 \end{multicols} 3143 3244 3144 \paragraph{Creation}3145 3146 Creation is measured by creating/deleting a specific kind of control-flow object.3147 Figure~\ref{f:creation} shows the code for \CFA with results in Table~\ref{t:creation}.3148 Note, the call stacks of \CFA coroutines are lazily created on the first resume, therefore the cost of creation with and without a stack are presented.3149 3150 \begin{multicols}{2}3151 \lstset{language=CFA,moredelim=**[is][\color{red}]{@}{@},deletedelim=**[is][]{`}{`}}3152 \begin{cfa}3153 @coroutine@ MyCoroutine {};3154 void ?{}( MyCoroutine & this ) {3155 #ifdef EAGER3156 resume( this );3157 #endif3158 }3159 void main( MyCoroutine & ) {}3160 int main() {3161 BENCH( for ( N ) { @MyCoroutine c;@ } )3162 sout | result;3163 }3164 \end{cfa}3165 \captionof{figure}{\CFA creation benchmark}3166 \label{f:creation}3167 3168 \columnbreak3169 3170 \vspace*{-16pt}3171 \captionof{table}{Creation comparison (nanoseconds)}3172 \label{t:creation}3173 3174 \begin{tabular}[t]{@{}r*{3}{D{.}{.}{5.2}}@{}}3175 \multicolumn{1}{@{}c}{} & \multicolumn{1}{c}{Median} & \multicolumn{1}{c}{Average} & \multicolumn{1}{c@{}}{Std Dev} \\3176 \CFA generator & 0.6 & 0.6 & 0.0 \\3177 \CFA coroutine lazy & 13.4 & 13.1 & 0.5 \\3178 \CFA coroutine eager & 144.7 & 143.9 & 1.5 \\3179 \CFA thread & 466.4 & 468.0 & 11.3 \\3180 \uC coroutine & 155.6 & 155.7 & 1.7 \\3181 \uC thread & 523.4 & 523.9 & 7.7 \\3182 Python generator & 123.2 & 124.3 & 4.1 \\3183 Node.js generator & 32.3 & 32.2 & 0.3 \\3184 Goroutine thread & 751.0 & 750.5 & 3.1 \\3185 Rust thread & 53801.0 & 53896.8 & 274.9 \\3186 Java thread & 120274.0 & 120722.9 & 2356.7 \\3187 Pthreads thread & 31465.5 & 31419.5 & 140.43188 \end{tabular}3189 \end{multicols}3190 3191 3245 3192 3246 \subsection{Discussion} 3193 3247 3194 Languages using 1:1 threading based on pthreads can at best meet or exceed (due to language overhead)the pthread results.3248 Languages using 1:1 threading based on pthreads can at best meet or exceed, due to language overhead, the pthread results. 3195 3249 Note, pthreads has a fast zero-contention mutex lock checked in user space. 3196 3250 Languages with M:N threading have better performance than 1:1 because there is no operating-system interactions. … … 3200 3254 3201 3255 3202 \section{Conclusion }3256 \section{Conclusions and Future Work} 3203 3257 3204 3258 Advanced control-flow will always be difficult, especially when there is temporal ordering and nondeterminism. … … 3207 3261 Combining these properties creates a number of high-level, efficient, and maintainable control-flow types: generator, coroutine, thread, each of which can be a monitor. 3208 3262 Eliminated from \CFA are barging and spurious wakeup, which are nonintuitive and lead to errors, and having to work with a bewildering set of low-level locks and acquisition techniques. 3209 \CFA high-level race-free monitors and threads provide the core mechanisms for mutual exclusion and synchronization, without having to resort to magic qualifiers like @volatile@/@atomic@.3263 \CFA high-level race-free monitors and threads, when used with mutex access function, provide the core mechanisms for mutual exclusion and synchronization, without having to resort to magic qualifiers like @volatile@ or @atomic@. 3210 3264 Extending these mechanisms to handle high-level deadlock-free bulk acquire across both mutual exclusion and synchronization is a unique contribution. 3211 3265 The \CFA runtime provides concurrency based on a preemptive M:N user-level threading-system, executing in clusters, which encapsulate scheduling of work on multiple kernel threads providing parallelism. 3212 3266 The M:N model is judged to be efficient and provide greater flexibility than a 1:1 threading model. 3213 3267 These concepts and the \CFA runtime-system are written in the \CFA language, extensively leveraging the \CFA type-system, which demonstrates the expressiveness of the \CFA language. 3214 Performance comparisons with other concurrent systems /languages show the \CFA approach is competitive across all basic operations, which translates directly into good performance in well-written applications with advanced control-flow.3268 Performance comparisons with other concurrent systems and languages show the \CFA approach is competitive across all basic operations, which translates directly into good performance in well-written applications with advanced control-flow. 3215 3269 C programmers should feel comfortable using these mechanisms for developing complex control-flow in applications, with the ability to obtain maximum available performance by selecting mechanisms at the appropriate level of need using only calling communication. 3216 3270 3217 3218 \section{Future Work}3219 3220 3271 While control flow in \CFA has a strong start, development is still underway to complete a number of missing features. 3221 3272 3273 \vspace{-5pt} 3222 3274 \paragraph{Flexible Scheduling} 3223 3275 \label{futur:sched} 3224 3276 3225 3277 An important part of concurrency is scheduling. 3226 Different scheduling algorithms can affect performance (both in terms of average and variation).3278 Different scheduling algorithms can affect performance, both in terms of average and variation. 3227 3279 However, no single scheduler is optimal for all workloads and therefore there is value in being able to change the scheduler for given programs. 3228 3280 One solution is to offer various tuning options, allowing the scheduler to be adjusted to the requirements of the workload. … … 3230 3282 Currently, the \CFA pluggable scheduler is too simple to handle complex scheduling, \eg quality of service and real-time, where the scheduler must interact with mutex objects to deal with issues like priority inversion~\cite{Buhr00b}. 3231 3283 3284 \vspace{-5pt} 3232 3285 \paragraph{Non-Blocking I/O} 3233 3286 \label{futur:nbio} 3234 3235 3287 Many modern workloads are not bound by computation but IO operations, common cases being web servers and XaaS~\cite{XaaS} (anything as a service). 3236 3288 These types of workloads require significant engineering to amortizing costs of blocking IO-operations. … … 3241 3293 A non-blocking I/O library is currently under development for \CFA. 3242 3294 3295 \vspace{-5pt} 3243 3296 \paragraph{Other Concurrency Tools} 3244 3297 \label{futur:tools} … … 3249 3302 As well, new \CFA extensions should make it possible to create a uniform interface for virtually all mutual exclusion, including monitors and low-level locks. 3250 3303 3304 \vspace{-5pt} 3251 3305 \paragraph{Implicit Threading} 3252 3306 \label{futur:implcit} 3253 3307 3254 Basic concurrent (embarrassingly parallel)applications can benefit greatly from implicit concurrency, where sequential programs are converted to concurrent, possibly with some help from pragmas to guide the conversion.3308 Basic \emph{embarrassingly parallel} applications can benefit greatly from implicit concurrency, where sequential programs are converted to concurrent, possibly with some help from pragmas to guide the conversion. 3255 3309 This type of concurrency can be achieved both at the language level and at the library level. 3256 3310 The canonical example of implicit concurrency is concurrent nested @for@ loops, which are amenable to divide and conquer algorithms~\cite{uC++book}. -
doc/papers/concurrency/mail2
r4e7c0fc0 r04b4a71 512 512 Software: Practice and Experience Editorial Office 513 513 514 515 516 Date: Sat, 18 Apr 2020 10:42:13 +0000 517 From: Richard Jones <onbehalfof@manuscriptcentral.com> 518 Reply-To: R.E.Jones@kent.ac.uk 519 To: tdelisle@uwaterloo.ca, pabuhr@uwaterloo.ca 520 Subject: Software: Practice and Experience - Decision on Manuscript ID 521 SPE-19-0219.R1 522 523 18-Apr-2020 524 525 Dear Dr Buhr, 526 527 Many thanks for submitting SPE-19-0219.R1 entitled "Advanced Control-flow and Concurrency in Cforall" to Software: Practice and Experience. The paper has now been reviewed and the comments of the referees are included at the bottom of this letter. 528 529 I believe that we are making progress here towards a paper that can be published in Software: Practice and Experience. However the referees still have significant concerns about the paper. The journal's focus is on practice and experience, and one of the the reviewers' concerns remains that your submission should focus the narrative more on the perspective of the programmer than the language designer. I agree that this would strengthen your submission, and I ask you to address this as well as the referees' other comments. 530 531 A revised version of your manuscript that takes into account the comments of the referee(s) will be reconsidered for publication. 532 533 Please note that submitting a revision of your manuscript does not guarantee eventual acceptance, and that your revision may be subject to re-review by the referees before a decision is rendered. 534 535 You have 90 days from the date of this email to submit your revision. If you are unable to complete the revision within this time, please contact me to request a short extension. 536 537 You can upload your revised manuscript and submit it through your Author Center. Log into https://mc.manuscriptcentral.com/spe and enter your Author Center, where you will find your manuscript title listed under "Manuscripts with Decisions". 538 539 When submitting your revised manuscript, you will be able to respond to the comments made by the referee(s) in the space provided. You can use this space to document any changes you make to the original manuscript. 540 541 If you would like help with English language editing, or other article preparation support, Wiley Editing Services offers expert help with English Language Editing, as well as translation, manuscript formatting, and figure formatting at www.wileyauthors.com/eeo/preparation. You can also check out our resources for Preparing Your Article for general guidance about writing and preparing your manuscript at www.wileyauthors.com/eeo/prepresources. 542 543 Once again, thank you for submitting your manuscript to Software: Practice and Experience and I look forward to receiving your revision. 544 545 Sincerely, 546 Richard 547 548 Prof. Richard Jones 549 Software: Practice and Experience 550 R.E.Jones@kent.ac.uk 551 552 553 Referee(s)' Comments to Author: 554 555 Reviewing: 1 556 557 Comments to the Author 558 (A relatively short second review) 559 560 I thank the authors for their revisions and comprehensive response to 561 reviewers' comments --- many of my comments have been successfully 562 addressed by the revisions. Here I'll structure my comments around 563 the main salient points in that response which I consider would 564 benefit from further explanation. 565 566 > Table 1 is moved to the start and explained in detail. 567 568 I consider this change makes a significant improvement to the paper, 569 laying out the landscape of language features at the start, and thus 570 addresses my main concerns about the paper. 571 572 I still have a couple of issues --- perhaps the largest is that it's 573 still not clear at this point in the paper what some of these options 574 are, or crucially how they would be used. I don't know if it's 575 possbile to give high-level examples or use cases to be clear about 576 these up front - or if that would duplicate too much information from 577 later in the paper - either way expanding out the discussion - even if 578 just two a couple of sentences for each row - would help me more. The 579 point is not just to define these categories but to ensure the 580 readers' understanding of these definitons agrees with that used in 581 the paper. 582 583 in a little more detail: 584 585 * 1st para section 2 begs the question: why not support each 586 dimension independently, and let the programmer or library designer 587 combiine features? 588 589 * "execution state" seems a relatively low-level description here. 590 I don't think of e.g. the lambda calculus that way. Perhaps it's as 591 good a term as any. 592 593 * Why must there "be language mechanisms to create, block/unblock, 594 and join with a thread"? There aren't in Smalltalk (although there 595 are in the runtime). Especially given in Cforall those mechanisms 596 are *implicit* on thread creation and destruction? 597 598 * "Case 1 is a function that borrows storage for its state (stack 599 frame/activation) and a thread from its invoker" 600 601 this much makes perfect sense to me, but I don't understand how a 602 non-stateful, non-theaded function can then retain 603 604 "this state across callees, ie, function local-variables are 605 retained on the stack across calls." 606 607 how can it retain function-local values *across calls* when it 608 doesn't have any functional-local state? 609 610 I'm not sure if I see two separate cases here - rougly equivalent 611 to C functions without static storage, and then C functions *with* 612 static storage. I assumed that was the distinction between cases 1 613 & 3; but perhpas the actual distinction is that 3 has a 614 suspend/resume point, and so the "state" in figure 1 is this 615 component of execution state (viz figs 1 & 2), not the state 616 representing the cross-call variables? 617 618 > but such evaluation isn't appropriate for garbage-collected or JITTed 619 languages like Java or Go. 620 621 For JITTed languages in particular, reporting peak performance needs 622 to "warm up" the JIT with a number of iterators before beginning 623 measurement. Actually for JIT's its even worse: see Edd Barrett et al 624 OOPSLA 2017. 625 626 627 628 minor issues: 629 630 * footnote A - I've looked at various other papers & the website to 631 try to understand how "object-oriented" Cforall is - I'm still not 632 sure. This footnote says Cforall has "virtuals" - presumably 633 virtual functions, i.e. dynamic dispatch - and inheritance: that 634 really is OO as far as I (and most OO people) are concerned. For 635 example Haskell doesn't have inheritance, so it's not OO; while 636 CLOS (the Common Lisp *Object* System) or things like Cecil and 637 Dylan are considered OO even though they have "multiple function 638 parameters as receivers", lack "lexical binding between a structure 639 and set of functions", and don't have explicit receiver invocation 640 syntax. Python has receiver syntax, but unlike Java or Smalltalk 641 or C++, method declarations still need to have an explicit "self" 642 receiver parameter. Seems to me that Go, for example, is 643 more-or-less OO with interfaces, methods, and dynamic dispatch (yes 644 also and an explicit receiver syntax but that's not 645 determiniative); while Rust lacks dynamic dispatch built-in. C is 646 not OO as a language, but as you say given it supports function 647 pointers with structures, it does support an OO programm style. 648 649 This is why I again recommend just not buying into this fight: not 650 making any claims about whether Cforall is OO or is not - because 651 as I see it, the rest of the paper doesn't depend on whether 652 Cforall is OO or not. That said: this is just a recommendation, 653 and I won't quibble over this any further. 654 655 * is a "monitor function" the same as a "mutex function"? 656 if so the paper should pick one term; if not, make the distinction clear. 657 658 659 * "As stated on line 1 because state declarations from the generator 660 type can be moved out of the coroutine type into the coroutine main" 661 662 OK sure, but again: *why* would a programmer want to do that? 663 (Other than, I guess, to show the difference between coroutines & 664 generators?) Perhaps another way to put this is that the first 665 para of 3.2 gives the disadvantages of coroutines vs-a-vs 666 generators, briefly describes the extended semantics, but never 667 actualy says why a programmer may want those extended semantics, 668 or how they would benefit. I don't mean to belabour the point, 669 but (generalist?) readers like me would generally benefit from 670 those kinds of discussions about each feature throughout the 671 paper: why might a programmer want to use them? 672 673 674 > p17 if the multiple-monitor entry procedure really is novel, write a paper 675 > about that, and only about that. 676 677 > We do not believe this is a practical suggestion. 678 679 * I'm honestly not trying to be snide here: I'm not an expert on 680 monitor or concurrent implementations. Brinch Hansen's original 681 monitors were single acquire; this draft does not cite any other 682 previous work that I could see. I'm not suggesting that the brief 683 mention of this mechanism necessarily be removed from this paper, 684 but if this is novel (and a clear advance over a classical OO 685 monitor a-la Java which only acquires the distinguished reciever) 686 then that would be worth another paper in itself. 687 688 > * conclusion should conclude the paper, not the related. 689 > We do not understand this comment.if ithis 690 691 My typo: the paper's conclusion should come at the end, after the 692 future work section. 693 694 695 696 697 To encourage accountability, I'm signing my reviews in 2020. 698 For the record, I am James Noble, kjx@ecs.vuw.ac.nz. 699 700 701 Reviewing: 2 702 703 Comments to the Author 704 I thank the authors for their detailed response. To respond to a couple of points raised in response to my review (number 2): 705 706 - on the Boehm paper and whether code is "all sequential to the compiler": I now understand the authors' position better and suspect we are in violent agreement, except for whether it's appropriate to use the rather breezy phrase "all sequential to the compiler". It would be straightforward to clarify that code not using the atomics features is optimized *as if* it were sequential, i.e. on the assumption of a lack of data races. 707 708 - on the distinction between "mutual exclusion" and "synchronization": the added citation does help, in that it makes a coherent case for the definition the authors prefer. However, the text could usefully clarify that this is a matter of definition not of fact, given especially that in my assessment the authors' preferred definition is not the most common one. (Although the mention of Hoare's apparent use of this definition is one data point, countervailing ones are found in many contemporaneous or later papers, e.g. Habermann's 1972 "Synchronization of Communicating Processes" (CACM 15(3)), Reed & Kanodia's 1979 "Synchronization with eventcounts and sequencers" (CACM (22(2)) and so on.) 709 710 I am glad to see that the authors have taken on board most of the straightforward improvements I suggested. 711 712 However, a recurring problem of unclear writing still remains through many parts of the paper, including much of sections 2, 3 and 6. To highlight a couple of problem patches (by no means exhaustive): 713 714 - section 2 (an expanded version of what was previously section 5.9) lacks examples and is generally obscure and allusory ("the most advanced feature" -- name it! "in triplets" -- there is only one triplet!; what are "execution locations"? "initialize" and "de-initialize" what? "borrowed from the invoker" is a concept in need of explaining or at least a fully explained example -- in what sense does a plain function borrow" its stack frame? "computation only" as opposed to what? in 2.2, in what way is a "request" fundamental to "synchronization"? and the "implicitly" versus "explicitly" point needs stating as elsewhere, with a concrete example e.g. Java built-in mutexes versus java.util.concurrent). 715 716 - section 6: 6.2 omits the most important facts in preference for otherwise inscrutable detail: "identify the kind of parameter" (first say *that there are* kinds of parameter, and what "kinds" means!); "mutex parameters are documentation" is misleading (they are also semantically significant!) and fails to say *what* they mean; the most important thing is surely that 'mutex' is a language feature for performing lock/unlock operations at function entry/exit. So say it! The meanings of examples f3 and f4 remain unclear. Meanwhile in 6.3, "urgent" is not introduced (we are supposed to infer its meaning from Figure 12, but that Figure is incomprehensible to me), and we are told of "external scheduling"'s long history in Ada but not clearly what it actually means; 6.4's description of "waitfor" tells us it is different from an if-else chain but tries to use two *different* inputs to tell us that the behavior is different; tell us an instance where *the same* values of C1 and C2 give different behavior (I even wrote out a truth table and still don't see the semantic difference) 717 718 The authors frequently use bracketed phrases, and sometimes slashes "/", in ways that are confusing and/or detrimental to readability. Page 13 line 2's "forward (backward)" is one particularly egregious example. In general I would recommend the the authors try to limit their use of parentheses and slashes as a means of forcing a clearer wording to emerge. Also, the use of "eg." is often cursory and does not explain the examples given, which are frequently a one- or two-word phrase of unclear referent. 719 720 Considering the revision more broadly, none of the more extensive or creative rewrites I suggested in my previous review have been attempted, nor any equivalent efforts to improve its readability. The hoisting of the former section 5.9 is a good idea, but the newly added material accompanying it (around Table 1) suffers fresh deficiencies in clarity. Overall the paper is longer than before, even though (as my previous review stated), I believe a shorter paper is required in order to serve the likely purpose of publication. (Indeed, the authors' letter implies that a key goal of publication is to build community and gain external users.) 721 722 Given this trajectory, I no longer see a path to an acceptable revision of the present submission. Instead I suggest the authors consider splitting the paper in two: one half about coroutines and stack management, the other about mutexes, monitors and the runtime. (A briefer presentation of the runtime may be helpful in the first paper also, and a brief recap of the generator and coroutine support is obviously needed in the second too.) Both of these new papers would need to be written with a strong emphasis on clarity, paying great care to issues of structure, wording, choices of example, and restraint (saying what's important, not everything that could be said). I am confident the authors could benefit from getting early feedback from others at their institution. For the performance experiments, of course these do not split evenly -- most (but not all) belong in the second of these two hypothetical papers. But the first of them would still have plenty of meat to it; for me, a clear and thorough study of the design space around coroutines is the most interesting and tantalizing prospect. 723 724 I do not buy the authors' defense of the limited practical experience or "non-micro" benchmarking presented. Yes, gaining external users is hard and I am sympathetic on that point. But building something at least *somewhat* substantial with your own system should be within reach, and without it the "practice and experience" aspects of the work have not been explored. Clearly C\/ is the product of a lot of work over an extended period, so it is a surprise that no such experience is readily available for inclusion. 725 726 Some smaller points: 727 728 It does not seem right to state that a stack is essential to Von Neumann architectures -- since the earliest Von Neumann machines (and indeed early Fortran) did not use one. 729 730 To elaborate on something another reviewer commented on: it is a surprise to find a "Future work" section *after* the "Conclusion" section. A "Conclusions and future work" section often works well. 731 732 733 Reviewing: 3 734 735 Comments to the Author 736 This is the second round of reviewing. 737 738 As in the first review, I found that the paper (and Cforall) contains 739 a lot of really interesting ideas, but it remains really difficult to 740 have a good sense of which idea I should use and when. This applies in 741 different ways to different features from the language: 742 743 * coroutines/generators/threads: here there is 744 some discussion, but it can be improved. 745 * interal/external scheduling: I didn't find any direct comparison 746 between these features, except by way of example. 747 748 I requested similar things in my previous review and I see that 749 content was added in response to those requests. Unfortunately, I'm 750 not sure that I can say it improved the paper's overall read. I think 751 in some sense the additions were "too much" -- I would have preferred 752 something more like a table or a few paragraphs highlighting the key 753 reasons one would pick one construct or the other. 754 755 In general, I do wonder if the paper is just trying to do too much. 756 The discussion of clusters and pre-emption in particular feels quite 757 rushed. 758 759 ## Summary 760 761 I make a number of suggestions below but the two most important 762 I think are: 763 764 * Recommend to shorten the comparison on coroutine/generator/threads 765 in Section 2 to a paragraph with a few examples, or possibly a table 766 explaining the trade-offs between the constructs 767 * Recommend to clarify the relationship between internal/external 768 scheduling -- is one more general but more error-prone or low-level? 769 770 ## Coroutines/generators/threads 771 772 There is obviously a lot of overlap between these features, and in 773 particular between coroutines and generators. As noted in the previous 774 review, many languages have chosen to offer *only* generators, and to 775 build coroutines by stacks of generators invoking one another. 776 777 I believe the newly introduced Section 2 of the paper is trying to 778 motivate why each of these constructs exist, but I did not find it 779 effective. It was dense and difficult to understand. I think the 780 problem is that Section 2 seems to be trying to derive "from first 781 principles" why each construct exists, but I think that a more "top 782 down" approach would be easier to understand. 783 784 In fact, the end of Section 2.1 (on page 5) contains a particular 785 paragraph that embodies this "top down" approach. It starts, 786 "programmers can now answer three basic questions", and thus gives 787 some practical advice for which construct you should use and when. I 788 think giving some examples of specific applications that this 789 paragraph, combined with some examples of cases where each construct 790 was needed, would be a better approach. 791 792 I don't think this compariosn needs to be very long. It seems clear 793 enough that one would 794 795 * prefer generators for simple computations that yield up many values, 796 * prefer coroutines for more complex processes that have significant 797 internal structure, 798 * prefer threads for cases where parallel execution is desired or 799 needed. 800 801 I did appreciate the comparison in Section 2.3 between async-await in 802 JS/Java and generators/coroutines. I agree with its premise that those 803 mechanisms are a poor replacement for generators (and, indeed, JS has 804 a distinct generator mechanism, for example, in part for this reason). 805 I believe I may have asked for this in a previous review, but having 806 read it, I wonder if it is really necessary, since those mechanisms 807 are so different in purpose. 808 809 ## Internal vs external scheduling 810 811 I find the motivation for supporting both internal and external 812 scheduling to be fairly implicit. After several reads through the 813 section, I came to the conclusion that internal scheduling is more 814 expressive than external scheduling, but sometimes less convenient or 815 clear. Is this correct? If not, it'd be useful to clarify where 816 external scheduling is more expressive. 817 818 The same is true, I think, of the `signal_block` function, which I 819 have not encountered before; it seems like its behavior can be modeled 820 with multiple condition variables, but that's clearly more complex. 821 822 One question I had about `signal_block`: what happens if one signals 823 but no other thread is waiting? Does it block until some other thread 824 waits? Or is that user error? 825 826 I would find it very interesting to try and capture some of the 827 properties that make internal vs external scheduling the better 828 choice. 829 830 For example, it seems to me that external scheduling works well if 831 there are only a few "key" operations, but that internal scheduling 832 might be better otherwise, simply because it would be useful to have 833 the ability to name a signal that can be referenced by many 834 methods. Consider the bounded buffer from Figure 13: if it had 835 multiple methods for removing elements, and not just `remove`, then 836 the `waitfor(remove)` call in `insert` might not be sufficient. 837 838 ## Comparison of external scheduling to messaging 839 840 I did enjoy the section comparing external scheduling to Go's 841 messaging mechanism, which I believe is a new addition. 842 843 I believe that one difference between the Go program and the Cforall 844 equivalent is that the Goroutine has an associated queue, so that 845 multiple messages could be enqueued, whereas the Cforall equivalent is 846 effectively a "bounded buffer" of length 1. Is that correct? I think 847 this should be stated explicitly. (Presumably, one could modify the 848 Cforall program to include an explicit vector of queued messages if 849 desired, but you would also be reimplementing the channel 850 abstraction.) 851 852 Also, in Figure 20, I believe that there is a missing `mutex` keyword. 853 The fiugre states: 854 855 ``` 856 void main(GoRtn & gortn) with(gortn) { 857 ``` 858 859 but I think it should probably be as follows: 860 861 ``` 862 void main(GoRtn & mutex gortn) with(gortn) { 863 ``` 864 865 Unless there is some implicit `mutex` associated with being a main 866 function for a `monitor thread`. 867 868 ## Atomic operations and race freedom 869 870 I was glad to see that the paper acknowledged that Cforall still had 871 low-level atomic operations, even if their use is discouraged in favor 872 of higher-level alternatives. 873 874 However, I still feel that the conclusion overstates the value of the 875 contribution here when it says that "Cforall high-level race-free 876 monitors and threads provide the core mechanisms for mutual exclusion 877 and synchronization, without the need for volatile and atomics". I 878 feel confident that Java programmers, for example, would be advised to 879 stick with synchronized methods whenever possible, and it seems to me 880 that they offer similar advantages -- but they sometimes wind up using 881 volatiles for performance reasons. 882 883 I was also confused by the term "race-free" in that sentence. In 884 particular, I don't think that Cforall has any mechanisms for 885 preventing *data races*, and it clearly doesn't prevent "race 886 conditions" (which would bar all sorts of useful programs). I suppose 887 that "race free" here might be referring to the improvements such as 888 removing barging behavior. 889 890 ## Performance comparisons 891 892 In my previous review, I requested comparisons against Rust and 893 node.js, and I see that the new version of the paper includes both, 894 which is a good addition. 895 896 One note on the Rust results: I believe that the results are comparing 897 against the threads found in Rust's standard library, which are 898 essentially a shallow wrapper around pthreads, and hence the 899 performance is quite close to pthread performance (as one would 900 expect). It would perhaps be more interesting to see a comparison 901 built using [tokio] or [async-std], two of the more prominent 902 user-space threading libraries that build on Rust's async-await 903 feature (which operates quite differently than Javascript's 904 async-await, in that it doesn't cause every aync function call to 905 schedule a distinct task). 906 907 [tokio]: https://tokio.rs/ 908 [async-std]: https://async.rs/ 909 910 That said, I am satisfied with the performance results as they are in 911 the current revision. 912 913 ## Minor notes and typos 914 915 Several figures used the `with` keyword. I deduced that `with(foo)` 916 permits one to write `bar` instead of `foo.bar`. It seems worth 917 introducing. Apologies if this is stated in the paper, if so I missed 918 it. 919 920 On page 20, section 6.3, "external scheduling and vice versus" should be 921 "external scheduling and vice versa". 922 923 On page 5, section 2.3, the paper states "we content" but it should be 924 "we contend". 925 926 Reviewing: Editor 927 928 A few small comments in addition to those of the referees. 929 930 Page 1. I don't believe that it s fair to imply that Scala is "research vehicle" as it is used by major players, Twitter being the most prominent example. 931 932 Page 15. Must Cforall threads start after construction (e.g. see your example on page 15, line 21)? I can think of examples where it is not desirable that threads start immediately after construction, e.g. a game with N players, each of whom is expensive to create, but all of whom should be started at the same time. 933 934 Page 18, line 17: is using 935
Note: See TracChangeset
for help on using the changeset viewer.