Changeset d5ccbe9 for doc/papers


Ignore:
Timestamp:
Mar 16, 2018, 1:57:38 PM (4 years ago)
Author:
Peter A. Buhr <pabuhr@…>
Branches:
aaron-thesis, arm-eh, cleanup-dtors, deferred_resn, demangler, jacob/cs343-translation, jenkins-sandbox, master, new-ast, new-ast-unique-expr, new-env, no_list, persistent-indexer, with_gc
Children:
fdfced6
Parents:
759f05f
git-author:
Peter A. Buhr <pabuhr@…> (03/16/18 12:14:02)
git-committer:
Peter A. Buhr <pabuhr@…> (03/16/18 13:57:38)
Message:

modify for SPE macros

Location:
doc/papers/concurrency
Files:
323 added
4 edited

Legend:

Unmodified
Added
Removed
  • doc/papers/concurrency/.gitignore

    r759f05f rd5ccbe9  
    33*.pdf
    44*.ps
     5
     6Paper.out.ps
     7WileyNJD-AMA.bst
  • doc/papers/concurrency/Makefile

    r759f05f rd5ccbe9  
    33Build = build
    44Figures = figures
    5 Macros = ../../LaTeXmacros
    6 TeXLIB = .:style:annex:${Macros}:${Build}:../../bibliography:
     5Macros = AMA/AMA-stix/ama
     6TeXLIB = .:style:annex:../../LaTeXmacros:${Macros}:${Build}:../../bibliography:
    77LaTeX  = TEXINPUTS=${TeXLIB} && export TEXINPUTS && latex -halt-on-error -output-directory=${Build}
    88BibTeX = BIBINPUTS=${TeXLIB} && export BIBINPUTS && bibtex
    99
    10 MAKEFLAGS = --no-print-directory --silent #
     10MAKEFLAGS = --no-print-directory # --silent
    1111VPATH = ${Build} ${Figures}
    1212
     
    4040
    4141DOCUMENT = Paper.pdf
     42BASE = ${basename ${DOCUMENT}}
    4243
    4344# Directives #
     
    4849
    4950clean :
    50         @rm -frv ${DOCUMENT} ${basename ${DOCUMENT}}.ps ${Build}
     51        @rm -frv ${DOCUMENT} ${BASE}.ps WileyNJD-AMA.bst ${BASE}.out.ps ${Build}
    5152
    5253# File Dependencies #
    5354
    54 ${DOCUMENT} : ${basename ${DOCUMENT}}.ps
     55${DOCUMENT} : ${BASE}.ps
    5556        ps2pdf $<
    5657
    57 ${basename ${DOCUMENT}}.ps : ${basename ${DOCUMENT}}.dvi
     58${BASE}.ps : ${BASE}.dvi
    5859        dvips ${Build}/$< -o $@
    5960
    60 ${basename ${DOCUMENT}}.dvi : Makefile ${Build} ${GRAPHS} ${PROGRAMS} ${PICTURES} ${FIGURES} ${SOURCES} \
    61                 ${Macros}/common.tex ${Macros}/indexstyle annex/local.bib ../../bibliography/pl.bib
     61${BASE}.dvi : Makefile ${Build} ${BASE}.out.ps WileyNJD-AMA.bst ${GRAPHS} ${PROGRAMS} ${PICTURES} ${FIGURES} ${SOURCES} \
     62                annex/local.bib ../../bibliography/pl.bib
    6263        # Must have *.aux file containing citations for bibtex
    6364        if [ ! -r ${basename $@}.aux ] ; then ${LaTeX} ${basename $@}.tex ; fi
    64         -${BibTeX} ${Build}/${basename $@}
     65        ${BibTeX} ${Build}/${basename $@}
    6566        # Some citations reference others so run again to resolve these citations
    6667        ${LaTeX} ${basename $@}.tex
    67         -${BibTeX} ${Build}/${basename $@}
     68        ${BibTeX} ${Build}/${basename $@}
    6869        # Run again to finish citations
    6970        ${LaTeX} ${basename $@}.tex
     
    7374${Build}:
    7475        mkdir -p ${Build}
     76
     77${BASE}.out.ps:
     78        ln -fs build/Paper.out.ps .
     79
     80WileyNJD-AMA.bst:
     81        ln -fs AMA/AMA-stix/ama/WileyNJD-AMA.bst .
    7582
    7683%.tex : %.fig
  • doc/papers/concurrency/Paper.tex

    r759f05f rd5ccbe9  
    77% math escape $...$ (dollar symbol)
    88
    9 \documentclass[10pt]{article}
     9\documentclass[AMA,STIX1COL]{WileyNJD-v2}
     10
     11\articletype{RESEARCH ARTICLE}%
     12
     13\received{26 April 2016}
     14\revised{6 June 2016}
     15\accepted{6 June 2016}
     16
     17\raggedbottom
    1018
    1119%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    1220
    1321% Latex packages used in the document.
     22
    1423\usepackage[T1]{fontenc}                                        % allow Latin1 (extended ASCII) characters
    1524\usepackage{textcomp}
    1625\usepackage[latin1]{inputenc}
    17 \usepackage{fullpage,times,comment}
     26
    1827\usepackage{epic,eepic}
     28\usepackage{xspace}
     29\usepackage{comment}
    1930\usepackage{upquote}                                            % switch curled `'" to straight
    20 \usepackage{calc}
    21 \usepackage{xspace}
     31\usepackage{listings}                                           % format program code
    2232\usepackage[labelformat=simple]{subfig}
    2333\renewcommand{\thesubfigure}{(\alph{subfigure})}
    24 \usepackage{graphicx}
    25 \usepackage{tabularx}
    26 \usepackage{multicol}
    27 \usepackage{varioref}
    28 \usepackage{listings}                                           % format program code
    29 \usepackage[flushmargin]{footmisc}                              % support label/reference in footnote
    30 \usepackage{latexsym}                                           % \Box glyph
    31 \usepackage{mathptmx}                                           % better math font with "times"
    32 \usepackage[usenames]{color}
     34\usepackage{siunitx}
     35\sisetup{ binary-units=true }
     36\input{style}                                                           % bespoke macros used in the document
     37
     38\hypersetup{breaklinks=true}
     39\definecolor{OliveGreen}{cmyk}{0.64 0 0.95 0.40}
     40\definecolor{Mahogany}{cmyk}{0 0.85 0.87 0.35}
     41\definecolor{Plum}{cmyk}{0.50 1 0 0}
     42
    3343\usepackage[pagewise]{lineno}
    3444\renewcommand{\linenumberfont}{\scriptsize\sffamily}
    35 \usepackage{fancyhdr}
    36 \usepackage{float}
    37 \usepackage{siunitx}
    38 \sisetup{ binary-units=true }
    39 \input{style}                                                   % bespoke macros used in the document
    40 \usepackage{url}
    41 \usepackage[dvips,plainpages=false,pdfpagelabels,pdfpagemode=UseNone,colorlinks=true,pagebackref=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pagebackref=true,breaklinks=true]{hyperref}
    42 \usepackage{breakurl}
    43 \urlstyle{rm}
    44 
    45 \setlength{\topmargin}{-0.45in}                         % move running title into header
    46 \setlength{\headsep}{0.25in}
     45
     46\lefthyphenmin=4                                                        % hyphen only after 4 characters
     47\righthyphenmin=4
    4748
    4849%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     
    5051% Names used in the document.
    5152
    52 \newcommand{\Version}{1.0.0}
    5353\newcommand{\CS}{C\raisebox{-0.9ex}{\large$^\sharp$}\xspace}
    5454
     
    6767%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    6868
    69 \setcounter{secnumdepth}{2}                           % number subsubsections
    70 \setcounter{tocdepth}{2}                              % subsubsections in table of contents
    71 % \linenumbers                                          % comment out to turn off line numbering
    72 
    73 \title{Concurrency in \CFA}
    74 \author{Thierry Delisle and Peter A. Buhr, Waterloo, Ontario, Canada}
     69\title{\texorpdfstring{Concurrency in \protect\CFA}{Concurrency in Cforall}}
     70
     71\author[1]{Thierry Delisle}
     72\author[1]{Peter A. Buhr*}
     73\authormark{Thierry Delisle \textsc{et al}}
     74
     75\address[1]{\orgdiv{David R. Cheriton School of Computer Science}, \orgname{University of Waterloo}, \orgaddress{\state{Ontario}, \country{Canada}}}
     76
     77\corres{*Peter A. Buhr, \email{pabuhr{\char`\@}uwaterloo.ca}}
     78\presentaddress{David R. Cheriton School of Computer Science, University of Waterloo, Waterloo, ON, N2L 3G1, Canada}
     79
     80
     81\abstract[Summary]{
     82\CFA is a modern, polymorphic, \emph{non-object-oriented} extension of the C programming language.
     83This paper discusses the design of the concurrency and parallelism features in \CFA, and the concurrent runtime-system that supports them.
     84These features are created from scratch as ISO C lacks concurrency, relying largely on pthreads.
     85Coroutines and lightweight (user) threads are introduced into the language.
     86In addition, monitors are added as a high-level mechanism for mutual exclusion and synchronization.
     87A unique contribution is allowing multiple monitors to be safely acquired simultaneously.
     88All features respect the expectations of C programmers, while being fully integrate with the \CFA polymorphic type-system and other language features.
     89Finally, experimental results are presented to validate several of the new features with other concurrent programming-languages.
     90}%
     91
     92\keywords{concurrency, runtime, coroutines, threads, C, Cforall}
    7593
    7694
    7795\begin{document}
     96\linenumbers                                            % comment out to turn off line numbering
     97
    7898\maketitle
    7999
    80 \begin{abstract}
    81 \CFA is a modern, \emph{non-object-oriented} extension of the C programming language.
    82 This paper serves as a definition and an implementation for the concurrency and parallelism \CFA offers. These features are created from scratch due to the lack of concurrency in ISO C. Lightweight threads are introduced into the language. In addition, monitors are introduced as a high-level tool for control-flow based synchronization and mutual-exclusion. The main contributions of this paper are two-fold: it extends the existing semantics of monitors introduce by~\cite{Hoare74} to handle monitors in groups and also details the engineering effort needed to introduce these features as core language features. Indeed, these features are added with respect to expectations of C programmers, and integrate with the \CFA type-system and other language features.
    83 \end{abstract}
    84 
    85 %----------------------------------------------------------------------
    86 % MAIN BODY
    87 %----------------------------------------------------------------------
    88 
     100% ======================================================================
    89101% ======================================================================
    90102\section{Introduction}
    91103% ======================================================================
    92 
    93 This paper provides a minimal concurrency \textbf{api} that is simple, efficient and can be reused to build higher-level features. The simplest possible concurrency system is a thread and a lock but this low-level approach is hard to master. An easier approach for users is to support higher-level constructs as the basis of concurrency. Indeed, for highly productive concurrent programming, high-level approaches are much more popular~\cite{HPP:Study}. Examples are task based, message passing and implicit threading. The high-level approach and its minimal \textbf{api} are tested in a dialect of C, called \CFA. Furthermore, the proposed \textbf{api} doubles as an early definition of the \CFA language and library. This paper also provides an implementation of the concurrency library for \CFA as well as all the required language features added to the source-to-source translator.
    94 
    95 There are actually two problems that need to be solved in the design of concurrency for a programming language: which concurrency and which parallelism tools are available to the programmer. While these two concepts are often combined, they are in fact distinct, requiring different tools~\cite{Buhr05a}. Concurrency tools need to handle mutual exclusion and synchronization, while parallelism tools are about performance, cost and resource utilization.
    96 
    97 In the context of this paper, a \textbf{thread} is a fundamental unit of execution that runs a sequence of code, generally on a program stack. Having multiple simultaneous threads gives rise to concurrency and generally requires some kind of locking mechanism to ensure proper execution. Correspondingly, \textbf{concurrency} is defined as the concepts and challenges that occur when multiple independent (sharing memory, timing dependencies, etc.) concurrent threads are introduced. Accordingly, \textbf{locking} (and by extension locks) are defined as a mechanism that prevents the progress of certain threads in order to avoid problems due to concurrency. Finally, in this paper \textbf{parallelism} is distinct from concurrency and is defined as running multiple threads simultaneously. More precisely, parallelism implies \emph{actual} simultaneous execution as opposed to concurrency which only requires \emph{apparent} simultaneous execution. As such, parallelism is only observable in the differences in performance or, more generally, differences in timing.
     104% ======================================================================
     105
     106This paper provides a minimal concurrency \textbf{api} that is simple, efficient and can be reused to build higher-level features.
     107The simplest possible concurrency system is a thread and a lock but this low-level approach is hard to master.
     108An easier approach for users is to support higher-level constructs as the basis of concurrency.
     109Indeed, for highly productive concurrent programming, high-level approaches are much more popular~\cite{HPP:Study}.
     110Examples are task based, message passing and implicit threading.
     111The high-level approach and its minimal \textbf{api} are tested in a dialect of C, called \CFA.
     112Furthermore, the proposed \textbf{api} doubles as an early definition of the \CFA language and library.
     113This paper also provides an implementation of the concurrency library for \CFA as well as all the required language features added to the source-to-source translator.
     114
     115There are actually two problems that need to be solved in the design of concurrency for a programming language: which concurrency and which parallelism tools are available to the programmer.
     116While these two concepts are often combined, they are in fact distinct, requiring different tools~\cite{Buhr05a}.
     117Concurrency tools need to handle mutual exclusion and synchronization, while parallelism tools are about performance, cost and resource utilization.
     118
     119In the context of this paper, a \textbf{thread} is a fundamental unit of execution that runs a sequence of code, generally on a program stack.
     120Having multiple simultaneous threads gives rise to concurrency and generally requires some kind of locking mechanism to ensure proper execution.
     121Correspondingly, \textbf{concurrency} is defined as the concepts and challenges that occur when multiple independent (sharing memory, timing dependencies, etc.) concurrent threads are introduced.
     122Accordingly, \textbf{locking} (and by extension locks) are defined as a mechanism that prevents the progress of certain threads in order to avoid problems due to concurrency.
     123Finally, in this paper \textbf{parallelism} is distinct from concurrency and is defined as running multiple threads simultaneously.
     124More precisely, parallelism implies \emph{actual} simultaneous execution as opposed to concurrency which only requires \emph{apparent} simultaneous execution.
     125As such, parallelism is only observable in the differences in performance or, more generally, differences in timing.
    98126
    99127% ======================================================================
     
    105133The following is a quick introduction to the \CFA language, specifically tailored to the features needed to support concurrency.
    106134
    107 \CFA is an extension of ISO-C and therefore supports all of the same paradigms as C. It is a non-object-oriented system-language, meaning most of the major abstractions have either no runtime overhead or can be opted out easily. Like C, the basics of \CFA revolve around structures and routines, which are thin abstractions over machine code. The vast majority of the code produced by the \CFA translator respects memory layouts and calling conventions laid out by C. Interestingly, while \CFA is not an object-oriented language, lacking the concept of a receiver (e.g., {\tt this}), it does have some notion of objects\footnote{C defines the term objects as : ``region of data storage in the execution environment, the contents of which can represent
    108 values''~\cite[3.15]{C11}}, most importantly construction and destruction of objects. Most of the following code examples can be found on the \CFA website~\cite{www-cfa}.
     135\CFA is an extension of ISO-C and therefore supports all of the same paradigms as C.
     136It is a non-object-oriented system-language, meaning most of the major abstractions have either no runtime overhead or can be opted out easily.
     137Like C, the basics of \CFA revolve around structures and routines, which are thin abstractions over machine code.
     138The vast majority of the code produced by the \CFA translator respects memory layouts and calling conventions laid out by C.
     139Interestingly, while \CFA is not an object-oriented language, lacking the concept of a receiver (e.g., {\tt this}), it does have some notion of objects\footnote{C defines the term objects as : ``region of data storage in the execution environment, the contents of which can represent
     140values''~\cite[3.15]{C11}}, most importantly construction and destruction of objects.
     141Most of the following code examples can be found on the \CFA website~\cite{www-cfa}.
    109142
    110143% ======================================================================
    111144\subsection{References}
    112145
    113 Like \CC, \CFA introduces rebind-able references providing multiple dereferencing as an alternative to pointers. In regards to concurrency, the semantic difference between pointers and references are not particularly relevant, but since this document uses mostly references, here is a quick overview of the semantics:
     146Like \CC, \CFA introduces rebind-able references providing multiple dereferencing as an alternative to pointers.
     147In regards to concurrency, the semantic difference between pointers and references are not particularly relevant, but since this document uses mostly references, here is a quick overview of the semantics:
    114148\begin{cfacode}
    115149int x, *p1 = &x, **p2 = &p1, ***p3 = &p2,
     
    130164\subsection{Overloading}
    131165
    132 Another important feature of \CFA is function overloading as in Java and \CC, where routines with the same name are selected based on the number and type of the arguments. As well, \CFA uses the return type as part of the selection criteria, as in Ada~\cite{Ada}. For routines with multiple parameters and returns, the selection is complex.
     166Another important feature of \CFA is function overloading as in Java and \CC, where routines with the same name are selected based on the number and type of the arguments.
     167As well, \CFA uses the return type as part of the selection criteria, as in Ada~\cite{Ada}.
     168For routines with multiple parameters and returns, the selection is complex.
    133169\begin{cfacode}
    134170//selection based on type and number of parameters
     
    146182double d = f(4);                //select (2)
    147183\end{cfacode}
    148 This feature is particularly important for concurrency since the runtime system relies on creating different types to represent concurrency objects. Therefore, overloading is necessary to prevent the need for long prefixes and other naming conventions that prevent name clashes. As seen in section \ref{basics}, routine \code{main} is an example that benefits from overloading.
     184This feature is particularly important for concurrency since the runtime system relies on creating different types to represent concurrency objects.
     185Therefore, overloading is necessary to prevent the need for long prefixes and other naming conventions that prevent name clashes.
     186As seen in section \ref{basics}, routine \code{main} is an example that benefits from overloading.
    149187
    150188% ======================================================================
    151189\subsection{Operators}
    152 Overloading also extends to operators. The syntax for denoting operator-overloading is to name a routine with the symbol of the operator and question marks where the arguments of the operation appear, e.g.:
     190Overloading also extends to operators.
     191The syntax for denoting operator-overloading is to name a routine with the symbol of the operator and question marks where the arguments of the operation appear, e.g.:
    153192\begin{cfacode}
    154193int ++? (int op);                       //unary prefix increment
     
    170209% ======================================================================
    171210\subsection{Constructors/Destructors}
    172 Object lifetime is often a challenge in concurrency. \CFA uses the approach of giving concurrent meaning to object lifetime as a means of synchronization and/or mutual exclusion. Since \CFA relies heavily on the lifetime of objects, constructors and destructors is a core feature required for concurrency and parallelism. \CFA uses the following syntax for constructors and destructors:
     211Object lifetime is often a challenge in concurrency. \CFA uses the approach of giving concurrent meaning to object lifetime as a means of synchronization and/or mutual exclusion.
     212Since \CFA relies heavily on the lifetime of objects, constructors and destructors is a core feature required for concurrency and parallelism. \CFA uses the following syntax for constructors and destructors:
    173213\begin{cfacode}
    174214struct S {
     
    191231}                                                               //implicit calls: ^?{}(y), ^?{}(x)
    192232\end{cfacode}
    193 The language guarantees that every object and all their fields are constructed. Like \CC, construction of an object is automatically done on allocation and destruction of the object is done on deallocation. Allocation and deallocation can occur on the stack or on the heap.
     233The language guarantees that every object and all their fields are constructed.
     234Like \CC, construction of an object is automatically done on allocation and destruction of the object is done on deallocation.
     235Allocation and deallocation can occur on the stack or on the heap.
    194236\begin{cfacode}
    195237{
     
    206248\subsection{Parametric Polymorphism}
    207249\label{s:ParametricPolymorphism}
    208 Routines in \CFA can also be reused for multiple types. This capability is done using the \code{forall} clauses, which allow separately compiled routines to support generic usage over multiple types. For example, the following sum function works for any type that supports construction from 0 and addition:
     250Routines in \CFA can also be reused for multiple types.
     251This capability is done using the \code{forall} clauses, which allow separately compiled routines to support generic usage over multiple types.
     252For example, the following sum function works for any type that supports construction from 0 and addition:
    209253\begin{cfacode}
    210254//constraint type, 0 and +
     
    221265\end{cfacode}
    222266
    223 Since writing constraints on types can become cumbersome for more constrained functions, \CFA also has the concept of traits. Traits are named collection of constraints that can be used both instead and in addition to regular constraints:
     267Since writing constraints on types can become cumbersome for more constrained functions, \CFA also has the concept of traits.
     268Traits are named collection of constraints that can be used both instead and in addition to regular constraints:
    224269\begin{cfacode}
    225270trait summable( otype T ) {
     
    234279\end{cfacode}
    235280
    236 Note that the type use for assertions can be either an \code{otype} or a \code{dtype}. Types declared as \code{otype} refer to ``complete'' objects, i.e., objects with a size, a default constructor, a copy constructor, a destructor and an assignment operator. Using \code{dtype,} on the other hand, has none of these assumptions but is extremely restrictive, it only guarantees the object is addressable.
     281Note that the type use for assertions can be either an \code{otype} or a \code{dtype}.
     282Types declared as \code{otype} refer to ``complete'' objects, i.e., objects with a size, a default constructor, a copy constructor, a destructor and an assignment operator.
     283Using \code{dtype,} on the other hand, has none of these assumptions but is extremely restrictive, it only guarantees the object is addressable.
    237284
    238285% ======================================================================
    239286\subsection{with Clause/Statement}
    240 Since \CFA lacks the concept of a receiver, certain functions end up needing to repeat variable names often. To remove this inconvenience, \CFA provides the \code{with} statement, which opens an aggregate scope making its fields directly accessible (like Pascal).
     287Since \CFA lacks the concept of a receiver, certain functions end up needing to repeat variable names often.
     288To remove this inconvenience, \CFA provides the \code{with} statement, which opens an aggregate scope making its fields directly accessible (like Pascal).
    241289\begin{cfacode}
    242290struct S { int i, j; };
     
    273321
    274322\section{Basics of concurrency}
    275 At its core, concurrency is based on having multiple call-stacks and scheduling among threads of execution executing on these stacks. Concurrency without parallelism only requires having multiple call stacks (or contexts) for a single thread of execution.
    276 
    277 Execution with a single thread and multiple stacks where the thread is self-scheduling deterministically across the stacks is called coroutining. Execution with a single and multiple stacks but where the thread is scheduled by an oracle (non-deterministic from the thread's perspective) across the stacks is called concurrency.
    278 
    279 Therefore, a minimal concurrency system can be achieved by creating coroutines (see Section \ref{coroutine}), which instead of context-switching among each other, always ask an oracle where to context-switch next. While coroutines can execute on the caller's stack-frame, stack-full coroutines allow full generality and are sufficient as the basis for concurrency. The aforementioned oracle is a scheduler and the whole system now follows a cooperative threading-model (a.k.a., non-preemptive scheduling). The oracle/scheduler can either be a stack-less or stack-full entity and correspondingly require one or two context-switches to run a different coroutine. In any case, a subset of concurrency related challenges start to appear. For the complete set of concurrency challenges to occur, the only feature missing is preemption.
    280 
    281 A scheduler introduces order of execution uncertainty, while preemption introduces uncertainty about where context switches occur. Mutual exclusion and synchronization are ways of limiting non-determinism in a concurrent system. Now it is important to understand that uncertainty is desirable; uncertainty can be used by runtime systems to significantly increase performance and is often the basis of giving a user the illusion that tasks are running in parallel. Optimal performance in concurrent applications is often obtained by having as much non-determinism as correctness allows.
     323At its core, concurrency is based on having multiple call-stacks and scheduling among threads of execution executing on these stacks.
     324Concurrency without parallelism only requires having multiple call stacks (or contexts) for a single thread of execution.
     325
     326Execution with a single thread and multiple stacks where the thread is self-scheduling deterministically across the stacks is called coroutining.
     327Execution with a single and multiple stacks but where the thread is scheduled by an oracle (non-deterministic from the thread's perspective) across the stacks is called concurrency.
     328
     329Therefore, a minimal concurrency system can be achieved by creating coroutines (see Section \ref{coroutine}), which instead of context-switching among each other, always ask an oracle where to context-switch next.
     330While coroutines can execute on the caller's stack-frame, stack-full coroutines allow full generality and are sufficient as the basis for concurrency.
     331The aforementioned oracle is a scheduler and the whole system now follows a cooperative threading-model (a.k.a., non-preemptive scheduling).
     332The oracle/scheduler can either be a stack-less or stack-full entity and correspondingly require one or two context-switches to run a different coroutine.
     333In any case, a subset of concurrency related challenges start to appear.
     334For the complete set of concurrency challenges to occur, the only feature missing is preemption.
     335
     336A scheduler introduces order of execution uncertainty, while preemption introduces uncertainty about where context switches occur.
     337Mutual exclusion and synchronization are ways of limiting non-determinism in a concurrent system.
     338Now it is important to understand that uncertainty is desirable; uncertainty can be used by runtime systems to significantly increase performance and is often the basis of giving a user the illusion that tasks are running in parallel.
     339Optimal performance in concurrent applications is often obtained by having as much non-determinism as correctness allows.
    282340
    283341\section{\protect\CFA's Thread Building Blocks}
    284 One of the important features that are missing in C is threading\footnote{While the C11 standard defines a ``threads.h'' header, it is minimal and defined as optional. As such, library support for threading is far from widespread. At the time of writing the paper, neither \texttt{gcc} nor \texttt{clang} support ``threads.h'' in their respective standard libraries.}. On modern architectures, a lack of threading is unacceptable~\cite{Sutter05, Sutter05b}, and therefore modern programming languages must have the proper tools to allow users to write efficient concurrent programs to take advantage of parallelism. As an extension of C, \CFA needs to express these concepts in a way that is as natural as possible to programmers familiar with imperative languages. And being a system-level language means programmers expect to choose precisely which features they need and which cost they are willing to pay.
     342One of the important features that are missing in C is threading\footnote{While the C11 standard defines a ``threads.h'' header, it is minimal and defined as optional.
     343As such, library support for threading is far from widespread.
     344At the time of writing the paper, neither \texttt{gcc} nor \texttt{clang} support ``threads.h'' in their respective standard libraries.}.
     345On modern architectures, a lack of threading is unacceptable~\cite{Sutter05, Sutter05b}, and therefore modern programming languages must have the proper tools to allow users to write efficient concurrent programs to take advantage of parallelism.
     346As an extension of C, \CFA needs to express these concepts in a way that is as natural as possible to programmers familiar with imperative languages.
     347And being a system-level language means programmers expect to choose precisely which features they need and which cost they are willing to pay.
    285348
    286349\section{Coroutines: A Stepping Stone}\label{coroutine}
    287 While the main focus of this proposal is concurrency and parallelism, it is important to address coroutines, which are actually a significant building block of a concurrency system. \textbf{Coroutine}s are generalized routines which have predefined points where execution is suspended and can be resumed at a later time. Therefore, they need to deal with context switches and other context-management operations. This proposal includes coroutines both as an intermediate step for the implementation of threads, and a first-class feature of \CFA. Furthermore, many design challenges of threads are at least partially present in designing coroutines, which makes the design effort that much more relevant. The core \textbf{api} of coroutines revolves around two features: independent call-stacks and \code{suspend}/\code{resume}.
     350While the main focus of this proposal is concurrency and parallelism, it is important to address coroutines, which are actually a significant building block of a concurrency system. \textbf{Coroutine}s are generalized routines which have predefined points where execution is suspended and can be resumed at a later time.
     351Therefore, they need to deal with context switches and other context-management operations.
     352This proposal includes coroutines both as an intermediate step for the implementation of threads, and a first-class feature of \CFA.
     353Furthermore, many design challenges of threads are at least partially present in designing coroutines, which makes the design effort that much more relevant.
     354The core \textbf{api} of coroutines revolves around two features: independent call-stacks and \code{suspend}/\code{resume}.
    288355
    289356\begin{table}
     
    399466\end{table}
    400467
    401 A good example of a problem made easier with coroutines is generators, e.g., generating the Fibonacci sequence. This problem comes with the challenge of decoupling how a sequence is generated and how it is used. Listing \ref{lst:fibonacci-c} shows conventional approaches to writing generators in C. All three of these approach suffer from strong coupling. The left and centre approaches require that the generator have knowledge of how the sequence is used, while the rightmost approach requires holding internal state between calls on behalf of the generator and makes it much harder to handle corner cases like the Fibonacci seed.
    402 
    403 Listing \ref{lst:fibonacci-cfa} is an example of a solution to the Fibonacci problem using \CFA coroutines, where the coroutine stack holds sufficient state for the next generation. This solution has the advantage of having very strong decoupling between how the sequence is generated and how it is used. Indeed, this version is as easy to use as the \code{fibonacci_state} solution, while the implementation is very similar to the \code{fibonacci_func} example.
     468A good example of a problem made easier with coroutines is generators, e.g., generating the Fibonacci sequence.
     469This problem comes with the challenge of decoupling how a sequence is generated and how it is used.
     470Listing \ref{lst:fibonacci-c} shows conventional approaches to writing generators in C.
     471All three of these approach suffer from strong coupling.
     472The left and centre approaches require that the generator have knowledge of how the sequence is used, while the rightmost approach requires holding internal state between calls on behalf of the generator and makes it much harder to handle corner cases like the Fibonacci seed.
     473
     474Listing \ref{lst:fibonacci-cfa} is an example of a solution to the Fibonacci problem using \CFA coroutines, where the coroutine stack holds sufficient state for the next generation.
     475This solution has the advantage of having very strong decoupling between how the sequence is generated and how it is used.
     476Indeed, this version is as easy to use as the \code{fibonacci_state} solution, while the implementation is very similar to the \code{fibonacci_func} example.
    404477
    405478\begin{figure}
     
    447520\end{figure}
    448521
    449 Listing \ref{lst:fmt-line} shows the \code{Format} coroutine for restructuring text into groups of character blocks of fixed size. The example takes advantage of resuming coroutines in the constructor to simplify the code and highlights the idea that interesting control flow can occur in the constructor.
     522Listing \ref{lst:fmt-line} shows the \code{Format} coroutine for restructuring text into groups of character blocks of fixed size.
     523The example takes advantage of resuming coroutines in the constructor to simplify the code and highlights the idea that interesting control flow can occur in the constructor.
    450524
    451525\begin{figure}
     
    497571
    498572\subsection{Construction}
    499 One important design challenge for implementing coroutines and threads (shown in section \ref{threads}) is that the runtime system needs to run code after the user-constructor runs to connect the fully constructed object into the system. In the case of coroutines, this challenge is simpler since there is no non-determinism from preemption or scheduling. However, the underlying challenge remains the same for coroutines and threads.
    500 
    501 The runtime system needs to create the coroutine's stack and, more importantly, prepare it for the first resumption. The timing of the creation is non-trivial since users expect both to have fully constructed objects once execution enters the coroutine main and to be able to resume the coroutine from the constructor. There are several solutions to this problem but the chosen option effectively forces the design of the coroutine.
    502 
    503 Furthermore, \CFA faces an extra challenge as polymorphic routines create invisible thunks when cast to non-polymorphic routines and these thunks have function scope. For example, the following code, while looking benign, can run into undefined behaviour because of thunks:
     573One important design challenge for implementing coroutines and threads (shown in section \ref{threads}) is that the runtime system needs to run code after the user-constructor runs to connect the fully constructed object into the system.
     574In the case of coroutines, this challenge is simpler since there is no non-determinism from preemption or scheduling.
     575However, the underlying challenge remains the same for coroutines and threads.
     576
     577The runtime system needs to create the coroutine's stack and, more importantly, prepare it for the first resumption.
     578The timing of the creation is non-trivial since users expect both to have fully constructed objects once execution enters the coroutine main and to be able to resume the coroutine from the constructor.
     579There are several solutions to this problem but the chosen option effectively forces the design of the coroutine.
     580
     581Furthermore, \CFA faces an extra challenge as polymorphic routines create invisible thunks when cast to non-polymorphic routines and these thunks have function scope.
     582For example, the following code, while looking benign, can run into undefined behaviour because of thunks:
    504583
    505584\begin{cfacode}
     
    534613}
    535614\end{ccode}
    536 The problem in this example is a storage management issue, the function pointer \code{_thunk0} is only valid until the end of the block, which limits the viable solutions because storing the function pointer for too long causes undefined behaviour; i.e., the stack-based thunk being destroyed before it can be used. This challenge is an extension of challenges that come with second-class routines. Indeed, GCC nested routines also have the limitation that nested routine cannot be passed outside of the declaration scope. The case of coroutines and threads is simply an extension of this problem to multiple call stacks.
     615The problem in this example is a storage management issue, the function pointer \code{_thunk0} is only valid until the end of the block, which limits the viable solutions because storing the function pointer for too long causes undefined behaviour; i.e., the stack-based thunk being destroyed before it can be used.
     616This challenge is an extension of challenges that come with second-class routines.
     617Indeed, GCC nested routines also have the limitation that nested routine cannot be passed outside of the declaration scope.
     618The case of coroutines and threads is simply an extension of this problem to multiple call stacks.
    537619
    538620\subsection{Alternative: Composition}
     
    555637}
    556638\end{cfacode}
    557 The downside of this approach is that users need to correctly construct the coroutine handle before using it. Like any other objects, the user must carefully choose construction order to prevent usage of objects not yet constructed. However, in the case of coroutines, users must also pass to the coroutine information about the coroutine main, like in the previous example. This opens the door for user errors and requires extra runtime storage to pass at runtime information that can be known statically.
     639The downside of this approach is that users need to correctly construct the coroutine handle before using it.
     640Like any other objects, the user must carefully choose construction order to prevent usage of objects not yet constructed.
     641However, in the case of coroutines, users must also pass to the coroutine information about the coroutine main, like in the previous example.
     642This opens the door for user errors and requires extra runtime storage to pass at runtime information that can be known statically.
    558643
    559644\subsection{Alternative: Reserved keyword}
     
    565650};
    566651\end{cfacode}
    567 The \code{coroutine} keyword means the compiler can find and inject code where needed. The downside of this approach is that it makes coroutine a special case in the language. Users wanting to extend coroutines or build their own for various reasons can only do so in ways offered by the language. Furthermore, implementing coroutines without language supports also displays the power of the programming language used. While this is ultimately the option used for idiomatic \CFA code, coroutines and threads can still be constructed by users without using the language support. The reserved keywords are only present to improve ease of use for the common cases.
     652The \code{coroutine} keyword means the compiler can find and inject code where needed.
     653The downside of this approach is that it makes coroutine a special case in the language.
     654Users wanting to extend coroutines or build their own for various reasons can only do so in ways offered by the language.
     655Furthermore, implementing coroutines without language supports also displays the power of the programming language used.
     656While this is ultimately the option used for idiomatic \CFA code, coroutines and threads can still be constructed by users without using the language support.
     657The reserved keywords are only present to improve ease of use for the common cases.
    568658
    569659\subsection{Alternative: Lambda Objects}
    570660
    571 For coroutines as for threads, many implementations are based on routine pointers or function objects~\cite{Butenhof97, C++14, MS:VisualC++, BoostCoroutines15}. For example, Boost implements coroutines in terms of four functor object types:
     661For coroutines as for threads, many implementations are based on routine pointers or function objects~\cite{Butenhof97, C++14, MS:VisualC++, BoostCoroutines15}.
     662For example, Boost implements coroutines in terms of four functor object types:
    572663\begin{cfacode}
    573664asymmetric_coroutine<>::pull_type
     
    576667symmetric_coroutine<>::yield_type
    577668\end{cfacode}
    578 Often, the canonical threading paradigm in languages is based on function pointers, \texttt{pthread} being one of the most well-known examples. The main problem of this approach is that the thread usage is limited to a generic handle that must otherwise be wrapped in a custom type. Since the custom type is simple to write in \CFA and solves several issues, added support for routine/lambda based coroutines adds very little.
     669Often, the canonical threading paradigm in languages is based on function pointers, \texttt{pthread} being one of the most well-known examples.
     670The main problem of this approach is that the thread usage is limited to a generic handle that must otherwise be wrapped in a custom type.
     671Since the custom type is simple to write in \CFA and solves several issues, added support for routine/lambda based coroutines adds very little.
    579672
    580673A variation of this would be to use a simple function pointer in the same way \texttt{pthread} does for threads:
     
    591684}
    592685\end{cfacode}
    593 This semantics is more common for thread interfaces but coroutines work equally well. As discussed in section \ref{threads}, this approach is superseded by static approaches in terms of expressivity.
     686This semantics is more common for thread interfaces but coroutines work equally well.
     687As discussed in section \ref{threads}, this approach is superseded by static approaches in terms of expressivity.
    594688
    595689\subsection{Alternative: Trait-Based Coroutines}
    596690
    597 Finally, the underlying approach, which is the one closest to \CFA idioms, is to use trait-based lazy coroutines. This approach defines a coroutine as anything that satisfies the trait \code{is_coroutine} (as defined below) and is used as a coroutine.
     691Finally, the underlying approach, which is the one closest to \CFA idioms, is to use trait-based lazy coroutines.
     692This approach defines a coroutine as anything that satisfies the trait \code{is_coroutine} (as defined below) and is used as a coroutine.
    598693
    599694\begin{cfacode}
     
    606701forall( dtype T | is_coroutine(T) ) void resume (T&);
    607702\end{cfacode}
    608 This ensures that an object is not a coroutine until \code{resume} is called on the object. Correspondingly, any object that is passed to \code{resume} is a coroutine since it must satisfy the \code{is_coroutine} trait to compile. The advantage of this approach is that users can easily create different types of coroutines, for example, changing the memory layout of a coroutine is trivial when implementing the \code{get_coroutine} routine. The \CFA keyword \code{coroutine} simply has the effect of implementing the getter and forward declarations required for users to implement the main routine.
     703This ensures that an object is not a coroutine until \code{resume} is called on the object.
     704Correspondingly, any object that is passed to \code{resume} is a coroutine since it must satisfy the \code{is_coroutine} trait to compile.
     705The advantage of this approach is that users can easily create different types of coroutines, for example, changing the memory layout of a coroutine is trivial when implementing the \code{get_coroutine} routine.
     706The \CFA keyword \code{coroutine} simply has the effect of implementing the getter and forward declarations required for users to implement the main routine.
    609707
    610708\begin{center}
     
    635733
    636734\section{Thread Interface}\label{threads}
    637 The basic building blocks of multithreading in \CFA are \textbf{cfathread}. Both user and kernel threads are supported, where user threads are the concurrency mechanism and kernel threads are the parallel mechanism. User threads offer a flexible and lightweight interface. A thread can be declared using a struct declaration \code{thread} as follows:
     735The basic building blocks of multithreading in \CFA are \textbf{cfathread}.
     736Both user and kernel threads are supported, where user threads are the concurrency mechanism and kernel threads are the parallel mechanism.
     737User threads offer a flexible and lightweight interface.
     738A thread can be declared using a struct declaration \code{thread} as follows:
    638739
    639740\begin{cfacode}
     
    651752\end{cfacode}
    652753
    653 Obviously, for this thread implementation to be useful it must run some user code. Several other threading interfaces use a function-pointer representation as the interface of threads (for example \Csharp~\cite{Csharp} and Scala~\cite{Scala}). However, this proposal considers that statically tying a \code{main} routine to a thread supersedes this approach. Since the \code{main} routine is already a special routine in \CFA (where the program begins), it is a natural extension of the semantics to use overloading to declare mains for different threads (the normal main being the main of the initial thread). As such the \code{main} routine of a thread can be defined as
     754Obviously, for this thread implementation to be useful it must run some user code.
     755Several other threading interfaces use a function-pointer representation as the interface of threads (for example \Csharp~\cite{Csharp} and Scala~\cite{Scala}).
     756However, this proposal considers that statically tying a \code{main} routine to a thread supersedes this approach.
     757Since the \code{main} routine is already a special routine in \CFA (where the program begins), it is a natural extension of the semantics to use overloading to declare mains for different threads (the normal main being the main of the initial thread).
     758As such the \code{main} routine of a thread can be defined as
    654759\begin{cfacode}
    655760thread foo {};
     
    660765\end{cfacode}
    661766
    662 In this example, threads of type \code{foo} start execution in the \code{void main(foo &)} routine, which prints \code{"Hello World!".} While this paper encourages this approach to enforce strongly typed programming, users may prefer to use the routine-based thread semantics for the sake of simplicity. With the static semantics it is trivial to write a thread type that takes a function pointer as a parameter and executes it on its stack asynchronously.
     767In this example, threads of type \code{foo} start execution in the \code{void main(foo &)} routine, which prints \code{"Hello World!".} While this paper encourages this approach to enforce strongly typed programming, users may prefer to use the routine-based thread semantics for the sake of simplicity.
     768With the static semantics it is trivial to write a thread type that takes a function pointer as a parameter and executes it on its stack asynchronously.
    663769\begin{cfacode}
    664770typedef void (*voidFunc)(int);
     
    691797A consequence of the strongly typed approach to main is that memory layout of parameters and return values to/from a thread are now explicitly specified in the \textbf{api}.
    692798
    693 Of course, for threads to be useful, it must be possible to start and stop threads and wait for them to complete execution. While using an \textbf{api} such as \code{fork} and \code{join} is relatively common in the literature, such an interface is unnecessary. Indeed, the simplest approach is to use \textbf{raii} principles and have threads \code{fork} after the constructor has completed and \code{join} before the destructor runs.
     799Of course, for threads to be useful, it must be possible to start and stop threads and wait for them to complete execution.
     800While using an \textbf{api} such as \code{fork} and \code{join} is relatively common in the literature, such an interface is unnecessary.
     801Indeed, the simplest approach is to use \textbf{raii} principles and have threads \code{fork} after the constructor has completed and \code{join} before the destructor runs.
    694802\begin{cfacode}
    695803thread World;
     
    732840\end{cfacode}
    733841
    734 However, one of the drawbacks of this approach is that threads always form a tree where nodes must always outlive their children, i.e., they are always destroyed in the opposite order of construction because of C scoping rules. This restriction is relaxed by using dynamic allocation, so threads can outlive the scope in which they are created, much like dynamically allocating memory lets objects outlive the scope in which they are created.
     842However, one of the drawbacks of this approach is that threads always form a tree where nodes must always outlive their children, i.e., they are always destroyed in the opposite order of construction because of C scoping rules.
     843This restriction is relaxed by using dynamic allocation, so threads can outlive the scope in which they are created, much like dynamically allocating memory lets objects outlive the scope in which they are created.
    735844
    736845\begin{cfacode}
     
    769878% ======================================================================
    770879% ======================================================================
    771 Several tools can be used to solve concurrency challenges. Since many of these challenges appear with the use of mutable shared state, some languages and libraries simply disallow mutable shared state (Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, Akka (Scala)~\cite{Akka}). In these paradigms, interaction among concurrent objects relies on message passing~\cite{Thoth,Harmony,V-Kernel} or other paradigms closely relate to networking concepts (channels~\cite{CSP,Go} for example). However, in languages that use routine calls as their core abstraction mechanism, these approaches force a clear distinction between concurrent and non-concurrent paradigms (i.e., message passing versus routine calls). This distinction in turn means that, in order to be effective, programmers need to learn two sets of design patterns. While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account.
    772 
    773 Approaches based on shared memory are more closely related to non-concurrent paradigms since they often rely on basic constructs like routine calls and shared objects. At the lowest level, concurrent paradigms are implemented as atomic operations and locks. Many such mechanisms have been proposed, including semaphores~\cite{Dijkstra68b} and path expressions~\cite{Campbell74}. However, for productivity reasons it is desirable to have a higher-level construct be the core concurrency paradigm~\cite{HPP:Study}.
    774 
    775 An approach that is worth mentioning because it is gaining in popularity is transactional memory~\cite{Herlihy93}. While this approach is even pursued by system languages like \CC~\cite{Cpp-Transactions}, the performance and feature set is currently too restrictive to be the main concurrency paradigm for system languages, which is why it was rejected as the core paradigm for concurrency in \CFA.
    776 
    777 One of the most natural, elegant, and efficient mechanisms for synchronization and communication, especially for shared-memory systems, is the \emph{monitor}. Monitors were first proposed by Brinch Hansen~\cite{Hansen73} and later described and extended by C.A.R.~Hoare~\cite{Hoare74}. Many programming languages---e.g., Concurrent Pascal~\cite{ConcurrentPascal}, Mesa~\cite{Mesa}, Modula~\cite{Modula-2}, Turing~\cite{Turing:old}, Modula-3~\cite{Modula-3}, NeWS~\cite{NeWS}, Emerald~\cite{Emerald}, \uC~\cite{Buhr92a} and Java~\cite{Java}---provide monitors as explicit language constructs. In addition, operating-system kernels and device drivers have a monitor-like structure, although they often use lower-level primitives such as semaphores or locks to simulate monitors. For these reasons, this project proposes monitors as the core concurrency construct.
     880Several tools can be used to solve concurrency challenges.
     881Since many of these challenges appear with the use of mutable shared state, some languages and libraries simply disallow mutable shared state (Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, Akka (Scala)~\cite{Akka}).
     882In these paradigms, interaction among concurrent objects relies on message passing~\cite{Thoth,Harmony,V-Kernel} or other paradigms closely relate to networking concepts (channels~\cite{CSP,Go} for example).
     883However, in languages that use routine calls as their core abstraction mechanism, these approaches force a clear distinction between concurrent and non-concurrent paradigms (i.e., message passing versus routine calls).
     884This distinction in turn means that, in order to be effective, programmers need to learn two sets of design patterns.
     885While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account.
     886
     887Approaches based on shared memory are more closely related to non-concurrent paradigms since they often rely on basic constructs like routine calls and shared objects.
     888At the lowest level, concurrent paradigms are implemented as atomic operations and locks.
     889Many such mechanisms have been proposed, including semaphores~\cite{Dijkstra68b} and path expressions~\cite{Campbell74}.
     890However, for productivity reasons it is desirable to have a higher-level construct be the core concurrency paradigm~\cite{HPP:Study}.
     891
     892An approach that is worth mentioning because it is gaining in popularity is transactional memory~\cite{Herlihy93}.
     893While this approach is even pursued by system languages like \CC~\cite{Cpp-Transactions}, the performance and feature set is currently too restrictive to be the main concurrency paradigm for system languages, which is why it was rejected as the core paradigm for concurrency in \CFA.
     894
     895One of the most natural, elegant, and efficient mechanisms for synchronization and communication, especially for shared-memory systems, is the \emph{monitor}.
     896Monitors were first proposed by Brinch Hansen~\cite{Hansen73} and later described and extended by C.A.R.~Hoare~\cite{Hoare74}.
     897Many programming languages---e.g., Concurrent Pascal~\cite{ConcurrentPascal}, Mesa~\cite{Mesa}, Modula~\cite{Modula-2}, Turing~\cite{Turing:old}, Modula-3~\cite{Modula-3}, NeWS~\cite{NeWS}, Emerald~\cite{Emerald}, \uC~\cite{Buhr92a} and Java~\cite{Java}---provide monitors as explicit language constructs.
     898In addition, operating-system kernels and device drivers have a monitor-like structure, although they often use lower-level primitives such as semaphores or locks to simulate monitors.
     899For these reasons, this project proposes monitors as the core concurrency construct.
    778900
    779901\section{Basics}
    780 Non-determinism requires concurrent systems to offer support for mutual-exclusion and synchronization. Mutual-exclusion is the concept that only a fixed number of threads can access a critical section at any given time, where a critical section is a group of instructions on an associated portion of data that requires the restricted access. On the other hand, synchronization enforces relative ordering of execution and synchronization tools provide numerous mechanisms to establish timing relationships among threads.
     902Non-determinism requires concurrent systems to offer support for mutual-exclusion and synchronization.
     903Mutual-exclusion is the concept that only a fixed number of threads can access a critical section at any given time, where a critical section is a group of instructions on an associated portion of data that requires the restricted access.
     904On the other hand, synchronization enforces relative ordering of execution and synchronization tools provide numerous mechanisms to establish timing relationships among threads.
    781905
    782906\subsection{Mutual-Exclusion}
    783 As mentioned above, mutual-exclusion is the guarantee that only a fix number of threads can enter a critical section at once. However, many solutions exist for mutual exclusion, which vary in terms of performance, flexibility and ease of use. Methods range from low-level locks, which are fast and flexible but require significant attention to be correct, to  higher-level concurrency techniques, which sacrifice some performance in order to improve ease of use. Ease of use comes by either guaranteeing some problems cannot occur (e.g., being deadlock free) or by offering a more explicit coupling between data and corresponding critical section. For example, the \CC \code{std::atomic<T>} offers an easy way to express mutual-exclusion on a restricted set of operations (e.g., reading/writing large types atomically). Another challenge with low-level locks is composability. Locks have restricted composability because it takes careful organizing for multiple locks to be used while preventing deadlocks. Easing composability is another feature higher-level mutual-exclusion mechanisms often offer.
     907As mentioned above, mutual-exclusion is the guarantee that only a fix number of threads can enter a critical section at once.
     908However, many solutions exist for mutual exclusion, which vary in terms of performance, flexibility and ease of use.
     909Methods range from low-level locks, which are fast and flexible but require significant attention to be correct, to  higher-level concurrency techniques, which sacrifice some performance in order to improve ease of use.
     910Ease of use comes by either guaranteeing some problems cannot occur (e.g., being deadlock free) or by offering a more explicit coupling between data and corresponding critical section.
     911For example, the \CC \code{std::atomic<T>} offers an easy way to express mutual-exclusion on a restricted set of operations (e.g., reading/writing large types atomically).
     912Another challenge with low-level locks is composability.
     913Locks have restricted composability because it takes careful organizing for multiple locks to be used while preventing deadlocks.
     914Easing composability is another feature higher-level mutual-exclusion mechanisms often offer.
    784915
    785916\subsection{Synchronization}
    786 As with mutual-exclusion, low-level synchronization primitives often offer good performance and good flexibility at the cost of ease of use. Again, higher-level mechanisms often simplify usage by adding either better coupling between synchronization and data (e.g., message passing) or offering a simpler solution to otherwise involved challenges. As mentioned above, synchronization can be expressed as guaranteeing that event \textit{X} always happens before \textit{Y}. Most of the time, synchronization happens within a critical section, where threads must acquire mutual-exclusion in a certain order. However, it may also be desirable to guarantee that event \textit{Z} does not occur between \textit{X} and \textit{Y}. Not satisfying this property is called \textbf{barging}. For example, where event \textit{X} tries to effect event \textit{Y} but another thread acquires the critical section and emits \textit{Z} before \textit{Y}. The classic example is the thread that finishes using a resource and unblocks a thread waiting to use the resource, but the unblocked thread must compete to acquire the resource. Preventing or detecting barging is an involved challenge with low-level locks, which can be made much easier by higher-level constructs. This challenge is often split into two different methods, barging avoidance and barging prevention. Algorithms that use flag variables to detect barging threads are said to be using barging avoidance, while algorithms that baton-pass locks~\cite{Andrews89} between threads instead of releasing the locks are said to be using barging prevention.
     917As with mutual-exclusion, low-level synchronization primitives often offer good performance and good flexibility at the cost of ease of use.
     918Again, higher-level mechanisms often simplify usage by adding either better coupling between synchronization and data (e.g., message passing) or offering a simpler solution to otherwise involved challenges.
     919As mentioned above, synchronization can be expressed as guaranteeing that event \textit{X} always happens before \textit{Y}.
     920Most of the time, synchronization happens within a critical section, where threads must acquire mutual-exclusion in a certain order.
     921However, it may also be desirable to guarantee that event \textit{Z} does not occur between \textit{X} and \textit{Y}.
     922Not satisfying this property is called \textbf{barging}.
     923For example, where event \textit{X} tries to effect event \textit{Y} but another thread acquires the critical section and emits \textit{Z} before \textit{Y}.
     924The classic example is the thread that finishes using a resource and unblocks a thread waiting to use the resource, but the unblocked thread must compete to acquire the resource.
     925Preventing or detecting barging is an involved challenge with low-level locks, which can be made much easier by higher-level constructs.
     926This challenge is often split into two different methods, barging avoidance and barging prevention.
     927Algorithms that use flag variables to detect barging threads are said to be using barging avoidance, while algorithms that baton-pass locks~\cite{Andrews89} between threads instead of releasing the locks are said to be using barging prevention.
    787928
    788929% ======================================================================
     
    791932% ======================================================================
    792933% ======================================================================
    793 A \textbf{monitor} is a set of routines that ensure mutual-exclusion when accessing shared state. More precisely, a monitor is a programming technique that associates mutual-exclusion to routine scopes, as opposed to mutex locks, where mutual-exclusion is defined by lock/release calls independently of any scoping of the calling routine. This strong association eases readability and maintainability, at the cost of flexibility. Note that both monitors and mutex locks, require an abstract handle to identify them. This concept is generally associated with object-oriented languages like Java~\cite{Java} or \uC~\cite{uC++book} but does not strictly require OO semantics. The only requirement is the ability to declare a handle to a shared object and a set of routines that act on it:
     934A \textbf{monitor} is a set of routines that ensure mutual-exclusion when accessing shared state.
     935More precisely, a monitor is a programming technique that associates mutual-exclusion to routine scopes, as opposed to mutex locks, where mutual-exclusion is defined by lock/release calls independently of any scoping of the calling routine.
     936This strong association eases readability and maintainability, at the cost of flexibility.
     937Note that both monitors and mutex locks, require an abstract handle to identify them.
     938This concept is generally associated with object-oriented languages like Java~\cite{Java} or \uC~\cite{uC++book} but does not strictly require OO semantics.
     939The only requirement is the ability to declare a handle to a shared object and a set of routines that act on it:
    794940\begin{cfacode}
    795941typedef /*some monitor type*/ monitor;
     
    807953% ======================================================================
    808954% ======================================================================
    809 The above monitor example displays some of the intrinsic characteristics. First, it is necessary to use pass-by-reference over pass-by-value for monitor routines. This semantics is important, because at their core, monitors are implicit mutual-exclusion objects (locks), and these objects cannot be copied. Therefore, monitors are non-copy-able objects (\code{dtype}).
    810 
    811 Another aspect to consider is when a monitor acquires its mutual exclusion. For example, a monitor may need to be passed through multiple helper routines that do not acquire the monitor mutual-exclusion on entry. Passthrough can occur for generic helper routines (\code{swap}, \code{sort}, etc.) or specific helper routines like the following to implement an atomic counter:
     955The above monitor example displays some of the intrinsic characteristics.
     956First, it is necessary to use pass-by-reference over pass-by-value for monitor routines.
     957This semantics is important, because at their core, monitors are implicit mutual-exclusion objects (locks), and these objects cannot be copied.
     958Therefore, monitors are non-copy-able objects (\code{dtype}).
     959
     960Another aspect to consider is when a monitor acquires its mutual exclusion.
     961For example, a monitor may need to be passed through multiple helper routines that do not acquire the monitor mutual-exclusion on entry.
     962Passthrough can occur for generic helper routines (\code{swap}, \code{sort}, etc.) or specific helper routines like the following to implement an atomic counter:
    812963
    813964\begin{cfacode}
     
    838989Notice how the counter is used without any explicit synchronization and yet supports thread-safe semantics for both reading and writing, which is similar in usage to the \CC template \code{std::atomic}.
    839990
    840 Here, the constructor (\code{?\{\}}) uses the \code{nomutex} keyword to signify that it does not acquire the monitor mutual-exclusion when constructing. This semantics is because an object not yet con\-structed should never be shared and therefore does not require mutual exclusion. Furthermore, it allows the implementation greater freedom when it initializes the monitor locking. The prefix increment operator uses \code{mutex} to protect the incrementing process from race conditions. Finally, there is a conversion operator from \code{counter_t} to \code{size_t}. This conversion may or may not require the \code{mutex} keyword depending on whether or not reading a \code{size_t} is an atomic operation.
    841 
    842 For maximum usability, monitors use \textbf{multi-acq} semantics, which means a single thread can acquire the same monitor multiple times without deadlock. For example, listing \ref{fig:search} uses recursion and \textbf{multi-acq} to print values inside a binary tree.
     991Here, the constructor (\code{?\{\}}) uses the \code{nomutex} keyword to signify that it does not acquire the monitor mutual-exclusion when constructing.
     992This semantics is because an object not yet con\-structed should never be shared and therefore does not require mutual exclusion.
     993Furthermore, it allows the implementation greater freedom when it initializes the monitor locking.
     994The prefix increment operator uses \code{mutex} to protect the incrementing process from race conditions.
     995Finally, there is a conversion operator from \code{counter_t} to \code{size_t}.
     996This conversion may or may not require the \code{mutex} keyword depending on whether or not reading a \code{size_t} is an atomic operation.
     997
     998For maximum usability, monitors use \textbf{multi-acq} semantics, which means a single thread can acquire the same monitor multiple times without deadlock.
     999For example, listing \ref{fig:search} uses recursion and \textbf{multi-acq} to print values inside a binary tree.
    8431000\begin{figure}
    8441001\begin{cfacode}[caption={Recursive printing algorithm using \textbf{multi-acq}.},label={fig:search}]
     
    8581015\end{figure}
    8591016
    860 Having both \code{mutex} and \code{nomutex} keywords can be redundant, depending on the meaning of a routine having neither of these keywords. For example, it is reasonable that it should default to the safest option (\code{mutex}) when given a routine without qualifiers \code{void foo(counter_t & this)}, whereas assuming \code{nomutex} is unsafe and may cause subtle errors. On the other hand, \code{nomutex} is the ``normal'' parameter behaviour, it effectively states explicitly that ``this routine is not special''. Another alternative is making exactly one of these keywords mandatory, which provides the same semantics but without the ambiguity of supporting routines with neither keyword. Mandatory keywords would also have the added benefit of being self-documented but at the cost of extra typing. While there are several benefits to mandatory keywords, they do bring a few challenges. Mandatory keywords in \CFA would imply that the compiler must know without doubt whether or not a parameter is a monitor or not. Since \CFA relies heavily on traits as an abstraction mechanism, the distinction between a type that is a monitor and a type that looks like a monitor can become blurred. For this reason, \CFA only has the \code{mutex} keyword and uses no keyword to mean \code{nomutex}.
    861 
    862 The next semantic decision is to establish when \code{mutex} may be used as a type qualifier. Consider the following declarations:
     1017Having both \code{mutex} and \code{nomutex} keywords can be redundant, depending on the meaning of a routine having neither of these keywords.
     1018For example, it is reasonable that it should default to the safest option (\code{mutex}) when given a routine without qualifiers \code{void foo(counter_t & this)}, whereas assuming \code{nomutex} is unsafe and may cause subtle errors.
     1019On the other hand, \code{nomutex} is the ``normal'' parameter behaviour, it effectively states explicitly that ``this routine is not special''.
     1020Another alternative is making exactly one of these keywords mandatory, which provides the same semantics but without the ambiguity of supporting routines with neither keyword.
     1021Mandatory keywords would also have the added benefit of being self-documented but at the cost of extra typing.
     1022While there are several benefits to mandatory keywords, they do bring a few challenges.
     1023Mandatory keywords in \CFA would imply that the compiler must know without doubt whether or not a parameter is a monitor or not.
     1024Since \CFA relies heavily on traits as an abstraction mechanism, the distinction between a type that is a monitor and a type that looks like a monitor can become blurred.
     1025For this reason, \CFA only has the \code{mutex} keyword and uses no keyword to mean \code{nomutex}.
     1026
     1027The next semantic decision is to establish when \code{mutex} may be used as a type qualifier.
     1028Consider the following declarations:
    8631029\begin{cfacode}
    8641030int f1(monitor & mutex m);
     
    8681034int f5(graph(monitor *) & mutex m);
    8691035\end{cfacode}
    870 The problem is to identify which object(s) should be acquired. Furthermore, each object needs to be acquired only once. In the case of simple routines like \code{f1} and \code{f2} it is easy to identify an exhaustive list of objects to acquire on entry. Adding indirections (\code{f3}) still allows the compiler and programmer to identify which object is acquired. However, adding in arrays (\code{f4}) makes it much harder. Array lengths are not necessarily known in C, and even then, making sure objects are only acquired once becomes none-trivial. This problem can be extended to absurd limits like \code{f5}, which uses a graph of monitors. To make the issue tractable, this project imposes the requirement that a routine may only acquire one monitor per parameter and it must be the type of the parameter with at most one level of indirection (ignoring potential qualifiers). Also note that while routine \code{f3} can be supported, meaning that monitor \code{**m} is acquired, passing an array to this routine would be type-safe and yet result in undefined behaviour because only the first element of the array is acquired. However, this ambiguity is part of the C type-system with respects to arrays. For this reason, \code{mutex} is disallowed in the context where arrays may be passed:
     1036The problem is to identify which object(s) should be acquired.
     1037Furthermore, each object needs to be acquired only once.
     1038In the case of simple routines like \code{f1} and \code{f2} it is easy to identify an exhaustive list of objects to acquire on entry.
     1039Adding indirections (\code{f3}) still allows the compiler and programmer to identify which object is acquired.
     1040However, adding in arrays (\code{f4}) makes it much harder.
     1041Array lengths are not necessarily known in C, and even then, making sure objects are only acquired once becomes none-trivial.
     1042This problem can be extended to absurd limits like \code{f5}, which uses a graph of monitors.
     1043To make the issue tractable, this project imposes the requirement that a routine may only acquire one monitor per parameter and it must be the type of the parameter with at most one level of indirection (ignoring potential qualifiers).
     1044Also note that while routine \code{f3} can be supported, meaning that monitor \code{**m} is acquired, passing an array to this routine would be type-safe and yet result in undefined behaviour because only the first element of the array is acquired.
     1045However, this ambiguity is part of the C type-system with respects to arrays.
     1046For this reason, \code{mutex} is disallowed in the context where arrays may be passed:
    8711047\begin{cfacode}
    8721048int f1(monitor & mutex m);    //Okay : recommended case
     
    8761052int f5(monitor * mutex m []); //Not Okay : Array of unknown length
    8771053\end{cfacode}
    878 Note that not all array functions are actually distinct in the type system. However, even if the code generation could tell the difference, the extra information is still not sufficient to extend meaningfully the monitor call semantic.
    879 
    880 Unlike object-oriented monitors, where calling a mutex member \emph{implicitly} acquires mutual-exclusion of the receiver object, \CFA uses an explicit mechanism to specify the object that acquires mutual-exclusion. A consequence of this approach is that it extends naturally to multi-monitor calls.
     1054Note that not all array functions are actually distinct in the type system.
     1055However, even if the code generation could tell the difference, the extra information is still not sufficient to extend meaningfully the monitor call semantic.
     1056
     1057Unlike object-oriented monitors, where calling a mutex member \emph{implicitly} acquires mutual-exclusion of the receiver object, \CFA uses an explicit mechanism to specify the object that acquires mutual-exclusion.
     1058A consequence of this approach is that it extends naturally to multi-monitor calls.
    8811059\begin{cfacode}
    8821060int f(MonitorA & mutex a, MonitorB & mutex b);
     
    8861064f(a,b);
    8871065\end{cfacode}
    888 While OO monitors could be extended with a mutex qualifier for multiple-monitor calls, no example of this feature could be found. The capability to acquire multiple locks before entering a critical section is called \emph{\textbf{bulk-acq}}. In practice, writing multi-locking routines that do not lead to deadlocks is tricky. Having language support for such a feature is therefore a significant asset for \CFA. In the case presented above, \CFA guarantees that the order of acquisition is consistent across calls to different routines using the same monitors as arguments. This consistent ordering means acquiring multiple monitors is safe from deadlock when using \textbf{bulk-acq}. However, users can still force the acquiring order. For example, notice which routines use \code{mutex}/\code{nomutex} and how this affects acquiring order:
     1066While OO monitors could be extended with a mutex qualifier for multiple-monitor calls, no example of this feature could be found.
     1067The capability to acquire multiple locks before entering a critical section is called \emph{\textbf{bulk-acq}}.
     1068In practice, writing multi-locking routines that do not lead to deadlocks is tricky.
     1069Having language support for such a feature is therefore a significant asset for \CFA.
     1070In the case presented above, \CFA guarantees that the order of acquisition is consistent across calls to different routines using the same monitors as arguments.
     1071This consistent ordering means acquiring multiple monitors is safe from deadlock when using \textbf{bulk-acq}.
     1072However, users can still force the acquiring order.
     1073For example, notice which routines use \code{mutex}/\code{nomutex} and how this affects acquiring order:
    8891074\begin{cfacode}
    8901075void foo(A& mutex a, B& mutex b) { //acquire a & b
     
    9001085}
    9011086\end{cfacode}
    902 The \textbf{multi-acq} monitor lock allows a monitor lock to be acquired by both \code{bar} or \code{baz} and acquired again in \code{foo}. In the calls to \code{bar} and \code{baz} the monitors are acquired in opposite order.
    903 
    904 However, such use leads to lock acquiring order problems. In the example above, the user uses implicit ordering in the case of function \code{foo} but explicit ordering in the case of \code{bar} and \code{baz}. This subtle difference means that calling these routines concurrently may lead to deadlock and is therefore undefined behaviour. As shown~\cite{Lister77}, solving this problem requires:
     1087The \textbf{multi-acq} monitor lock allows a monitor lock to be acquired by both \code{bar} or \code{baz} and acquired again in \code{foo}.
     1088In the calls to \code{bar} and \code{baz} the monitors are acquired in opposite order.
     1089
     1090However, such use leads to lock acquiring order problems.
     1091In the example above, the user uses implicit ordering in the case of function \code{foo} but explicit ordering in the case of \code{bar} and \code{baz}.
     1092This subtle difference means that calling these routines concurrently may lead to deadlock and is therefore undefined behaviour.
     1093As shown~\cite{Lister77}, solving this problem requires:
    9051094\begin{enumerate}
    9061095        \item Dynamically tracking the monitor-call order.
    9071096        \item Implement rollback semantics.
    9081097\end{enumerate}
    909 While the first requirement is already a significant constraint on the system, implementing a general rollback semantics in a C-like language is still prohibitively complex~\cite{Dice10}. In \CFA, users simply need to be careful when acquiring multiple monitors at the same time or only use \textbf{bulk-acq} of all the monitors. While \CFA provides only a partial solution, most systems provide no solution and the \CFA partial solution handles many useful cases.
     1098While the first requirement is already a significant constraint on the system, implementing a general rollback semantics in a C-like language is still prohibitively complex~\cite{Dice10}.
     1099In \CFA, users simply need to be careful when acquiring multiple monitors at the same time or only use \textbf{bulk-acq} of all the monitors.
     1100While \CFA provides only a partial solution, most systems provide no solution and the \CFA partial solution handles many useful cases.
    9101101
    9111102For example, \textbf{multi-acq} and \textbf{bulk-acq} can be used together in interesting ways:
     
    9201111}
    9211112\end{cfacode}
    922 This example shows a trivial solution to the bank-account transfer problem~\cite{BankTransfer}. Without \textbf{multi-acq} and \textbf{bulk-acq}, the solution to this problem is much more involved and requires careful engineering.
     1113This example shows a trivial solution to the bank-account transfer problem~\cite{BankTransfer}.
     1114Without \textbf{multi-acq} and \textbf{bulk-acq}, the solution to this problem is much more involved and requires careful engineering.
    9231115
    9241116\subsection{\code{mutex} statement} \label{mutex-stmt}
    9251117
    926 The call semantics discussed above have one software engineering issue: only a routine can acquire the mutual-exclusion of a set of monitor. \CFA offers the \code{mutex} statement to work around the need for unnecessary names, avoiding a major software engineering problem~\cite{2FTwoHardThings}. Table \ref{lst:mutex-stmt} shows an example of the \code{mutex} statement, which introduces a new scope in which the mutual-exclusion of a set of monitor is acquired. Beyond naming, the \code{mutex} statement has no semantic difference from a routine call with \code{mutex} parameters.
     1118The call semantics discussed above have one software engineering issue: only a routine can acquire the mutual-exclusion of a set of monitor. \CFA offers the \code{mutex} statement to work around the need for unnecessary names, avoiding a major software engineering problem~\cite{2FTwoHardThings}.
     1119Table \ref{lst:mutex-stmt} shows an example of the \code{mutex} statement, which introduces a new scope in which the mutual-exclusion of a set of monitor is acquired.
     1120Beyond naming, the \code{mutex} statement has no semantic difference from a routine call with \code{mutex} parameters.
    9271121
    9281122\begin{table}
     
    9611155% ======================================================================
    9621156% ======================================================================
    963 Once the call semantics are established, the next step is to establish data semantics. Indeed, until now a monitor is used simply as a generic handle but in most cases monitors contain shared data. This data should be intrinsic to the monitor declaration to prevent any accidental use of data without its appropriate protection. For example, here is a complete version of the counter shown in section \ref{call}:
     1157Once the call semantics are established, the next step is to establish data semantics.
     1158Indeed, until now a monitor is used simply as a generic handle but in most cases monitors contain shared data.
     1159This data should be intrinsic to the monitor declaration to prevent any accidental use of data without its appropriate protection.
     1160For example, here is a complete version of the counter shown in section \ref{call}:
    9641161\begin{cfacode}
    9651162monitor counter_t {
     
    9811178\end{cfacode}
    9821179
    983 Like threads and coroutines, monitors are defined in terms of traits with some additional language support in the form of the \code{monitor} keyword. The monitor trait is:
     1180Like threads and coroutines, monitors are defined in terms of traits with some additional language support in the form of the \code{monitor} keyword.
     1181The monitor trait is:
    9841182\begin{cfacode}
    9851183trait is_monitor(dtype T) {
     
    9881186};
    9891187\end{cfacode}
    990 Note that the destructor of a monitor must be a \code{mutex} routine to prevent deallocation while a thread is accessing the monitor. As with any object, calls to a monitor, using \code{mutex} or otherwise, is undefined behaviour after the destructor has run.
     1188Note that the destructor of a monitor must be a \code{mutex} routine to prevent deallocation while a thread is accessing the monitor.
     1189As with any object, calls to a monitor, using \code{mutex} or otherwise, is undefined behaviour after the destructor has run.
    9911190
    9921191% ======================================================================
     
    9951194% ======================================================================
    9961195% ======================================================================
    997 In addition to mutual exclusion, the monitors at the core of \CFA's concurrency can also be used to achieve synchronization. With monitors, this capability is generally achieved with internal or external scheduling as in~\cite{Hoare74}. With \textbf{scheduling} loosely defined as deciding which thread acquires the critical section next, \textbf{internal scheduling} means making the decision from inside the critical section (i.e., with access to the shared state), while \textbf{external scheduling} means making the decision when entering the critical section (i.e., without access to the shared state). Since internal scheduling within a single monitor is mostly a solved problem, this paper concentrates on extending internal scheduling to multiple monitors. Indeed, like the \textbf{bulk-acq} semantics, internal scheduling extends to multiple monitors in a way that is natural to the user but requires additional complexity on the implementation side.
     1196In addition to mutual exclusion, the monitors at the core of \CFA's concurrency can also be used to achieve synchronization.
     1197With monitors, this capability is generally achieved with internal or external scheduling as in~\cite{Hoare74}.
     1198With \textbf{scheduling} loosely defined as deciding which thread acquires the critical section next, \textbf{internal scheduling} means making the decision from inside the critical section (i.e., with access to the shared state), while \textbf{external scheduling} means making the decision when entering the critical section (i.e., without access to the shared state).
     1199Since internal scheduling within a single monitor is mostly a solved problem, this paper concentrates on extending internal scheduling to multiple monitors.
     1200Indeed, like the \textbf{bulk-acq} semantics, internal scheduling extends to multiple monitors in a way that is natural to the user but requires additional complexity on the implementation side.
    9981201
    9991202First, here is a simple example of internal scheduling:
     
    10181221}
    10191222\end{cfacode}
    1020 There are two details to note here. First, \code{signal} is a delayed operation; it only unblocks the waiting thread when it reaches the end of the critical section. This semantics is needed to respect mutual-exclusion, i.e., the signaller and signalled thread cannot be in the monitor simultaneously. The alternative is to return immediately after the call to \code{signal}, which is significantly more restrictive. Second, in \CFA, while it is common to store a \code{condition} as a field of the monitor, a \code{condition} variable can be stored/created independently of a monitor. Here routine \code{foo} waits for the \code{signal} from \code{bar} before making further progress, ensuring a basic ordering.
    1021 
    1022 An important aspect of the implementation is that \CFA does not allow barging, which means that once function \code{bar} releases the monitor, \code{foo} is guaranteed to be the next thread to acquire the monitor (unless some other thread waited on the same condition). This guarantee offers the benefit of not having to loop around waits to recheck that a condition is met. The main reason \CFA offers this guarantee is that users can easily introduce barging if it becomes a necessity but adding barging prevention or barging avoidance is more involved without language support. Supporting barging prevention as well as extending internal scheduling to multiple monitors is the main source of complexity in the design and implementation of \CFA concurrency.
     1223There are two details to note here.
     1224First, \code{signal} is a delayed operation; it only unblocks the waiting thread when it reaches the end of the critical section.
     1225This semantics is needed to respect mutual-exclusion, i.e., the signaller and signalled thread cannot be in the monitor simultaneously.
     1226The alternative is to return immediately after the call to \code{signal}, which is significantly more restrictive.
     1227Second, in \CFA, while it is common to store a \code{condition} as a field of the monitor, a \code{condition} variable can be stored/created independently of a monitor.
     1228Here routine \code{foo} waits for the \code{signal} from \code{bar} before making further progress, ensuring a basic ordering.
     1229
     1230An important aspect of the implementation is that \CFA does not allow barging, which means that once function \code{bar} releases the monitor, \code{foo} is guaranteed to be the next thread to acquire the monitor (unless some other thread waited on the same condition).
     1231This guarantee offers the benefit of not having to loop around waits to recheck that a condition is met.
     1232The main reason \CFA offers this guarantee is that users can easily introduce barging if it becomes a necessity but adding barging prevention or barging avoidance is more involved without language support.
     1233Supporting barging prevention as well as extending internal scheduling to multiple monitors is the main source of complexity in the design and implementation of \CFA concurrency.
    10231234
    10241235% ======================================================================
     
    10271238% ======================================================================
    10281239% ======================================================================
    1029 It is easy to understand the problem of multi-monitor scheduling using a series of pseudo-code examples. Note that for simplicity in the following snippets of pseudo-code, waiting and signalling is done using an implicit condition variable, like Java built-in monitors. Indeed, \code{wait} statements always use the implicit condition variable as parameters and explicitly name the monitors (A and B) associated with the condition. Note that in \CFA, condition variables are tied to a \emph{group} of monitors on first use (called branding), which means that using internal scheduling with distinct sets of monitors requires one condition variable per set of monitors. The example below shows the simple case of having two threads (one for each column) and a single monitor A.
     1240It is easy to understand the problem of multi-monitor scheduling using a series of pseudo-code examples.
     1241Note that for simplicity in the following snippets of pseudo-code, waiting and signalling is done using an implicit condition variable, like Java built-in monitors.
     1242Indeed, \code{wait} statements always use the implicit condition variable as parameters and explicitly name the monitors (A and B) associated with the condition.
     1243Note that in \CFA, condition variables are tied to a \emph{group} of monitors on first use (called branding), which means that using internal scheduling with distinct sets of monitors requires one condition variable per set of monitors.
     1244The example below shows the simple case of having two threads (one for each column) and a single monitor A.
    10301245
    10311246\begin{multicols}{2}
     
    10461261\end{pseudo}
    10471262\end{multicols}
    1048 One thread acquires before waiting (atomically blocking and releasing A) and the other acquires before signalling. It is important to note here that both \code{wait} and \code{signal} must be called with the proper monitor(s) already acquired. This semantic is a logical requirement for barging prevention.
     1263One thread acquires before waiting (atomically blocking and releasing A) and the other acquires before signalling.
     1264It is important to note here that both \code{wait} and \code{signal} must be called with the proper monitor(s) already acquired.
     1265This semantic is a logical requirement for barging prevention.
    10491266
    10501267A direct extension of the previous example is a \textbf{bulk-acq} version:
     
    10621279\end{pseudo}
    10631280\end{multicols}
    1064 \noindent This version uses \textbf{bulk-acq} (denoted using the {\sf\&} symbol), but the presence of multiple monitors does not add a particularly new meaning. Synchronization happens between the two threads in exactly the same way and order. The only difference is that mutual exclusion covers a group of monitors. On the implementation side, handling multiple monitors does add a degree of complexity as the next few examples demonstrate.
    1065 
    1066 While deadlock issues can occur when nesting monitors, these issues are only a symptom of the fact that locks, and by extension monitors, are not perfectly composable. For monitors, a well-known deadlock problem is the Nested Monitor Problem~\cite{Lister77}, which occurs when a \code{wait} is made by a thread that holds more than one monitor. For example, the following pseudo-code runs into the nested-monitor problem:
     1281\noindent This version uses \textbf{bulk-acq} (denoted using the {\sf\&} symbol), but the presence of multiple monitors does not add a particularly new meaning.
     1282Synchronization happens between the two threads in exactly the same way and order.
     1283The only difference is that mutual exclusion covers a group of monitors.
     1284On the implementation side, handling multiple monitors does add a degree of complexity as the next few examples demonstrate.
     1285
     1286While deadlock issues can occur when nesting monitors, these issues are only a symptom of the fact that locks, and by extension monitors, are not perfectly composable.
     1287For monitors, a well-known deadlock problem is the Nested Monitor Problem~\cite{Lister77}, which occurs when a \code{wait} is made by a thread that holds more than one monitor.
     1288For example, the following pseudo-code runs into the nested-monitor problem:
    10671289\begin{multicols}{2}
    10681290\begin{pseudo}
     
    10841306\end{pseudo}
    10851307\end{multicols}
    1086 \noindent The \code{wait} only releases monitor \code{B} so the signalling thread cannot acquire monitor \code{A} to get to the \code{signal}. Attempting release of all acquired monitors at the \code{wait} introduces a different set of problems, such as releasing monitor \code{C}, which has nothing to do with the \code{signal}.
    1087 
    1088 However, for monitors as for locks, it is possible to write a program using nesting without encountering any problems if nesting is done correctly. For example, the next pseudo-code snippet acquires monitors {\sf A} then {\sf B} before waiting, while only acquiring {\sf B} when signalling, effectively avoiding the Nested Monitor Problem~\cite{Lister77}.
     1308\noindent The \code{wait} only releases monitor \code{B} so the signalling thread cannot acquire monitor \code{A} to get to the \code{signal}.
     1309Attempting release of all acquired monitors at the \code{wait} introduces a different set of problems, such as releasing monitor \code{C}, which has nothing to do with the \code{signal}.
     1310
     1311However, for monitors as for locks, it is possible to write a program using nesting without encountering any problems if nesting is done correctly.
     1312For example, the next pseudo-code snippet acquires monitors {\sf A} then {\sf B} before waiting, while only acquiring {\sf B} when signalling, effectively avoiding the Nested Monitor Problem~\cite{Lister77}.
    10891313
    10901314\begin{multicols}{2}
     
    11161340% ======================================================================
    11171341
    1118 A larger example is presented to show complex issues for \textbf{bulk-acq} and its implementation options are analyzed. Listing \ref{lst:int-bulk-pseudo} shows an example where \textbf{bulk-acq} adds a significant layer of complexity to the internal signalling semantics, and listing \ref{lst:int-bulk-cfa} shows the corresponding \CFA code to implement the pseudo-code in listing \ref{lst:int-bulk-pseudo}. For the purpose of translating the given pseudo-code into \CFA-code, any method of introducing a monitor is acceptable, e.g., \code{mutex} parameters, global variables, pointer parameters, or using locals with the \code{mutex} statement.
     1342A larger example is presented to show complex issues for \textbf{bulk-acq} and its implementation options are analyzed.
     1343Listing \ref{lst:int-bulk-pseudo} shows an example where \textbf{bulk-acq} adds a significant layer of complexity to the internal signalling semantics, and listing \ref{lst:int-bulk-cfa} shows the corresponding \CFA code to implement the pseudo-code in listing \ref{lst:int-bulk-pseudo}.
     1344For the purpose of translating the given pseudo-code into \CFA-code, any method of introducing a monitor is acceptable, e.g., \code{mutex} parameters, global variables, pointer parameters, or using locals with the \code{mutex} statement.
    11191345
    11201346\begin{figure}[!t]
     
    12111437\end{figure}
    12121438
    1213 The complexity begins at code sections 4 and 8 in listing \ref{lst:int-bulk-pseudo}, which are where the existing semantics of internal scheduling needs to be extended for multiple monitors. The root of the problem is that \textbf{bulk-acq} is used in a context where one of the monitors is already acquired, which is why it is important to define the behaviour of the previous pseudo-code. When the signaller thread reaches the location where it should ``release \code{A & B}'' (listing \ref{lst:int-bulk-pseudo} line \ref{line:releaseFirst}), it must actually transfer ownership of monitor \code{B} to the waiting thread. This ownership transfer is required in order to prevent barging into \code{B} by another thread, since both the signalling and signalled threads still need monitor \code{A}. There are three options:
     1439The complexity begins at code sections 4 and 8 in listing \ref{lst:int-bulk-pseudo}, which are where the existing semantics of internal scheduling needs to be extended for multiple monitors.
     1440The root of the problem is that \textbf{bulk-acq} is used in a context where one of the monitors is already acquired, which is why it is important to define the behaviour of the previous pseudo-code.
     1441When the signaller thread reaches the location where it should ``release \code{A & B}'' (listing \ref{lst:int-bulk-pseudo} line \ref{line:releaseFirst}), it must actually transfer ownership of monitor \code{B} to the waiting thread.
     1442This ownership transfer is required in order to prevent barging into \code{B} by another thread, since both the signalling and signalled threads still need monitor \code{A}.
     1443There are three options:
    12141444
    12151445\subsubsection{Delaying Signals}
    1216 The obvious solution to the problem of multi-monitor scheduling is to keep ownership of all locks until the last lock is ready to be transferred. It can be argued that that moment is when the last lock is no longer needed, because this semantics fits most closely to the behaviour of single-monitor scheduling. This solution has the main benefit of transferring ownership of groups of monitors, which simplifies the semantics from multiple objects to a single group of objects, effectively making the existing single-monitor semantic viable by simply changing monitors to monitor groups. This solution releases the monitors once every monitor in a group can be released. However, since some monitors are never released (e.g., the monitor of a thread), this interpretation means a group might never be released. A more interesting interpretation is to transfer the group until all its monitors are released, which means the group is not passed further and a thread can retain its locks.
    1217 
    1218 However, listing \ref{lst:int-secret} shows this solution can become much more complicated depending on what is executed while secretly holding B at line \ref{line:secret}, while avoiding the need to transfer ownership of a subset of the condition monitors. Listing \ref{lst:dependency} shows a slightly different example where a third thread is waiting on monitor \code{A}, using a different condition variable. Because the third thread is signalled when secretly holding \code{B}, the goal  becomes unreachable. Depending on the order of signals (listing \ref{lst:dependency} line \ref{line:signal-ab} and \ref{line:signal-a}) two cases can happen:
     1446The obvious solution to the problem of multi-monitor scheduling is to keep ownership of all locks until the last lock is ready to be transferred.
     1447It can be argued that that moment is when the last lock is no longer needed, because this semantics fits most closely to the behaviour of single-monitor scheduling.
     1448This solution has the main benefit of transferring ownership of groups of monitors, which simplifies the semantics from multiple objects to a single group of objects, effectively making the existing single-monitor semantic viable by simply changing monitors to monitor groups.
     1449This solution releases the monitors once every monitor in a group can be released.
     1450However, since some monitors are never released (e.g., the monitor of a thread), this interpretation means a group might never be released.
     1451A more interesting interpretation is to transfer the group until all its monitors are released, which means the group is not passed further and a thread can retain its locks.
     1452
     1453However, listing \ref{lst:int-secret} shows this solution can become much more complicated depending on what is executed while secretly holding B at line \ref{line:secret}, while avoiding the need to transfer ownership of a subset of the condition monitors.
     1454Listing \ref{lst:dependency} shows a slightly different example where a third thread is waiting on monitor \code{A}, using a different condition variable.
     1455Because the third thread is signalled when secretly holding \code{B}, the goal  becomes unreachable.
     1456Depending on the order of signals (listing \ref{lst:dependency} line \ref{line:signal-ab} and \ref{line:signal-a}) two cases can happen:
    12191457
    12201458\paragraph{Case 1: thread $\alpha$ goes first.} In this case, the problem is that monitor \code{A} needs to be passed to thread $\beta$ when thread $\alpha$ is done with it.
     
    12221460\\
    12231461
    1224 Note that ordering is not determined by a race condition but by whether signalled threads are enqueued in FIFO or FILO order. However, regardless of the answer, users can move line \ref{line:signal-a} before line \ref{line:signal-ab} and get the reverse effect for listing \ref{lst:dependency}.
     1462Note that ordering is not determined by a race condition but by whether signalled threads are enqueued in FIFO or FILO order.
     1463However, regardless of the answer, users can move line \ref{line:signal-a} before line \ref{line:signal-ab} and get the reverse effect for listing \ref{lst:dependency}.
    12251464
    12261465In both cases, the threads need to be able to distinguish, on a per monitor basis, which ones need to be released and which ones need to be transferred, which means knowing when to release a group becomes complex and inefficient (see next section) and therefore effectively precludes this approach.
     
    12661505\end{figure}
    12671506
    1268 In listing \ref{lst:int-bulk-pseudo}, there is a solution that satisfies both barging prevention and mutual exclusion. If ownership of both monitors is transferred to the waiter when the signaller releases \code{A & B} and then the waiter transfers back ownership of \code{A} back to the signaller when it releases it, then the problem is solved (\code{B} is no longer in use at this point). Dynamically finding the correct order is therefore the second possible solution. The problem is effectively resolving a dependency graph of ownership requirements. Here even the simplest of code snippets requires two transfers and has a super-linear complexity. This complexity can be seen in listing \ref{lst:explosion}, which is just a direct extension to three monitors, requires at least three ownership transfer and has multiple solutions. Furthermore, the presence of multiple solutions for ownership transfer can cause deadlock problems if a specific solution is not consistently picked; In the same way that multiple lock acquiring order can cause deadlocks.
     1507In listing \ref{lst:int-bulk-pseudo}, there is a solution that satisfies both barging prevention and mutual exclusion.
     1508If ownership of both monitors is transferred to the waiter when the signaller releases \code{A & B} and then the waiter transfers back ownership of \code{A} back to the signaller when it releases it, then the problem is solved (\code{B} is no longer in use at this point).
     1509Dynamically finding the correct order is therefore the second possible solution.
     1510The problem is effectively resolving a dependency graph of ownership requirements.
     1511Here even the simplest of code snippets requires two transfers and has a super-linear complexity.
     1512This complexity can be seen in listing \ref{lst:explosion}, which is just a direct extension to three monitors, requires at least three ownership transfer and has multiple solutions.
     1513Furthermore, the presence of multiple solutions for ownership transfer can cause deadlock problems if a specific solution is not consistently picked; In the same way that multiple lock acquiring order can cause deadlocks.
    12691514\begin{figure}
    12701515\begin{multicols}{2}
     
    12951540\end{figure}
    12961541
    1297 Given the three threads example in listing \ref{lst:dependency}, figure \ref{fig:dependency} shows the corresponding dependency graph that results, where every node is a statement of one of the three threads, and the arrows the dependency of that statement (e.g., $\alpha1$ must happen before $\alpha2$). The extra challenge is that this dependency graph is effectively post-mortem, but the runtime system needs to be able to build and solve these graphs as the dependencies unfold. Resolving dependency graphs being a complex and expensive endeavour, this solution is not the preferred one.
     1542Given the three threads example in listing \ref{lst:dependency}, figure \ref{fig:dependency} shows the corresponding dependency graph that results, where every node is a statement of one of the three threads, and the arrows the dependency of that statement (e.g., $\alpha1$ must happen before $\alpha2$).
     1543The extra challenge is that this dependency graph is effectively post-mortem, but the runtime system needs to be able to build and solve these graphs as the dependencies unfold.
     1544Resolving dependency graphs being a complex and expensive endeavour, this solution is not the preferred one.
    12981545
    12991546\subsubsection{Partial Signalling} \label{partial-sig}
    1300 Finally, the solution that is chosen for \CFA is to use partial signalling. Again using listing \ref{lst:int-bulk-pseudo}, the partial signalling solution transfers ownership of monitor \code{B} at lines \ref{line:signal1} to the waiter but does not wake the waiting thread since it is still using monitor \code{A}. Only when it reaches line \ref{line:lastRelease} does it actually wake up the waiting thread. This solution has the benefit that complexity is encapsulated into only two actions: passing monitors to the next owner when they should be released and conditionally waking threads if all conditions are met. This solution has a much simpler implementation than a dependency graph solving algorithms, which is why it was chosen. Furthermore, after being fully implemented, this solution does not appear to have any significant downsides.
     1547Finally, the solution that is chosen for \CFA is to use partial signalling.
     1548Again using listing \ref{lst:int-bulk-pseudo}, the partial signalling solution transfers ownership of monitor \code{B} at lines \ref{line:signal1} to the waiter but does not wake the waiting thread since it is still using monitor \code{A}.
     1549Only when it reaches line \ref{line:lastRelease} does it actually wake up the waiting thread.
     1550This solution has the benefit that complexity is encapsulated into only two actions: passing monitors to the next owner when they should be released and conditionally waking threads if all conditions are met.
     1551This solution has a much simpler implementation than a dependency graph solving algorithms, which is why it was chosen.
     1552Furthermore, after being fully implemented, this solution does not appear to have any significant downsides.
    13011553
    13021554Using partial signalling, listing \ref{lst:dependency} can be solved easily:
     
    14151667\label{tbl:datingservice}
    14161668\end{table}
    1417 An important note is that, until now, signalling a monitor was a delayed operation. The ownership of the monitor is transferred only when the monitor would have otherwise been released, not at the point of the \code{signal} statement. However, in some cases, it may be more convenient for users to immediately transfer ownership to the thread that is waiting for cooperation, which is achieved using the \code{signal_block} routine.
    1418 
    1419 The example in table \ref{tbl:datingservice} highlights the difference in behaviour. As mentioned, \code{signal} only transfers ownership once the current critical section exits; this behaviour requires additional synchronization when a two-way handshake is needed. To avoid this explicit synchronization, the \code{condition} type offers the \code{signal_block} routine, which handles the two-way handshake as shown in the example. This feature removes the need for a second condition variables and simplifies programming. Like every other monitor semantic, \code{signal_block} uses barging prevention, which means mutual-exclusion is baton-passed both on the front end and the back end of the call to \code{signal_block}, meaning no other thread can acquire the monitor either before or after the call.
     1669An important note is that, until now, signalling a monitor was a delayed operation.
     1670The ownership of the monitor is transferred only when the monitor would have otherwise been released, not at the point of the \code{signal} statement.
     1671However, in some cases, it may be more convenient for users to immediately transfer ownership to the thread that is waiting for cooperation, which is achieved using the \code{signal_block} routine.
     1672
     1673The example in table \ref{tbl:datingservice} highlights the difference in behaviour.
     1674As mentioned, \code{signal} only transfers ownership once the current critical section exits; this behaviour requires additional synchronization when a two-way handshake is needed.
     1675To avoid this explicit synchronization, the \code{condition} type offers the \code{signal_block} routine, which handles the two-way handshake as shown in the example.
     1676This feature removes the need for a second condition variables and simplifies programming.
     1677Like every other monitor semantic, \code{signal_block} uses barging prevention, which means mutual-exclusion is baton-passed both on the front end and the back end of the call to \code{signal_block}, meaning no other thread can acquire the monitor either before or after the call.
    14201678
    14211679% ======================================================================
     
    14881746\label{tbl:sched}
    14891747\end{table}
    1490 This method is more constrained and explicit, which helps users reduce the non-deterministic nature of concurrency. Indeed, as the following examples demonstrate, external scheduling allows users to wait for events from other threads without the concern of unrelated events occurring. External scheduling can generally be done either in terms of control flow (e.g., Ada with \code{accept}, \uC with \code{_Accept}) or in terms of data (e.g., Go with channels). Of course, both of these paradigms have their own strengths and weaknesses, but for this project, control-flow semantics was chosen to stay consistent with the rest of the languages semantics. Two challenges specific to \CFA arise when trying to add external scheduling with loose object definitions and multiple-monitor routines. The previous example shows a simple use \code{_Accept} versus \code{wait}/\code{signal} and its advantages. Note that while other languages often use \code{accept}/\code{select} as the core external scheduling keyword, \CFA uses \code{waitfor} to prevent name collisions with existing socket \textbf{api}s.
    1491 
    1492 For the \code{P} member above using internal scheduling, the call to \code{wait} only guarantees that \code{V} is the last routine to access the monitor, allowing a third routine, say \code{isInUse()}, acquire mutual exclusion several times while routine \code{P} is waiting. On the other hand, external scheduling guarantees that while routine \code{P} is waiting, no other routine than \code{V} can acquire the monitor.
     1748This method is more constrained and explicit, which helps users reduce the non-deterministic nature of concurrency.
     1749Indeed, as the following examples demonstrate, external scheduling allows users to wait for events from other threads without the concern of unrelated events occurring.
     1750External scheduling can generally be done either in terms of control flow (e.g., Ada with \code{accept}, \uC with \code{_Accept}) or in terms of data (e.g., Go with channels).
     1751Of course, both of these paradigms have their own strengths and weaknesses, but for this project, control-flow semantics was chosen to stay consistent with the rest of the languages semantics.
     1752Two challenges specific to \CFA arise when trying to add external scheduling with loose object definitions and multiple-monitor routines.
     1753The previous example shows a simple use \code{_Accept} versus \code{wait}/\code{signal} and its advantages.
     1754Note that while other languages often use \code{accept}/\code{select} as the core external scheduling keyword, \CFA uses \code{waitfor} to prevent name collisions with existing socket \textbf{api}s.
     1755
     1756For the \code{P} member above using internal scheduling, the call to \code{wait} only guarantees that \code{V} is the last routine to access the monitor, allowing a third routine, say \code{isInUse()}, acquire mutual exclusion several times while routine \code{P} is waiting.
     1757On the other hand, external scheduling guarantees that while routine \code{P} is waiting, no other routine than \code{V} can acquire the monitor.
    14931758
    14941759% ======================================================================
     
    14971762% ======================================================================
    14981763% ======================================================================
    1499 In \uC, a monitor class declaration includes an exhaustive list of monitor operations. Since \CFA is not object oriented, monitors become both more difficult to implement and less clear for a user:
     1764In \uC, a monitor class declaration includes an exhaustive list of monitor operations.
     1765Since \CFA is not object oriented, monitors become both more difficult to implement and less clear for a user:
    15001766
    15011767\begin{cfacode}
     
    15131779\end{cfacode}
    15141780
    1515 Furthermore, external scheduling is an example where implementation constraints become visible from the interface. Here is the pseudo-code for the entering phase of a monitor:
     1781Furthermore, external scheduling is an example where implementation constraints become visible from the interface.
     1782Here is the pseudo-code for the entering phase of a monitor:
    15161783\begin{center}
    15171784\begin{tabular}{l}
     
    15281795\end{tabular}
    15291796\end{center}
    1530 For the first two conditions, it is easy to implement a check that can evaluate the condition in a few instructions. However, a fast check for \pscode{monitor accepts me} is much harder to implement depending on the constraints put on the monitors. Indeed, monitors are often expressed as an entry queue and some acceptor queue as in Figure~\ref{fig:ClassicalMonitor}.
     1797For the first two conditions, it is easy to implement a check that can evaluate the condition in a few instructions.
     1798However, a fast check for \pscode{monitor accepts me} is much harder to implement depending on the constraints put on the monitors.
     1799Indeed, monitors are often expressed as an entry queue and some acceptor queue as in Figure~\ref{fig:ClassicalMonitor}.
    15311800
    15321801\begin{figure}
     
    15441813\end{figure}
    15451814
    1546 There are other alternatives to these pictures, but in the case of the left picture, implementing a fast accept check is relatively easy. Restricted to a fixed number of mutex members, N, the accept check reduces to updating a bitmask when the acceptor queue changes, a check that executes in a single instruction even with a fairly large number (e.g., 128) of mutex members. This approach requires a unique dense ordering of routines with an upper-bound and that ordering must be consistent across translation units. For OO languages these constraints are common, since objects only offer adding member routines consistently across translation units via inheritance. However, in \CFA users can extend objects with mutex routines that are only visible in certain translation unit. This means that establishing a program-wide dense-ordering among mutex routines can only be done in the program linking phase, and still could have issues when using dynamically shared objects.
     1815There are other alternatives to these pictures, but in the case of the left picture, implementing a fast accept check is relatively easy.
     1816Restricted to a fixed number of mutex members, N, the accept check reduces to updating a bitmask when the acceptor queue changes, a check that executes in a single instruction even with a fairly large number (e.g., 128) of mutex members.
     1817This approach requires a unique dense ordering of routines with an upper-bound and that ordering must be consistent across translation units.
     1818For OO languages these constraints are common, since objects only offer adding member routines consistently across translation units via inheritance.
     1819However, in \CFA users can extend objects with mutex routines that are only visible in certain translation unit.
     1820This means that establishing a program-wide dense-ordering among mutex routines can only be done in the program linking phase, and still could have issues when using dynamically shared objects.
    15471821
    15481822The alternative is to alter the implementation as in Figure~\ref{fig:BulkMonitor}.
    1549 Here, the mutex routine called is associated with a thread on the entry queue while a list of acceptable routines is kept separate. Generating a mask dynamically means that the storage for the mask information can vary between calls to \code{waitfor}, allowing for more flexibility and extensions. Storing an array of accepted function pointers replaces the single instruction bitmask comparison with dereferencing a pointer followed by a linear search. Furthermore, supporting nested external scheduling (e.g., listing \ref{lst:nest-ext}) may now require additional searches for the \code{waitfor} statement to check if a routine is already queued.
     1823Here, the mutex routine called is associated with a thread on the entry queue while a list of acceptable routines is kept separate.
     1824Generating a mask dynamically means that the storage for the mask information can vary between calls to \code{waitfor}, allowing for more flexibility and extensions.
     1825Storing an array of accepted function pointers replaces the single instruction bitmask comparison with dereferencing a pointer followed by a linear search.
     1826Furthermore, supporting nested external scheduling (e.g., listing \ref{lst:nest-ext}) may now require additional searches for the \code{waitfor} statement to check if a routine is already queued.
    15501827
    15511828\begin{figure}
     
    15641841\end{figure}
    15651842
    1566 Note that in the right picture, tasks need to always keep track of the monitors associated with mutex routines, and the routine mask needs to have both a function pointer and a set of monitors, as is discussed in the next section. These details are omitted from the picture for the sake of simplicity.
    1567 
    1568 At this point, a decision must be made between flexibility and performance. Many design decisions in \CFA achieve both flexibility and performance, for example polymorphic routines add significant flexibility but inlining them means the optimizer can easily remove any runtime cost. Here, however, the cost of flexibility cannot be trivially removed. In the end, the most flexible approach has been chosen since it allows users to write programs that would otherwise be  hard to write. This decision is based on the assumption that writing fast but inflexible locks is closer to a solved problem than writing locks that are as flexible as external scheduling in \CFA.
     1843Note that in the right picture, tasks need to always keep track of the monitors associated with mutex routines, and the routine mask needs to have both a function pointer and a set of monitors, as is discussed in the next section.
     1844These details are omitted from the picture for the sake of simplicity.
     1845
     1846At this point, a decision must be made between flexibility and performance.
     1847Many design decisions in \CFA achieve both flexibility and performance, for example polymorphic routines add significant flexibility but inlining them means the optimizer can easily remove any runtime cost.
     1848Here, however, the cost of flexibility cannot be trivially removed.
     1849In the end, the most flexible approach has been chosen since it allows users to write programs that would otherwise be  hard to write.
     1850This decision is based on the assumption that writing fast but inflexible locks is closer to a solved problem than writing locks that are as flexible as external scheduling in \CFA.
    15691851
    15701852% ======================================================================
     
    15741856% ======================================================================
    15751857
    1576 External scheduling, like internal scheduling, becomes significantly more complex when introducing multi-monitor syntax. Even in the simplest possible case, some new semantics needs to be established:
     1858External scheduling, like internal scheduling, becomes significantly more complex when introducing multi-monitor syntax.
     1859Even in the simplest possible case, some new semantics needs to be established:
    15771860\begin{cfacode}
    15781861monitor M {};
     
    15961879}
    15971880\end{cfacode}
    1598 This syntax is unambiguous. Both locks are acquired and kept by \code{g}. When routine \code{f} is called, the lock for monitor \code{b} is temporarily transferred from \code{g} to \code{f} (while \code{g} still holds lock \code{a}). This behaviour can be extended to the multi-monitor \code{waitfor} statement as follows.
     1881This syntax is unambiguous.
     1882Both locks are acquired and kept by \code{g}.
     1883When routine \code{f} is called, the lock for monitor \code{b} is temporarily transferred from \code{g} to \code{f} (while \code{g} still holds lock \code{a}).
     1884This behaviour can be extended to the multi-monitor \code{waitfor} statement as follows.
    15991885
    16001886\begin{cfacode}
     
    16331919}
    16341920\end{cfacode}
    1635 While the equivalent can happen when using internal scheduling, the fact that conditions are specific to a set of monitors means that users have to use two different condition variables. In both cases, partially matching monitor sets does not wakeup the waiting thread. It is also important to note that in the case of external scheduling the order of parameters is irrelevant; \code{waitfor(f,a,b)} and \code{waitfor(f,b,a)} are indistinguishable waiting condition.
     1921While the equivalent can happen when using internal scheduling, the fact that conditions are specific to a set of monitors means that users have to use two different condition variables.
     1922In both cases, partially matching monitor sets does not wakeup the waiting thread.
     1923It is also important to note that in the case of external scheduling the order of parameters is irrelevant; \code{waitfor(f,a,b)} and \code{waitfor(f,b,a)} are indistinguishable waiting condition.
    16361924
    16371925% ======================================================================
     
    16411929% ======================================================================
    16421930
    1643 Syntactically, the \code{waitfor} statement takes a function identifier and a set of monitors. While the set of monitors can be any list of expressions, the function name is more restricted because the compiler validates at compile time the validity of the function type and the parameters used with the \code{waitfor} statement. It checks that the set of monitors passed in matches the requirements for a function call. Listing \ref{lst:waitfor} shows various usages of the waitfor statement and which are acceptable. The choice of the function type is made ignoring any non-\code{mutex} parameter. One limitation of the current implementation is that it does not handle overloading, but overloading is possible.
     1931Syntactically, the \code{waitfor} statement takes a function identifier and a set of monitors.
     1932While the set of monitors can be any list of expressions, the function name is more restricted because the compiler validates at compile time the validity of the function type and the parameters used with the \code{waitfor} statement.
     1933It checks that the set of monitors passed in matches the requirements for a function call.
     1934Listing \ref{lst:waitfor} shows various usages of the waitfor statement and which are acceptable.
     1935The choice of the function type is made ignoring any non-\code{mutex} parameter.
     1936One limitation of the current implementation is that it does not handle overloading, but overloading is possible.
    16441937\begin{figure}
    16451938\begin{cfacode}[caption={Various correct and incorrect uses of the waitfor statement},label={lst:waitfor}]
     
    16751968\end{figure}
    16761969
    1677 Finally, for added flexibility, \CFA supports constructing a complex \code{waitfor} statement using the \code{or}, \code{timeout} and \code{else}. Indeed, multiple \code{waitfor} clauses can be chained together using \code{or}; this chain forms a single statement that uses baton pass to any function that fits one of the function+monitor set passed in. To enable users to tell which accepted function executed, \code{waitfor}s are followed by a statement (including the null statement \code{;}) or a compound statement, which is executed after the clause is triggered. A \code{waitfor} chain can also be followed by a \code{timeout}, to signify an upper bound on the wait, or an \code{else}, to signify that the call should be non-blocking, which checks for a matching function call already arrived and otherwise continues. Any and all of these clauses can be preceded by a \code{when} condition to dynamically toggle the accept clauses on or off based on some current state. Listing \ref{lst:waitfor2} demonstrates several complex masks and some incorrect ones.
     1970Finally, for added flexibility, \CFA supports constructing a complex \code{waitfor} statement using the \code{or}, \code{timeout} and \code{else}.
     1971Indeed, multiple \code{waitfor} clauses can be chained together using \code{or}; this chain forms a single statement that uses baton pass to any function that fits one of the function+monitor set passed in.
     1972To enable users to tell which accepted function executed, \code{waitfor}s are followed by a statement (including the null statement \code{;}) or a compound statement, which is executed after the clause is triggered.
     1973A \code{waitfor} chain can also be followed by a \code{timeout}, to signify an upper bound on the wait, or an \code{else}, to signify that the call should be non-blocking, which checks for a matching function call already arrived and otherwise continues.
     1974Any and all of these clauses can be preceded by a \code{when} condition to dynamically toggle the accept clauses on or off based on some current state.
     1975Listing \ref{lst:waitfor2} demonstrates several complex masks and some incorrect ones.
    16781976
    16791977\begin{figure}
     
    17432041% ======================================================================
    17442042% ======================================================================
    1745 An interesting use for the \code{waitfor} statement is destructor semantics. Indeed, the \code{waitfor} statement can accept any \code{mutex} routine, which includes the destructor (see section \ref{data}). However, with the semantics discussed until now, waiting for the destructor does not make any sense, since using an object after its destructor is called is undefined behaviour. The simplest approach is to disallow \code{waitfor} on a destructor. However, a more expressive approach is to flip ordering of execution when waiting for the destructor, meaning that waiting for the destructor allows the destructor to run after the current \code{mutex} routine, similarly to how a condition is signalled.
     2043An interesting use for the \code{waitfor} statement is destructor semantics.
     2044Indeed, the \code{waitfor} statement can accept any \code{mutex} routine, which includes the destructor (see section \ref{data}).
     2045However, with the semantics discussed until now, waiting for the destructor does not make any sense, since using an object after its destructor is called is undefined behaviour.
     2046The simplest approach is to disallow \code{waitfor} on a destructor.
     2047However, a more expressive approach is to flip ordering of execution when waiting for the destructor, meaning that waiting for the destructor allows the destructor to run after the current \code{mutex} routine, similarly to how a condition is signalled.
    17462048\begin{figure}
    17472049\begin{cfacode}[caption={Example of an executor which executes action in series until the destructor is called.},label={lst:dtor-order}]
     
    17612063\end{cfacode}
    17622064\end{figure}
    1763 For example, listing \ref{lst:dtor-order} shows an example of an executor with an infinite loop, which waits for the destructor to break out of this loop. Switching the semantic meaning introduces an idiomatic way to terminate a task and/or wait for its termination via destruction.
     2065For example, listing \ref{lst:dtor-order} shows an example of an executor with an infinite loop, which waits for the destructor to break out of this loop.
     2066Switching the semantic meaning introduces an idiomatic way to terminate a task and/or wait for its termination via destruction.
    17642067
    17652068
     
    17722075% #       #     # #     # #     # ####### ####### ####### ####### ###  #####  #     #
    17732076\section{Parallelism}
    1774 Historically, computer performance was about processor speeds and instruction counts. However, with heat dissipation being a direct consequence of speed increase, parallelism has become the new source for increased performance~\cite{Sutter05, Sutter05b}. In this decade, it is no longer reasonable to create a high-performance application without caring about parallelism. Indeed, parallelism is an important aspect of performance and more specifically throughput and hardware utilization. The lowest-level approach of parallelism is to use \textbf{kthread} in combination with semantics like \code{fork}, \code{join}, etc. However, since these have significant costs and limitations, \textbf{kthread} are now mostly used as an implementation tool rather than a user oriented one. There are several alternatives to solve these issues that all have strengths and weaknesses. While there are many variations of the presented paradigms, most of these variations do not actually change the guarantees or the semantics, they simply move costs in order to achieve better performance for certain workloads.
     2077Historically, computer performance was about processor speeds and instruction counts.
     2078However, with heat dissipation being a direct consequence of speed increase, parallelism has become the new source for increased performance~\cite{Sutter05, Sutter05b}.
     2079In this decade, it is no longer reasonable to create a high-performance application without caring about parallelism.
     2080Indeed, parallelism is an important aspect of performance and more specifically throughput and hardware utilization.
     2081The lowest-level approach of parallelism is to use \textbf{kthread} in combination with semantics like \code{fork}, \code{join}, etc.
     2082However, since these have significant costs and limitations, \textbf{kthread} are now mostly used as an implementation tool rather than a user oriented one.
     2083There are several alternatives to solve these issues that all have strengths and weaknesses.
     2084While there are many variations of the presented paradigms, most of these variations do not actually change the guarantees or the semantics, they simply move costs in order to achieve better performance for certain workloads.
    17752085
    17762086\section{Paradigms}
    17772087\subsection{User-Level Threads}
    1778 A direct improvement on the \textbf{kthread} approach is to use \textbf{uthread}. These threads offer most of the same features that the operating system already provides but can be used on a much larger scale. This approach is the most powerful solution as it allows all the features of multithreading, while removing several of the more expensive costs of kernel threads. The downside is that almost none of the low-level threading problems are hidden; users still have to think about data races, deadlocks and synchronization issues. These issues can be somewhat alleviated by a concurrency toolkit with strong guarantees, but the parallelism toolkit offers very little to reduce complexity in itself.
     2088A direct improvement on the \textbf{kthread} approach is to use \textbf{uthread}.
     2089These threads offer most of the same features that the operating system already provides but can be used on a much larger scale.
     2090This approach is the most powerful solution as it allows all the features of multithreading, while removing several of the more expensive costs of kernel threads.
     2091The downside is that almost none of the low-level threading problems are hidden; users still have to think about data races, deadlocks and synchronization issues.
     2092These issues can be somewhat alleviated by a concurrency toolkit with strong guarantees, but the parallelism toolkit offers very little to reduce complexity in itself.
    17792093
    17802094Examples of languages that support \textbf{uthread} are Erlang~\cite{Erlang} and \uC~\cite{uC++book}.
    17812095
    17822096\subsection{Fibers : User-Level Threads Without Preemption} \label{fibers}
    1783 A popular variant of \textbf{uthread} is what is often referred to as \textbf{fiber}. However, \textbf{fiber} do not present meaningful semantic differences with \textbf{uthread}. The significant difference between \textbf{uthread} and \textbf{fiber} is the lack of \textbf{preemption} in the latter. Advocates of \textbf{fiber} list their high performance and ease of implementation as major strengths, but the performance difference between \textbf{uthread} and \textbf{fiber} is controversial, and the ease of implementation, while true, is a weak argument in the context of language design. Therefore this proposal largely ignores fibers.
     2097A popular variant of \textbf{uthread} is what is often referred to as \textbf{fiber}.
     2098However, \textbf{fiber} do not present meaningful semantic differences with \textbf{uthread}.
     2099The significant difference between \textbf{uthread} and \textbf{fiber} is the lack of \textbf{preemption} in the latter.
     2100Advocates of \textbf{fiber} list their high performance and ease of implementation as major strengths, but the performance difference between \textbf{uthread} and \textbf{fiber} is controversial, and the ease of implementation, while true, is a weak argument in the context of language design.
     2101Therefore this proposal largely ignores fibers.
    17842102
    17852103An example of a language that uses fibers is Go~\cite{Go}
    17862104
    17872105\subsection{Jobs and Thread Pools}
    1788 An approach on the opposite end of the spectrum is to base parallelism on \textbf{pool}. Indeed, \textbf{pool} offer limited flexibility but at the benefit of a simpler user interface. In \textbf{pool} based systems, users express parallelism as units of work, called jobs, and a dependency graph (either explicit or implicit) that ties them together. This approach means users need not worry about concurrency but significantly limit the interaction that can occur among jobs. Indeed, any \textbf{job} that blocks also block the underlying worker, which effectively means the CPU utilization, and therefore throughput, suffers noticeably. It can be argued that a solution to this problem is to use more workers than available cores. However, unless the number of jobs and the number of workers are comparable, having a significant number of blocked jobs always results in idles cores.
     2106An approach on the opposite end of the spectrum is to base parallelism on \textbf{pool}.
     2107Indeed, \textbf{pool} offer limited flexibility but at the benefit of a simpler user interface.
     2108In \textbf{pool} based systems, users express parallelism as units of work, called jobs, and a dependency graph (either explicit or implicit) that ties them together.
     2109This approach means users need not worry about concurrency but significantly limit the interaction that can occur among jobs.
     2110Indeed, any \textbf{job} that blocks also block the underlying worker, which effectively means the CPU utilization, and therefore throughput, suffers noticeably.
     2111It can be argued that a solution to this problem is to use more workers than available cores.
     2112However, unless the number of jobs and the number of workers are comparable, having a significant number of blocked jobs always results in idles cores.
    17892113
    17902114The gold standard of this implementation is Intel's TBB library~\cite{TBB}.
    17912115
    17922116\subsection{Paradigm Performance}
    1793 While the choice between the three paradigms listed above may have significant performance implications, it is difficult to pin down the performance implications of choosing a model at the language level. Indeed, in many situations one of these paradigms may show better performance but it all strongly depends on the workload. Having a large amount of mostly independent units of work to execute almost guarantees equivalent performance across paradigms and that the \textbf{pool}-based system has the best efficiency thanks to the lower memory overhead (i.e., no thread stack per job). However, interactions among jobs can easily exacerbate contention. User-level threads allow fine-grain context switching, which results in better resource utilization, but a context switch is more expensive and the extra control means users need to tweak more variables to get the desired performance. Finally, if the units of uninterrupted work are large, enough the paradigm choice is largely amortized by the actual work done.
     2117While the choice between the three paradigms listed above may have significant performance implications, it is difficult to pin down the performance implications of choosing a model at the language level.
     2118Indeed, in many situations one of these paradigms may show better performance but it all strongly depends on the workload.
     2119Having a large amount of mostly independent units of work to execute almost guarantees equivalent performance across paradigms and that the \textbf{pool}-based system has the best efficiency thanks to the lower memory overhead (i.e., no thread stack per job).
     2120However, interactions among jobs can easily exacerbate contention.
     2121User-level threads allow fine-grain context switching, which results in better resource utilization, but a context switch is more expensive and the extra control means users need to tweak more variables to get the desired performance.
     2122Finally, if the units of uninterrupted work are large, enough the paradigm choice is largely amortized by the actual work done.
    17942123
    17952124\section{The \protect\CFA\ Kernel : Processors, Clusters and Threads}\label{kernel}
    1796 A \textbf{cfacluster} is a group of \textbf{kthread} executed in isolation. \textbf{uthread} are scheduled on the \textbf{kthread} of a given \textbf{cfacluster}, allowing organization between \textbf{uthread} and \textbf{kthread}. It is important that \textbf{kthread} belonging to a same \textbf{cfacluster} have homogeneous settings, otherwise migrating a \textbf{uthread} from one \textbf{kthread} to the other can cause issues. A \textbf{cfacluster} also offers a pluggable scheduler that can optimize the workload generated by the \textbf{uthread}.
    1797 
    1798 \textbf{cfacluster} have not been fully implemented in the context of this paper. Currently \CFA only supports one \textbf{cfacluster}, the initial one.
     2125A \textbf{cfacluster} is a group of \textbf{kthread} executed in isolation. \textbf{uthread} are scheduled on the \textbf{kthread} of a given \textbf{cfacluster}, allowing organization between \textbf{uthread} and \textbf{kthread}.
     2126It is important that \textbf{kthread} belonging to a same \textbf{cfacluster} have homogeneous settings, otherwise migrating a \textbf{uthread} from one \textbf{kthread} to the other can cause issues.
     2127A \textbf{cfacluster} also offers a pluggable scheduler that can optimize the workload generated by the \textbf{uthread}.
     2128
     2129\textbf{cfacluster} have not been fully implemented in the context of this paper.
     2130Currently \CFA only supports one \textbf{cfacluster}, the initial one.
    17992131
    18002132\subsection{Future Work: Machine Setup}\label{machine}
    1801 While this was not done in the context of this paper, another important aspect of clusters is affinity. While many common desktop and laptop PCs have homogeneous CPUs, other devices often have more heterogeneous setups. For example, a system using \textbf{numa} configurations may benefit from users being able to tie clusters and/or kernel threads to certain CPU cores. OS support for CPU affinity is now common~\cite{affinityLinux, affinityWindows, affinityFreebsd, affinityNetbsd, affinityMacosx}, which means it is both possible and desirable for \CFA to offer an abstraction mechanism for portable CPU affinity.
     2133While this was not done in the context of this paper, another important aspect of clusters is affinity.
     2134While many common desktop and laptop PCs have homogeneous CPUs, other devices often have more heterogeneous setups.
     2135For example, a system using \textbf{numa} configurations may benefit from users being able to tie clusters and/or kernel threads to certain CPU cores.
     2136OS support for CPU affinity is now common~\cite{affinityLinux, affinityWindows, affinityFreebsd, affinityNetbsd, affinityMacosx}, which means it is both possible and desirable for \CFA to offer an abstraction mechanism for portable CPU affinity.
    18022137
    18032138\subsection{Paradigms}\label{cfaparadigms}
    1804 Given these building blocks, it is possible to reproduce all three of the popular paradigms. Indeed, \textbf{uthread} is the default paradigm in \CFA. However, disabling \textbf{preemption} on the \textbf{cfacluster} means \textbf{cfathread} effectively become \textbf{fiber}. Since several \textbf{cfacluster} with different scheduling policy can coexist in the same application, this allows \textbf{fiber} and \textbf{uthread} to coexist in the runtime of an application. Finally, it is possible to build executors for thread pools from \textbf{uthread} or \textbf{fiber}, which includes specialized jobs like actors~\cite{Actors}.
     2139Given these building blocks, it is possible to reproduce all three of the popular paradigms.
     2140Indeed, \textbf{uthread} is the default paradigm in \CFA.
     2141However, disabling \textbf{preemption} on the \textbf{cfacluster} means \textbf{cfathread} effectively become \textbf{fiber}.
     2142Since several \textbf{cfacluster} with different scheduling policy can coexist in the same application, this allows \textbf{fiber} and \textbf{uthread} to coexist in the runtime of an application.
     2143Finally, it is possible to build executors for thread pools from \textbf{uthread} or \textbf{fiber}, which includes specialized jobs like actors~\cite{Actors}.
    18052144
    18062145
    18072146
    18082147\section{Behind the Scenes}
    1809 There are several challenges specific to \CFA when implementing concurrency. These challenges are a direct result of \textbf{bulk-acq} and loose object definitions. These two constraints are the root cause of most design decisions in the implementation. Furthermore, to avoid contention from dynamically allocating memory in a concurrent environment, the internal-scheduling design is (almost) entirely free of mallocs. This approach avoids the chicken and egg problem~\cite{Chicken} of having a memory allocator that relies on the threading system and a threading system that relies on the runtime. This extra goal means that memory management is a constant concern in the design of the system.
    1810 
    1811 The main memory concern for concurrency is queues. All blocking operations are made by parking threads onto queues and all queues are designed with intrusive nodes, where each node has pre-allocated link fields for chaining, to avoid the need for memory allocation. Since several concurrency operations can use an unbound amount of memory (depending on \textbf{bulk-acq}), statically defining information in the intrusive fields of threads is insufficient.The only way to use a variable amount of memory without requiring memory allocation is to pre-allocate large buffers of memory eagerly and store the information in these buffers. Conveniently, the call stack fits that description and is easy to use, which is why it is used heavily in the implementation of internal scheduling, particularly variable-length arrays. Since stack allocation is based on scopes, the first step of the implementation is to identify the scopes that are available to store the information, and which of these can have a variable-length array. The threads and the condition both have a fixed amount of memory, while \code{mutex} routines and blocking calls allow for an unbound amount, within the stack size.
     2148There are several challenges specific to \CFA when implementing concurrency.
     2149These challenges are a direct result of \textbf{bulk-acq} and loose object definitions.
     2150These two constraints are the root cause of most design decisions in the implementation.
     2151Furthermore, to avoid contention from dynamically allocating memory in a concurrent environment, the internal-scheduling design is (almost) entirely free of mallocs.
     2152This approach avoids the chicken and egg problem~\cite{Chicken} of having a memory allocator that relies on the threading system and a threading system that relies on the runtime.
     2153This extra goal means that memory management is a constant concern in the design of the system.
     2154
     2155The main memory concern for concurrency is queues.
     2156All blocking operations are made by parking threads onto queues and all queues are designed with intrusive nodes, where each node has pre-allocated link fields for chaining, to avoid the need for memory allocation.
     2157Since several concurrency operations can use an unbound amount of memory (depending on \textbf{bulk-acq}), statically defining information in the intrusive fields of threads is insufficient.The only way to use a variable amount of memory without requiring memory allocation is to pre-allocate large buffers of memory eagerly and store the information in these buffers.
     2158Conveniently, the call stack fits that description and is easy to use, which is why it is used heavily in the implementation of internal scheduling, particularly variable-length arrays.
     2159Since stack allocation is based on scopes, the first step of the implementation is to identify the scopes that are available to store the information, and which of these can have a variable-length array.
     2160The threads and the condition both have a fixed amount of memory, while \code{mutex} routines and blocking calls allow for an unbound amount, within the stack size.
    18122161
    18132162Note that since the major contributions of this paper are extending monitor semantics to \textbf{bulk-acq} and loose object definitions, any challenges that are not resulting of these characteristics of \CFA are considered as solved problems and therefore not discussed.
     
    18192168% ======================================================================
    18202169
    1821 The first step towards the monitor implementation is simple \code{mutex} routines. In the single monitor case, mutual-exclusion is done using the entry/exit procedure in listing \ref{lst:entry1}. The entry/exit procedures do not have to be extended to support multiple monitors. Indeed it is sufficient to enter/leave monitors one-by-one as long as the order is correct to prevent deadlock~\cite{Havender68}. In \CFA, ordering of monitor acquisition relies on memory ordering. This approach is sufficient because all objects are guaranteed to have distinct non-overlapping memory layouts and mutual-exclusion for a monitor is only defined for its lifetime, meaning that destroying a monitor while it is acquired is undefined behaviour. When a mutex call is made, the concerned monitors are aggregated into a variable-length pointer array and sorted based on pointer values. This array persists for the entire duration of the mutual-exclusion and its ordering reused extensively.
     2170The first step towards the monitor implementation is simple \code{mutex} routines.
     2171In the single monitor case, mutual-exclusion is done using the entry/exit procedure in listing \ref{lst:entry1}.
     2172The entry/exit procedures do not have to be extended to support multiple monitors.
     2173Indeed it is sufficient to enter/leave monitors one-by-one as long as the order is correct to prevent deadlock~\cite{Havender68}.
     2174In \CFA, ordering of monitor acquisition relies on memory ordering.
     2175This approach is sufficient because all objects are guaranteed to have distinct non-overlapping memory layouts and mutual-exclusion for a monitor is only defined for its lifetime, meaning that destroying a monitor while it is acquired is undefined behaviour.
     2176When a mutex call is made, the concerned monitors are aggregated into a variable-length pointer array and sorted based on pointer values.
     2177This array persists for the entire duration of the mutual-exclusion and its ordering reused extensively.
    18222178\begin{figure}
    18232179\begin{multicols}{2}
     
    18462202
    18472203\subsection{Details: Interaction with polymorphism}
    1848 Depending on the choice of semantics for when monitor locks are acquired, interaction between monitors and \CFA's concept of polymorphism can be more complex to support. However, it is shown that entry-point locking solves most of the issues.
    1849 
    1850 First of all, interaction between \code{otype} polymorphism (see Section~\ref{s:ParametricPolymorphism}) and monitors is impossible since monitors do not support copying. Therefore, the main question is how to support \code{dtype} polymorphism. It is important to present the difference between the two acquiring options: \textbf{callsite-locking} and entry-point locking, i.e., acquiring the monitors before making a mutex routine-call or as the first operation of the mutex routine-call. For example:
     2204Depending on the choice of semantics for when monitor locks are acquired, interaction between monitors and \CFA's concept of polymorphism can be more complex to support.
     2205However, it is shown that entry-point locking solves most of the issues.
     2206
     2207First of all, interaction between \code{otype} polymorphism (see Section~\ref{s:ParametricPolymorphism}) and monitors is impossible since monitors do not support copying.
     2208Therefore, the main question is how to support \code{dtype} polymorphism.
     2209It is important to present the difference between the two acquiring options: \textbf{callsite-locking} and entry-point locking, i.e., acquiring the monitors before making a mutex routine-call or as the first operation of the mutex routine-call.
     2210For example:
    18512211\begin{table}[H]
    18522212\begin{center}
     
    19152275\end{cfacode}
    19162276
    1917 Both entry point and \textbf{callsite-locking} are feasible implementations. The current \CFA implementation uses entry-point locking because it requires less work when using \textbf{raii}, effectively transferring the burden of implementation to object construction/destruction. It is harder to use \textbf{raii} for call-site locking, as it does not necessarily have an existing scope that matches exactly the scope of the mutual exclusion, i.e., the function body. For example, the monitor call can appear in the middle of an expression. Furthermore, entry-point locking requires less code generation since any useful routine is called multiple times but there is only one entry point for many call sites.
     2277Both entry point and \textbf{callsite-locking} are feasible implementations.
     2278The current \CFA implementation uses entry-point locking because it requires less work when using \textbf{raii}, effectively transferring the burden of implementation to object construction/destruction.
     2279It is harder to use \textbf{raii} for call-site locking, as it does not necessarily have an existing scope that matches exactly the scope of the mutual exclusion, i.e., the function body.
     2280For example, the monitor call can appear in the middle of an expression.
     2281Furthermore, entry-point locking requires less code generation since any useful routine is called multiple times but there is only one entry point for many call sites.
    19182282
    19192283% ======================================================================
     
    19232287% ======================================================================
    19242288
    1925 Figure \ref{fig:system1} shows a high-level picture if the \CFA runtime system in regards to concurrency. Each component of the picture is explained in detail in the flowing sections.
     2289Figure \ref{fig:system1} shows a high-level picture if the \CFA runtime system in regards to concurrency.
     2290Each component of the picture is explained in detail in the flowing sections.
    19262291
    19272292\begin{figure}
     
    19342299
    19352300\subsection{Processors}
    1936 Parallelism in \CFA is built around using processors to specify how much parallelism is desired. \CFA processors are object wrappers around kernel threads, specifically \texttt{pthread}s in the current implementation of \CFA. Indeed, any parallelism must go through operating-system libraries. However, \textbf{uthread} are still the main source of concurrency, processors are simply the underlying source of parallelism. Indeed, processor \textbf{kthread} simply fetch a \textbf{uthread} from the scheduler and run it; they are effectively executers for user-threads. The main benefit of this approach is that it offers a well-defined boundary between kernel code and user code, for example, kernel thread quiescing, scheduling and interrupt handling. Processors internally use coroutines to take advantage of the existing context-switching semantics.
     2301Parallelism in \CFA is built around using processors to specify how much parallelism is desired. \CFA processors are object wrappers around kernel threads, specifically \texttt{pthread}s in the current implementation of \CFA.
     2302Indeed, any parallelism must go through operating-system libraries.
     2303However, \textbf{uthread} are still the main source of concurrency, processors are simply the underlying source of parallelism.
     2304Indeed, processor \textbf{kthread} simply fetch a \textbf{uthread} from the scheduler and run it; they are effectively executers for user-threads.
     2305The main benefit of this approach is that it offers a well-defined boundary between kernel code and user code, for example, kernel thread quiescing, scheduling and interrupt handling.
     2306Processors internally use coroutines to take advantage of the existing context-switching semantics.
    19372307
    19382308\subsection{Stack Management}
    1939 One of the challenges of this system is to reduce the footprint as much as possible. Specifically, all \texttt{pthread}s created also have a stack created with them, which should be used as much as possible. Normally, coroutines also create their own stack to run on, however, in the case of the coroutines used for processors, these coroutines run directly on the \textbf{kthread} stack, effectively stealing the processor stack. The exception to this rule is the Main Processor, i.e., the initial \textbf{kthread} that is given to any program. In order to respect C user expectations, the stack of the initial kernel thread, the main stack of the program, is used by the main user thread rather than the main processor, which can grow very large.
     2309One of the challenges of this system is to reduce the footprint as much as possible.
     2310Specifically, all \texttt{pthread}s created also have a stack created with them, which should be used as much as possible.
     2311Normally, coroutines also create their own stack to run on, however, in the case of the coroutines used for processors, these coroutines run directly on the \textbf{kthread} stack, effectively stealing the processor stack.
     2312The exception to this rule is the Main Processor, i.e., the initial \textbf{kthread} that is given to any program.
     2313In order to respect C user expectations, the stack of the initial kernel thread, the main stack of the program, is used by the main user thread rather than the main processor, which can grow very large.
    19402314
    19412315\subsection{Context Switching}
    1942 As mentioned in section \ref{coroutine}, coroutines are a stepping stone for implementing threading, because they share the same mechanism for context-switching between different stacks. To improve performance and simplicity, context-switching is implemented using the following assumption: all context-switches happen inside a specific function call. This assumption means that the context-switch only has to copy the callee-saved registers onto the stack and then switch the stack registers with the ones of the target coroutine/thread. Note that the instruction pointer can be left untouched since the context-switch is always inside the same function. Threads, however, do not context-switch between each other directly. They context-switch to the scheduler. This method is called a 2-step context-switch and has the advantage of having a clear distinction between user code and the kernel where scheduling and other system operations happen. Obviously, this doubles the context-switch cost because threads must context-switch to an intermediate stack. The alternative 1-step context-switch uses the stack of the ``from'' thread to schedule and then context-switches directly to the ``to'' thread. However, the performance of the 2-step context-switch is still superior to a \code{pthread_yield} (see section \ref{results}). Additionally, for users in need for optimal performance, it is important to note that having a 2-step context-switch as the default does not prevent \CFA from offering a 1-step context-switch (akin to the Microsoft \code{SwitchToFiber}~\cite{switchToWindows} routine). This option is not currently present in \CFA, but the changes required to add it are strictly additive.
     2316As mentioned in section \ref{coroutine}, coroutines are a stepping stone for implementing threading, because they share the same mechanism for context-switching between different stacks.
     2317To improve performance and simplicity, context-switching is implemented using the following assumption: all context-switches happen inside a specific function call.
     2318This assumption means that the context-switch only has to copy the callee-saved registers onto the stack and then switch the stack registers with the ones of the target coroutine/thread.
     2319Note that the instruction pointer can be left untouched since the context-switch is always inside the same function.
     2320Threads, however, do not context-switch between each other directly.
     2321They context-switch to the scheduler.
     2322This method is called a 2-step context-switch and has the advantage of having a clear distinction between user code and the kernel where scheduling and other system operations happen.
     2323Obviously, this doubles the context-switch cost because threads must context-switch to an intermediate stack.
     2324The alternative 1-step context-switch uses the stack of the ``from'' thread to schedule and then context-switches directly to the ``to'' thread.
     2325However, the performance of the 2-step context-switch is still superior to a \code{pthread_yield} (see section \ref{results}).
     2326Additionally, for users in need for optimal performance, it is important to note that having a 2-step context-switch as the default does not prevent \CFA from offering a 1-step context-switch (akin to the Microsoft \code{SwitchToFiber}~\cite{switchToWindows} routine).
     2327This option is not currently present in \CFA, but the changes required to add it are strictly additive.
    19432328
    19442329\subsection{Preemption} \label{preemption}
    1945 Finally, an important aspect for any complete threading system is preemption. As mentioned in section \ref{basics}, preemption introduces an extra degree of uncertainty, which enables users to have multiple threads interleave transparently, rather than having to cooperate among threads for proper scheduling and CPU distribution. Indeed, preemption is desirable because it adds a degree of isolation among threads. In a fully cooperative system, any thread that runs a long loop can starve other threads, while in a preemptive system, starvation can still occur but it does not rely on every thread having to yield or block on a regular basis, which reduces significantly a programmer burden. Obviously, preemption is not optimal for every workload. However any preemptive system can become a cooperative system by making the time slices extremely large. Therefore, \CFA uses a preemptive threading system.
    1946 
    1947 Preemption in \CFA\footnote{Note that the implementation of preemption is strongly tied with the underlying threading system. For this reason, only the Linux implementation is cover, \CFA does not run on Windows at the time of writting} is based on kernel timers, which are used to run a discrete-event simulation. Every processor keeps track of the current time and registers an expiration time with the preemption system. When the preemption system receives a change in preemption, it inserts the time in a sorted order and sets a kernel timer for the closest one, effectively stepping through preemption events on each signal sent by the timer. These timers use the Linux signal {\tt SIGALRM}, which is delivered to the process rather than the kernel-thread. This results in an implementation problem, because when delivering signals to a process, the kernel can deliver the signal to any kernel thread for which the signal is not blocked, i.e.:
     2330Finally, an important aspect for any complete threading system is preemption.
     2331As mentioned in section \ref{basics}, preemption introduces an extra degree of uncertainty, which enables users to have multiple threads interleave transparently, rather than having to cooperate among threads for proper scheduling and CPU distribution.
     2332Indeed, preemption is desirable because it adds a degree of isolation among threads.
     2333In a fully cooperative system, any thread that runs a long loop can starve other threads, while in a preemptive system, starvation can still occur but it does not rely on every thread having to yield or block on a regular basis, which reduces significantly a programmer burden.
     2334Obviously, preemption is not optimal for every workload.
     2335However any preemptive system can become a cooperative system by making the time slices extremely large.
     2336Therefore, \CFA uses a preemptive threading system.
     2337
     2338Preemption in \CFA\footnote{Note that the implementation of preemption is strongly tied with the underlying threading system.
     2339For this reason, only the Linux implementation is cover, \CFA does not run on Windows at the time of writting} is based on kernel timers, which are used to run a discrete-event simulation.
     2340Every processor keeps track of the current time and registers an expiration time with the preemption system.
     2341When the preemption system receives a change in preemption, it inserts the time in a sorted order and sets a kernel timer for the closest one, effectively stepping through preemption events on each signal sent by the timer.
     2342These timers use the Linux signal {\tt SIGALRM}, which is delivered to the process rather than the kernel-thread.
     2343This results in an implementation problem, because when delivering signals to a process, the kernel can deliver the signal to any kernel thread for which the signal is not blocked, i.e.:
    19482344\begin{quote}
    1949 A process-directed signal may be delivered to any one of the threads that does not currently have the signal blocked. If more than one of the threads has the signal unblocked, then the kernel chooses an arbitrary thread to which to deliver the signal.
     2345A process-directed signal may be delivered to any one of the threads that does not currently have the signal blocked.
     2346If more than one of the threads has the signal unblocked, then the kernel chooses an arbitrary thread to which to deliver the signal.
    19502347SIGNAL(7) - Linux Programmer's Manual
    19512348\end{quote}
    19522349For the sake of simplicity, and in order to prevent the case of having two threads receiving alarms simultaneously, \CFA programs block the {\tt SIGALRM} signal on every kernel thread except one.
    19532350
    1954 Now because of how involuntary context-switches are handled, the kernel thread handling {\tt SIGALRM} cannot also be a processor thread. Hence, involuntary context-switching is done by sending signal {\tt SIGUSR1} to the corresponding proces\-sor and having the thread yield from inside the signal handler. This approach effectively context-switches away from the signal handler back to the kernel and the signal handler frame is eventually unwound when the thread is scheduled again. As a result, a signal handler can start on one kernel thread and terminate on a second kernel thread (but the same user thread). It is important to note that signal handlers save and restore signal masks because user-thread migration can cause a signal mask to migrate from one kernel thread to another. This behaviour is only a problem if all kernel threads, among which a user thread can migrate, differ in terms of signal masks\footnote{Sadly, official POSIX documentation is silent on what distinguishes ``async-signal-safe'' functions from other functions.}. However, since the kernel thread handling preemption requires a different signal mask, executing user threads on the kernel-alarm thread can cause deadlocks. For this reason, the alarm thread is in a tight loop around a system call to \code{sigwaitinfo}, requiring very little CPU time for preemption. One final detail about the alarm thread is how to wake it when additional communication is required (e.g., on thread termination). This unblocking is also done using {\tt SIGALRM}, but sent through the \code{pthread_sigqueue}. Indeed, \code{sigwait} can differentiate signals sent from \code{pthread_sigqueue} from signals sent from alarms or the kernel.
     2351Now because of how involuntary context-switches are handled, the kernel thread handling {\tt SIGALRM} cannot also be a processor thread.
     2352Hence, involuntary context-switching is done by sending signal {\tt SIGUSR1} to the corresponding proces\-sor and having the thread yield from inside the signal handler.
     2353This approach effectively context-switches away from the signal handler back to the kernel and the signal handler frame is eventually unwound when the thread is scheduled again.
     2354As a result, a signal handler can start on one kernel thread and terminate on a second kernel thread (but the same user thread).
     2355It is important to note that signal handlers save and restore signal masks because user-thread migration can cause a signal mask to migrate from one kernel thread to another.
     2356This behaviour is only a problem if all kernel threads, among which a user thread can migrate, differ in terms of signal masks\footnote{Sadly, official POSIX documentation is silent on what distinguishes ``async-signal-safe'' functions from other functions.}.
     2357However, since the kernel thread handling preemption requires a different signal mask, executing user threads on the kernel-alarm thread can cause deadlocks.
     2358For this reason, the alarm thread is in a tight loop around a system call to \code{sigwaitinfo}, requiring very little CPU time for preemption.
     2359One final detail about the alarm thread is how to wake it when additional communication is required (e.g., on thread termination).
     2360This unblocking is also done using {\tt SIGALRM}, but sent through the \code{pthread_sigqueue}.
     2361Indeed, \code{sigwait} can differentiate signals sent from \code{pthread_sigqueue} from signals sent from alarms or the kernel.
    19552362
    19562363\subsection{Scheduler}
    1957 Finally, an aspect that was not mentioned yet is the scheduling algorithm. Currently, the \CFA scheduler uses a single ready queue for all processors, which is the simplest approach to scheduling. Further discussion on scheduling is present in section \ref{futur:sched}.
     2364Finally, an aspect that was not mentioned yet is the scheduling algorithm.
     2365Currently, the \CFA scheduler uses a single ready queue for all processors, which is the simplest approach to scheduling.
     2366Further discussion on scheduling is present in section \ref{futur:sched}.
    19582367
    19592368% ======================================================================
     
    19712380\end{figure}
    19722381
    1973 This picture has several components, the two most important being the entry queue and the AS-stack. The entry queue is an (almost) FIFO list where threads waiting to enter are parked, while the acceptor/signaller (AS) stack is a FILO list used for threads that have been signalled or otherwise marked as running next.
    1974 
    1975 For \CFA, this picture does not have support for blocking multiple monitors on a single condition. To support \textbf{bulk-acq} two changes to this picture are required. First, it is no longer helpful to attach the condition to \emph{a single} monitor. Secondly, the thread waiting on the condition has to be separated across multiple monitors, seen in figure \ref{fig:monitor_cfa}.
     2382This picture has several components, the two most important being the entry queue and the AS-stack.
     2383The entry queue is an (almost) FIFO list where threads waiting to enter are parked, while the acceptor/signaller (AS) stack is a FILO list used for threads that have been signalled or otherwise marked as running next.
     2384
     2385For \CFA, this picture does not have support for blocking multiple monitors on a single condition.
     2386To support \textbf{bulk-acq} two changes to this picture are required.
     2387First, it is no longer helpful to attach the condition to \emph{a single} monitor.
     2388Secondly, the thread waiting on the condition has to be separated across multiple monitors, seen in figure \ref{fig:monitor_cfa}.
    19762389
    19772390\begin{figure}[H]
     
    19832396\end{figure}
    19842397
    1985 This picture and the proper entry and leave algorithms (see listing \ref{lst:entry2}) is the fundamental implementation of internal scheduling. Note that when a thread is moved from the condition to the AS-stack, it is conceptually split into N pieces, where N is the number of monitors specified in the parameter list. The thread is woken up when all the pieces have popped from the AS-stacks and made active. In this picture, the threads are split into halves but this is only because there are two monitors. For a specific signalling operation every monitor needs a piece of thread on its AS-stack.
     2398This picture and the proper entry and leave algorithms (see listing \ref{lst:entry2}) is the fundamental implementation of internal scheduling.
     2399Note that when a thread is moved from the condition to the AS-stack, it is conceptually split into N pieces, where N is the number of monitors specified in the parameter list.
     2400The thread is woken up when all the pieces have popped from the AS-stacks and made active.
     2401In this picture, the threads are split into halves but this is only because there are two monitors.
     2402For a specific signalling operation every monitor needs a piece of thread on its AS-stack.
    19862403
    19872404\begin{figure}[b]
     
    20162433\end{figure}
    20172434
    2018 The solution discussed in \ref{intsched} can be seen in the exit routine of listing \ref{lst:entry2}. Basically, the solution boils down to having a separate data structure for the condition queue and the AS-stack, and unconditionally transferring ownership of the monitors but only unblocking the thread when the last monitor has transferred ownership. This solution is deadlock safe as well as preventing any potential barging. The data structures used for the AS-stack are reused extensively for external scheduling, but in the case of internal scheduling, the data is allocated using variable-length arrays on the call stack of the \code{wait} and \code{signal_block} routines.
     2435The solution discussed in \ref{intsched} can be seen in the exit routine of listing \ref{lst:entry2}.
     2436Basically, the solution boils down to having a separate data structure for the condition queue and the AS-stack, and unconditionally transferring ownership of the monitors but only unblocking the thread when the last monitor has transferred ownership.
     2437This solution is deadlock safe as well as preventing any potential barging.
     2438The data structures used for the AS-stack are reused extensively for external scheduling, but in the case of internal scheduling, the data is allocated using variable-length arrays on the call stack of the \code{wait} and \code{signal_block} routines.
    20192439
    20202440\begin{figure}[H]
     
    20262446\end{figure}
    20272447
    2028 Figure \ref{fig:structs} shows a high-level representation of these data structures. The main idea behind them is that, a thread cannot contain an arbitrary number of intrusive ``next'' pointers for linking onto monitors. The \code{condition node} is the data structure that is queued onto a condition variable and, when signalled, the condition queue is popped and each \code{condition criterion} is moved to the AS-stack. Once all the criteria have been popped from their respective AS-stacks, the thread is woken up, which is what is shown in listing \ref{lst:entry2}.
     2448Figure \ref{fig:structs} shows a high-level representation of these data structures.
     2449The main idea behind them is that, a thread cannot contain an arbitrary number of intrusive ``next'' pointers for linking onto monitors.
     2450The \code{condition node} is the data structure that is queued onto a condition variable and, when signalled, the condition queue is popped and each \code{condition criterion} is moved to the AS-stack.
     2451Once all the criteria have been popped from their respective AS-stacks, the thread is woken up, which is what is shown in listing \ref{lst:entry2}.
    20292452
    20302453% ======================================================================
     
    20332456% ======================================================================
    20342457% ======================================================================
    2035 Similarly to internal scheduling, external scheduling for multiple monitors relies on the idea that waiting-thread queues are no longer specific to a single monitor, as mentioned in section \ref{extsched}. For internal scheduling, these queues are part of condition variables, which are still unique for a given scheduling operation (i.e., no signal statement uses multiple conditions). However, in the case of external scheduling, there is no equivalent object which is associated with \code{waitfor} statements. This absence means the queues holding the waiting threads must be stored inside at least one of the monitors that is acquired. These monitors being the only objects that have sufficient lifetime and are available on both sides of the \code{waitfor} statement. This requires an algorithm to choose which monitor holds the relevant queue. It is also important that said algorithm be independent of the order in which users list parameters. The proposed algorithm is to fall back on monitor lock ordering (sorting by address) and specify that the monitor that is acquired first is the one with the relevant waiting queue. This assumes that the lock acquiring order is static for the lifetime of all concerned objects but that is a reasonable constraint.
     2458Similarly to internal scheduling, external scheduling for multiple monitors relies on the idea that waiting-thread queues are no longer specific to a single monitor, as mentioned in section \ref{extsched}.
     2459For internal scheduling, these queues are part of condition variables, which are still unique for a given scheduling operation (i.e., no signal statement uses multiple conditions).
     2460However, in the case of external scheduling, there is no equivalent object which is associated with \code{waitfor} statements.
     2461This absence means the queues holding the waiting threads must be stored inside at least one of the monitors that is acquired.
     2462These monitors being the only objects that have sufficient lifetime and are available on both sides of the \code{waitfor} statement.
     2463This requires an algorithm to choose which monitor holds the relevant queue.
     2464It is also important that said algorithm be independent of the order in which users list parameters.
     2465The proposed algorithm is to fall back on monitor lock ordering (sorting by address) and specify that the monitor that is acquired first is the one with the relevant waiting queue.
     2466This assumes that the lock acquiring order is static for the lifetime of all concerned objects but that is a reasonable constraint.
    20362467
    20372468This algorithm choice has two consequences:
    20382469\begin{itemize}
    2039         \item The queue of the monitor with the lowest address is no longer a true FIFO queue because threads can be moved to the front of the queue. These queues need to contain a set of monitors for each of the waiting threads. Therefore, another thread whose set contains the same lowest address monitor but different lower priority monitors may arrive first but enter the critical section after a thread with the correct pairing.
    2040         \item The queue of the lowest priority monitor is both required and potentially unused. Indeed, since it is not known at compile time which monitor is the monitor which has the lowest address, every monitor needs to have the correct queues even though it is possible that some queues go unused for the entire duration of the program, for example if a monitor is only used in a specific pair.
     2470        \item The queue of the monitor with the lowest address is no longer a true FIFO queue because threads can be moved to the front of the queue.
     2471These queues need to contain a set of monitors for each of the waiting threads.
     2472Therefore, another thread whose set contains the same lowest address monitor but different lower priority monitors may arrive first but enter the critical section after a thread with the correct pairing.
     2473        \item The queue of the lowest priority monitor is both required and potentially unused.
     2474Indeed, since it is not known at compile time which monitor is the monitor which has the lowest address, every monitor needs to have the correct queues even though it is possible that some queues go unused for the entire duration of the program, for example if a monitor is only used in a specific pair.
    20412475\end{itemize}
    20422476Therefore, the following modifications need to be made to support external scheduling:
    20432477\begin{itemize}
    2044         \item The threads waiting on the entry queue need to keep track of which routine they are trying to enter, and using which set of monitors. The \code{mutex} routine already has all the required information on its stack, so the thread only needs to keep a pointer to that information.
    2045         \item The monitors need to keep a mask of acceptable routines. This mask contains for each acceptable routine, a routine pointer and an array of monitors to go with it. It also needs storage to keep track of which routine was accepted. Since this information is not specific to any monitor, the monitors actually contain a pointer to an integer on the stack of the waiting thread. Note that if a thread has acquired two monitors but executes a \code{waitfor} with only one monitor as a parameter, setting the mask of acceptable routines to both monitors will not cause any problems since the extra monitor will not change ownership regardless. This becomes relevant when \code{when} clauses affect the number of monitors passed to a \code{waitfor} statement.
     2478        \item The threads waiting on the entry queue need to keep track of which routine they are trying to enter, and using which set of monitors.
     2479The \code{mutex} routine already has all the required information on its stack, so the thread only needs to keep a pointer to that information.
     2480        \item The monitors need to keep a mask of acceptable routines.
     2481This mask contains for each acceptable routine, a routine pointer and an array of monitors to go with it.
     2482It also needs storage to keep track of which routine was accepted.
     2483Since this information is not specific to any monitor, the monitors actually contain a pointer to an integer on the stack of the waiting thread.
     2484Note that if a thread has acquired two monitors but executes a \code{waitfor} with only one monitor as a parameter, setting the mask of acceptable routines to both monitors will not cause any problems since the extra monitor will not change ownership regardless.
     2485This becomes relevant when \code{when} clauses affect the number of monitors passed to a \code{waitfor} statement.
    20462486        \item The entry/exit routines need to be updated as shown in listing \ref{lst:entry3}.
    20472487\end{itemize}
    20482488
    20492489\subsection{External Scheduling - Destructors}
    2050 Finally, to support the ordering inversion of destructors, the code generation needs to be modified to use a special entry routine. This routine is needed because of the storage requirements of the call order inversion. Indeed, when waiting for the destructors, storage is needed for the waiting context and the lifetime of said storage needs to outlive the waiting operation it is needed for. For regular \code{waitfor} statements, the call stack of the routine itself matches this requirement but it is no longer the case when waiting for the destructor since it is pushed on to the AS-stack for later. The \code{waitfor} semantics can then be adjusted correspondingly, as seen in listing \ref{lst:entry-dtor}
     2490Finally, to support the ordering inversion of destructors, the code generation needs to be modified to use a special entry routine.
     2491This routine is needed because of the storage requirements of the call order inversion.
     2492Indeed, when waiting for the destructors, storage is needed for the waiting context and the lifetime of said storage needs to outlive the waiting operation it is needed for.
     2493For regular \code{waitfor} statements, the call stack of the routine itself matches this requirement but it is no longer the case when waiting for the destructor since it is pushed on to the AS-stack for later.
     2494The \code{waitfor} semantics can then be adjusted correspondingly, as seen in listing \ref{lst:entry-dtor}
    20512495
    20522496\begin{figure}
     
    21412585
    21422586\section{Threads As Monitors}
    2143 As it was subtly alluded in section \ref{threads}, \code{thread}s in \CFA are in fact monitors, which means that all monitor features are available when using threads. For example, here is a very simple two thread pipeline that could be used for a simulator of a game engine:
     2587As it was subtly alluded in section \ref{threads}, \code{thread}s in \CFA are in fact monitors, which means that all monitor features are available when using threads.
     2588For example, here is a very simple two thread pipeline that could be used for a simulator of a game engine:
    21442589\begin{figure}[H]
    21452590\begin{cfacode}[caption={Toy simulator using \code{thread}s and \code{monitor}s.},label={lst:engine-v1}]
     
    21722617\end{cfacode}
    21732618\end{figure}
    2174 One of the obvious complaints of the previous code snippet (other than its toy-like simplicity) is that it does not handle exit conditions and just goes on forever. Luckily, the monitor semantics can also be used to clearly enforce a shutdown order in a concise manner:
     2619One of the obvious complaints of the previous code snippet (other than its toy-like simplicity) is that it does not handle exit conditions and just goes on forever.
     2620Luckily, the monitor semantics can also be used to clearly enforce a shutdown order in a concise manner:
    21752621\begin{figure}[H]
    21762622\begin{cfacode}[caption={Same toy simulator with proper termination condition.},label={lst:engine-v2}]
     
    22162662
    22172663\section{Fibers \& Threads}
    2218 As mentioned in section \ref{preemption}, \CFA uses preemptive threads by default but can use fibers on demand. Currently, using fibers is done by adding the following line of code to the program~:
     2664As mentioned in section \ref{preemption}, \CFA uses preemptive threads by default but can use fibers on demand.
     2665Currently, using fibers is done by adding the following line of code to the program~:
    22192666\begin{cfacode}
    22202667unsigned int default_preemption() {
     
    22222669}
    22232670\end{cfacode}
    2224 This function is called by the kernel to fetch the default preemption rate, where 0 signifies an infinite time-slice, i.e., no preemption. However, once clusters are fully implemented, it will be possible to create fibers and \textbf{uthread} in the same system, as in listing \ref{lst:fiber-uthread}
     2671This function is called by the kernel to fetch the default preemption rate, where 0 signifies an infinite time-slice, i.e., no preemption.
     2672However, once clusters are fully implemented, it will be possible to create fibers and \textbf{uthread} in the same system, as in listing \ref{lst:fiber-uthread}
    22252673\begin{figure}
    22262674\begin{cfacode}[caption={Using fibers and \textbf{uthread} side-by-side in \CFA},label={lst:fiber-uthread}]
     
    22812729% ======================================================================
    22822730\section{Machine Setup}
    2283 Table \ref{tab:machine} shows the characteristics of the machine used to run the benchmarks. All tests were made on this machine.
     2731Table \ref{tab:machine} shows the characteristics of the machine used to run the benchmarks.
     2732All tests were made on this machine.
    22842733\begin{table}[H]
    22852734\begin{center}
     
    23142763
    23152764\section{Micro Benchmarks}
    2316 All benchmarks are run using the same harness to produce the results, seen as the \code{BENCH()} macro in the following examples. This macro uses the following logic to benchmark the code:
     2765All benchmarks are run using the same harness to produce the results, seen as the \code{BENCH()} macro in the following examples.
     2766This macro uses the following logic to benchmark the code:
    23172767\begin{pseudo}
    23182768#define BENCH(run, result) \
     
    23222772        result = (after - before) / N;
    23232773\end{pseudo}
    2324 The method used to get time is \code{clock_gettime(CLOCK_THREAD_CPUTIME_ID);}. Each benchmark is using many iterations of a simple call to measure the cost of the call. The specific number of iterations depends on the specific benchmark.
     2774The method used to get time is \code{clock_gettime(CLOCK_THREAD_CPUTIME_ID);}.
     2775Each benchmark is using many iterations of a simple call to measure the cost of the call.
     2776The specific number of iterations depends on the specific benchmark.
    23252777
    23262778\subsection{Context-Switching}
    2327 The first interesting benchmark is to measure how long context-switches take. The simplest approach to do this is to yield on a thread, which executes a 2-step context switch. Yielding causes the thread to context-switch to the scheduler and back, more precisely: from the \textbf{uthread} to the \textbf{kthread} then from the \textbf{kthread} back to the same \textbf{uthread} (or a different one in the general case). In order to make the comparison fair, coroutines also execute a 2-step context-switch by resuming another coroutine which does nothing but suspending in a tight loop, which is a resume/suspend cycle instead of a yield. Listing \ref{lst:ctx-switch} shows the code for coroutines and threads with the results in table \ref{tab:ctx-switch}. All omitted tests are functionally identical to one of these tests. The difference between coroutines and threads can be attributed to the cost of scheduling.
     2779The first interesting benchmark is to measure how long context-switches take.
     2780The simplest approach to do this is to yield on a thread, which executes a 2-step context switch.
     2781Yielding causes the thread to context-switch to the scheduler and back, more precisely: from the \textbf{uthread} to the \textbf{kthread} then from the \textbf{kthread} back to the same \textbf{uthread} (or a different one in the general case).
     2782In order to make the comparison fair, coroutines also execute a 2-step context-switch by resuming another coroutine which does nothing but suspending in a tight loop, which is a resume/suspend cycle instead of a yield.
     2783Listing \ref{lst:ctx-switch} shows the code for coroutines and threads with the results in table \ref{tab:ctx-switch}.
     2784All omitted tests are functionally identical to one of these tests.
     2785The difference between coroutines and threads can be attributed to the cost of scheduling.
    23282786\begin{figure}
    23292787\begin{multicols}{2}
     
    23862844\end{tabular}
    23872845\end{center}
    2388 \caption{Context Switch comparison. All numbers are in nanoseconds(\si{\nano\second})}
     2846\caption{Context Switch comparison.
     2847All numbers are in nanoseconds(\si{\nano\second})}
    23892848\label{tab:ctx-switch}
    23902849\end{table}
    23912850
    23922851\subsection{Mutual-Exclusion}
    2393 The next interesting benchmark is to measure the overhead to enter/leave a critical-section. For monitors, the simplest approach is to measure how long it takes to enter and leave a monitor routine. Listing \ref{lst:mutex} shows the code for \CFA. To put the results in context, the cost of entering a non-inline function and the cost of acquiring and releasing a \code{pthread_mutex} lock is also measured. The results can be shown in table \ref{tab:mutex}.
     2852The next interesting benchmark is to measure the overhead to enter/leave a critical-section.
     2853For monitors, the simplest approach is to measure how long it takes to enter and leave a monitor routine.
     2854Listing \ref{lst:mutex} shows the code for \CFA.
     2855To put the results in context, the cost of entering a non-inline function and the cost of acquiring and releasing a \code{pthread_mutex} lock is also measured.
     2856The results can be shown in table \ref{tab:mutex}.
    23942857
    23952858\begin{figure}
     
    24282891\end{tabular}
    24292892\end{center}
    2430 \caption{Mutex routine comparison. All numbers are in nanoseconds(\si{\nano\second})}
     2893\caption{Mutex routine comparison.
     2894All numbers are in nanoseconds(\si{\nano\second})}
    24312895\label{tab:mutex}
    24322896\end{table}
    24332897
    24342898\subsection{Internal Scheduling}
    2435 The internal-scheduling benchmark measures the cost of waiting on and signalling a condition variable. Listing \ref{lst:int-sched} shows the code for \CFA, with results table \ref{tab:int-sched}. As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
     2899The internal-scheduling benchmark measures the cost of waiting on and signalling a condition variable.
     2900Listing \ref{lst:int-sched} shows the code for \CFA, with results table \ref{tab:int-sched}.
     2901As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
    24362902
    24372903\begin{figure}
     
    24842950\end{tabular}
    24852951\end{center}
    2486 \caption{Internal scheduling comparison. All numbers are in nanoseconds(\si{\nano\second})}
     2952\caption{Internal scheduling comparison.
     2953All numbers are in nanoseconds(\si{\nano\second})}
    24872954\label{tab:int-sched}
    24882955\end{table}
    24892956
    24902957\subsection{External Scheduling}
    2491 The Internal scheduling benchmark measures the cost of the \code{waitfor} statement (\code{_Accept} in \uC). Listing \ref{lst:ext-sched} shows the code for \CFA, with results in table \ref{tab:ext-sched}. As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
     2958The Internal scheduling benchmark measures the cost of the \code{waitfor} statement (\code{_Accept} in \uC).
     2959Listing \ref{lst:ext-sched} shows the code for \CFA, with results in table \ref{tab:ext-sched}.
     2960As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
    24922961
    24932962\begin{figure}
     
    25373006\end{tabular}
    25383007\end{center}
    2539 \caption{External scheduling comparison. All numbers are in nanoseconds(\si{\nano\second})}
     3008\caption{External scheduling comparison.
     3009All numbers are in nanoseconds(\si{\nano\second})}
    25403010\label{tab:ext-sched}
    25413011\end{table}
    25423012
    25433013\subsection{Object Creation}
    2544 Finally, the last benchmark measures the cost of creation for concurrent objects. Listing \ref{lst:creation} shows the code for \texttt{pthread}s and \CFA threads, with results shown in table \ref{tab:creation}. As with all other benchmarks, all omitted tests are functionally identical to one of these tests. The only note here is that the call stacks of \CFA coroutines are lazily created, therefore without priming the coroutine, the creation cost is very low.
     3014Finally, the last benchmark measures the cost of creation for concurrent objects.
     3015Listing \ref{lst:creation} shows the code for \texttt{pthread}s and \CFA threads, with results shown in table \ref{tab:creation}.
     3016As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
     3017The only note here is that the call stacks of \CFA coroutines are lazily created, therefore without priming the coroutine, the creation cost is very low.
    25453018
    25463019\begin{figure}
     
    26043077\end{tabular}
    26053078\end{center}
    2606 \caption{Creation comparison. All numbers are in nanoseconds(\si{\nano\second}).}
     3079\caption{Creation comparison.
     3080All numbers are in nanoseconds(\si{\nano\second}).}
    26073081\label{tab:creation}
    26083082\end{table}
     
    26113085
    26123086\section{Conclusion}
    2613 This paper has achieved a minimal concurrency \textbf{api} that is simple, efficient and usable as the basis for higher-level features. The approach presented is based on a lightweight thread-system for parallelism, which sits on top of clusters of processors. This M:N model is judged to be both more efficient and allow more flexibility for users. Furthermore, this document introduces monitors as the main concurrency tool for users. This paper also offers a novel approach allowing multiple monitors to be accessed simultaneously without running into the Nested Monitor Problem~\cite{Lister77}. It also offers a full implementation of the concurrency runtime written entirely in \CFA, effectively the largest \CFA code base to date.
     3087This paper has achieved a minimal concurrency \textbf{api} that is simple, efficient and usable as the basis for higher-level features.
     3088The approach presented is based on a lightweight thread-system for parallelism, which sits on top of clusters of processors.
     3089This M:N model is judged to be both more efficient and allow more flexibility for users.
     3090Furthermore, this document introduces monitors as the main concurrency tool for users.
     3091This paper also offers a novel approach allowing multiple monitors to be accessed simultaneously without running into the Nested Monitor Problem~\cite{Lister77}.
     3092It also offers a full implementation of the concurrency runtime written entirely in \CFA, effectively the largest \CFA code base to date.
    26143093
    26153094
     
    26213100
    26223101\subsection{Performance} \label{futur:perf}
    2623 This paper presents a first implementation of the \CFA concurrency runtime. Therefore, there is still significant work to improve performance. Many of the data structures and algorithms may change in the future to more efficient versions. For example, the number of monitors in a single \textbf{bulk-acq} is only bound by the stack size, this is probably unnecessarily generous. It may be possible that limiting the number helps increase performance. However, it is not obvious that the benefit would be significant.
     3102This paper presents a first implementation of the \CFA concurrency runtime.
     3103Therefore, there is still significant work to improve performance.
     3104Many of the data structures and algorithms may change in the future to more efficient versions.
     3105For example, the number of monitors in a single \textbf{bulk-acq} is only bound by the stack size, this is probably unnecessarily generous.
     3106It may be possible that limiting the number helps increase performance.
     3107However, it is not obvious that the benefit would be significant.
    26243108
    26253109\subsection{Flexible Scheduling} \label{futur:sched}
    2626 An important part of concurrency is scheduling. Different scheduling algorithms can affect performance (both in terms of average and variation). However, no single scheduler is optimal for all workloads and therefore there is value in being able to change the scheduler for given programs. One solution is to offer various tweaking options to users, allowing the scheduler to be adjusted to the requirements of the workload. However, in order to be truly flexible, it would be interesting to allow users to add arbitrary data and arbitrary scheduling algorithms. For example, a web server could attach Type-of-Service information to threads and have a ``ToS aware'' scheduling algorithm tailored to this specific web server. This path of flexible schedulers will be explored for \CFA.
     3110An important part of concurrency is scheduling.
     3111Different scheduling algorithms can affect performance (both in terms of average and variation).
     3112However, no single scheduler is optimal for all workloads and therefore there is value in being able to change the scheduler for given programs.
     3113One solution is to offer various tweaking options to users, allowing the scheduler to be adjusted to the requirements of the workload.
     3114However, in order to be truly flexible, it would be interesting to allow users to add arbitrary data and arbitrary scheduling algorithms.
     3115For example, a web server could attach Type-of-Service information to threads and have a ``ToS aware'' scheduling algorithm tailored to this specific web server.
     3116This path of flexible schedulers will be explored for \CFA.
    26273117
    26283118\subsection{Non-Blocking I/O} \label{futur:nbio}
    2629 While most of the parallelism tools are aimed at data parallelism and control-flow parallelism, many modern workloads are not bound on computation but on IO operations, a common case being web servers and XaaS (anything as a service). These types of workloads often require significant engineering around amortizing costs of blocking IO operations. At its core, non-blocking I/O is an operating system level feature that allows queuing IO operations (e.g., network operations) and registering for notifications instead of waiting for requests to complete. In this context, the role of the language makes Non-Blocking IO easily available and with low overhead. The current trend is to use asynchronous programming using tools like callbacks and/or futures and promises, which can be seen in frameworks like Node.js~\cite{NodeJs} for JavaScript, Spring MVC~\cite{SpringMVC} for Java and Django~\cite{Django} for Python. However, while these are valid solutions, they lead to code that is harder to read and maintain because it is much less linear.
     3119While most of the parallelism tools are aimed at data parallelism and control-flow parallelism, many modern workloads are not bound on computation but on IO operations, a common case being web servers and XaaS (anything as a service).
     3120These types of workloads often require significant engineering around amortizing costs of blocking IO operations.
     3121At its core, non-blocking I/O is an operating system level feature that allows queuing IO operations (e.g., network operations) and registering for notifications instead of waiting for requests to complete.
     3122In this context, the role of the language makes Non-Blocking IO easily available and with low overhead.
     3123The current trend is to use asynchronous programming using tools like callbacks and/or futures and promises, which can be seen in frameworks like Node.js~\cite{NodeJs} for JavaScript, Spring MVC~\cite{SpringMVC} for Java and Django~\cite{Django} for Python.
     3124However, while these are valid solutions, they lead to code that is harder to read and maintain because it is much less linear.
    26303125
    26313126\subsection{Other Concurrency Tools} \label{futur:tools}
    2632 While monitors offer a flexible and powerful concurrent core for \CFA, other concurrency tools are also necessary for a complete multi-paradigm concurrency package. Examples of such tools can include simple locks and condition variables, futures and promises~\cite{promises}, executors and actors. These additional features are useful when monitors offer a level of abstraction that is inadequate for certain tasks.
     3127While monitors offer a flexible and powerful concurrent core for \CFA, other concurrency tools are also necessary for a complete multi-paradigm concurrency package.
     3128Examples of such tools can include simple locks and condition variables, futures and promises~\cite{promises}, executors and actors.
     3129These additional features are useful when monitors offer a level of abstraction that is inadequate for certain tasks.
    26333130
    26343131\subsection{Implicit Threading} \label{futur:implcit}
    2635 Simpler applications can benefit greatly from having implicit parallelism. That is, parallelism that does not rely on the user to write concurrency. This type of parallelism can be achieved both at the language level and at the library level. The canonical example of implicit parallelism is parallel for loops, which are the simplest example of a divide and conquer algorithms~\cite{uC++book}. Table \ref{lst:parfor} shows three different code examples that accomplish point-wise sums of large arrays. Note that none of these examples explicitly declare any concurrency or parallelism objects.
     3132Simpler applications can benefit greatly from having implicit parallelism.
     3133That is, parallelism that does not rely on the user to write concurrency.
     3134This type of parallelism can be achieved both at the language level and at the library level.
     3135The canonical example of implicit parallelism is parallel for loops, which are the simplest example of a divide and conquer algorithms~\cite{uC++book}.
     3136Table \ref{lst:parfor} shows three different code examples that accomplish point-wise sums of large arrays.
     3137Note that none of these examples explicitly declare any concurrency or parallelism objects.
    26363138
    26373139\begin{table}
     
    27183220\end{table}
    27193221
    2720 Implicit parallelism is a restrictive solution and therefore has its limitations. However, it is a quick and simple approach to parallelism, which may very well be sufficient for smaller applications and reduces the amount of boilerplate needed to start benefiting from parallelism in modern CPUs.
     3222Implicit parallelism is a restrictive solution and therefore has its limitations.
     3223However, it is a quick and simple approach to parallelism, which may very well be sufficient for smaller applications and reduces the amount of boilerplate needed to start benefiting from parallelism in modern CPUs.
    27213224
    27223225
     
    27313234% B I B L I O G R A P H Y
    27323235% -----------------------------
    2733 \bibliographystyle{plain}
     3236%\bibliographystyle{plain}
    27343237\bibliography{pl,local}
    27353238
  • doc/papers/concurrency/style/cfa-format.tex

    r759f05f rd5ccbe9  
    1 \usepackage[usenames,dvipsnames]{xcolor}
     1%\usepackage[usenames,dvipsnames]{xcolor}
    22\usepackage{listings}
    33\usepackage{inconsolata}
     
    149149\lstdefinestyle{cfaStyle}{
    150150  escapeinside={@@},
    151   basicstyle=\linespread{0.9}\tt\footnotesize,          % reduce line spacing and use typewriter font
    152   keywordstyle=\bfseries\color{blue},
    153   keywordstyle=[2]\bfseries\color{Plum},
    154   commentstyle=\sf\itshape\color{OliveGreen},             % green and italic comments
     151  basicstyle=\linespread{0.9}\sf,               % reduce line spacing and use typewriter font
     152%  keywordstyle=\bfseries\color{blue},
     153  keywordstyle=[2]\bfseries\color{red},
     154%  commentstyle=\sf\itshape\color{OliveGreen},            % green and italic comments
    155155  identifierstyle=\color{identifierCol},
    156   stringstyle=\sf\color{Mahogany},                                % use sanserif font
     156%  stringstyle=\sf\color{Mahogany},                               % use sanserif font
     157  stringstyle=\tt,                                                                              % use typewriter font
    157158  mathescape=true,
    158159  columns=fixed,
Note: See TracChangeset for help on using the changeset viewer.