Ignore:
Timestamp:
Apr 25, 2018, 4:55:53 PM (7 years ago)
Author:
Aaron Moss <a3moss@…>
Branches:
new-env, with_gc
Children:
42107b4
Parents:
2efe4b8 (diff), 9d5fb67 (diff)
Note: this is a merge changeset, the changes displayed below correspond to the merge itself.
Use the (diff) links above to see all the changes relative to each parent.
Message:

Merge remote-tracking branch 'origin/master' into with_gc

File:
1 edited

Legend:

Unmodified
Added
Removed

  • doc/papers/concurrency/Paper.tex

    r2efe4b8 r1cdfa82  
    1 % inline code ©...© (copyright symbol) emacs: C-q M-)
    2 % red highlighting ®...® (registered trademark symbol) emacs: C-q M-.
    3 % blue highlighting ß...ß (sharp s symbol) emacs: C-q M-_
    4 % green highlighting ¢...¢ (cent symbol) emacs: C-q M-"
    5 % LaTex escape §...§ (section symbol) emacs: C-q M-'
    6 % keyword escape ¶...¶ (pilcrow symbol) emacs: C-q M-^
    7 % math escape $...$ (dollar symbol)
    8 
    9 \documentclass[10pt]{article}
     1\documentclass[AMA,STIX1COL]{WileyNJD-v2}
     2
     3\articletype{RESEARCH ARTICLE}%
     4
     5\received{26 April 2016}
     6\revised{6 June 2016}
     7\accepted{6 June 2016}
     8
     9\raggedbottom
    1010
    1111%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    1212
    1313% Latex packages used in the document.
    14 \usepackage[T1]{fontenc}                                        % allow Latin1 (extended ASCII) characters
    15 \usepackage{textcomp}
    16 \usepackage[latin1]{inputenc}
    17 \usepackage{fullpage,times,comment}
     14
    1815\usepackage{epic,eepic}
     16\usepackage{xspace}
     17\usepackage{comment}
    1918\usepackage{upquote}                                            % switch curled `'" to straight
    20 \usepackage{calc}
    21 \usepackage{xspace}
    22 \usepackage[labelformat=simple]{subfig}
    23 \renewcommand{\thesubfigure}{(\alph{subfigure})}
    24 \usepackage{graphicx}
    25 \usepackage{tabularx}
    26 \usepackage{multicol}
    27 \usepackage{varioref}
    2819\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}
     20\usepackage[labelformat=simple,aboveskip=0pt,farskip=0pt]{subfig}
     21\renewcommand{\thesubfigure}{(\Alph{subfigure})}
     22\captionsetup{justification=raggedright,singlelinecheck=false}
     23\usepackage{siunitx}
     24\sisetup{ binary-units=true }
     25
     26\hypersetup{breaklinks=true}
     27\definecolor{OliveGreen}{cmyk}{0.64 0 0.95 0.40}
     28\definecolor{Mahogany}{cmyk}{0 0.85 0.87 0.35}
     29\definecolor{Plum}{cmyk}{0.50 1 0 0}
     30
    3331\usepackage[pagewise]{lineno}
    3432\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}
     33
     34\renewcommand{\textfraction}{0.0}       % the entire page maybe devoted to floats with no text on the page at all
     35
     36\lefthyphenmin=3                                                        % hyphen only after 4 characters
     37\righthyphenmin=3
    4738
    4839%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    … …  
    5041% Names used in the document.
    5142
    52 \newcommand{\Version}{1.0.0}
    53 \newcommand{\CS}{C\raisebox{-0.9ex}{\large$^\sharp$}\xspace}
     43\newcommand{\CFAIcon}{\textsf{C}\raisebox{\depth}{\rotatebox{180}{\textsf{A}}}\xspace} % Cforall symbolic name
     44\newcommand{\CFA}{\protect\CFAIcon}             % safe for section/caption
     45\newcommand{\CFL}{\textrm{Cforall}\xspace}      % Cforall symbolic name
     46\newcommand{\Celeven}{\textrm{C11}\xspace}      % C11 symbolic name
     47\newcommand{\CC}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}\xspace} % C++ symbolic name
     48\newcommand{\CCeleven}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}11\xspace} % C++11 symbolic name
     49\newcommand{\CCfourteen}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}14\xspace} % C++14 symbolic name
     50\newcommand{\CCseventeen}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}17\xspace} % C++17 symbolic name
     51\newcommand{\CCtwenty}{\textrm{C}\kern-.1em\hbox{+\kern-.25em+}20\xspace} % C++20 symbolic name
     52\newcommand{\Csharp}{C\raisebox{-0.7ex}{\Large$^\sharp$}\xspace} % C# symbolic name
     53
     54%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    5455
    5556\newcommand{\Textbf}[2][red]{{\color{#1}{\textbf{#2}}}}
    … …  
    6263\newcommand{\TODO}{{\Textbf{TODO}}}
    6364
    64 
    65 \newsavebox{\LstBox}
    66 
    6765%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    6866
    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}
     67% Default underscore is too low and wide. Cannot use lstlisting "literate" as replacing underscore
     68% removes it as a variable-name character so keywords in variables are highlighted. MUST APPEAR
     69% AFTER HYPERREF.
     70%\DeclareTextCommandDefault{\textunderscore}{\leavevmode\makebox[1.2ex][c]{\rule{1ex}{0.1ex}}}
     71\renewcommand{\textunderscore}{\leavevmode\makebox[1.2ex][c]{\rule{1ex}{0.075ex}}}
     72
     73\makeatletter
     74% parindent is relative, i.e., toggled on/off in environments like itemize, so store the value for
     75% use rather than use \parident directly.
     76\newlength{\parindentlnth}
     77\setlength{\parindentlnth}{\parindent}
     78
     79\newcommand{\LstBasicStyle}[1]{{\lst@basicstyle{\lst@basicstyle{#1}}}}
     80\newcommand{\LstKeywordStyle}[1]{{\lst@basicstyle{\lst@keywordstyle{#1}}}}
     81\newcommand{\LstCommentStyle}[1]{{\lst@basicstyle{\lst@commentstyle{#1}}}}
     82
     83\newlength{\gcolumnposn}                                        % temporary hack because lstlisting does not handle tabs correctly
     84\newlength{\columnposn}
     85\setlength{\gcolumnposn}{3.5in}
     86\setlength{\columnposn}{\gcolumnposn}
     87\newcommand{\C}[2][\@empty]{\ifx#1\@empty\else\global\setlength{\columnposn}{#1}\global\columnposn=\columnposn\fi\hfill\makebox[\textwidth-\columnposn][l]{\lst@basicstyle{\LstCommentStyle{#2}}}}
     88\newcommand{\CRT}{\global\columnposn=\gcolumnposn}
     89
     90% Denote newterms in particular font and index them without particular font and in lowercase, e.g., \newterm{abc}.
     91% The option parameter provides an index term different from the new term, e.g., \newterm[\texttt{abc}]{abc}
     92% The star version does not lowercase the index information, e.g., \newterm*{IBM}.
     93\newcommand{\newtermFontInline}{\emph}
     94\newcommand{\newterm}{\@ifstar\@snewterm\@newterm}
     95\newcommand{\@newterm}[2][\@empty]{\lowercase{\def\temp{#2}}{\newtermFontInline{#2}}\ifx#1\@empty\index{\temp}\else\index{#1@{\protect#2}}\fi}
     96\newcommand{\@snewterm}[2][\@empty]{{\newtermFontInline{#2}}\ifx#1\@empty\index{#2}\else\index{#1@{\protect#2}}\fi}
     97
     98% Latin abbreviation
     99\newcommand{\abbrevFont}{\textit}                       % set empty for no italics
     100\@ifundefined{eg}{
     101\newcommand{\EG}{\abbrevFont{e}\abbrevFont{g}}
     102\newcommand*{\eg}{%
     103        \@ifnextchar{,}{\EG}%
     104                {\@ifnextchar{:}{\EG}%
     105                        {\EG,\xspace}}%
     106}}{}%
     107\@ifundefined{ie}{
     108\newcommand{\IE}{\abbrevFont{i}\abbrevFont{e}}
     109\newcommand*{\ie}{%
     110        \@ifnextchar{,}{\IE}%
     111                {\@ifnextchar{:}{\IE}%
     112                        {\IE,\xspace}}%
     113}}{}%
     114\@ifundefined{etc}{
     115\newcommand{\ETC}{\abbrevFont{etc}}
     116\newcommand*{\etc}{%
     117        \@ifnextchar{.}{\ETC}%
     118        {\ETC.\xspace}%
     119}}{}%
     120\@ifundefined{etal}{
     121\newcommand{\ETAL}{\abbrevFont{et}~\abbrevFont{al}}
     122\newcommand*{\etal}{%
     123        \@ifnextchar{.}{\protect\ETAL}%
     124                {\protect\ETAL.\xspace}%
     125}}{}%
     126\@ifundefined{viz}{
     127\newcommand{\VIZ}{\abbrevFont{viz}}
     128\newcommand*{\viz}{%
     129        \@ifnextchar{.}{\VIZ}%
     130                {\VIZ.\xspace}%
     131}}{}%
     132\makeatother
     133
     134\newenvironment{cquote}{%
     135        \list{}{\lstset{resetmargins=true,aboveskip=0pt,belowskip=0pt}\topsep=3pt\parsep=0pt\leftmargin=\parindentlnth\rightmargin\leftmargin}%
     136        \item\relax
     137}{%
     138        \endlist
     139}% cquote
     140
     141% CFA programming language, based on ANSI C (with some gcc additions)
     142\lstdefinelanguage{CFA}[ANSI]{C}{
     143        morekeywords={
     144                _Alignas, _Alignof, __alignof, __alignof__, asm, __asm, __asm__, __attribute, __attribute__,
     145                auto, _Bool, catch, catchResume, choose, _Complex, __complex, __complex__, __const, __const__,
     146                coroutine, disable, dtype, enable, exception, __extension__, fallthrough, fallthru, finally,
     147                __float80, float80, __float128, float128, forall, ftype, _Generic, _Imaginary, __imag, __imag__,
     148                inline, __inline, __inline__, __int128, int128, __label__, monitor, mutex, _Noreturn, one_t, or,
     149                otype, restrict, __restrict, __restrict__, __signed, __signed__, _Static_assert, thread,
     150                _Thread_local, throw, throwResume, timeout, trait, try, ttype, typeof, __typeof, __typeof__,
     151                virtual, __volatile, __volatile__, waitfor, when, with, zero_t},
     152        moredirectives={defined,include_next}%
     153}
     154
     155\lstset{
     156language=CFA,
     157columns=fullflexible,
     158basicstyle=\linespread{0.9}\sf,                                                 % reduce line spacing and use sanserif font
     159stringstyle=\tt,                                                                                % use typewriter font
     160tabsize=5,                                                                                              % N space tabbing
     161xleftmargin=\parindentlnth,                                                             % indent code to paragraph indentation
     162%mathescape=true,                                                                               % LaTeX math escape in CFA code $...$
     163escapechar=\$,                                                                                  % LaTeX escape in CFA code
     164keepspaces=true,                                                                                %
     165showstringspaces=false,                                                                 % do not show spaces with cup
     166showlines=true,                                                                                 % show blank lines at end of code
     167aboveskip=4pt,                                                                                  % spacing above/below code block
     168belowskip=3pt,
     169% replace/adjust listing characters that look bad in sanserif
     170literate={-}{\makebox[1ex][c]{\raisebox{0.4ex}{\rule{0.8ex}{0.1ex}}}}1 {^}{\raisebox{0.6ex}{$\scriptstyle\land\,$}}1
     171        {~}{\raisebox{0.3ex}{$\scriptstyle\sim\,$}}1 % {`}{\ttfamily\upshape\hspace*{-0.1ex}`}1
     172        {<-}{$\leftarrow$}2 {=>}{$\Rightarrow$}2 {->}{\makebox[1ex][c]{\raisebox{0.4ex}{\rule{0.8ex}{0.075ex}}}\kern-0.2ex{\textgreater}}2,
     173moredelim=**[is][\color{red}]{`}{`},
     174}% lstset
     175
     176% uC++ programming language, based on ANSI C++
     177\lstdefinelanguage{uC++}[ANSI]{C++}{
     178        morekeywords={
     179                _Accept, _AcceptReturn, _AcceptWait, _Actor, _At, _CatchResume, _Cormonitor, _Coroutine, _Disable,
     180                _Else, _Enable, _Event, _Finally, _Monitor, _Mutex, _Nomutex, _PeriodicTask, _RealTimeTask,
     181                _Resume, _Select, _SporadicTask, _Task, _Timeout, _When, _With, _Throw},
     182}
     183\lstdefinelanguage{Golang}{
     184        morekeywords=[1]{package,import,func,type,struct,return,defer,panic,recover,select,var,const,iota,},
     185        morekeywords=[2]{string,uint,uint8,uint16,uint32,uint64,int,int8,int16,int32,int64,
     186                bool,float32,float64,complex64,complex128,byte,rune,uintptr, error,interface},
     187        morekeywords=[3]{map,slice,make,new,nil,len,cap,copy,close,true,false,delete,append,real,imag,complex,chan,},
     188        morekeywords=[4]{for,break,continue,range,goto,switch,case,fallthrough,if,else,default,},
     189        morekeywords=[5]{Println,Printf,Error,},
     190        sensitive=true,
     191        morecomment=[l]{//},
     192        morecomment=[s]{/*}{*/},
     193        morestring=[b]',
     194        morestring=[b]",
     195        morestring=[s]{`}{`},
     196}
     197
     198\lstnewenvironment{cfa}[1][]
     199{\lstset{#1}}
     200{}
     201\lstnewenvironment{C++}[1][]                            % use C++ style
     202{\lstset{language=C++,moredelim=**[is][\protect\color{red}]{`}{`},#1}\lstset{#1}}
     203{}
     204\lstnewenvironment{uC++}[1][]
     205{\lstset{#1}}
     206{}
     207\lstnewenvironment{Go}[1][]
     208{\lstset{#1}}
     209{}
     210
     211% inline code @...@
     212\lstMakeShortInline@%
     213
     214
     215\title{\texorpdfstring{Concurrency in \protect\CFA}{Concurrency in Cforall}}
     216
     217\author[1]{Thierry Delisle}
     218\author[1]{Peter A. Buhr*}
     219\authormark{DELISLE \textsc{et al.}}
     220
     221\address[1]{\orgdiv{Cheriton School of Computer Science}, \orgname{University of Waterloo}, \orgaddress{\state{Waterloo, ON}, \country{Canada}}}
     222
     223\corres{*Peter A. Buhr, Cheriton School of Computer Science, University of Waterloo, 200 University Avenue West, Waterloo, ON, N2L 3G1, Canada. \email{pabuhr{\char`\@}uwaterloo.ca}}
     224
     225\fundingInfo{Natural Sciences and Engineering Research Council of Canada}
     226
     227\abstract[Summary]{
     228\CFA is a modern, polymorphic, \emph{non-object-oriented} extension of the C programming language.
     229This paper discusses the design of the concurrency and parallelism features in \CFA, and the concurrent runtime-system.
     230These features are created from scratch as ISO C lacks concurrency, relying largely on pthreads library.
     231Coroutines and lightweight (user) threads are introduced into the language.
     232In addition, monitors are added as a high-level mechanism for mutual exclusion and synchronization.
     233A unique contribution is allowing multiple monitors to be safely acquired simultaneously.
     234All features respect the expectations of C programmers, while being fully integrate with the \CFA polymorphic type-system and other language features.
     235Finally, experimental results are presented to compare the performance of the new features with similar mechanisms in other concurrent programming-languages.
     236}%
     237
     238\keywords{concurrency, parallelism, coroutines, threads, monitors, runtime, C, Cforall}
    75239
    76240
    77241\begin{document}
     242\linenumbers                                            % comment out to turn off line numbering
     243
    78244\maketitle
    79245
    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 
     246% ======================================================================
    89247% ======================================================================
    90248\section{Introduction}
    91249% ======================================================================
    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.
     250% ======================================================================
     251
     252This paper provides a minimal concurrency \newterm{Abstract Program Interface} (API) that is simple, efficient and can be used to build other concurrency features.
     253While the simplest concurrency system is a thread and a lock, this low-level approach is hard to master.
     254An easier approach for programmers is to support higher-level constructs as the basis of concurrency.
     255Indeed, for highly productive concurrent programming, high-level approaches are much more popular~\cite{Hochstein05}.
     256Examples of high-level approaches are task based~\cite{TBB}, message passing~\cite{Erlang,MPI}, and implicit threading~\cite{OpenMP}.
     257
     258This paper uses the following terminology.
     259A \newterm{thread} is a fundamental unit of execution that runs a sequence of code and requires a stack to maintain state.
     260Multiple simultaneous threads give rise to \newterm{concurrency}, which requires locking to ensure safe communication and access to shared data.
     261% Correspondingly, concurrency is defined as the concepts and challenges that occur when multiple independent (sharing memory, timing dependencies, \etc) concurrent threads are introduced.
     262\newterm{Locking}, and by extension locks, are defined as a mechanism to prevent progress of threads to provide safety.
     263\newterm{Parallelism} is running multiple threads simultaneously.
     264Parallelism implies \emph{actual} simultaneous execution, where concurrency only requires \emph{apparent} simultaneous execution.
     265As such, parallelism only affects performance, which is observed through differences in space and/or time.
     266
     267Hence, there are two problems to be solved in the design of concurrency for a programming language: concurrency and parallelism.
     268While these two concepts are often combined, they are distinct, requiring different tools~\cite[\S~2]{Buhr05a}.
     269Concurrency tools handle synchronization and mutual exclusion, while parallelism tools handle performance, cost and resource utilization.
     270
     271The proposed concurrency API is implemented in a dialect of C, called \CFA.
     272The paper discusses how the language features are added to the \CFA translator with respect to parsing, semantic, and type checking, and the corresponding high-perforamnce runtime-library to implement the concurrency features.
    98273
    99274% ======================================================================
    … …  
    104279
    105280The following is a quick introduction to the \CFA language, specifically tailored to the features needed to support concurrency.
    106 
    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}.
    109 
    110 % ======================================================================
     281Most of the following code examples can be found on the \CFA website~\cite{Cforall}.
     282
     283\CFA is an extension of ISO-C, and therefore, supports all of the same paradigms as C.
     284%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.
     285Like C, the basics of \CFA revolve around structures and routines, which are thin abstractions over machine code.
     286The vast majority of the code produced by the \CFA translator respects memory layouts and calling conventions laid out by C.
     287Interestingly, while \CFA is not an object-oriented language, lacking the concept of a receiver (\eg @this@) and inheritance, 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
     288values''~\cite[3.15]{C11}}, most importantly construction and destruction of objects.
     289
     290
    111291\subsection{References}
    112292
    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:
    114 \begin{cfacode}
    115 int x, *p1 = &x, **p2 = &p1, ***p3 = &p2,
    116         &r1 = x,    &&r2 = r1,   &&&r3 = r2;
    117 ***p3 = 3;                                                      //change x
    118 r3    = 3;                                                      //change x, ***r3
    119 **p3  = ...;                                            //change p1
    120 *p3   = ...;                                            //change p2
    121 int y, z, & ar[3] = {x, y, z};          //initialize array of references
    122 typeof( ar[1]) p;                                       //is int, referenced object type
    123 typeof(&ar[1]) q;                                       //is int &, reference type
    124 sizeof( ar[1]) == sizeof(int);          //is true, referenced object size
    125 sizeof(&ar[1]) == sizeof(int *);        //is true, reference size
    126 \end{cfacode}
     293Like \CC, \CFA introduces rebind-able references providing multiple dereferencing as an alternative to pointers.
     294In 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:
     295\begin{cfa}
     296int x, y, z;
     297int * p1 = &x, ** p2 = &p1, *** p3 = &p2,       $\C{// pointers to x}$
     298        & r1 = x,   && r2 = r1, &&& r3 = r2;    $\C{// references to x}$
     299
     300*p1 = 3; **p2 = 3; ***p3 = 3;                           $\C{// change x}$
     301  r1 = 3;    r2 = 3;      r3 = 3;                       $\C{// change x}$
     302**p3 = &y; *p3 = &z;                                            $\C{// change p1, p2}$
     303&&r3 = &y; &r3 = &z;                                            $\C{// change p1, p2}$
     304int & ar[3] = {x, y, z};                                        $\C{// initialize array of references}$
     305
     306typeof( ar[1]) p;                                                       $\C{// is int, referenced object type}$
     307typeof(&ar[1]) q;                                                       $\C{// is int \&, reference type}$
     308sizeof( ar[1]) == sizeof(int);                          $\C{// is true, referenced object size}$
     309sizeof(&ar[1]) == sizeof(int *);                        $\C{// is true, reference size}$
     310\end{cfa}
    127311The important take away from this code example is that a reference offers a handle to an object, much like a pointer, but which is automatically dereferenced for convenience.
    128312
    … …  
    130314\subsection{Overloading}
    131315
    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.
    133 \begin{cfacode}
    134 //selection based on type and number of parameters
    135 void f(void);                   //(1)
    136 void f(char);                   //(2)
    137 void f(int, double);    //(3)
    138 f();                                    //select (1)
    139 f('a');                                 //select (2)
    140 f(3, 5.2);                              //select (3)
    141 
    142 //selection based on  type and number of returns
    143 char   f(int);                  //(1)
    144 double f(int);                  //(2)
    145 char   c = f(3);                //select (1)
    146 double d = f(4);                //select (2)
    147 \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.
     316Another 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.
     317As well, \CFA uses the return type as part of the selection criteria, as in Ada~\cite{Ada}.
     318For routines with multiple parameters and returns, the selection is complex.
     319\begin{cfa}
     320// selection based on type and number of parameters
     321void f(void);                   $\C{// (1)}$
     322void f(char);                   $\C{// (2)}$
     323void f(int, double);    $\C{// (3)}$
     324f();                                    $\C{// select (1)}$
     325f('a');                                 $\C{// select (2)}$
     326f(3, 5.2);                              $\C{// select (3)}$
     327
     328// selection based on  type and number of returns
     329char   f(int);                  $\C{// (1)}$
     330double f(int);                  $\C{// (2)}$
     331char   c = f(3);                $\C{// select (1)}$
     332double d = f(4);                $\C{// select (2)}$
     333\end{cfa}
     334This feature is particularly important for concurrency since the runtime system relies on creating different types to represent concurrency objects.
     335Therefore, overloading is necessary to prevent the need for long prefixes and other naming conventions that prevent name clashes.
     336As seen in section \ref{basics}, routine @main@ is an example that benefits from overloading.
    149337
    150338% ======================================================================
    151339\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.:
    153 \begin{cfacode}
    154 int ++? (int op);                       //unary prefix increment
    155 int ?++ (int op);                       //unary postfix increment
    156 int ?+? (int op1, int op2);             //binary plus
    157 int ?<=?(int op1, int op2);             //binary less than
    158 int ?=? (int & op1, int op2);           //binary assignment
    159 int ?+=?(int & op1, int op2);           //binary plus-assignment
     340Overloading also extends to operators.
     341The 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, \eg:
     342\begin{cfa}
     343int ++? (int op);                       $\C{// unary prefix increment}$
     344int ?++ (int op);                       $\C{// unary postfix increment}$
     345int ?+? (int op1, int op2);             $\C{// binary plus}$
     346int ?<=?(int op1, int op2);             $\C{// binary less than}$
     347int ?=? (int & op1, int op2);           $\C{// binary assignment}$
     348int ?+=?(int & op1, int op2);           $\C{// binary plus-assignment}$
    160349
    161350struct S {int i, j;};
    162 S ?+?(S op1, S op2) {                           //add two structures
     351S ?+?(S op1, S op2) {                           $\C{// add two structures}$
    163352        return (S){op1.i + op2.i, op1.j + op2.j};
    164353}
    165354S s1 = {1, 2}, s2 = {2, 3}, s3;
    166 s3 = s1 + s2;                                           //compute sum: s3 == {2, 5}
    167 \end{cfacode}
     355s3 = s1 + s2;                                           $\C{// compute sum: s3 == {2, 5}}$
     356\end{cfa}
    168357While concurrency does not use operator overloading directly, this feature is more important as an introduction for the syntax of constructors.
    169358
    170359% ======================================================================
    171360\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:
    173 \begin{cfacode}
     361Object 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.
     362Since \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:
     363\begin{cfa}
    174364struct S {
    175365        size_t size;
    176366        int * ia;
    177367};
    178 void ?{}(S & s, int asize) {    //constructor operator
    179         s.size = asize;                         //initialize fields
     368void ?{}(S & s, int asize) {    $\C{// constructor operator}$
     369        s.size = asize;                         $\C{// initialize fields}$
    180370        s.ia = calloc(size, sizeof(S));
    181371}
    182 void ^?{}(S & s) {                              //destructor operator
    183         free(ia);                                       //de-initialization fields
     372void ^?{}(S & s) {                              $\C{// destructor operator}$
     373        free(ia);                                       $\C{// de-initialization fields}$
    184374}
    185375int main() {
    186         S x = {10}, y = {100};          //implicit calls: ?{}(x, 10), ?{}(y, 100)
    187         ...                                                     //use x and y
    188         ^x{};  ^y{};                            //explicit calls to de-initialize
    189         x{20};  y{200};                         //explicit calls to reinitialize
    190         ...                                                     //reuse x and y
    191 }                                                               //implicit calls: ^?{}(y), ^?{}(x)
    192 \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.
    194 \begin{cfacode}
     376        S x = {10}, y = {100};          $\C{// implicit calls: ?\{\}(x, 10), ?\{\}(y, 100)}$
     377        ...                                                     $\C{// use x and y}$
     378        ^x{};  ^y{};                            $\C{// explicit calls to de-initialize}$
     379        x{20};  y{200};                         $\C{// explicit calls to reinitialize}$
     380        ...                                                     $\C{// reuse x and y}$
     381}                                                               $\C{// implicit calls: \^?\{\}(y), \^?\{\}(x)}$
     382\end{cfa}
     383The language guarantees that every object and all their fields are constructed.
     384Like \CC, construction of an object is automatically done on allocation and destruction of the object is done on deallocation.
     385Allocation and deallocation can occur on the stack or on the heap.
     386\begin{cfa}
    195387{
    196         struct S s = {10};      //allocation, call constructor
     388        struct S s = {10};      $\C{// allocation, call constructor}$
    197389        ...
    198 }                                               //deallocation, call destructor
    199 struct S * s = new();   //allocation, call constructor
     390}                                               $\C{// deallocation, call destructor}$
     391struct S * s = new();   $\C{// allocation, call constructor}$
    200392...
    201 delete(s);                              //deallocation, call destructor
    202 \end{cfacode}
    203 Note that like \CC, \CFA introduces \code{new} and \code{delete}, which behave like \code{malloc} and \code{free} in addition to constructing and destructing objects, after calling \code{malloc} and before calling \code{free}, respectively.
     393delete(s);                              $\C{// deallocation, call destructor}$
     394\end{cfa}
     395Note that like \CC, \CFA introduces @new@ and @delete@, which behave like @malloc@ and @free@ in addition to constructing and destructing objects, after calling @malloc@ and before calling @free@, respectively.
    204396
    205397% ======================================================================
    206398\subsection{Parametric Polymorphism}
    207399\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:
    209 \begin{cfacode}
    210 //constraint type, 0 and +
     400Routines in \CFA can also be reused for multiple types.
     401This capability is done using the @forall@ clauses, which allow separately compiled routines to support generic usage over multiple types.
     402For example, the following sum function works for any type that supports construction from 0 and addition:
     403\begin{cfa}
     404// constraint type, 0 and +
    211405forall(otype T | { void ?{}(T *, zero_t); T ?+?(T, T); })
    212406T sum(T a[ ], size_t size) {
    213         T total = 0;                            //construct T from 0
     407        T total = 0;                            $\C{// construct T from 0}$
    214408        for(size_t i = 0; i < size; i++)
    215                 total = total + a[i];   //select appropriate +
     409                total = total + a[i];   $\C{// select appropriate +}$
    216410        return total;
    217411}
    218412
    219413S sa[5];
    220 int i = sum(sa, 5);                             //use S's 0 construction and +
    221 \end{cfacode}
    222 
    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:
    224 \begin{cfacode}
     414int i = sum(sa, 5);                             $\C{// use S's 0 construction and +}$
     415\end{cfa}
     416
     417Since writing constraints on types can become cumbersome for more constrained functions, \CFA also has the concept of traits.
     418Traits are named collection of constraints that can be used both instead and in addition to regular constraints:
     419\begin{cfa}
    225420trait summable( otype T ) {
    226         void ?{}(T *, zero_t);          //constructor from 0 literal
    227         T ?+?(T, T);                            //assortment of additions
     421        void ?{}(T *, zero_t);          $\C{// constructor from 0 literal}$
     422        T ?+?(T, T);                            $\C{// assortment of additions}$
    228423        T ?+=?(T *, T);
    229424        T ++?(T *);
    230425        T ?++(T *);
    231426};
    232 forall( otype T | summable(T) ) //use trait
     427forall( otype T | summable(T) ) $\C{// use trait}$
    233428T sum(T a[], size_t size);
    234 \end{cfacode}
    235 
    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.
     429\end{cfa}
     430
     431Note that the type use for assertions can be either an @otype@ or a @dtype@.
     432Types declared as @otype@ refer to ``complete'' objects, \ie objects with a size, a default constructor, a copy constructor, a destructor and an assignment operator.
     433Using @dtype@, on the other hand, has none of these assumptions but is extremely restrictive, it only guarantees the object is addressable.
    237434
    238435% ======================================================================
    239436\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).
    241 \begin{cfacode}
     437Since \CFA lacks the concept of a receiver, certain functions end up needing to repeat variable names often.
     438To remove this inconvenience, \CFA provides the @with@ statement, which opens an aggregate scope making its fields directly accessible (like Pascal).
     439\begin{cfa}
    242440struct S { int i, j; };
    243 int mem(S & this) with (this)           //with clause
    244         i = 1;                                                  //this->i
    245         j = 2;                                                  //this->j
     441int mem(S & this) with (this)           $\C{// with clause}$
     442        i = 1;                                                  $\C{// this->i}$
     443        j = 2;                                                  $\C{// this->j}$
    246444}
    247445int foo() {
    248446        struct S1 { ... } s1;
    249447        struct S2 { ... } s2;
    250         with (s1)                                               //with statement
     448        with (s1)                                               $\C{// with statement}$
    251449        {
    252                 //access fields of s1 without qualification
    253                 with (s2)                                       //nesting
     450                // access fields of s1 without qualification
     451                with (s2)                                       $\C{// nesting}$
    254452                {
    255                         //access fields of s1 and s2 without qualification
     453                        // access fields of s1 and s2 without qualification
    256454                }
    257455        }
    258         with (s1, s2)                                   //scopes open in parallel
     456        with (s1, s2)                                   $\C{// scopes open in parallel}$
    259457        {
    260                 //access fields of s1 and s2 without qualification
     458                // access fields of s1 and s2 without qualification
    261459        }
    262460}
    263 \end{cfacode}
    264 
    265 For more information on \CFA see \cite{cforall-ug,rob-thesis,www-cfa}.
     461\end{cfa}
     462
     463For more information on \CFA see \cite{cforall-ug,Schluntz17,www-cfa}.
    266464
    267465% ======================================================================
    … …  
    270468% ======================================================================
    271469% ======================================================================
    272 Before any detailed discussion of the concurrency and parallelism in \CFA, it is important to describe the basics of concurrency and how they are expressed in \CFA user code.
    273 
    274 \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.
    282 
    283 \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.
    285 
    286 \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}.
    288 
    289 \begin{table}
    290 \begin{center}
    291 \begin{tabular}{c @{\hskip 0.025in}|@{\hskip 0.025in} c @{\hskip 0.025in}|@{\hskip 0.025in} c}
    292 \begin{ccode}[tabsize=2]
    293 //Using callbacks
    294 void fibonacci_func(
    295         int n,
    296         void (*callback)(int)
    297 ) {
    298         int first = 0;
    299         int second = 1;
    300         int next, i;
    301         for(i = 0; i < n; i++)
    302         {
    303                 if(i <= 1)
    304                         next = i;
    305                 else {
    306                         next = f1 + f2;
    307                         f1 = f2;
    308                         f2 = next;
    309                 }
    310                 callback(next);
     470
     471At its core, concurrency is based on having multiple call-stacks and scheduling among threads of execution executing on these stacks.
     472Multiple call stacks (or contexts) and a single thread of execution does \emph{not} imply concurrency.
     473Execution with a single thread and multiple stacks where the thread is deterministically self-scheduling across the stacks is called \newterm{coroutining};
     474execution with a single thread 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~\cite[\S~3]{Buhr05a}.
     475Therefore, a minimal concurrency system can be achieved using coroutines (see Section \ref{coroutine}), which instead of context-switching among each other, always defer to an oracle for where to context-switch next.
     476
     477While coroutines can execute on the caller's stack-frame, stack-full coroutines allow full generality and are sufficient as the basis for concurrency.
     478The aforementioned oracle is a scheduler and the whole system now follows a cooperative threading-model (a.k.a., non-preemptive scheduling).
     479The 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.
     480In any case, a subset of concurrency related challenges start to appear.
     481For the complete set of concurrency challenges to occur, the only feature missing is preemption.
     482
     483A scheduler introduces order of execution uncertainty, while preemption introduces uncertainty about where context switches occur.
     484Mutual exclusion and synchronization are ways of limiting non-determinism in a concurrent system.
     485Now 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.
     486Optimal performance in concurrent applications is often obtained by having as much non-determinism as correctness allows.
     487
     488
     489\subsection{\protect\CFA's Thread Building Blocks}
     490
     491One 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.
     492As such, library support for threading is far from widespread.
     493At the time of writing the paper, neither \protect\lstinline|gcc| nor \protect\lstinline|clang| support ``threads.h'' in their standard libraries.}.
     494On 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.
     495As 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.
     496And being a system-level language means programmers expect to choose precisely which features they need and which cost they are willing to pay.
     497
     498
     499\subsection{Coroutines: A Stepping Stone}\label{coroutine}
     500
     501While the focus of this proposal is concurrency and parallelism, it is important to address coroutines, which are a significant building block of a concurrency system.
     502\newterm{Coroutine}s are generalized routines with points where execution is suspended and resumed at a later time.
     503Suspend/resume is a context switche and coroutines have other context-management operations.
     504Many design challenges of threads are partially present in designing coroutines, which makes the design effort relevant.
     505The core \textbf{api} of coroutines has two features: independent call-stacks and @suspend@/@resume@.
     506
     507A coroutine handles the class of problems that need to retain state between calls (\eg plugin, device driver, finite-state machine).
     508For example, a problem made easier with coroutines is unbounded generators, \eg generating an infinite sequence of Fibonacci numbers:
     509\begin{displaymath}
     510f(n) = \left \{
     511\begin{array}{ll}
     5120                               & n = 0         \\
     5131                               & n = 1         \\
     514f(n-1) + f(n-2) & n \ge 2       \\
     515\end{array}
     516\right.
     517\end{displaymath}
     518Figure~\ref{f:C-fibonacci} shows conventional approaches for writing a Fibonacci generator in C.
     519
     520Figure~\ref{f:GlobalVariables} illustrates the following problems:
     521unencapsulated global variables necessary to retain state between calls;
     522only one fibonacci generator can run at a time;
     523execution state must be explicitly retained.
     524Figure~\ref{f:ExternalState} addresses these issues:
     525unencapsulated program global variables become encapsulated structure variables;
     526multiple fibonacci generators can run at a time by declaring multiple fibonacci objects;
     527explicit execution state is removed by precomputing the first two Fibonacci numbers and returning $f(n-2)$.
     528
     529\begin{figure}
     530\centering
     531\newbox\myboxA
     532\begin{lrbox}{\myboxA}
     533\begin{lstlisting}[aboveskip=0pt,belowskip=0pt]
     534`int f1, f2, state = 1;`   // single global variables
     535int fib() {
     536        int fn;
     537        `switch ( state )` {  // explicit execution state
     538          case 1: fn = 0;  f1 = fn;  state = 2;  break;
     539          case 2: fn = 1;  f2 = f1;  f1 = fn;  state = 3;  break;
     540          case 3: fn = f1 + f2;  f2 = f1;  f1 = fn;  break;
    311541        }
    312 }
    313 
     542        return fn;
     543}
    314544int main() {
    315         void print_fib(int n) {
    316                 printf("%d\n", n);
     545
     546        for ( int i = 0; i < 10; i += 1 ) {
     547                printf( "%d\n", fib() );
    317548        }
    318 
    319         fibonacci_func(
    320                 10, print_fib
    321         );
    322 
    323 
    324 
    325 }
    326 \end{ccode}&\begin{ccode}[tabsize=2]
    327 //Using output array
    328 void fibonacci_array(
    329         int n,
    330         int* array
    331 ) {
    332         int f1 = 0; int f2 = 1;
    333         int next, i;
    334         for(i = 0; i < n; i++)
    335         {
    336                 if(i <= 1)
    337                         next = i;
    338                 else {
    339                         next = f1 + f2;
    340                         f1 = f2;
    341                         f2 = next;
    342                 }
    343                 array[i] = next;
     549}
     550\end{lstlisting}
     551\end{lrbox}
     552
     553\newbox\myboxB
     554\begin{lrbox}{\myboxB}
     555\begin{lstlisting}[aboveskip=0pt,belowskip=0pt]
     556#define FIB_INIT `{ 0, 1 }`
     557typedef struct { int f2, f1; } Fib;
     558int fib( Fib * f ) {
     559
     560        int ret = f->f2;
     561        int fn = f->f1 + f->f2;
     562        f->f2 = f->f1; f->f1 = fn;
     563
     564        return ret;
     565}
     566int main() {
     567        Fib f1 = FIB_INIT, f2 = FIB_INIT;
     568        for ( int i = 0; i < 10; i += 1 ) {
     569                printf( "%d %d\n", fib( &f1 ), fib( &f2 ) );
    344570        }
    345571}
    346 
    347 
     572\end{lstlisting}
     573\end{lrbox}
     574
     575\subfloat[3 States: global variables]{\label{f:GlobalVariables}\usebox\myboxA}
     576\qquad
     577\subfloat[1 State: external variables]{\label{f:ExternalState}\usebox\myboxB}
     578\caption{C Fibonacci Implementations}
     579\label{f:C-fibonacci}
     580
     581\bigskip
     582
     583\newbox\myboxA
     584\begin{lrbox}{\myboxA}
     585\begin{lstlisting}[aboveskip=0pt,belowskip=0pt]
     586`coroutine` Fib { int fn; };
     587void main( Fib & f ) with( f ) {
     588        int f1, f2;
     589        fn = 0;  f1 = fn;  `suspend()`;
     590        fn = 1;  f2 = f1;  f1 = fn;  `suspend()`;
     591        for ( ;; ) {
     592                fn = f1 + f2;  f2 = f1;  f1 = fn;  `suspend()`;
     593        }
     594}
     595int next( Fib & fib ) with( fib ) {
     596        `resume( fib );`
     597        return fn;
     598}
    348599int main() {
    349         int a[10];
    350 
    351         fibonacci_func(
    352                 10, a
    353         );
    354 
    355         for(int i=0;i<10;i++){
    356                 printf("%d\n", a[i]);
    357         }
    358 
    359 }
    360 \end{ccode}&\begin{ccode}[tabsize=2]
    361 //Using external state
    362 typedef struct {
    363         int f1, f2;
    364 } Iterator_t;
    365 
    366 int fibonacci_state(
    367         Iterator_t* it
    368 ) {
    369         int f;
    370         f = it->f1 + it->f2;
    371         it->f2 = it->f1;
    372         it->f1 = max(f,1);
    373         return f;
    374 }
    375 
    376 
    377 
    378 
    379 
    380 
    381 
    382 int main() {
    383         Iterator_t it={0,0};
    384 
    385         for(int i=0;i<10;i++){
    386                 printf("%d\n",
    387                         fibonacci_state(
    388                                 &it
    389                         );
    390                 );
    391         }
    392 
    393 }
    394 \end{ccode}
    395 \end{tabular}
    396 \end{center}
    397 \caption{Different implementations of a Fibonacci sequence generator in C.}
    398 \label{lst:fibonacci-c}
    399 \end{table}
    400 
    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.
    404 
    405 \begin{figure}
    406 \begin{cfacode}[caption={Implementation of Fibonacci using coroutines},label={lst:fibonacci-cfa}]
    407 coroutine Fibonacci {
    408         int fn; //used for communication
    409 };
    410 
    411 void ?{}(Fibonacci& this) { //constructor
    412         this.fn = 0;
    413 }
    414 
    415 //main automatically called on first resume
    416 void main(Fibonacci& this) with (this) {
    417         int fn1, fn2;           //retained between resumes
    418         fn  = 0;
    419         fn1 = fn;
    420         suspend(this);          //return to last resume
    421 
    422         fn  = 1;
    423         fn2 = fn1;
    424         fn1 = fn;
    425         suspend(this);          //return to last resume
    426 
    427         for ( ;; ) {
    428                 fn  = fn1 + fn2;
    429                 fn2 = fn1;
    430                 fn1 = fn;
    431                 suspend(this);  //return to last resume
    432         }
    433 }
    434 
    435 int next(Fibonacci& this) {
    436         resume(this); //transfer to last suspend
    437         return this.fn;
    438 }
    439 
    440 void main() { //regular program main
    441         Fibonacci f1, f2;
     600        Fib f1, f2;
    442601        for ( int i = 1; i <= 10; i += 1 ) {
    443602                sout | next( f1 ) | next( f2 ) | endl;
    444603        }
    445604}
    446 \end{cfacode}
     605\end{lstlisting}
     606\end{lrbox}
     607\newbox\myboxB
     608\begin{lrbox}{\myboxB}
     609\begin{lstlisting}[aboveskip=0pt,belowskip=0pt]
     610`coroutine` Fib { int ret; };
     611void main( Fib & f ) with( f ) {
     612        int fn, f1 = 1, f2 = 0;
     613        for ( ;; ) {
     614                ret = f2;
     615
     616                fn = f1 + f2;  f2 = f1;  f1 = fn; `suspend();`
     617        }
     618}
     619int next( Fib & fib ) with( fib ) {
     620        `resume( fib );`
     621        return ret;
     622}
     623
     624
     625
     626
     627
     628
     629\end{lstlisting}
     630\end{lrbox}
     631\subfloat[3 States, internal variables]{\label{f:Coroutine3States}\usebox\myboxA}
     632\qquad\qquad
     633\subfloat[1 State, internal variables]{\label{f:Coroutine1State}\usebox\myboxB}
     634\caption{\CFA Coroutine Fibonacci Implementations}
     635\label{f:fibonacci-cfa}
    447636\end{figure}
    448637
    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.
     638Figure~\ref{f:Coroutine3States} creates a @coroutine@ type, which provides communication for multiple interface functions, and the \newterm{coroutine main}, which runs on the coroutine stack.
     639\begin{cfa}
     640`coroutine C { char c; int i; _Bool s; };`      $\C{// used for communication}$
     641void ?{}( C & c ) { s = false; }                        $\C{// constructor}$
     642void main( C & cor ) with( cor ) {                      $\C{// actual coroutine}$
     643        while ( ! s ) // process c
     644        if ( v == ... ) s = false;
     645}
     646// interface functions
     647char cont( C & cor, char ch ) { c = ch; resume( cor ); return c; }
     648_Bool stop( C & cor, int v ) { s = true; i = v; resume( cor ); return s; }
     649\end{cfa}
     650
     651encapsulates the Fibonacci state in the  shows is an example of a solution to the Fibonacci problem using \CFA coroutines, where the coroutine stack holds sufficient state for the next generation.
     652This solution has the advantage of having very strong decoupling between how the sequence is generated and how it is used.
     653Indeed, this version is as easy to use as the @fibonacci_state@ solution, while the implementation is very similar to the @fibonacci_func@ example.
     654
     655Figure~\ref{f:fmt-line} shows the @Format@ coroutine for restructuring text into groups of character blocks of fixed size.
     656The 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.
    450657
    451658\begin{figure}
    452 \begin{cfacode}[tabsize=3,caption={Formatting text into lines of 5 blocks of 4 characters.},label={lst:fmt-line}]
    453 //format characters into blocks of 4 and groups of 5 blocks per line
    454 coroutine Format {
    455         char ch;                                                                        //used for communication
    456         int g, b;                                                               //global because used in destructor
     659\begin{cfa}[xleftmargin=4\parindentlnth]
     660`coroutine` Format {
     661        char ch;                                                                $\C{// used for communication}$
     662        int g, b;                                                               $\C{// global because used in destructor}$
    457663};
    458 
    459 void  ?{}(Format& fmt) {
    460         resume( fmt );                                                  //prime (start) coroutine
    461 }
    462 
    463 void ^?{}(Format& fmt) with fmt {
    464         if ( fmt.g != 0 || fmt.b != 0 )
    465         sout | endl;
    466 }
    467 
    468 void main(Format& fmt) with fmt {
    469         for ( ;; ) {                                                    //for as many characters
    470                 for(g = 0; g < 5; g++) {                //groups of 5 blocks
    471                         for(b = 0; b < 4; fb++) {       //blocks of 4 characters
    472                                 suspend();
    473                                 sout | ch;                                      //print character
     664void ?{}( Format & fmt ) { `resume( fmt );` } $\C{// prime (start) coroutine}$
     665void ^?{}( Format & fmt ) with( fmt ) { if ( g != 0 || b != 0 ) sout | endl; }
     666void main( Format & fmt ) with( fmt ) {
     667        for ( ;; ) {                                                    $\C{// for as many characters}$
     668                for ( g = 0; g < 5; g += 1 ) {          $\C{// groups of 5 blocks}$
     669                        for ( b = 0; b < 4; b += 1 ) {  $\C{// blocks of 4 characters}$
     670                                `suspend();`
     671                                sout | ch;                                      $\C{// print character}$
    474672                        }
    475                         sout | "  ";                                    //print block separator
     673                        sout | "  ";                                    $\C{// print block separator}$
    476674                }
    477                 sout | endl;                                            //print group separator
     675                sout | endl;                                            $\C{// print group separator}$
    478676        }
    479677}
    480 
    481 void prt(Format & fmt, char ch) {
     678void prt( Format & fmt, char ch ) {
    482679        fmt.ch = ch;
    483         resume(fmt);
    484 }
    485 
     680        `resume( fmt );`
     681}
    486682int main() {
    487683        Format fmt;
    488684        char ch;
    489         Eof: for ( ;; ) {                                               //read until end of file
    490                 sin | ch;                                                       //read one character
    491                 if(eof(sin)) break Eof;                 //eof ?
    492                 prt(fmt, ch);                                           //push character for formatting
     685        for ( ;; ) {                                                    $\C{// read until end of file}$
     686                sin | ch;                                                       $\C{// read one character}$
     687          if ( eof( sin ) ) break;                              $\C{// eof ?}$
     688                prt( fmt, ch );                                         $\C{// push character for formatting}$
    493689        }
    494690}
    495 \end{cfacode}
     691\end{cfa}
     692\caption{Formatting text into lines of 5 blocks of 4 characters.}
     693\label{f:fmt-line}
    496694\end{figure}
    497695
    498 \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:
    504 
    505 \begin{cfacode}
    506 //async: Runs function asynchronously on another thread
     696\begin{figure}
     697\centering
     698\lstset{language=CFA,escapechar={},moredelim=**[is][\protect\color{red}]{`}{`}}
     699\begin{tabular}{@{}l@{\hspace{2\parindentlnth}}l@{}}
     700\begin{cfa}
     701`coroutine` Prod {
     702        Cons & c;
     703        int N, money, receipt;
     704};
     705void main( Prod & prod ) with( prod ) {
     706        // 1st resume starts here
     707        for ( int i = 0; i < N; i += 1 ) {
     708                int p1 = random( 100 ), p2 = random( 100 );
     709                sout | p1 | " " | p2 | endl;
     710                int status = delivery( c, p1, p2 );
     711                sout | " $" | money | endl | status | endl;
     712                receipt += 1;
     713        }
     714        stop( c );
     715        sout | "prod stops" | endl;
     716}
     717int payment( Prod & prod, int money ) {
     718        prod.money = money;
     719        `resume( prod );`
     720        return prod.receipt;
     721}
     722void start( Prod & prod, int N, Cons &c ) {
     723        &prod.c = &c;
     724        prod.[N, receipt] = [N, 0];
     725        `resume( prod );`
     726}
     727int main() {
     728        Prod prod;
     729        Cons cons = { prod };
     730        srandom( getpid() );
     731        start( prod, 5, cons );
     732}
     733\end{cfa}
     734&
     735\begin{cfa}
     736`coroutine` Cons {
     737        Prod & p;
     738        int p1, p2, status;
     739        _Bool done;
     740};
     741void ?{}( Cons & cons, Prod & p ) {
     742        &cons.p = &p;
     743        cons.[status, done ] = [0, false];
     744}
     745void ^?{}( Cons & cons ) {}
     746void main( Cons & cons ) with( cons ) {
     747        // 1st resume starts here
     748        int money = 1, receipt;
     749        for ( ; ! done; ) {
     750                sout | p1 | " " | p2 | endl | " $" | money | endl;
     751                status += 1;
     752                receipt = payment( p, money );
     753                sout | " #" | receipt | endl;
     754                money += 1;
     755        }
     756        sout | "cons stops" | endl;
     757}
     758int delivery( Cons & cons, int p1, int p2 ) {
     759        cons.[p1, p2] = [p1, p2];
     760        `resume( cons );`
     761        return cons.status;
     762}
     763void stop( Cons & cons ) {
     764        cons.done = true;
     765        `resume( cons );`
     766}
     767
     768\end{cfa}
     769\end{tabular}
     770\caption{Producer / consumer: resume-resume cycle, bi-directional communication}
     771\label{f:ProdCons}
     772\end{figure}
     773
     774
     775\subsubsection{Construction}
     776
     777One 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.
     778In the case of coroutines, this challenge is simpler since there is no non-determinism from preemption or scheduling.
     779However, the underlying challenge remains the same for coroutines and threads.
     780
     781The runtime system needs to create the coroutine's stack and, more importantly, prepare it for the first resumption.
     782The 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.
     783There are several solutions to this problem but the chosen option effectively forces the design of the coroutine.
     784
     785Furthermore, \CFA faces an extra challenge as polymorphic routines create invisible thunks when cast to non-polymorphic routines and these thunks have function scope.
     786For example, the following code, while looking benign, can run into undefined behaviour because of thunks:
     787
     788\begin{cfa}
     789// async: Runs function asynchronously on another thread
    507790forall(otype T)
    508791extern void async(void (*func)(T*), T* obj);
    … …  
    513796void bar() {
    514797        int a;
    515         async(noop, &a); //start thread running noop with argument a
    516 }
    517 \end{cfacode}
     798        async(noop, &a); // start thread running noop with argument a
     799}
     800\end{cfa}
    518801
    519802The generated C code\footnote{Code trimmed down for brevity} creates a local thunk to hold type information:
    520803
    521 \begin{ccode}
     804\begin{cfa}
    522805extern void async(/* omitted */, void (*func)(void*), void* obj);
    523806
    … …  
    533816        async(/* omitted */, ((void (*)(void*))(&_thunk0)), (&a));
    534817}
    535 \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.
    537 
    538 \subsection{Alternative: Composition}
     818\end{cfa}
     819The problem in this example is a storage management issue, the function pointer @_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; \ie the stack-based thunk being destroyed before it can be used.
     820This challenge is an extension of challenges that come with second-class routines.
     821Indeed, GCC nested routines also have the limitation that nested routine cannot be passed outside of the declaration scope.
     822The case of coroutines and threads is simply an extension of this problem to multiple call stacks.
     823
     824
     825\subsubsection{Alternative: Composition}
     826
    539827One solution to this challenge is to use composition/containment, where coroutine fields are added to manage the coroutine.
    540828
    541 \begin{cfacode}
     829\begin{cfa}
    542830struct Fibonacci {
    543         int fn; //used for communication
    544         coroutine c; //composition
     831        int fn; // used for communication
     832        coroutine c; // composition
    545833};
    546834
    … …  
    551839void ?{}(Fibonacci& this) {
    552840        this.fn = 0;
    553         //Call constructor to initialize coroutine
     841        // Call constructor to initialize coroutine
    554842        (this.c){myMain};
    555843}
    556 \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.
    558 
    559 \subsection{Alternative: Reserved keyword}
     844\end{cfa}
     845The downside of this approach is that users need to correctly construct the coroutine handle before using it.
     846Like any other objects, the user must carefully choose construction order to prevent usage of objects not yet constructed.
     847However, in the case of coroutines, users must also pass to the coroutine information about the coroutine main, like in the previous example.
     848This opens the door for user errors and requires extra runtime storage to pass at runtime information that can be known statically.
     849
     850
     851\subsubsection{Alternative: Reserved keyword}
     852
    560853The next alternative is to use language support to annotate coroutines as follows:
    561 
    562 \begin{cfacode}
     854\begin{cfa}
    563855coroutine Fibonacci {
    564         int fn; //used for communication
     856        int fn; // used for communication
    565857};
    566 \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.
    568 
    569 \subsection{Alternative: Lambda Objects}
    570 
    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:
    572 \begin{cfacode}
     858\end{cfa}
     859The @coroutine@ keyword means the compiler can find and inject code where needed.
     860The downside of this approach is that it makes coroutine a special case in the language.
     861Users wanting to extend coroutines or build their own for various reasons can only do so in ways offered by the language.
     862Furthermore, implementing coroutines without language supports also displays the power of the programming language used.
     863While this is ultimately the option used for idiomatic \CFA code, coroutines and threads can still be constructed by users without using the language support.
     864The reserved keywords are only present to improve ease of use for the common cases.
     865
     866
     867\subsubsection{Alternative: Lambda Objects}
     868
     869For coroutines as for threads, many implementations are based on routine pointers or function objects~\cite{Butenhof97, C++14, MS:VisualC++, BoostCoroutines15}.
     870For example, Boost implements coroutines in terms of four functor object types:
     871\begin{cfa}
    573872asymmetric_coroutine<>::pull_type
    574873asymmetric_coroutine<>::push_type
    575874symmetric_coroutine<>::call_type
    576875symmetric_coroutine<>::yield_type
    577 \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.
    579 
    580 A variation of this would be to use a simple function pointer in the same way \texttt{pthread} does for threads:
    581 \begin{cfacode}
     876\end{cfa}
     877Often, the canonical threading paradigm in languages is based on function pointers, @pthread@ being one of the most well-known examples.
     878The 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.
     879Since the custom type is simple to write in \CFA and solves several issues, added support for routine/lambda based coroutines adds very little.
     880
     881A variation of this would be to use a simple function pointer in the same way @pthread@ does for threads:
     882\begin{cfa}
    582883void foo( coroutine_t cid, void* arg ) {
    583884        int* value = (int*)arg;
    584         //Coroutine body
     885        // Coroutine body
    585886}
    586887
    … …  
    590891        coroutine_resume( &cid );
    591892}
    592 \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.
    594 
    595 \subsection{Alternative: Trait-Based Coroutines}
    596 
    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.
    598 
    599 \begin{cfacode}
     893\end{cfa}
     894This semantics is more common for thread interfaces but coroutines work equally well.
     895As discussed in section \ref{threads}, this approach is superseded by static approaches in terms of expressivity.
     896
     897
     898\subsubsection{Alternative: Trait-Based Coroutines}
     899
     900Finally, the underlying approach, which is the one closest to \CFA idioms, is to use trait-based lazy coroutines.
     901This approach defines a coroutine as anything that satisfies the trait @is_coroutine@ (as defined below) and is used as a coroutine.
     902
     903\begin{cfa}
    600904trait is_coroutine(dtype T) {
    601905      void main(T& this);
    … …  
    605909forall( dtype T | is_coroutine(T) ) void suspend(T&);
    606910forall( dtype T | is_coroutine(T) ) void resume (T&);
    607 \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.
     911\end{cfa}
     912This ensures that an object is not a coroutine until @resume@ is called on the object.
     913Correspondingly, any object that is passed to @resume@ is a coroutine since it must satisfy the @is_coroutine@ trait to compile.
     914The 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 @get_coroutine@ routine.
     915The \CFA keyword @coroutine@ simply has the effect of implementing the getter and forward declarations required for users to implement the main routine.
    609916
    610917\begin{center}
    611918\begin{tabular}{c c c}
    612 \begin{cfacode}[tabsize=3]
     919\begin{cfa}[tabsize=3]
    613920coroutine MyCoroutine {
    614921        int someValue;
    615922};
    616 \end{cfacode} & == & \begin{cfacode}[tabsize=3]
     923\end{cfa} & == & \begin{cfa}[tabsize=3]
    617924struct MyCoroutine {
    618925        int someValue;
    … …  
    628935
    629936void main(struct MyCoroutine* this);
    630 \end{cfacode}
     937\end{cfa}
    631938\end{tabular}
    632939\end{center}
    … …  
    634941The combination of these two approaches allows users new to coroutining and concurrency to have an easy and concise specification, while more advanced users have tighter control on memory layout and initialization.
    635942
    636 \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:
    638 
    639 \begin{cfacode}
     943\subsection{Thread Interface}\label{threads}
     944The basic building blocks of multithreading in \CFA are \textbf{cfathread}.
     945Both user and kernel threads are supported, where user threads are the concurrency mechanism and kernel threads are the parallel mechanism.
     946User threads offer a flexible and lightweight interface.
     947A thread can be declared using a struct declaration @thread@ as follows:
     948
     949\begin{cfa}
    640950thread foo {};
    641 \end{cfacode}
     951\end{cfa}
    642952
    643953As for coroutines, the keyword is a thin wrapper around a \CFA trait:
    644954
    645 \begin{cfacode}
     955\begin{cfa}
    646956trait is_thread(dtype T) {
    647957      void ^?{}(T & mutex this);
    … …  
    649959      thread_desc* get_thread(T & this);
    650960};
    651 \end{cfacode}
    652 
    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
    654 \begin{cfacode}
     961\end{cfa}
     962
     963Obviously, for this thread implementation to be useful it must run some user code.
     964Several other threading interfaces use a function-pointer representation as the interface of threads (for example \Csharp~\cite{Csharp} and Scala~\cite{Scala}).
     965However, this proposal considers that statically tying a @main@ routine to a thread supersedes this approach.
     966Since the @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).
     967As such the @main@ routine of a thread can be defined as
     968\begin{cfa}
    655969thread foo {};
    656970
    … …  
    658972        sout | "Hello World!" | endl;
    659973}
    660 \end{cfacode}
    661 
    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.
    663 \begin{cfacode}
     974\end{cfa}
     975
     976In this example, threads of type @foo@ start execution in the @void main(foo &)@ routine, which prints @"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.
     977With 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.
     978\begin{cfa}
    664979typedef void (*voidFunc)(int);
    665980
    … …  
    675990
    676991void main(FuncRunner & this) {
    677         //thread starts here and runs the function
     992        // thread starts here and runs the function
    678993        this.func( this.arg );
    679994}
    … …  
    6871002        return 0?
    6881003}
    689 \end{cfacode}
     1004\end{cfa}
    6901005
    6911006A 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}.
    6921007
    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.
    694 \begin{cfacode}
     1008Of course, for threads to be useful, it must be possible to start and stop threads and wait for them to complete execution.
     1009While using an \textbf{api} such as @fork@ and @join@ is relatively common in the literature, such an interface is unnecessary.
     1010Indeed, the simplest approach is to use \textbf{raii} principles and have threads @fork@ after the constructor has completed and @join@ before the destructor runs.
     1011\begin{cfa}
    6951012thread World;
    6961013
    … …  
    7011018void main() {
    7021019        World w;
    703         //Thread forks here
    704 
    705         //Printing "Hello " and "World!" are run concurrently
     1020        // Thread forks here
     1021
     1022        // Printing "Hello " and "World!" are run concurrently
    7061023        sout | "Hello " | endl;
    7071024
    708         //Implicit join at end of scope
    709 }
    710 \end{cfacode}
     1025        // Implicit join at end of scope
     1026}
     1027\end{cfa}
    7111028
    7121029This semantic has several advantages over explicit semantics: a thread is always started and stopped exactly once, users cannot make any programming errors, and it naturally scales to multiple threads meaning basic synchronization is very simple.
    7131030
    714 \begin{cfacode}
     1031\begin{cfa}
    7151032thread MyThread {
    7161033        //...
    7171034};
    7181035
    719 //main
     1036// main
    7201037void main(MyThread& this) {
    7211038        //...
    … …  
    7241041void foo() {
    7251042        MyThread thrds[10];
    726         //Start 10 threads at the beginning of the scope
     1043        // Start 10 threads at the beginning of the scope
    7271044
    7281045        DoStuff();
    7291046
    730         //Wait for the 10 threads to finish
    731 }
    732 \end{cfacode}
    733 
    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.
    735 
    736 \begin{cfacode}
     1047        // Wait for the 10 threads to finish
     1048}
     1049\end{cfa}
     1050
     1051However, one of the drawbacks of this approach is that threads always form a tree where nodes must always outlive their children, \ie they are always destroyed in the opposite order of construction because of C scoping rules.
     1052This 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.
     1053
     1054\begin{cfa}
    7371055thread MyThread {
    7381056        //...
    … …  
    7461064        MyThread* long_lived;
    7471065        {
    748                 //Start a thread at the beginning of the scope
     1066                // Start a thread at the beginning of the scope
    7491067                MyThread short_lived;
    7501068
    751                 //create another thread that will outlive the thread in this scope
     1069                // create another thread that will outlive the thread in this scope
    7521070                long_lived = new MyThread;
    7531071
    7541072                DoStuff();
    7551073
    756                 //Wait for the thread short_lived to finish
     1074                // Wait for the thread short_lived to finish
    7571075        }
    7581076        DoMoreStuff();
    7591077
    760         //Now wait for the long_lived to finish
     1078        // Now wait for the long_lived to finish
    7611079        delete long_lived;
    7621080}
    763 \end{cfacode}
     1081\end{cfa}
    7641082
    7651083
    … …  
    7691087% ======================================================================
    7701088% ======================================================================
    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.
    778 
    779 \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.
    781 
    782 \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.
    784 
    785 \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.
     1089Several tools can be used to solve concurrency challenges.
     1090Since 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}).
     1091In 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).
     1092However, in languages that use routine calls as their core abstraction mechanism, these approaches force a clear distinction between concurrent and non-concurrent paradigms (\ie message passing versus routine calls).
     1093This distinction in turn means that, in order to be effective, programmers need to learn two sets of design patterns.
     1094While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account.
     1095
     1096Approaches 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.
     1097At the lowest level, concurrent paradigms are implemented as atomic operations and locks.
     1098Many such mechanisms have been proposed, including semaphores~\cite{Dijkstra68b} and path expressions~\cite{Campbell74}.
     1099However, for productivity reasons it is desirable to have a higher-level construct be the core concurrency paradigm~\cite{Hochstein05}.
     1100
     1101An approach that is worth mentioning because it is gaining in popularity is transactional memory~\cite{Herlihy93}.
     1102While 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.
     1103
     1104One of the most natural, elegant, and efficient mechanisms for synchronization and communication, especially for shared-memory systems, is the \emph{monitor}.
     1105Monitors were first proposed by Brinch Hansen~\cite{Hansen73} and later described and extended by C.A.R.~Hoare~\cite{Hoare74}.
     1106Many programming languages---\eg 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.
     1107In 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.
     1108For these reasons, this project proposes monitors as the core concurrency construct.
     1109
     1110
     1111\subsection{Basics}
     1112
     1113Non-determinism requires concurrent systems to offer support for mutual-exclusion and synchronization.
     1114Mutual-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.
     1115On the other hand, synchronization enforces relative ordering of execution and synchronization tools provide numerous mechanisms to establish timing relationships among threads.
     1116
     1117
     1118\subsubsection{Mutual-Exclusion}
     1119
     1120As mentioned above, mutual-exclusion is the guarantee that only a fix number of threads can enter a critical section at once.
     1121However, many solutions exist for mutual exclusion, which vary in terms of performance, flexibility and ease of use.
     1122Methods 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.
     1123Ease of use comes by either guaranteeing some problems cannot occur (\eg being deadlock free) or by offering a more explicit coupling between data and corresponding critical section.
     1124For example, the \CC @std::atomic<T>@ offers an easy way to express mutual-exclusion on a restricted set of operations (\eg reading/writing large types atomically).
     1125Another challenge with low-level locks is composability.
     1126Locks have restricted composability because it takes careful organizing for multiple locks to be used while preventing deadlocks.
     1127Easing composability is another feature higher-level mutual-exclusion mechanisms often offer.
     1128
     1129
     1130\subsubsection{Synchronization}
     1131
     1132As with mutual-exclusion, low-level synchronization primitives often offer good performance and good flexibility at the cost of ease of use.
     1133Again, higher-level mechanisms often simplify usage by adding either better coupling between synchronization and data (\eg message passing) or offering a simpler solution to otherwise involved challenges.
     1134As mentioned above, synchronization can be expressed as guaranteeing that event \textit{X} always happens before \textit{Y}.
     1135Most of the time, synchronization happens within a critical section, where threads must acquire mutual-exclusion in a certain order.
     1136However, it may also be desirable to guarantee that event \textit{Z} does not occur between \textit{X} and \textit{Y}.
     1137Not satisfying this property is called \textbf{barging}.
     1138For 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}.
     1139The 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.
     1140Preventing or detecting barging is an involved challenge with low-level locks, which can be made much easier by higher-level constructs.
     1141This challenge is often split into two different methods, barging avoidance and barging prevention.
     1142Algorithms 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.
     1143
    7871144
    7881145% ======================================================================
    … …  
    7911148% ======================================================================
    7921149% ======================================================================
    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:
    794 \begin{cfacode}
     1150A \textbf{monitor} is a set of routines that ensure mutual-exclusion when accessing shared state.
     1151More 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.
     1152This strong association eases readability and maintainability, at the cost of flexibility.
     1153Note that both monitors and mutex locks, require an abstract handle to identify them.
     1154This concept is generally associated with object-oriented languages like Java~\cite{Java} or \uC~\cite{uC++book} but does not strictly require OO semantics.
     1155The only requirement is the ability to declare a handle to a shared object and a set of routines that act on it:
     1156\begin{cfa}
    7951157typedef /*some monitor type*/ monitor;
    7961158int f(monitor & m);
    7971159
    7981160int main() {
    799         monitor m;  //Handle m
    800         f(m);       //Routine using handle
    801 }
    802 \end{cfacode}
     1161        monitor m;  // Handle m
     1162        f(m);       // Routine using handle
     1163}
     1164\end{cfa}
    8031165
    8041166% ======================================================================
    … …  
    8071169% ======================================================================
    8081170% ======================================================================
    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:
    812 
    813 \begin{cfacode}
     1171The above monitor example displays some of the intrinsic characteristics.
     1172First, it is necessary to use pass-by-reference over pass-by-value for monitor routines.
     1173This semantics is important, because at their core, monitors are implicit mutual-exclusion objects (locks), and these objects cannot be copied.
     1174Therefore, monitors are non-copy-able objects (@dtype@).
     1175
     1176Another aspect to consider is when a monitor acquires its mutual exclusion.
     1177For example, a monitor may need to be passed through multiple helper routines that do not acquire the monitor mutual-exclusion on entry.
     1178Passthrough can occur for generic helper routines (@swap@, @sort@, \etc) or specific helper routines like the following to implement an atomic counter:
     1179
     1180\begin{cfa}
    8141181monitor counter_t { /*...see section $\ref{data}$...*/ };
    8151182
    816 void ?{}(counter_t & nomutex this); //constructor
    817 size_t ++?(counter_t & mutex this); //increment
    818 
    819 //need for mutex is platform dependent
    820 void ?{}(size_t * this, counter_t & mutex cnt); //conversion
    821 \end{cfacode}
     1183void ?{}(counter_t & nomutex this); // constructor
     1184size_t ++?(counter_t & mutex this); // increment
     1185
     1186// need for mutex is platform dependent
     1187void ?{}(size_t * this, counter_t & mutex cnt); // conversion
     1188\end{cfa}
    8221189This counter is used as follows:
    8231190\begin{center}
    8241191\begin{tabular}{c @{\hskip 0.35in} c @{\hskip 0.35in} c}
    825 \begin{cfacode}
    826 //shared counter
     1192\begin{cfa}
     1193// shared counter
    8271194counter_t cnt1, cnt2;
    8281195
    829 //multiple threads access counter
     1196// multiple threads access counter
    8301197thread 1 : cnt1++; cnt2++;
    8311198thread 2 : cnt1++; cnt2++;
    … …  
    8331200        ...
    8341201thread N : cnt1++; cnt2++;
    835 \end{cfacode}
     1202\end{cfa}
    8361203\end{tabular}
    8371204\end{center}
    838 Notice 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}.
    839 
    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.
     1205Notice 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 @std::atomic@.
     1206
     1207Here, the constructor (@?{}@) uses the @nomutex@ keyword to signify that it does not acquire the monitor mutual-exclusion when constructing.
     1208This semantics is because an object not yet constructed should never be shared and therefore does not require mutual exclusion.
     1209Furthermore, it allows the implementation greater freedom when it initializes the monitor locking.
     1210The prefix increment operator uses @mutex@ to protect the incrementing process from race conditions.
     1211Finally, there is a conversion operator from @counter_t@ to @size_t@.
     1212This conversion may or may not require the @mutex@ keyword depending on whether or not reading a @size_t@ is an atomic operation.
     1213
     1214For maximum usability, monitors use \textbf{multi-acq} semantics, which means a single thread can acquire the same monitor multiple times without deadlock.
     1215For example, listing \ref{fig:search} uses recursion and \textbf{multi-acq} to print values inside a binary tree.
    8431216\begin{figure}
    844 \begin{cfacode}[caption={Recursive printing algorithm using \textbf{multi-acq}.},label={fig:search}]
     1217\begin{cfa}[caption={Recursive printing algorithm using \textbf{multi-acq}.},label={fig:search}]
    8451218monitor printer { ... };
    8461219struct tree {
    … …  
    8551228        print(p, t->right);
    8561229}
    857 \end{cfacode}
     1230\end{cfa}
    8581231\end{figure}
    8591232
    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:
    863 \begin{cfacode}
     1233Having both @mutex@ and @nomutex@ keywords can be redundant, depending on the meaning of a routine having neither of these keywords.
     1234For example, it is reasonable that it should default to the safest option (@mutex@) when given a routine without qualifiers @void foo(counter_t & this)@, whereas assuming @nomutex@ is unsafe and may cause subtle errors.
     1235On the other hand, @nomutex@ is the ``normal'' parameter behaviour, it effectively states explicitly that ``this routine is not special''.
     1236Another alternative is making exactly one of these keywords mandatory, which provides the same semantics but without the ambiguity of supporting routines with neither keyword.
     1237Mandatory keywords would also have the added benefit of being self-documented but at the cost of extra typing.
     1238While there are several benefits to mandatory keywords, they do bring a few challenges.
     1239Mandatory keywords in \CFA would imply that the compiler must know without doubt whether or not a parameter is a monitor or not.
     1240Since \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.
     1241For this reason, \CFA only has the @mutex@ keyword and uses no keyword to mean @nomutex@.
     1242
     1243The next semantic decision is to establish when @mutex@ may be used as a type qualifier.
     1244Consider the following declarations:
     1245\begin{cfa}
    8641246int f1(monitor & mutex m);
    8651247int f2(const monitor & mutex m);
    … …  
    8671249int f4(monitor * mutex m []);
    8681250int f5(graph(monitor *) & mutex m);
    869 \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:
    871 \begin{cfacode}
    872 int f1(monitor & mutex m);    //Okay : recommended case
    873 int f2(monitor * mutex m);    //Not Okay : Could be an array
    874 int f3(monitor mutex m []);  //Not Okay : Array of unknown length
    875 int f4(monitor ** mutex m);   //Not Okay : Could be an array
    876 int f5(monitor * mutex m []); //Not Okay : Array of unknown length
    877 \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.
    881 \begin{cfacode}
     1251\end{cfa}
     1252The problem is to identify which object(s) should be acquired.
     1253Furthermore, each object needs to be acquired only once.
     1254In the case of simple routines like @f1@ and @f2@ it is easy to identify an exhaustive list of objects to acquire on entry.
     1255Adding indirections (@f3@) still allows the compiler and programmer to identify which object is acquired.
     1256However, adding in arrays (@f4@) makes it much harder.
     1257Array lengths are not necessarily known in C, and even then, making sure objects are only acquired once becomes none-trivial.
     1258This problem can be extended to absurd limits like @f5@, which uses a graph of monitors.
     1259To 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).
     1260Also note that while routine @f3@ can be supported, meaning that monitor @**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.
     1261However, this ambiguity is part of the C type-system with respects to arrays.
     1262For this reason, @mutex@ is disallowed in the context where arrays may be passed:
     1263\begin{cfa}
     1264int f1(monitor & mutex m);    // Okay : recommended case
     1265int f2(monitor * mutex m);    // Not Okay : Could be an array
     1266int f3(monitor mutex m []);  // Not Okay : Array of unknown length
     1267int f4(monitor ** mutex m);   // Not Okay : Could be an array
     1268int f5(monitor * mutex m []); // Not Okay : Array of unknown length
     1269\end{cfa}
     1270Note that not all array functions are actually distinct in the type system.
     1271However, even if the code generation could tell the difference, the extra information is still not sufficient to extend meaningfully the monitor call semantic.
     1272
     1273Unlike 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.
     1274A consequence of this approach is that it extends naturally to multi-monitor calls.
     1275\begin{cfa}
    8821276int f(MonitorA & mutex a, MonitorB & mutex b);
    8831277
    … …  
    8851279MonitorB b;
    8861280f(a,b);
    887 \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:
    889 \begin{cfacode}
    890 void foo(A& mutex a, B& mutex b) { //acquire a & b
     1281\end{cfa}
     1282While OO monitors could be extended with a mutex qualifier for multiple-monitor calls, no example of this feature could be found.
     1283The capability to acquire multiple locks before entering a critical section is called \emph{\textbf{bulk-acq}}.
     1284In practice, writing multi-locking routines that do not lead to deadlocks is tricky.
     1285Having language support for such a feature is therefore a significant asset for \CFA.
     1286In the case presented above, \CFA guarantees that the order of acquisition is consistent across calls to different routines using the same monitors as arguments.
     1287This consistent ordering means acquiring multiple monitors is safe from deadlock when using \textbf{bulk-acq}.
     1288However, users can still force the acquiring order.
     1289For example, notice which routines use @mutex@/@nomutex@ and how this affects acquiring order:
     1290\begin{cfa}
     1291void foo(A& mutex a, B& mutex b) { // acquire a & b
    8911292        ...
    8921293}
    8931294
    894 void bar(A& mutex a, B& /*nomutex*/ b) { //acquire a
    895         ... foo(a, b); ... //acquire b
    896 }
    897 
    898 void baz(A& /*nomutex*/ a, B& mutex b) { //acquire b
    899         ... foo(a, b); ... //acquire a
    900 }
    901 \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:
     1295void bar(A& mutex a, B& /*nomutex*/ b) { // acquire a
     1296        ... foo(a, b); ... // acquire b
     1297}
     1298
     1299void baz(A& /*nomutex*/ a, B& mutex b) { // acquire b
     1300        ... foo(a, b); ... // acquire a
     1301}
     1302\end{cfa}
     1303The \textbf{multi-acq} monitor lock allows a monitor lock to be acquired by both @bar@ or @baz@ and acquired again in @foo@.
     1304In the calls to @bar@ and @baz@ the monitors are acquired in opposite order.
     1305
     1306However, such use leads to lock acquiring order problems.
     1307In the example above, the user uses implicit ordering in the case of function @foo@ but explicit ordering in the case of @bar@ and @baz@.
     1308This subtle difference means that calling these routines concurrently may lead to deadlock and is therefore undefined behaviour.
     1309As shown~\cite{Lister77}, solving this problem requires:
    9051310\begin{enumerate}
    9061311        \item Dynamically tracking the monitor-call order.
    9071312        \item Implement rollback semantics.
    9081313\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.
     1314While 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}.
     1315In \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.
     1316While \CFA provides only a partial solution, most systems provide no solution and the \CFA partial solution handles many useful cases.
    9101317
    9111318For example, \textbf{multi-acq} and \textbf{bulk-acq} can be used together in interesting ways:
    912 \begin{cfacode}
     1319\begin{cfa}
    9131320monitor bank { ... };
    9141321
    … …  
    9191326        deposit( yourbank, me2you );
    9201327}
    921 \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.
    923 
    924 \subsection{\code{mutex} statement} \label{mutex-stmt}
    925 
    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.
     1328\end{cfa}
     1329This example shows a trivial solution to the bank-account transfer problem~\cite{BankTransfer}.
     1330Without \textbf{multi-acq} and \textbf{bulk-acq}, the solution to this problem is much more involved and requires careful engineering.
     1331
     1332
     1333\subsection{\protect\lstinline|mutex| statement} \label{mutex-stmt}
     1334
     1335The 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 @mutex@ statement to work around the need for unnecessary names, avoiding a major software engineering problem~\cite{2FTwoHardThings}.
     1336Table \ref{f:mutex-stmt} shows an example of the @mutex@ statement, which introduces a new scope in which the mutual-exclusion of a set of monitor is acquired.
     1337Beyond naming, the @mutex@ statement has no semantic difference from a routine call with @mutex@ parameters.
    9271338
    9281339\begin{table}
    9291340\begin{center}
    9301341\begin{tabular}{|c|c|}
    931 function call & \code{mutex} statement \\
     1342function call & @mutex@ statement \\
    9321343\hline
    933 \begin{cfacode}[tabsize=3]
     1344\begin{cfa}[tabsize=3]
    9341345monitor M {};
    9351346void foo( M & mutex m1, M & mutex m2 ) {
    936         //critical section
     1347        // critical section
    9371348}
    9381349
    … …  
    9401351        foo( m1, m2 );
    9411352}
    942 \end{cfacode}&\begin{cfacode}[tabsize=3]
     1353\end{cfa}&\begin{cfa}[tabsize=3]
    9431354monitor M {};
    9441355void bar( M & m1, M & m2 ) {
    9451356        mutex(m1, m2) {
    946                 //critical section
     1357                // critical section
    9471358        }
    9481359}
    9491360
    9501361
    951 \end{cfacode}
     1362\end{cfa}
    9521363\end{tabular}
    9531364\end{center}
    954 \caption{Regular call semantics vs. \code{mutex} statement}
    955 \label{lst:mutex-stmt}
     1365\caption{Regular call semantics vs. \protect\lstinline|mutex| statement}
     1366\label{f:mutex-stmt}
    9561367\end{table}
    9571368
    … …  
    9611372% ======================================================================
    9621373% ======================================================================
    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}:
    964 \begin{cfacode}
     1374Once the call semantics are established, the next step is to establish data semantics.
     1375Indeed, until now a monitor is used simply as a generic handle but in most cases monitors contain shared data.
     1376This data should be intrinsic to the monitor declaration to prevent any accidental use of data without its appropriate protection.
     1377For example, here is a complete version of the counter shown in section \ref{call}:
     1378\begin{cfa}
    9651379monitor counter_t {
    9661380        int value;
    … …  
    9751389}
    9761390
    977 //need for mutex is platform dependent here
     1391// need for mutex is platform dependent here
    9781392void ?{}(int * this, counter_t & mutex cnt) {
    9791393        *this = (int)cnt;
    9801394}
    981 \end{cfacode}
    982 
    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:
    984 \begin{cfacode}
     1395\end{cfa}
     1396
     1397Like threads and coroutines, monitors are defined in terms of traits with some additional language support in the form of the @monitor@ keyword.
     1398The monitor trait is:
     1399\begin{cfa}
    9851400trait is_monitor(dtype T) {
    9861401        monitor_desc * get_monitor( T & );
    9871402        void ^?{}( T & mutex );
    9881403};
    989 \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.
     1404\end{cfa}
     1405Note that the destructor of a monitor must be a @mutex@ routine to prevent deallocation while a thread is accessing the monitor.
     1406As with any object, calls to a monitor, using @mutex@ or otherwise, is undefined behaviour after the destructor has run.
    9911407
    9921408% ======================================================================
    … …  
    9951411% ======================================================================
    9961412% ======================================================================
    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.
     1413In addition to mutual exclusion, the monitors at the core of \CFA's concurrency can also be used to achieve synchronization.
     1414With monitors, this capability is generally achieved with internal or external scheduling as in~\cite{Hoare74}.
     1415With \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 (\ie with access to the shared state), while \textbf{external scheduling} means making the decision when entering the critical section (\ie without access to the shared state).
     1416Since internal scheduling within a single monitor is mostly a solved problem, this paper concentrates on extending internal scheduling to multiple monitors.
     1417Indeed, 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.
    9981418
    9991419First, here is a simple example of internal scheduling:
    10001420
    1001 \begin{cfacode}
     1421\begin{cfa}
    10021422monitor A {
    10031423        condition e;
    … …  
    10061426void foo(A& mutex a1, A& mutex a2) {
    10071427        ...
    1008         //Wait for cooperation from bar()
     1428        // Wait for cooperation from bar()
    10091429        wait(a1.e);
    10101430        ...
    … …  
    10121432
    10131433void bar(A& mutex a1, A& mutex a2) {
    1014         //Provide cooperation for foo()
     1434        // Provide cooperation for foo()
    10151435        ...
    1016         //Unblock foo
     1436        // Unblock foo
    10171437        signal(a1.e);
    10181438}
    1019 \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.
     1439\end{cfa}
     1440There are two details to note here.
     1441First, @signal@ is a delayed operation; it only unblocks the waiting thread when it reaches the end of the critical section.
     1442This semantics is needed to respect mutual-exclusion, \ie the signaller and signalled thread cannot be in the monitor simultaneously.
     1443The alternative is to return immediately after the call to @signal@, which is significantly more restrictive.
     1444Second, in \CFA, while it is common to store a @condition@ as a field of the monitor, a @condition@ variable can be stored/created independently of a monitor.
     1445Here routine @foo@ waits for the @signal@ from @bar@ before making further progress, ensuring a basic ordering.
     1446
     1447An important aspect of the implementation is that \CFA does not allow barging, which means that once function @bar@ releases the monitor, @foo@ is guaranteed to be the next thread to acquire the monitor (unless some other thread waited on the same condition).
     1448This guarantee offers the benefit of not having to loop around waits to recheck that a condition is met.
     1449The 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.
     1450Supporting 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.
    10231451
    10241452% ======================================================================
    … …  
    10271455% ======================================================================
    10281456% ======================================================================
    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.
     1457It is easy to understand the problem of multi-monitor scheduling using a series of pseudo-code examples.
     1458Note 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.
     1459Indeed, @wait@ statements always use the implicit condition variable as parameters and explicitly name the monitors (A and B) associated with the condition.
     1460Note 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.
     1461The example below shows the simple case of having two threads (one for each column) and a single monitor A.
    10301462
    10311463\begin{multicols}{2}
    10321464thread 1
    1033 \begin{pseudo}
     1465\begin{cfa}
    10341466acquire A
    10351467        wait A
    10361468release A
    1037 \end{pseudo}
     1469\end{cfa}
    10381470
    10391471\columnbreak
    10401472
    10411473thread 2
    1042 \begin{pseudo}
     1474\begin{cfa}
    10431475acquire A
    10441476        signal A
    10451477release A
    1046 \end{pseudo}
     1478\end{cfa}
    10471479\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.
     1480One thread acquires before waiting (atomically blocking and releasing A) and the other acquires before signalling.
     1481It is important to note here that both @wait@ and @signal@ must be called with the proper monitor(s) already acquired.
     1482This semantic is a logical requirement for barging prevention.
    10491483
    10501484A direct extension of the previous example is a \textbf{bulk-acq} version:
    10511485\begin{multicols}{2}
    1052 \begin{pseudo}
     1486\begin{cfa}
    10531487acquire A & B
    10541488        wait A & B
    10551489release A & B
    1056 \end{pseudo}
     1490\end{cfa}
    10571491\columnbreak
    1058 \begin{pseudo}
     1492\begin{cfa}
    10591493acquire A & B
    10601494        signal A & B
    10611495release A & B
    1062 \end{pseudo}
     1496\end{cfa}
    10631497\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:
     1498\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.
     1499Synchronization happens between the two threads in exactly the same way and order.
     1500The only difference is that mutual exclusion covers a group of monitors.
     1501On the implementation side, handling multiple monitors does add a degree of complexity as the next few examples demonstrate.
     1502
     1503While 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.
     1504For monitors, a well-known deadlock problem is the Nested Monitor Problem~\cite{Lister77}, which occurs when a @wait@ is made by a thread that holds more than one monitor.
     1505For example, the following cfa-code runs into the nested-monitor problem:
    10671506\begin{multicols}{2}
    1068 \begin{pseudo}
     1507\begin{cfa}
    10691508acquire A
    10701509        acquire B
    … …  
    10721511        release B
    10731512release A
    1074 \end{pseudo}
     1513\end{cfa}
    10751514
    10761515\columnbreak
    10771516
    1078 \begin{pseudo}
     1517\begin{cfa}
    10791518acquire A
    10801519        acquire B
    … …  
    10821521        release B
    10831522release A
    1084 \end{pseudo}
     1523\end{cfa}
    10851524\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}.
     1525\noindent The @wait@ only releases monitor @B@ so the signalling thread cannot acquire monitor @A@ to get to the @signal@.
     1526Attempting release of all acquired monitors at the @wait@ introduces a different set of problems, such as releasing monitor @C@, which has nothing to do with the @signal@.
     1527
     1528However, for monitors as for locks, it is possible to write a program using nesting without encountering any problems if nesting is done correctly.
     1529For example, the next cfa-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}.
    10891530
    10901531\begin{multicols}{2}
    1091 \begin{pseudo}
     1532\begin{cfa}
    10921533acquire A
    10931534        acquire B
    … …  
    10951536        release B
    10961537release A
    1097 \end{pseudo}
     1538\end{cfa}
    10981539
    10991540\columnbreak
    11001541
    1101 \begin{pseudo}
     1542\begin{cfa}
    11021543
    11031544acquire B
    … …  
    11051546release B
    11061547
    1107 \end{pseudo}
     1548\end{cfa}
    11081549\end{multicols}
    11091550
    … …  
    11161557% ======================================================================
    11171558
    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.
    1119 
    1120 \begin{figure}[!t]
     1559A larger example is presented to show complex issues for \textbf{bulk-acq} and its implementation options are analyzed.
     1560Figure~\ref{f:int-bulk-cfa} shows an example where \textbf{bulk-acq} adds a significant layer of complexity to the internal signalling semantics, and listing \ref{f:int-bulk-cfa} shows the corresponding \CFA code to implement the cfa-code in listing \ref{f:int-bulk-cfa}.
     1561For the purpose of translating the given cfa-code into \CFA-code, any method of introducing a monitor is acceptable, \eg @mutex@ parameters, global variables, pointer parameters, or using locals with the @mutex@ statement.
     1562
     1563\begin{figure}
    11211564\begin{multicols}{2}
    11221565Waiting thread
    1123 \begin{pseudo}[numbers=left]
     1566\begin{cfa}[numbers=left]
    11241567acquire A
    1125         //Code Section 1
     1568        // Code Section 1
    11261569        acquire A & B
    1127                 //Code Section 2
     1570                // Code Section 2
    11281571                wait A & B
    1129                 //Code Section 3
     1572                // Code Section 3
    11301573        release A & B
    1131         //Code Section 4
     1574        // Code Section 4
    11321575release A
    1133 \end{pseudo}
     1576\end{cfa}
    11341577\columnbreak
    11351578Signalling thread
    1136 \begin{pseudo}[numbers=left, firstnumber=10,escapechar=|]
     1579\begin{cfa}[numbers=left, firstnumber=10,escapechar=|]
    11371580acquire A
    1138         //Code Section 5
     1581        // Code Section 5
    11391582        acquire A & B
    1140                 //Code Section 6
     1583                // Code Section 6
    11411584                |\label{line:signal1}|signal A & B
    1142                 //Code Section 7
     1585                // Code Section 7
    11431586        |\label{line:releaseFirst}|release A & B
    1144         //Code Section 8
     1587        // Code Section 8
    11451588|\label{line:lastRelease}|release A
    1146 \end{pseudo}
     1589\end{cfa}
    11471590\end{multicols}
    1148 \begin{cfacode}[caption={Internal scheduling with \textbf{bulk-acq}},label={lst:int-bulk-pseudo}]
    1149 \end{cfacode}
     1591\begin{cfa}[caption={Internal scheduling with \textbf{bulk-acq}},label={f:int-bulk-cfa}]
     1592\end{cfa}
    11501593\begin{center}
    1151 \begin{cfacode}[xleftmargin=.4\textwidth]
     1594\begin{cfa}[xleftmargin=.4\textwidth]
    11521595monitor A a;
    11531596monitor B b;
    11541597condition c;
    1155 \end{cfacode}
     1598\end{cfa}
    11561599\end{center}
    11571600\begin{multicols}{2}
    11581601Waiting thread
    1159 \begin{cfacode}
     1602\begin{cfa}
    11601603mutex(a) {
    1161         //Code Section 1
     1604        // Code Section 1
    11621605        mutex(a, b) {
    1163                 //Code Section 2
     1606                // Code Section 2
    11641607                wait(c);
    1165                 //Code Section 3
     1608                // Code Section 3
    11661609        }
    1167         //Code Section 4
    1168 }
    1169 \end{cfacode}
     1610        // Code Section 4
     1611}
     1612\end{cfa}
    11701613\columnbreak
    11711614Signalling thread
    1172 \begin{cfacode}
     1615\begin{cfa}
    11731616mutex(a) {
    1174         //Code Section 5
     1617        // Code Section 5
    11751618        mutex(a, b) {
    1176                 //Code Section 6
     1619                // Code Section 6
    11771620                signal(c);
    1178                 //Code Section 7
     1621                // Code Section 7
    11791622        }
    1180         //Code Section 8
    1181 }
    1182 \end{cfacode}
     1623        // Code Section 8
     1624}
     1625\end{cfa}
    11831626\end{multicols}
    1184 \begin{cfacode}[caption={Equivalent \CFA code for listing \ref{lst:int-bulk-pseudo}},label={lst:int-bulk-cfa}]
    1185 \end{cfacode}
     1627\begin{cfa}[caption={Equivalent \CFA code for listing \ref{f:int-bulk-cfa}},label={f:int-bulk-cfa}]
     1628\end{cfa}
    11861629\begin{multicols}{2}
    11871630Waiter
    1188 \begin{pseudo}[numbers=left]
     1631\begin{cfa}[numbers=left]
    11891632acquire A
    11901633        acquire A & B
    … …  
    11921635        release A & B
    11931636release A
    1194 \end{pseudo}
     1637\end{cfa}
    11951638
    11961639\columnbreak
    11971640
    11981641Signaller
    1199 \begin{pseudo}[numbers=left, firstnumber=6,escapechar=|]
     1642\begin{cfa}[numbers=left, firstnumber=6,escapechar=|]
    12001643acquire A
    12011644        acquire A & B
    12021645                signal A & B
    12031646        release A & B
    1204         |\label{line:secret}|//Secretly keep B here
     1647        |\label{line:secret}|// Secretly keep B here
    12051648release A
    1206 //Wakeup waiter and transfer A & B
    1207 \end{pseudo}
     1649// Wakeup waiter and transfer A & B
     1650\end{cfa}
    12081651\end{multicols}
    1209 \begin{cfacode}[caption={Listing \ref{lst:int-bulk-pseudo}, with delayed signalling comments},label={lst:int-secret}]
    1210 \end{cfacode}
     1652\begin{cfa}[caption={Figure~\ref{f:int-bulk-cfa}, with delayed signalling comments},label={f:int-secret}]
     1653\end{cfa}
    12111654\end{figure}
    12121655
    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:
     1656The complexity begins at code sections 4 and 8 in listing \ref{f:int-bulk-cfa}, which are where the existing semantics of internal scheduling needs to be extended for multiple monitors.
     1657The 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 cfa-code.
     1658When the signaller thread reaches the location where it should ``release @A & B@'' (listing \ref{f:int-bulk-cfa} line \ref{line:releaseFirst}), it must actually transfer ownership of monitor @B@ to the waiting thread.
     1659This ownership transfer is required in order to prevent barging into @B@ by another thread, since both the signalling and signalled threads still need monitor @A@.
     1660There are three options:
    12141661
    12151662\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:
    1219 
    1220 \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.
    1221 \paragraph{Case 2: thread $\beta$ goes first.} In this case, the problem is that monitor \code{B} needs to be retained and passed to thread $\alpha$ along with monitor \code{A}, which can be done directly or possibly using thread $\beta$ as an intermediate.
     1663The 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.
     1664It 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.
     1665This 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.
     1666This solution releases the monitors once every monitor in a group can be released.
     1667However, since some monitors are never released (\eg the monitor of a thread), this interpretation means a group might never be released.
     1668A 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.
     1669
     1670However, listing \ref{f: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.
     1671Figure~\ref{f:dependency} shows a slightly different example where a third thread is waiting on monitor @A@, using a different condition variable.
     1672Because the third thread is signalled when secretly holding @B@, the goal  becomes unreachable.
     1673Depending on the order of signals (listing \ref{f:dependency} line \ref{line:signal-ab} and \ref{line:signal-a}) two cases can happen:
     1674
     1675\paragraph{Case 1: thread $\alpha$ goes first.} In this case, the problem is that monitor @A@ needs to be passed to thread $\beta$ when thread $\alpha$ is done with it.
     1676\paragraph{Case 2: thread $\beta$ goes first.} In this case, the problem is that monitor @B@ needs to be retained and passed to thread $\alpha$ along with monitor @A@, which can be done directly or possibly using thread $\beta$ as an intermediate.
    12221677\\
    12231678
    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}.
     1679Note that ordering is not determined by a race condition but by whether signalled threads are enqueued in FIFO or FILO order.
     1680However, 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{f:dependency}.
    12251681
    12261682In 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.
    … …  
    12321688\begin{multicols}{3}
    12331689Thread $\alpha$
    1234 \begin{pseudo}[numbers=left, firstnumber=1]
     1690\begin{cfa}[numbers=left, firstnumber=1]
    12351691acquire A
    12361692        acquire A & B
    … …  
    12381694        release A & B
    12391695release A
    1240 \end{pseudo}
     1696\end{cfa}
    12411697\columnbreak
    12421698Thread $\gamma$
    1243 \begin{pseudo}[numbers=left, firstnumber=6, escapechar=|]
     1699\begin{cfa}[numbers=left, firstnumber=6, escapechar=|]
    12441700acquire A
    12451701        acquire A & B
    … …  
    12481704        |\label{line:signal-a}|signal A
    12491705|\label{line:release-a}|release A
    1250 \end{pseudo}
     1706\end{cfa}
    12511707\columnbreak
    12521708Thread $\beta$
    1253 \begin{pseudo}[numbers=left, firstnumber=12, escapechar=|]
     1709\begin{cfa}[numbers=left, firstnumber=12, escapechar=|]
    12541710acquire A
    12551711        wait A
    12561712|\label{line:release-aa}|release A
    1257 \end{pseudo}
     1713\end{cfa}
    12581714\end{multicols}
    1259 \begin{cfacode}[caption={Pseudo-code for the three thread example.},label={lst:dependency}]
    1260 \end{cfacode}
     1715\begin{cfa}[caption={Pseudo-code for the three thread example.},label={f:dependency}]
     1716\end{cfa}
    12611717\begin{center}
    12621718\input{dependency}
    12631719\end{center}
    1264 \caption{Dependency graph of the statements in listing \ref{lst:dependency}}
     1720\caption{Dependency graph of the statements in listing \ref{f:dependency}}
    12651721\label{fig:dependency}
    12661722\end{figure}
    12671723
    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.
     1724In listing \ref{f:int-bulk-cfa}, there is a solution that satisfies both barging prevention and mutual exclusion.
     1725If ownership of both monitors is transferred to the waiter when the signaller releases @A & B@ and then the waiter transfers back ownership of @A@ back to the signaller when it releases it, then the problem is solved (@B@ is no longer in use at this point).
     1726Dynamically finding the correct order is therefore the second possible solution.
     1727The problem is effectively resolving a dependency graph of ownership requirements.
     1728Here even the simplest of code snippets requires two transfers and has a super-linear complexity.
     1729This complexity can be seen in listing \ref{f:explosion}, which is just a direct extension to three monitors, requires at least three ownership transfer and has multiple solutions.
     1730Furthermore, 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.
    12691731\begin{figure}
    12701732\begin{multicols}{2}
    1271 \begin{pseudo}
     1733\begin{cfa}
    12721734acquire A
    12731735        acquire B
    … …  
    12771739        release B
    12781740release A
    1279 \end{pseudo}
     1741\end{cfa}
    12801742
    12811743\columnbreak
    12821744
    1283 \begin{pseudo}
     1745\begin{cfa}
    12841746acquire A
    12851747        acquire B
    … …  
    12891751        release B
    12901752release A
    1291 \end{pseudo}
     1753\end{cfa}
    12921754\end{multicols}
    1293 \begin{cfacode}[caption={Extension to three monitors of listing \ref{lst:int-bulk-pseudo}},label={lst:explosion}]
    1294 \end{cfacode}
     1755\begin{cfa}[caption={Extension to three monitors of listing \ref{f:int-bulk-cfa}},label={f:explosion}]
     1756\end{cfa}
    12951757\end{figure}
    12961758
    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.
     1759Given the three threads example in listing \ref{f: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 (\eg $\alpha1$ must happen before $\alpha2$).
     1760The 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.
     1761Resolving dependency graphs being a complex and expensive endeavour, this solution is not the preferred one.
    12981762
    12991763\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.
    1301 
    1302 Using partial signalling, listing \ref{lst:dependency} can be solved easily:
     1764Finally, the solution that is chosen for \CFA is to use partial signalling.
     1765Again using listing \ref{f:int-bulk-cfa}, the partial signalling solution transfers ownership of monitor @B@ at lines \ref{line:signal1} to the waiter but does not wake the waiting thread since it is still using monitor @A@.
     1766Only when it reaches line \ref{line:lastRelease} does it actually wake up the waiting thread.
     1767This 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.
     1768This solution has a much simpler implementation than a dependency graph solving algorithms, which is why it was chosen.
     1769Furthermore, after being fully implemented, this solution does not appear to have any significant downsides.
     1770
     1771Using partial signalling, listing \ref{f:dependency} can be solved easily:
    13031772\begin{itemize}
    1304         \item When thread $\gamma$ reaches line \ref{line:release-ab} it transfers monitor \code{B} to thread $\alpha$ and continues to hold monitor \code{A}.
    1305         \item When thread $\gamma$ reaches line \ref{line:release-a}  it transfers monitor \code{A} to thread $\beta$  and wakes it up.
    1306         \item When thread $\beta$  reaches line \ref{line:release-aa} it transfers monitor \code{A} to thread $\alpha$ and wakes it up.
     1773        \item When thread $\gamma$ reaches line \ref{line:release-ab} it transfers monitor @B@ to thread $\alpha$ and continues to hold monitor @A@.
     1774        \item When thread $\gamma$ reaches line \ref{line:release-a}  it transfers monitor @A@ to thread $\beta$  and wakes it up.
     1775        \item When thread $\beta$  reaches line \ref{line:release-aa} it transfers monitor @A@ to thread $\alpha$ and wakes it up.
    13071776\end{itemize}
    13081777
    … …  
    13141783\begin{table}
    13151784\begin{tabular}{|c|c|}
    1316 \code{signal} & \code{signal_block} \\
     1785@signal@ & @signal_block@ \\
    13171786\hline
    1318 \begin{cfacode}[tabsize=3]
    1319 monitor DatingService
    1320 {
    1321         //compatibility codes
     1787\begin{cfa}[tabsize=3]
     1788monitor DatingService {
     1789        // compatibility codes
    13221790        enum{ CCodes = 20 };
    13231791
    … …  
    13301798condition exchange;
    13311799
    1332 int girl(int phoneNo, int ccode)
    1333 {
    1334         //no compatible boy ?
    1335         if(empty(boys[ccode]))
    1336         {
    1337                 //wait for boy
    1338                 wait(girls[ccode]);
    1339 
    1340                 //make phone number available
    1341                 girlPhoneNo = phoneNo;
    1342 
    1343                 //wake boy from chair
    1344                 signal(exchange);
    1345         }
    1346         else
    1347         {
    1348                 //make phone number available
    1349                 girlPhoneNo = phoneNo;
    1350 
    1351                 //wake boy
    1352                 signal(boys[ccode]);
    1353 
    1354                 //sit in chair
    1355                 wait(exchange);
     1800int girl(int phoneNo, int cfa) {
     1801        // no compatible boy ?
     1802        if(empty(boys[cfa])) {
     1803                wait(girls[cfa]);               // wait for boy
     1804                girlPhoneNo = phoneNo;          // make phone number available
     1805                signal(exchange);               // wake boy from chair
     1806        } else {
     1807                girlPhoneNo = phoneNo;          // make phone number available
     1808                signal(boys[cfa]);              // wake boy
     1809                wait(exchange);         // sit in chair
    13561810        }
    13571811        return boyPhoneNo;
    13581812}
    1359 
    1360 int boy(int phoneNo, int ccode)
    1361 {
    1362         //same as above
    1363         //with boy/girl interchanged
    1364 }
    1365 \end{cfacode}&\begin{cfacode}[tabsize=3]
    1366 monitor DatingService
    1367 {
    1368         //compatibility codes
    1369         enum{ CCodes = 20 };
     1813int boy(int phoneNo, int cfa) {
     1814        // same as above
     1815        // with boy/girl interchanged
     1816}
     1817\end{cfa}&\begin{cfa}[tabsize=3]
     1818monitor DatingService {
     1819
     1820        enum{ CCodes = 20 };    // compatibility codes
    13701821
    13711822        int girlPhoneNo;
    … …  
    13751826condition girls[CCodes];
    13761827condition boys [CCodes];
    1377 //exchange is not needed
    1378 
    1379 int girl(int phoneNo, int ccode)
    1380 {
    1381         //no compatible boy ?
    1382         if(empty(boys[ccode]))
    1383         {
    1384                 //wait for boy
    1385                 wait(girls[ccode]);
    1386 
    1387                 //make phone number available
    1388                 girlPhoneNo = phoneNo;
    1389 
    1390                 //wake boy from chair
    1391                 signal(exchange);
    1392         }
    1393         else
    1394         {
    1395                 //make phone number available
    1396                 girlPhoneNo = phoneNo;
    1397 
    1398                 //wake boy
    1399                 signal_block(boys[ccode]);
    1400 
    1401                 //second handshake unnecessary
     1828// exchange is not needed
     1829
     1830int girl(int phoneNo, int cfa) {
     1831        // no compatible boy ?
     1832        if(empty(boys[cfa])) {
     1833                wait(girls[cfa]);               // wait for boy
     1834                girlPhoneNo = phoneNo;          // make phone number available
     1835                signal(exchange);               // wake boy from chair
     1836        } else {
     1837                girlPhoneNo = phoneNo;          // make phone number available
     1838                signal_block(boys[cfa]);                // wake boy
     1839
     1840                // second handshake unnecessary
    14021841
    14031842        }
    … …  
    14051844}
    14061845
    1407 int boy(int phoneNo, int ccode)
    1408 {
    1409         //same as above
    1410         //with boy/girl interchanged
    1411 }
    1412 \end{cfacode}
     1846int boy(int phoneNo, int cfa) {
     1847        // same as above
     1848        // with boy/girl interchanged
     1849}
     1850\end{cfa}
    14131851\end{tabular}
    1414 \caption{Dating service example using \code{signal} and \code{signal_block}. }
     1852\caption{Dating service example using \protect\lstinline|signal| and \protect\lstinline|signal_block|. }
    14151853\label{tbl:datingservice}
    14161854\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.
     1855An important note is that, until now, signalling a monitor was a delayed operation.
     1856The ownership of the monitor is transferred only when the monitor would have otherwise been released, not at the point of the @signal@ statement.
     1857However, 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 @signal_block@ routine.
     1858
     1859The example in table \ref{tbl:datingservice} highlights the difference in behaviour.
     1860As mentioned, @signal@ only transfers ownership once the current critical section exits; this behaviour requires additional synchronization when a two-way handshake is needed.
     1861To avoid this explicit synchronization, the @condition@ type offers the @signal_block@ routine, which handles the two-way handshake as shown in the example.
     1862This feature removes the need for a second condition variables and simplifies programming.
     1863Like every other monitor semantic, @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 @signal_block@, meaning no other thread can acquire the monitor either before or after the call.
    14201864
    14211865% ======================================================================
    … …  
    14291873Internal Scheduling & External Scheduling & Go\\
    14301874\hline
    1431 \begin{ucppcode}[tabsize=3]
     1875\begin{uC++}[tabsize=3]
    14321876_Monitor Semaphore {
    14331877        condition c;
    … …  
    14441888        }
    14451889}
    1446 \end{ucppcode}&\begin{ucppcode}[tabsize=3]
     1890\end{uC++}&\begin{uC++}[tabsize=3]
    14471891_Monitor Semaphore {
    14481892
    … …  
    14591903        }
    14601904}
    1461 \end{ucppcode}&\begin{gocode}[tabsize=3]
     1905\end{uC++}&\begin{Go}[tabsize=3]
    14621906type MySem struct {
    14631907        inUse bool
    … …  
    14791923        s.inUse = false
    14801924
    1481         //This actually deadlocks
    1482         //when single thread
     1925        // This actually deadlocks
     1926        // when single thread
    14831927        s.c <- false
    14841928}
    1485 \end{gocode}
     1929\end{Go}
    14861930\end{tabular}
    14871931\caption{Different forms of scheduling.}
    14881932\label{tbl:sched}
    14891933\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.
     1934This method is more constrained and explicit, which helps users reduce the non-deterministic nature of concurrency.
     1935Indeed, as the following examples demonstrate, external scheduling allows users to wait for events from other threads without the concern of unrelated events occurring.
     1936External scheduling can generally be done either in terms of control flow (\eg Ada with @accept@, \uC with @_Accept@) or in terms of data (\eg Go with channels).
     1937Of 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.
     1938Two challenges specific to \CFA arise when trying to add external scheduling with loose object definitions and multiple-monitor routines.
     1939The previous example shows a simple use @_Accept@ versus @wait@/@signal@ and its advantages.
     1940Note that while other languages often use @accept@/@select@ as the core external scheduling keyword, \CFA uses @waitfor@ to prevent name collisions with existing socket \textbf{api}s.
     1941
     1942For the @P@ member above using internal scheduling, the call to @wait@ only guarantees that @V@ is the last routine to access the monitor, allowing a third routine, say @isInUse()@, acquire mutual exclusion several times while routine @P@ is waiting.
     1943On the other hand, external scheduling guarantees that while routine @P@ is waiting, no other routine than @V@ can acquire the monitor.
    14931944
    14941945% ======================================================================
    … …  
    14971948% ======================================================================
    14981949% ======================================================================
    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:
    1500 
    1501 \begin{cfacode}
     1950In \uC, a monitor class declaration includes an exhaustive list of monitor operations.
     1951Since \CFA is not object oriented, monitors become both more difficult to implement and less clear for a user:
     1952
     1953\begin{cfa}
    15021954monitor A {};
    15031955
    15041956void f(A & mutex a);
    15051957void g(A & mutex a) {
    1506         waitfor(f); //Obvious which f() to wait for
    1507 }
    1508 
    1509 void f(A & mutex a, int); //New different F added in scope
     1958        waitfor(f); // Obvious which f() to wait for
     1959}
     1960
     1961void f(A & mutex a, int); // New different F added in scope
    15101962void h(A & mutex a) {
    1511         waitfor(f); //Less obvious which f() to wait for
    1512 }
    1513 \end{cfacode}
    1514 
    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:
     1963        waitfor(f); // Less obvious which f() to wait for
     1964}
     1965\end{cfa}
     1966
     1967Furthermore, external scheduling is an example where implementation constraints become visible from the interface.
     1968Here is the cfa-code for the entering phase of a monitor:
    15161969\begin{center}
    15171970\begin{tabular}{l}
    1518 \begin{pseudo}
     1971\begin{cfa}
    15191972        if monitor is free
    15201973                enter
    … …  
    15251978        else
    15261979                block
    1527 \end{pseudo}
     1980\end{cfa}
    15281981\end{tabular}
    15291982\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}.
     1983For the first two conditions, it is easy to implement a check that can evaluate the condition in a few instructions.
     1984However, a fast check for @monitor accepts me@ is much harder to implement depending on the constraints put on the monitors.
     1985Indeed, monitors are often expressed as an entry queue and some acceptor queue as in Figure~\ref{fig:ClassicalMonitor}.
    15311986
    15321987\begin{figure}
    … …  
    15441999\end{figure}
    15452000
    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.
     2001There are other alternatives to these pictures, but in the case of the left picture, implementing a fast accept check is relatively easy.
     2002Restricted 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 (\eg 128) of mutex members.
     2003This approach requires a unique dense ordering of routines with an upper-bound and that ordering must be consistent across translation units.
     2004For OO languages these constraints are common, since objects only offer adding member routines consistently across translation units via inheritance.
     2005However, in \CFA users can extend objects with mutex routines that are only visible in certain translation unit.
     2006This 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.
    15472007
    15482008The 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.
     2009Here, the mutex routine called is associated with a thread on the entry queue while a list of acceptable routines is kept separate.
     2010Generating a mask dynamically means that the storage for the mask information can vary between calls to @waitfor@, allowing for more flexibility and extensions.
     2011Storing an array of accepted function pointers replaces the single instruction bitmask comparison with dereferencing a pointer followed by a linear search.
     2012Furthermore, supporting nested external scheduling (\eg listing \ref{f:nest-ext}) may now require additional searches for the @waitfor@ statement to check if a routine is already queued.
    15502013
    15512014\begin{figure}
    1552 \begin{cfacode}[caption={Example of nested external scheduling},label={lst:nest-ext}]
     2015\begin{cfa}[caption={Example of nested external scheduling},label={f:nest-ext}]
    15532016monitor M {};
    15542017void foo( M & mutex a ) {}
    15552018void bar( M & mutex b ) {
    1556         //Nested in the waitfor(bar, c) call
     2019        // Nested in the waitfor(bar, c) call
    15572020        waitfor(foo, b);
    15582021}
    … …  
    15612024}
    15622025
    1563 \end{cfacode}
     2026\end{cfa}
    15642027\end{figure}
    15652028
    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.
     2029Note 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.
     2030These details are omitted from the picture for the sake of simplicity.
     2031
     2032At this point, a decision must be made between flexibility and performance.
     2033Many 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.
     2034Here, however, the cost of flexibility cannot be trivially removed.
     2035In the end, the most flexible approach has been chosen since it allows users to write programs that would otherwise be  hard to write.
     2036This 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.
    15692037
    15702038% ======================================================================
    … …  
    15742042% ======================================================================
    15752043
    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:
    1577 \begin{cfacode}
     2044External scheduling, like internal scheduling, becomes significantly more complex when introducing multi-monitor syntax.
     2045Even in the simplest possible case, some new semantics needs to be established:
     2046\begin{cfa}
    15782047monitor M {};
    15792048
    … …  
    15812050
    15822051void g(M & mutex b, M & mutex c) {
    1583         waitfor(f); //two monitors M => unknown which to pass to f(M & mutex)
    1584 }
    1585 \end{cfacode}
     2052        waitfor(f); // two monitors M => unknown which to pass to f(M & mutex)
     2053}
     2054\end{cfa}
    15862055The obvious solution is to specify the correct monitor as follows:
    15872056
    1588 \begin{cfacode}
     2057\begin{cfa}
    15892058monitor M {};
    15902059
    … …  
    15922061
    15932062void g(M & mutex a, M & mutex b) {
    1594         //wait for call to f with argument b
     2063        // wait for call to f with argument b
    15952064        waitfor(f, b);
    15962065}
    1597 \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.
    1599 
    1600 \begin{cfacode}
     2066\end{cfa}
     2067This syntax is unambiguous.
     2068Both locks are acquired and kept by @g@.
     2069When routine @f@ is called, the lock for monitor @b@ is temporarily transferred from @g@ to @f@ (while @g@ still holds lock @a@).
     2070This behaviour can be extended to the multi-monitor @waitfor@ statement as follows.
     2071
     2072\begin{cfa}
    16012073monitor M {};
    16022074
    … …  
    16042076
    16052077void g(M & mutex a, M & mutex b) {
    1606         //wait for call to f with arguments a and b
     2078        // wait for call to f with arguments a and b
    16072079        waitfor(f, a, b);
    16082080}
    1609 \end{cfacode}
    1610 
    1611 Note that the set of monitors passed to the \code{waitfor} statement must be entirely contained in the set of monitors already acquired in the routine. \code{waitfor} used in any other context is undefined behaviour.
     2081\end{cfa}
     2082
     2083Note that the set of monitors passed to the @waitfor@ statement must be entirely contained in the set of monitors already acquired in the routine. @waitfor@ used in any other context is undefined behaviour.
    16122084
    16132085An important behaviour to note is when a set of monitors only match partially:
    16142086
    1615 \begin{cfacode}
     2087\begin{cfa}
    16162088mutex struct A {};
    16172089
    … …  
    16262098
    16272099void foo() {
    1628         g(a1, b); //block on accept
     2100        g(a1, b); // block on accept
    16292101}
    16302102
    16312103void bar() {
    1632         f(a2, b); //fulfill cooperation
    1633 }
    1634 \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.
    1636 
    1637 % ======================================================================
    1638 % ======================================================================
    1639 \subsection{\code{waitfor} Semantics}
    1640 % ======================================================================
    1641 % ======================================================================
    1642 
    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.
     2104        f(a2, b); // fulfill cooperation
     2105}
     2106\end{cfa}
     2107While 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.
     2108In both cases, partially matching monitor sets does not wakeup the waiting thread.
     2109It is also important to note that in the case of external scheduling the order of parameters is irrelevant; @waitfor(f,a,b)@ and @waitfor(f,b,a)@ are indistinguishable waiting condition.
     2110
     2111% ======================================================================
     2112% ======================================================================
     2113\subsection{\protect\lstinline|waitfor| Semantics}
     2114% ======================================================================
     2115% ======================================================================
     2116
     2117Syntactically, the @waitfor@ statement takes a function identifier and a set of monitors.
     2118While 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 @waitfor@ statement.
     2119It checks that the set of monitors passed in matches the requirements for a function call.
     2120Figure~\ref{f:waitfor} shows various usages of the waitfor statement and which are acceptable.
     2121The choice of the function type is made ignoring any non-@mutex@ parameter.
     2122One limitation of the current implementation is that it does not handle overloading, but overloading is possible.
    16442123\begin{figure}
    1645 \begin{cfacode}[caption={Various correct and incorrect uses of the waitfor statement},label={lst:waitfor}]
     2124\begin{cfa}[caption={Various correct and incorrect uses of the waitfor statement},label={f:waitfor}]
    16462125monitor A{};
    16472126monitor B{};
    … …  
    16572136        void (*fp)( A & mutex ) = f1;
    16582137
    1659         waitfor(f1, a1);     //Correct : 1 monitor case
    1660         waitfor(f2, a1, b1); //Correct : 2 monitor case
    1661         waitfor(f3, a1);     //Correct : non-mutex arguments are ignored
    1662         waitfor(f1, *ap);    //Correct : expression as argument
    1663 
    1664         waitfor(f1, a1, b1); //Incorrect : Too many mutex arguments
    1665         waitfor(f2, a1);     //Incorrect : Too few mutex arguments
    1666         waitfor(f2, a1, a2); //Incorrect : Mutex arguments don't match
    1667         waitfor(f1, 1);      //Incorrect : 1 not a mutex argument
    1668         waitfor(f9, a1);     //Incorrect : f9 function does not exist
    1669         waitfor(*fp, a1 );   //Incorrect : fp not an identifier
    1670         waitfor(f4, a1);     //Incorrect : f4 ambiguous
    1671 
    1672         waitfor(f2, a1, b2); //Undefined behaviour : b2 not mutex
    1673 }
    1674 \end{cfacode}
     2138        waitfor(f1, a1);     // Correct : 1 monitor case
     2139        waitfor(f2, a1, b1); // Correct : 2 monitor case
     2140        waitfor(f3, a1);     // Correct : non-mutex arguments are ignored
     2141        waitfor(f1, *ap);    // Correct : expression as argument
     2142
     2143        waitfor(f1, a1, b1); // Incorrect : Too many mutex arguments
     2144        waitfor(f2, a1);     // Incorrect : Too few mutex arguments
     2145        waitfor(f2, a1, a2); // Incorrect : Mutex arguments don't match
     2146        waitfor(f1, 1);      // Incorrect : 1 not a mutex argument
     2147        waitfor(f9, a1);     // Incorrect : f9 function does not exist
     2148        waitfor(*fp, a1 );   // Incorrect : fp not an identifier
     2149        waitfor(f4, a1);     // Incorrect : f4 ambiguous
     2150
     2151        waitfor(f2, a1, b2); // Undefined behaviour : b2 not mutex
     2152}
     2153\end{cfa}
    16752154\end{figure}
    16762155
    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.
     2156Finally, for added flexibility, \CFA supports constructing a complex @waitfor@ statement using the @or@, @timeout@ and @else@.
     2157Indeed, multiple @waitfor@ clauses can be chained together using @or@; this chain forms a single statement that uses baton pass to any function that fits one of the function+monitor set passed in.
     2158To enable users to tell which accepted function executed, @waitfor@s are followed by a statement (including the null statement @;@) or a compound statement, which is executed after the clause is triggered.
     2159A @waitfor@ chain can also be followed by a @timeout@, to signify an upper bound on the wait, or an @else@, to signify that the call should be non-blocking, which checks for a matching function call already arrived and otherwise continues.
     2160Any and all of these clauses can be preceded by a @when@ condition to dynamically toggle the accept clauses on or off based on some current state.
     2161Figure~\ref{f:waitfor2} demonstrates several complex masks and some incorrect ones.
    16782162
    16792163\begin{figure}
    1680 \begin{cfacode}[caption={Various correct and incorrect uses of the or, else, and timeout clause around a waitfor statement},label={lst:waitfor2}]
     2164\lstset{language=CFA,deletedelim=**[is][]{`}{`}}
     2165\begin{cfa}
    16812166monitor A{};
    16822167
    … …  
    16852170
    16862171void foo( A & mutex a, bool b, int t ) {
    1687         //Correct : blocking case
    1688         waitfor(f1, a);
    1689 
    1690         //Correct : block with statement
    1691         waitfor(f1, a) {
     2172        waitfor(f1, a);                                                 $\C{// Correct : blocking case}$
     2173
     2174        waitfor(f1, a) {                                                $\C{// Correct : block with statement}$
    16922175                sout | "f1" | endl;
    16932176        }
    1694 
    1695         //Correct : block waiting for f1 or f2
    1696         waitfor(f1, a) {
     2177        waitfor(f1, a) {                                                $\C{// Correct : block waiting for f1 or f2}$
    16972178                sout | "f1" | endl;
    16982179        } or waitfor(f2, a) {
    16992180                sout | "f2" | endl;
    17002181        }
    1701 
    1702         //Correct : non-blocking case
    1703         waitfor(f1, a); or else;
    1704 
    1705         //Correct : non-blocking case
    1706         waitfor(f1, a) {
     2182        waitfor(f1, a); or else;                                $\C{// Correct : non-blocking case}$
     2183
     2184        waitfor(f1, a) {                                                $\C{// Correct : non-blocking case}$
    17072185                sout | "blocked" | endl;
    17082186        } or else {
    17092187                sout | "didn't block" | endl;
    17102188        }
    1711 
    1712         //Correct : block at most 10 seconds
    1713         waitfor(f1, a) {
     2189        waitfor(f1, a) {                                                $\C{// Correct : block at most 10 seconds}$
    17142190                sout | "blocked" | endl;
    17152191        } or timeout( 10`s) {
    17162192                sout | "didn't block" | endl;
    17172193        }
    1718 
    1719         //Correct : block only if b == true
    1720         //if b == false, don't even make the call
     2194        // Correct : block only if b == true if b == false, don't even make the call
    17212195        when(b) waitfor(f1, a);
    17222196
    1723         //Correct : block only if b == true
    1724         //if b == false, make non-blocking call
     2197        // Correct : block only if b == true if b == false, make non-blocking call
    17252198        waitfor(f1, a); or when(!b) else;
    17262199
    1727         //Correct : block only of t > 1
     2200        // Correct : block only of t > 1
    17282201        waitfor(f1, a); or when(t > 1) timeout(t); or else;
    17292202
    1730         //Incorrect : timeout clause is dead code
     2203        // Incorrect : timeout clause is dead code
    17312204        waitfor(f1, a); or timeout(t); or else;
    17322205
    1733         //Incorrect : order must be
    1734         //waitfor [or waitfor... [or timeout] [or else]]
     2206        // Incorrect : order must be waitfor [or waitfor... [or timeout] [or else]]
    17352207        timeout(t); or waitfor(f1, a); or else;
    17362208}
    1737 \end{cfacode}
     2209\end{cfa}
     2210\caption{Correct and incorrect uses of the or, else, and timeout clause around a waitfor statement}
     2211\label{f:waitfor2}
    17382212\end{figure}
    17392213
    … …  
    17432217% ======================================================================
    17442218% ======================================================================
    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.
     2219An interesting use for the @waitfor@ statement is destructor semantics.
     2220Indeed, the @waitfor@ statement can accept any @mutex@ routine, which includes the destructor (see section \ref{data}).
     2221However, 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.
     2222The simplest approach is to disallow @waitfor@ on a destructor.
     2223However, 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 @mutex@ routine, similarly to how a condition is signalled.
    17462224\begin{figure}
    1747 \begin{cfacode}[caption={Example of an executor which executes action in series until the destructor is called.},label={lst:dtor-order}]
     2225\begin{cfa}[caption={Example of an executor which executes action in series until the destructor is called.},label={f:dtor-order}]
    17482226monitor Executer {};
    17492227struct  Action;
    … …  
    17592237        }
    17602238}
    1761 \end{cfacode}
     2239\end{cfa}
    17622240\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.
     2241For example, listing \ref{f:dtor-order} shows an example of an executor with an infinite loop, which waits for the destructor to break out of this loop.
     2242Switching the semantic meaning introduces an idiomatic way to terminate a task and/or wait for its termination via destruction.
    17642243
    17652244
    … …  
    17722251% #       #     # #     # #     # ####### ####### ####### ####### ###  #####  #     #
    17732252\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.
     2253Historically, computer performance was about processor speeds and instruction counts.
     2254However, with heat dissipation being a direct consequence of speed increase, parallelism has become the new source for increased performance~\cite{Sutter05, Sutter05b}.
     2255In this decade, it is no longer reasonable to create a high-performance application without caring about parallelism.
     2256Indeed, parallelism is an important aspect of performance and more specifically throughput and hardware utilization.
     2257The lowest-level approach of parallelism is to use \textbf{kthread} in combination with semantics like @fork@, @join@, \etc.
     2258However, since these have significant costs and limitations, \textbf{kthread} are now mostly used as an implementation tool rather than a user oriented one.
     2259There are several alternatives to solve these issues that all have strengths and weaknesses.
     2260While 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.
    17752261
    17762262\section{Paradigms}
    17772263\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.
     2264A direct improvement on the \textbf{kthread} approach is to use \textbf{uthread}.
     2265These threads offer most of the same features that the operating system already provides but can be used on a much larger scale.
     2266This 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.
     2267The 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.
     2268These issues can be somewhat alleviated by a concurrency toolkit with strong guarantees, but the parallelism toolkit offers very little to reduce complexity in itself.
    17792269
    17802270Examples of languages that support \textbf{uthread} are Erlang~\cite{Erlang} and \uC~\cite{uC++book}.
    17812271
    17822272\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.
     2273A popular variant of \textbf{uthread} is what is often referred to as \textbf{fiber}.
     2274However, \textbf{fiber} do not present meaningful semantic differences with \textbf{uthread}.
     2275The significant difference between \textbf{uthread} and \textbf{fiber} is the lack of \textbf{preemption} in the latter.
     2276Advocates 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.
     2277Therefore this proposal largely ignores fibers.
    17842278
    17852279An example of a language that uses fibers is Go~\cite{Go}
    17862280
    17872281\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.
     2282An approach on the opposite end of the spectrum is to base parallelism on \textbf{pool}.
     2283Indeed, \textbf{pool} offer limited flexibility but at the benefit of a simpler user interface.
     2284In \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.
     2285This approach means users need not worry about concurrency but significantly limit the interaction that can occur among jobs.
     2286Indeed, any \textbf{job} that blocks also block the underlying worker, which effectively means the CPU utilization, and therefore throughput, suffers noticeably.
     2287It can be argued that a solution to this problem is to use more workers than available cores.
     2288However, unless the number of jobs and the number of workers are comparable, having a significant number of blocked jobs always results in idles cores.
    17892289
    17902290The gold standard of this implementation is Intel's TBB library~\cite{TBB}.
    17912291
    17922292\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.
     2293While 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.
     2294Indeed, in many situations one of these paradigms may show better performance but it all strongly depends on the workload.
     2295Having 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 (\ie no thread stack per job).
     2296However, interactions among jobs can easily exacerbate contention.
     2297User-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.
     2298Finally, if the units of uninterrupted work are large, enough the paradigm choice is largely amortized by the actual work done.
    17942299
    17952300\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.
     2301A \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}.
     2302It 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.
     2303A \textbf{cfacluster} also offers a pluggable scheduler that can optimize the workload generated by the \textbf{uthread}.
     2304
     2305\textbf{cfacluster} have not been fully implemented in the context of this paper.
     2306Currently \CFA only supports one \textbf{cfacluster}, the initial one.
    17992307
    18002308\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.
     2309While this was not done in the context of this paper, another important aspect of clusters is affinity.
     2310While many common desktop and laptop PCs have homogeneous CPUs, other devices often have more heterogeneous setups.
     2311For example, a system using \textbf{numa} configurations may benefit from users being able to tie clusters and/or kernel threads to certain CPU cores.
     2312OS 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.
    18022313
    18032314\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}.
     2315Given these building blocks, it is possible to reproduce all three of the popular paradigms.
     2316Indeed, \textbf{uthread} is the default paradigm in \CFA.
     2317However, disabling \textbf{preemption} on the \textbf{cfacluster} means \textbf{cfathread} effectively become \textbf{fiber}.
     2318Since 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.
     2319Finally, it is possible to build executors for thread pools from \textbf{uthread} or \textbf{fiber}, which includes specialized jobs like actors~\cite{Actors}.
    18052320
    18062321
    18072322
    18082323\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.
     2324There are several challenges specific to \CFA when implementing concurrency.
     2325These challenges are a direct result of \textbf{bulk-acq} and loose object definitions.
     2326These two constraints are the root cause of most design decisions in the implementation.
     2327Furthermore, to avoid contention from dynamically allocating memory in a concurrent environment, the internal-scheduling design is (almost) entirely free of mallocs.
     2328This 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.
     2329This extra goal means that memory management is a constant concern in the design of the system.
     2330
     2331The main memory concern for concurrency is queues.
     2332All 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.
     2333Since 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.
     2334Conveniently, 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.
     2335Since 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.
     2336The threads and the condition both have a fixed amount of memory, while @mutex@ routines and blocking calls allow for an unbound amount, within the stack size.
    18122337
    18132338Note 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.
    … …  
    18192344% ======================================================================
    18202345
    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.
     2346The first step towards the monitor implementation is simple @mutex@ routines.
     2347In the single monitor case, mutual-exclusion is done using the entry/exit procedure in listing \ref{f:entry1}.
     2348The entry/exit procedures do not have to be extended to support multiple monitors.
     2349Indeed it is sufficient to enter/leave monitors one-by-one as long as the order is correct to prevent deadlock~\cite{Havender68}.
     2350In \CFA, ordering of monitor acquisition relies on memory ordering.
     2351This 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.
     2352When a mutex call is made, the concerned monitors are aggregated into a variable-length pointer array and sorted based on pointer values.
     2353This array persists for the entire duration of the mutual-exclusion and its ordering reused extensively.
    18222354\begin{figure}
    18232355\begin{multicols}{2}
    18242356Entry
    1825 \begin{pseudo}
     2357\begin{cfa}
    18262358if monitor is free
    18272359        enter
    … …  
    18312363        block
    18322364increment recursions
    1833 \end{pseudo}
     2365\end{cfa}
    18342366\columnbreak
    18352367Exit
    1836 \begin{pseudo}
     2368\begin{cfa}
    18372369decrement recursion
    18382370if recursion == 0
    18392371        if entry queue not empty
    18402372                wake-up thread
    1841 \end{pseudo}
     2373\end{cfa}
    18422374\end{multicols}
    1843 \begin{pseudo}[caption={Initial entry and exit routine for monitors},label={lst:entry1}]
    1844 \end{pseudo}
     2375\begin{cfa}[caption={Initial entry and exit routine for monitors},label={f:entry1}]
     2376\end{cfa}
    18452377\end{figure}
    18462378
    18472379\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:
    1851 \begin{table}[H]
     2380Depending 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.
     2381However, it is shown that entry-point locking solves most of the issues.
     2382
     2383First of all, interaction between @otype@ polymorphism (see Section~\ref{s:ParametricPolymorphism}) and monitors is impossible since monitors do not support copying.
     2384Therefore, the main question is how to support @dtype@ polymorphism.
     2385It is important to present the difference between the two acquiring options: \textbf{callsite-locking} and entry-point locking, \ie acquiring the monitors before making a mutex routine-call or as the first operation of the mutex routine-call.
     2386For example:
     2387\begin{table}
    18522388\begin{center}
    18532389\begin{tabular}{|c|c|c|}
    18542390Mutex & \textbf{callsite-locking} & \textbf{entry-point-locking} \\
    1855 call & pseudo-code & pseudo-code \\
     2391call & cfa-code & cfa-code \\
    18562392\hline
    1857 \begin{cfacode}[tabsize=3]
     2393\begin{cfa}[tabsize=3]
    18582394void foo(monitor& mutex a){
    18592395
    1860         //Do Work
     2396        // Do Work
    18612397        //...
    18622398
    … …  
    18692405
    18702406}
    1871 \end{cfacode} & \begin{pseudo}[tabsize=3]
     2407\end{cfa} & \begin{cfa}[tabsize=3]
    18722408foo(& a) {
    18732409
    1874         //Do Work
     2410        // Do Work
    18752411        //...
    18762412
    … …  
    18832419        release(a);
    18842420}
    1885 \end{pseudo} & \begin{pseudo}[tabsize=3]
     2421\end{cfa} & \begin{cfa}[tabsize=3]
    18862422foo(& a) {
    18872423        acquire(a);
    1888         //Do Work
     2424        // Do Work
    18892425        //...
    18902426        release(a);
    … …  
    18972433
    18982434}
    1899 \end{pseudo}
     2435\end{cfa}
    19002436\end{tabular}
    19012437\end{center}
    … …  
    19042440\end{table}
    19052441
    1906 Note the \code{mutex} keyword relies on the type system, which means that in cases where a generic monitor-routine is desired, writing the mutex routine is possible with the proper trait, e.g.:
    1907 \begin{cfacode}
    1908 //Incorrect: T may not be monitor
     2442Note the @mutex@ keyword relies on the type system, which means that in cases where a generic monitor-routine is desired, writing the mutex routine is possible with the proper trait, \eg:
     2443\begin{cfa}
     2444// Incorrect: T may not be monitor
    19092445forall(dtype T)
    19102446void foo(T * mutex t);
    19112447
    1912 //Correct: this function only works on monitors (any monitor)
     2448// Correct: this function only works on monitors (any monitor)
    19132449forall(dtype T | is_monitor(T))
    19142450void bar(T * mutex t));
    1915 \end{cfacode}
    1916 
    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.
     2451\end{cfa}
     2452
     2453Both entry point and \textbf{callsite-locking} are feasible implementations.
     2454The 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.
     2455It 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, \ie the function body.
     2456For example, the monitor call can appear in the middle of an expression.
     2457Furthermore, 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.
    19182458
    19192459% ======================================================================
    … …  
    19232463% ======================================================================
    19242464
    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.
     2465Figure \ref{fig:system1} shows a high-level picture if the \CFA runtime system in regards to concurrency.
     2466Each component of the picture is explained in detail in the flowing sections.
    19262467
    19272468\begin{figure}
    … …  
    19342475
    19352476\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.
     2477Parallelism in \CFA is built around using processors to specify how much parallelism is desired. \CFA processors are object wrappers around kernel threads, specifically @pthread@s in the current implementation of \CFA.
     2478Indeed, any parallelism must go through operating-system libraries.
     2479However, \textbf{uthread} are still the main source of concurrency, processors are simply the underlying source of parallelism.
     2480Indeed, processor \textbf{kthread} simply fetch a \textbf{uthread} from the scheduler and run it; they are effectively executers for user-threads.
     2481The 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.
     2482Processors internally use coroutines to take advantage of the existing context-switching semantics.
    19372483
    19382484\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.
     2485One of the challenges of this system is to reduce the footprint as much as possible.
     2486Specifically, all @pthread@s created also have a stack created with them, which should be used as much as possible.
     2487Normally, 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.
     2488The exception to this rule is the Main Processor, \ie the initial \textbf{kthread} that is given to any program.
     2489In 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.
    19402490
    19412491\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.
     2492As 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.
     2493To improve performance and simplicity, context-switching is implemented using the following assumption: all context-switches happen inside a specific function call.
     2494This 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.
     2495Note that the instruction pointer can be left untouched since the context-switch is always inside the same function.
     2496Threads, however, do not context-switch between each other directly.
     2497They context-switch to the scheduler.
     2498This 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.
     2499Obviously, this doubles the context-switch cost because threads must context-switch to an intermediate stack.
     2500The alternative 1-step context-switch uses the stack of the ``from'' thread to schedule and then context-switches directly to the ``to'' thread.
     2501However, the performance of the 2-step context-switch is still superior to a @pthread_yield@ (see section \ref{results}).
     2502Additionally, 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 @SwitchToFiber@~\cite{switchToWindows} routine).
     2503This option is not currently present in \CFA, but the changes required to add it are strictly additive.
    19432504
    19442505\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.:
     2506Finally, an important aspect for any complete threading system is preemption.
     2507As 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.
     2508Indeed, preemption is desirable because it adds a degree of isolation among threads.
     2509In 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.
     2510Obviously, preemption is not optimal for every workload.
     2511However any preemptive system can become a cooperative system by making the time slices extremely large.
     2512Therefore, \CFA uses a preemptive threading system.
     2513
     2514Preemption in \CFA\footnote{Note that the implementation of preemption is strongly tied with the underlying threading system.
     2515For 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.
     2516Every processor keeps track of the current time and registers an expiration time with the preemption system.
     2517When 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.
     2518These timers use the Linux signal {\tt SIGALRM}, which is delivered to the process rather than the kernel-thread.
     2519This 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, \ie:
    19482520\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.
     2521A process-directed signal may be delivered to any one of the threads that does not currently have the signal blocked.
     2522If more than one of the threads has the signal unblocked, then the kernel chooses an arbitrary thread to which to deliver the signal.
    19502523SIGNAL(7) - Linux Programmer's Manual
    19512524\end{quote}
    19522525For 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.
    19532526
    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.
     2527Now because of how involuntary context-switches are handled, the kernel thread handling {\tt SIGALRM} cannot also be a processor thread.
     2528Hence, 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.
     2529This 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.
     2530As a result, a signal handler can start on one kernel thread and terminate on a second kernel thread (but the same user thread).
     2531It 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.
     2532This 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.}.
     2533However, since the kernel thread handling preemption requires a different signal mask, executing user threads on the kernel-alarm thread can cause deadlocks.
     2534For this reason, the alarm thread is in a tight loop around a system call to @sigwaitinfo@, requiring very little CPU time for preemption.
     2535One final detail about the alarm thread is how to wake it when additional communication is required (\eg on thread termination).
     2536This unblocking is also done using {\tt SIGALRM}, but sent through the @pthread_sigqueue@.
     2537Indeed, @sigwait@ can differentiate signals sent from @pthread_sigqueue@ from signals sent from alarms or the kernel.
    19552538
    19562539\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}.
     2540Finally, an aspect that was not mentioned yet is the scheduling algorithm.
     2541Currently, the \CFA scheduler uses a single ready queue for all processors, which is the simplest approach to scheduling.
     2542Further discussion on scheduling is present in section \ref{futur:sched}.
    19582543
    19592544% ======================================================================
    … …  
    19642549The following figure is the traditional illustration of a monitor (repeated from page~\pageref{fig:ClassicalMonitor} for convenience):
    19652550
    1966 \begin{figure}[H]
     2551\begin{figure}
    19672552\begin{center}
    19682553{\resizebox{0.4\textwidth}{!}{\input{monitor}}}
    … …  
    19712556\end{figure}
    19722557
    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}.
    1976 
    1977 \begin{figure}[H]
     2558This picture has several components, the two most important being the entry queue and the AS-stack.
     2559The 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.
     2560
     2561For \CFA, this picture does not have support for blocking multiple monitors on a single condition.
     2562To support \textbf{bulk-acq} two changes to this picture are required.
     2563First, it is no longer helpful to attach the condition to \emph{a single} monitor.
     2564Secondly, the thread waiting on the condition has to be separated across multiple monitors, seen in figure \ref{fig:monitor_cfa}.
     2565
     2566\begin{figure}
    19782567\begin{center}
    19792568{\resizebox{0.8\textwidth}{!}{\input{int_monitor}}}
    … …  
    19832572\end{figure}
    19842573
    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.
    1986 
    1987 \begin{figure}[b]
     2574This picture and the proper entry and leave algorithms (see listing \ref{f:entry2}) is the fundamental implementation of internal scheduling.
     2575Note 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.
     2576The thread is woken up when all the pieces have popped from the AS-stacks and made active.
     2577In this picture, the threads are split into halves but this is only because there are two monitors.
     2578For a specific signalling operation every monitor needs a piece of thread on its AS-stack.
     2579
     2580\begin{figure}
    19882581\begin{multicols}{2}
    19892582Entry
    1990 \begin{pseudo}
     2583\begin{cfa}
    19912584if monitor is free
    19922585        enter
    … …  
    19972590increment recursion
    19982591
    1999 \end{pseudo}
     2592\end{cfa}
    20002593\columnbreak
    20012594Exit
    2002 \begin{pseudo}
     2595\begin{cfa}
    20032596decrement recursion
    20042597if recursion == 0
    … …  
    20102603        if entry queue not empty
    20112604                wake-up thread
    2012 \end{pseudo}
     2605\end{cfa}
    20132606\end{multicols}
    2014 \begin{pseudo}[caption={Entry and exit routine for monitors with internal scheduling},label={lst:entry2}]
    2015 \end{pseudo}
     2607\begin{cfa}[caption={Entry and exit routine for monitors with internal scheduling},label={f:entry2}]
     2608\end{cfa}
    20162609\end{figure}
    20172610
    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.
    2019 
    2020 \begin{figure}[H]
     2611The solution discussed in \ref{intsched} can be seen in the exit routine of listing \ref{f:entry2}.
     2612Basically, 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.
     2613This solution is deadlock safe as well as preventing any potential barging.
     2614The 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 @wait@ and @signal_block@ routines.
     2615
     2616\begin{figure}
    20212617\begin{center}
    20222618{\resizebox{0.8\textwidth}{!}{\input{monitor_structs.pstex_t}}}
    … …  
    20262622\end{figure}
    20272623
    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}.
     2624Figure \ref{fig:structs} shows a high-level representation of these data structures.
     2625The main idea behind them is that, a thread cannot contain an arbitrary number of intrusive ``next'' pointers for linking onto monitors.
     2626The @condition node@ is the data structure that is queued onto a condition variable and, when signalled, the condition queue is popped and each @condition criterion@ is moved to the AS-stack.
     2627Once all the criteria have been popped from their respective AS-stacks, the thread is woken up, which is what is shown in listing \ref{f:entry2}.
    20292628
    20302629% ======================================================================
    … …  
    20332632% ======================================================================
    20342633% ======================================================================
    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.
     2634Similarly 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}.
     2635For internal scheduling, these queues are part of condition variables, which are still unique for a given scheduling operation (\ie no signal statement uses multiple conditions).
     2636However, in the case of external scheduling, there is no equivalent object which is associated with @waitfor@ statements.
     2637This absence means the queues holding the waiting threads must be stored inside at least one of the monitors that is acquired.
     2638These monitors being the only objects that have sufficient lifetime and are available on both sides of the @waitfor@ statement.
     2639This requires an algorithm to choose which monitor holds the relevant queue.
     2640It is also important that said algorithm be independent of the order in which users list parameters.
     2641The 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.
     2642This assumes that the lock acquiring order is static for the lifetime of all concerned objects but that is a reasonable constraint.
    20362643
    20372644This algorithm choice has two consequences:
    20382645\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.
     2646        \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.
     2647These queues need to contain a set of monitors for each of the waiting threads.
     2648Therefore, 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.
     2649        \item The queue of the lowest priority monitor is both required and potentially unused.
     2650Indeed, 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.
    20412651\end{itemize}
    20422652Therefore, the following modifications need to be made to support external scheduling:
    20432653\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.
    2046         \item The entry/exit routines need to be updated as shown in listing \ref{lst:entry3}.
     2654        \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.
     2655The @mutex@ routine already has all the required information on its stack, so the thread only needs to keep a pointer to that information.
     2656        \item The monitors need to keep a mask of acceptable routines.
     2657This mask contains for each acceptable routine, a routine pointer and an array of monitors to go with it.
     2658It also needs storage to keep track of which routine was accepted.
     2659Since this information is not specific to any monitor, the monitors actually contain a pointer to an integer on the stack of the waiting thread.
     2660Note that if a thread has acquired two monitors but executes a @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.
     2661This becomes relevant when @when@ clauses affect the number of monitors passed to a @waitfor@ statement.
     2662        \item The entry/exit routines need to be updated as shown in listing \ref{f:entry3}.
    20472663\end{itemize}
    20482664
    20492665\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}
     2666Finally, to support the ordering inversion of destructors, the code generation needs to be modified to use a special entry routine.
     2667This routine is needed because of the storage requirements of the call order inversion.
     2668Indeed, 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.
     2669For regular @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.
     2670The @waitfor@ semantics can then be adjusted correspondingly, as seen in listing \ref{f:entry-dtor}
    20512671
    20522672\begin{figure}
    20532673\begin{multicols}{2}
    20542674Entry
    2055 \begin{pseudo}
     2675\begin{cfa}
    20562676if monitor is free
    20572677        enter
    … …  
    20642684        block
    20652685increment recursion
    2066 \end{pseudo}
     2686\end{cfa}
    20672687\columnbreak
    20682688Exit
    2069 \begin{pseudo}
     2689\begin{cfa}
    20702690decrement recursion
    20712691if recursion == 0
    … …  
    20802700                wake-up thread
    20812701        endif
    2082 \end{pseudo}
     2702\end{cfa}
    20832703\end{multicols}
    2084 \begin{pseudo}[caption={Entry and exit routine for monitors with internal scheduling and external scheduling},label={lst:entry3}]
    2085 \end{pseudo}
     2704\begin{cfa}[caption={Entry and exit routine for monitors with internal scheduling and external scheduling},label={f:entry3}]
     2705\end{cfa}
    20862706\end{figure}
    20872707
    … …  
    20892709\begin{multicols}{2}
    20902710Destructor Entry
    2091 \begin{pseudo}
     2711\begin{cfa}
    20922712if monitor is free
    20932713        enter
    … …  
    21032723        wait
    21042724increment recursion
    2105 \end{pseudo}
     2725\end{cfa}
    21062726\columnbreak
    21072727Waitfor
    2108 \begin{pseudo}
     2728\begin{cfa}
    21092729if matching thread is already there
    21102730        if found destructor
    … …  
    21262746block
    21272747return
    2128 \end{pseudo}
     2748\end{cfa}
    21292749\end{multicols}
    2130 \begin{pseudo}[caption={Pseudo code for the \code{waitfor} routine and the \code{mutex} entry routine for destructors},label={lst:entry-dtor}]
    2131 \end{pseudo}
     2750\begin{cfa}[caption={Pseudo code for the \protect\lstinline|waitfor| routine and the \protect\lstinline|mutex| entry routine for destructors},label={f:entry-dtor}]
     2751\end{cfa}
    21322752\end{figure}
    21332753
    … …  
    21412761
    21422762\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:
    2144 \begin{figure}[H]
    2145 \begin{cfacode}[caption={Toy simulator using \code{thread}s and \code{monitor}s.},label={lst:engine-v1}]
     2763As it was subtly alluded in section \ref{threads}, @thread@s in \CFA are in fact monitors, which means that all monitor features are available when using threads.
     2764For example, here is a very simple two thread pipeline that could be used for a simulator of a game engine:
     2765\begin{figure}
     2766\begin{cfa}[caption={Toy simulator using \protect\lstinline|thread|s and \protect\lstinline|monitor|s.},label={f:engine-v1}]
    21462767// Visualization declaration
    21472768thread Renderer {} renderer;
    … …  
    21702791        }
    21712792}
    2172 \end{cfacode}
     2793\end{cfa}
    21732794\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:
    2175 \begin{figure}[H]
    2176 \begin{cfacode}[caption={Same toy simulator with proper termination condition.},label={lst:engine-v2}]
     2795One 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.
     2796Luckily, the monitor semantics can also be used to clearly enforce a shutdown order in a concise manner:
     2797\begin{figure}
     2798\begin{cfa}[caption={Same toy simulator with proper termination condition.},label={f:engine-v2}]
    21772799// Visualization declaration
    21782800thread Renderer {} renderer;
    … …  
    22122834// Call destructor for simulator once simulator finishes
    22132835// Call destructor for renderer to signify shutdown
    2214 \end{cfacode}
     2836\end{cfa}
    22152837\end{figure}
    22162838
    22172839\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~:
    2219 \begin{cfacode}
     2840As mentioned in section \ref{preemption}, \CFA uses preemptive threads by default but can use fibers on demand.
     2841Currently, using fibers is done by adding the following line of code to the program~:
     2842\begin{cfa}
    22202843unsigned int default_preemption() {
    22212844        return 0;
    22222845}
    2223 \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}
     2846\end{cfa}
     2847This function is called by the kernel to fetch the default preemption rate, where 0 signifies an infinite time-slice, \ie no preemption.
     2848However, once clusters are fully implemented, it will be possible to create fibers and \textbf{uthread} in the same system, as in listing \ref{f:fiber-uthread}
    22252849\begin{figure}
    2226 \begin{cfacode}[caption={Using fibers and \textbf{uthread} side-by-side in \CFA},label={lst:fiber-uthread}]
    2227 //Cluster forward declaration
     2850\lstset{language=CFA,deletedelim=**[is][]{`}{`}}
     2851\begin{cfa}[caption={Using fibers and \textbf{uthread} side-by-side in \CFA},label={f:fiber-uthread}]
     2852// Cluster forward declaration
    22282853struct cluster;
    22292854
    2230 //Processor forward declaration
     2855// Processor forward declaration
    22312856struct processor;
    22322857
    2233 //Construct clusters with a preemption rate
     2858// Construct clusters with a preemption rate
    22342859void ?{}(cluster& this, unsigned int rate);
    2235 //Construct processor and add it to cluster
     2860// Construct processor and add it to cluster
    22362861void ?{}(processor& this, cluster& cluster);
    2237 //Construct thread and schedule it on cluster
     2862// Construct thread and schedule it on cluster
    22382863void ?{}(thread& this, cluster& cluster);
    22392864
    2240 //Declare two clusters
    2241 cluster thread_cluster = { 10`ms };                     //Preempt every 10 ms
    2242 cluster fibers_cluster = { 0 };                         //Never preempt
    2243 
    2244 //Construct 4 processors
     2865// Declare two clusters
     2866cluster thread_cluster = { 10`ms };                     // Preempt every 10 ms
     2867cluster fibers_cluster = { 0 };                         // Never preempt
     2868
     2869// Construct 4 processors
    22452870processor processors[4] = {
    22462871        //2 for the thread cluster
    … …  
    22522877};
    22532878
    2254 //Declares thread
     2879// Declares thread
    22552880thread UThread {};
    22562881void ?{}(UThread& this) {
    2257         //Construct underlying thread to automatically
    2258         //be scheduled on the thread cluster
     2882        // Construct underlying thread to automatically
     2883        // be scheduled on the thread cluster
    22592884        (this){ thread_cluster }
    22602885}
    … …  
    22622887void main(UThread & this);
    22632888
    2264 //Declares fibers
     2889// Declares fibers
    22652890thread Fiber {};
    22662891void ?{}(Fiber& this) {
    2267         //Construct underlying thread to automatically
    2268         //be scheduled on the fiber cluster
     2892        // Construct underlying thread to automatically
     2893        // be scheduled on the fiber cluster
    22692894        (this.__thread){ fibers_cluster }
    22702895}
    22712896
    22722897void main(Fiber & this);
    2273 \end{cfacode}
     2898\end{cfa}
    22742899\end{figure}
    22752900
    … …  
    22812906% ======================================================================
    22822907\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.
    2284 \begin{table}[H]
     2908Table \ref{tab:machine} shows the characteristics of the machine used to run the benchmarks.
     2909All tests were made on this machine.
     2910\begin{table}
    22852911\begin{center}
    22862912\begin{tabular}{| l | r | l | r |}
    … …  
    23142940
    23152941\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:
    2317 \begin{pseudo}
     2942All benchmarks are run using the same harness to produce the results, seen as the @BENCH()@ macro in the following examples.
     2943This macro uses the following logic to benchmark the code:
     2944\begin{cfa}
    23182945#define BENCH(run, result) \
    23192946        before = gettime(); \
    … …  
    23212948        after  = gettime(); \
    23222949        result = (after - before) / N;
    2323 \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.
     2950\end{cfa}
     2951The method used to get time is @clock_gettime(CLOCK_THREAD_CPUTIME_ID);@.
     2952Each benchmark is using many iterations of a simple call to measure the cost of the call.
     2953The specific number of iterations depends on the specific benchmark.
    23252954
    23262955\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.
     2956The first interesting benchmark is to measure how long context-switches take.
     2957The simplest approach to do this is to yield on a thread, which executes a 2-step context switch.
     2958Yielding 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).
     2959In 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.
     2960Figure~\ref{f:ctx-switch} shows the code for coroutines and threads with the results in table \ref{tab:ctx-switch}.
     2961All omitted tests are functionally identical to one of these tests.
     2962The difference between coroutines and threads can be attributed to the cost of scheduling.
    23282963\begin{figure}
    23292964\begin{multicols}{2}
    23302965\CFA Coroutines
    2331 \begin{cfacode}
     2966\begin{cfa}
    23322967coroutine GreatSuspender {};
    23332968void main(GreatSuspender& this) {
    … …  
    23452980        printf("%llu\n", result);
    23462981}
    2347 \end{cfacode}
     2982\end{cfa}
    23482983\columnbreak
    23492984\CFA Threads
    2350 \begin{cfacode}
     2985\begin{cfa}
    23512986
    23522987
    … …  
    23642999        printf("%llu\n", result);
    23653000}
    2366 \end{cfacode}
     3001\end{cfa}
    23673002\end{multicols}
    2368 \begin{cfacode}[caption={\CFA benchmark code used to measure context-switches for coroutines and threads.},label={lst:ctx-switch}]
    2369 \end{cfacode}
     3003\begin{cfa}[caption={\CFA benchmark code used to measure context-switches for coroutines and threads.},label={f:ctx-switch}]
     3004\end{cfa}
    23703005\end{figure}
    23713006
    … …  
    23863021\end{tabular}
    23873022\end{center}
    2388 \caption{Context Switch comparison. All numbers are in nanoseconds(\si{\nano\second})}
     3023\caption{Context Switch comparison.
     3024All numbers are in nanoseconds(\si{\nano\second})}
    23893025\label{tab:ctx-switch}
    23903026\end{table}
    23913027
    23923028\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}.
     3029The next interesting benchmark is to measure the overhead to enter/leave a critical-section.
     3030For monitors, the simplest approach is to measure how long it takes to enter and leave a monitor routine.
     3031Figure~\ref{f:mutex} shows the code for \CFA.
     3032To put the results in context, the cost of entering a non-inline function and the cost of acquiring and releasing a @pthread_mutex@ lock is also measured.
     3033The results can be shown in table \ref{tab:mutex}.
    23943034
    23953035\begin{figure}
    2396 \begin{cfacode}[caption={\CFA benchmark code used to measure mutex routines.},label={lst:mutex}]
     3036\begin{cfa}[caption={\CFA benchmark code used to measure mutex routines.},label={f:mutex}]
    23973037monitor M {};
    23983038void __attribute__((noinline)) call( M & mutex m /*, m2, m3, m4*/ ) {}
    … …  
    24083048        printf("%llu\n", result);
    24093049}
    2410 \end{cfacode}
     3050\end{cfa}
    24113051\end{figure}
    24123052
    … …  
    24203060FetchAdd + FetchSub                             & 26            & 26            & 0    \\
    24213061Pthreads Mutex Lock                             & 31            & 31.86 & 0.99 \\
    2422 \uC \code{monitor} member routine               & 30            & 30            & 0    \\
    2423 \CFA \code{mutex} routine, 1 argument   & 41            & 41.57 & 0.9  \\
    2424 \CFA \code{mutex} routine, 2 argument   & 76            & 76.96 & 1.57 \\
    2425 \CFA \code{mutex} routine, 4 argument   & 145           & 146.68        & 3.85 \\
     3062\uC @monitor@ member routine            & 30            & 30            & 0    \\
     3063\CFA @mutex@ routine, 1 argument        & 41            & 41.57 & 0.9  \\
     3064\CFA @mutex@ routine, 2 argument        & 76            & 76.96 & 1.57 \\
     3065\CFA @mutex@ routine, 4 argument        & 145           & 146.68        & 3.85 \\
    24263066Java synchronized routine                       & 27            & 28.57 & 2.6  \\
    24273067\hline
    24283068\end{tabular}
    24293069\end{center}
    2430 \caption{Mutex routine comparison. All numbers are in nanoseconds(\si{\nano\second})}
     3070\caption{Mutex routine comparison.
     3071All numbers are in nanoseconds(\si{\nano\second})}
    24313072\label{tab:mutex}
    24323073\end{table}
    24333074
    24343075\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.
     3076The internal-scheduling benchmark measures the cost of waiting on and signalling a condition variable.
     3077Figure~\ref{f:int-sched} shows the code for \CFA, with results table \ref{tab:int-sched}.
     3078As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
    24363079
    24373080\begin{figure}
    2438 \begin{cfacode}[caption={Benchmark code for internal scheduling},label={lst:int-sched}]
     3081\begin{cfa}[caption={Benchmark code for internal scheduling},label={f:int-sched}]
    24393082volatile int go = 0;
    24403083condition c;
    … …  
    24663109        return do_wait(m1);
    24673110}
    2468 \end{cfacode}
     3111\end{cfa}
    24693112\end{figure}
    24703113
    … …  
    24763119\hline
    24773120Pthreads Condition Variable                     & 5902.5        & 6093.29       & 714.78 \\
    2478 \uC \code{signal}                                       & 322           & 323   & 3.36   \\
    2479 \CFA \code{signal}, 1 \code{monitor}    & 352.5 & 353.11        & 3.66   \\
    2480 \CFA \code{signal}, 2 \code{monitor}    & 430           & 430.29        & 8.97   \\
    2481 \CFA \code{signal}, 4 \code{monitor}    & 594.5 & 606.57        & 18.33  \\
    2482 Java \code{notify}                              & 13831.5       & 15698.21      & 4782.3 \\
     3121\uC @signal@                                    & 322           & 323   & 3.36   \\
     3122\CFA @signal@, 1 @monitor@      & 352.5 & 353.11        & 3.66   \\
     3123\CFA @signal@, 2 @monitor@      & 430           & 430.29        & 8.97   \\
     3124\CFA @signal@, 4 @monitor@      & 594.5 & 606.57        & 18.33  \\
     3125Java @notify@                           & 13831.5       & 15698.21      & 4782.3 \\
    24833126\hline
    24843127\end{tabular}
    24853128\end{center}
    2486 \caption{Internal scheduling comparison. All numbers are in nanoseconds(\si{\nano\second})}
     3129\caption{Internal scheduling comparison.
     3130All numbers are in nanoseconds(\si{\nano\second})}
    24873131\label{tab:int-sched}
    24883132\end{table}
    24893133
    24903134\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.
     3135The Internal scheduling benchmark measures the cost of the @waitfor@ statement (@_Accept@ in \uC).
     3136Figure~\ref{f:ext-sched} shows the code for \CFA, with results in table \ref{tab:ext-sched}.
     3137As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
    24923138
    24933139\begin{figure}
    2494 \begin{cfacode}[caption={Benchmark code for external scheduling},label={lst:ext-sched}]
     3140\begin{cfa}[caption={Benchmark code for external scheduling},label={f:ext-sched}]
    24953141volatile int go = 0;
    24963142monitor M {};
    … …  
    25213167        return do_wait(m1);
    25223168}
    2523 \end{cfacode}
     3169\end{cfa}
    25243170\end{figure}
    25253171
    … …  
    25303176\multicolumn{1}{c |}{} & \multicolumn{1}{c |}{ Median } &\multicolumn{1}{c |}{ Average } & \multicolumn{1}{c |}{ Standard Deviation} \\
    25313177\hline
    2532 \uC \code{Accept}                                       & 350           & 350.61        & 3.11  \\
    2533 \CFA \code{waitfor}, 1 \code{monitor}   & 358.5 & 358.36        & 3.82  \\
    2534 \CFA \code{waitfor}, 2 \code{monitor}   & 422           & 426.79        & 7.95  \\
    2535 \CFA \code{waitfor}, 4 \code{monitor}   & 579.5 & 585.46        & 11.25 \\
     3178\uC @Accept@                                    & 350           & 350.61        & 3.11  \\
     3179\CFA @waitfor@, 1 @monitor@     & 358.5 & 358.36        & 3.82  \\
     3180\CFA @waitfor@, 2 @monitor@     & 422           & 426.79        & 7.95  \\
     3181\CFA @waitfor@, 4 @monitor@     & 579.5 & 585.46        & 11.25 \\
    25363182\hline
    25373183\end{tabular}
    25383184\end{center}
    2539 \caption{External scheduling comparison. All numbers are in nanoseconds(\si{\nano\second})}
     3185\caption{External scheduling comparison.
     3186All numbers are in nanoseconds(\si{\nano\second})}
    25403187\label{tab:ext-sched}
    25413188\end{table}
    25423189
     3190
    25433191\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.
     3192Finally, the last benchmark measures the cost of creation for concurrent objects.
     3193Figure~\ref{f:creation} shows the code for @pthread@s and \CFA threads, with results shown in table \ref{tab:creation}.
     3194As with all other benchmarks, all omitted tests are functionally identical to one of these tests.
     3195The 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.
    25453196
    25463197\begin{figure}
    25473198\begin{center}
    2548 \texttt{pthread}
    2549 \begin{ccode}
     3199@pthread@
     3200\begin{cfa}
    25503201int main() {
    25513202        BENCH(
    … …  
    25663217        printf("%llu\n", result);
    25673218}
    2568 \end{ccode}
     3219\end{cfa}
    25693220
    25703221
    25713222
    25723223\CFA Threads
    2573 \begin{cfacode}
     3224\begin{cfa}
    25743225int main() {
    25753226        BENCH(
    … …  
    25813232        printf("%llu\n", result);
    25823233}
    2583 \end{cfacode}
     3234\end{cfa}
    25843235\end{center}
    2585 \begin{cfacode}[caption={Benchmark code for \texttt{pthread}s and \CFA to measure object creation},label={lst:creation}]
    2586 \end{cfacode}
     3236\caption{Benchmark code for \protect\lstinline|pthread|s and \CFA to measure object creation}
     3237\label{f:creation}
    25873238\end{figure}
    25883239
    … …  
    26043255\end{tabular}
    26053256\end{center}
    2606 \caption{Creation comparison. All numbers are in nanoseconds(\si{\nano\second}).}
     3257\caption{Creation comparison.
     3258All numbers are in nanoseconds(\si{\nano\second}).}
    26073259\label{tab:creation}
    26083260\end{table}
    … …  
    26113263
    26123264\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.
     3265This paper has achieved a minimal concurrency \textbf{api} that is simple, efficient and usable as the basis for higher-level features.
     3266The approach presented is based on a lightweight thread-system for parallelism, which sits on top of clusters of processors.
     3267This M:N model is judged to be both more efficient and allow more flexibility for users.
     3268Furthermore, this document introduces monitors as the main concurrency tool for users.
     3269This paper also offers a novel approach allowing multiple monitors to be accessed simultaneously without running into the Nested Monitor Problem~\cite{Lister77}.
     3270It also offers a full implementation of the concurrency runtime written entirely in \CFA, effectively the largest \CFA code base to date.
    26143271
    26153272
    … …  
    26213278
    26223279\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.
     3280This paper presents a first implementation of the \CFA concurrency runtime.
     3281Therefore, there is still significant work to improve performance.
     3282Many of the data structures and algorithms may change in the future to more efficient versions.
     3283For example, the number of monitors in a single \textbf{bulk-acq} is only bound by the stack size, this is probably unnecessarily generous.
     3284It may be possible that limiting the number helps increase performance.
     3285However, it is not obvious that the benefit would be significant.
    26243286
    26253287\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.
     3288An important part of concurrency is scheduling.
     3289Different scheduling algorithms can affect performance (both in terms of average and variation).
     3290However, no single scheduler is optimal for all workloads and therefore there is value in being able to change the scheduler for given programs.
     3291One solution is to offer various tweaking options to users, allowing the scheduler to be adjusted to the requirements of the workload.
     3292However, in order to be truly flexible, it would be interesting to allow users to add arbitrary data and arbitrary scheduling algorithms.
     3293For 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.
     3294This path of flexible schedulers will be explored for \CFA.
    26273295
    26283296\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.
     3297While 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).
     3298These types of workloads often require significant engineering around amortizing costs of blocking IO operations.
     3299At its core, non-blocking I/O is an operating system level feature that allows queuing IO operations (\eg network operations) and registering for notifications instead of waiting for requests to complete.
     3300In this context, the role of the language makes Non-Blocking IO easily available and with low overhead.
     3301The 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.
     3302However, while these are valid solutions, they lead to code that is harder to read and maintain because it is much less linear.
    26303303
    26313304\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.
     3305While monitors offer a flexible and powerful concurrent core for \CFA, other concurrency tools are also necessary for a complete multi-paradigm concurrency package.
     3306Examples of such tools can include simple locks and condition variables, futures and promises~\cite{promises}, executors and actors.
     3307These additional features are useful when monitors offer a level of abstraction that is inadequate for certain tasks.
    26333308
    26343309\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.
     3310Simpler applications can benefit greatly from having implicit parallelism.
     3311That is, parallelism that does not rely on the user to write concurrency.
     3312This type of parallelism can be achieved both at the language level and at the library level.
     3313The canonical example of implicit parallelism is parallel for loops, which are the simplest example of a divide and conquer algorithms~\cite{uC++book}.
     3314Table \ref{f:parfor} shows three different code examples that accomplish point-wise sums of large arrays.
     3315Note that none of these examples explicitly declare any concurrency or parallelism objects.
    26363316
    26373317\begin{table}
    … …  
    26393319\begin{tabular}[t]{|c|c|c|}
    26403320Sequential & Library Parallel & Language Parallel \\
    2641 \begin{cfacode}[tabsize=3]
     3321\begin{cfa}[tabsize=3]
    26423322void big_sum(
    26433323        int* a, int* b,
    … …  
    26633343//... fill in a & b
    26643344big_sum(a,b,c,10000);
    2665 \end{cfacode} &\begin{cfacode}[tabsize=3]
     3345\end{cfa} &\begin{cfa}[tabsize=3]
    26663346void big_sum(
    26673347        int* a, int* b,
    … …  
    26873367//... fill in a & b
    26883368big_sum(a,b,c,10000);
    2689 \end{cfacode}&\begin{cfacode}[tabsize=3]
     3369\end{cfa}&\begin{cfa}[tabsize=3]
    26903370void big_sum(
    26913371        int* a, int* b,
    … …  
    27113391//... fill in a & b
    27123392big_sum(a,b,c,10000);
    2713 \end{cfacode}
     3393\end{cfa}
    27143394\end{tabular}
    27153395\end{center}
    27163396\caption{For loop to sum numbers: Sequential, using library parallelism and language parallelism.}
    2717 \label{lst:parfor}
     3397\label{f:parfor}
    27183398\end{table}
    27193399
    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.
     3400Implicit parallelism is a restrictive solution and therefore has its limitations.
     3401However, 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.
    27213402
    27223403
    … …  
    27313412% B I B L I O G R A P H Y
    27323413% -----------------------------
    2733 \bibliographystyle{plain}
     3414%\bibliographystyle{plain}
    27343415\bibliography{pl,local}
    27353416
Note: See TracChangeset for help on using the changeset viewer.