source: doc/proposals/concurrency/text/concurrency.tex @ 64b272a

aaron-thesisarm-ehcleanup-dtorsdeferred_resndemanglerjacob/cs343-translationjenkins-sandboxnew-astnew-ast-unique-exprnew-envno_listpersistent-indexerresolv-newwith_gc
Last change on this file since 64b272a was 64b272a, checked in by Thierry Delisle <tdelisle@…>, 4 years ago

Prereview commit

  • Property mode set to 100644
File size: 50.5 KB
Line 
1% ======================================================================
2% ======================================================================
3\chapter{Concurrency}
4% ======================================================================
5% ======================================================================
6Several tool 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\cit 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 call). This distinction in turn means that, in order to be effective, programmers need to learn two sets of designs patterns. While this distinction can be hidden away in library code, effective use of the librairy still has to take both paradigms into account.
7
8Approaches 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 desireable to have a higher-level construct be the core concurrency paradigm~\cite{HPP:Study}.
9
10An approach that is worth mentionning because it is gaining in popularity is transactionnal memory~\cite{Dice10}[Check citation]. While this approach is even pursued by system languages like \CC\cit, the performance and feature set is currently too restrictive to be the main concurrency paradigm for systems language, which is why it was rejected as the core paradigm for concurrency in \CFA.
11
12One 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.
13
14\section{Basics}
15Non-determinism requires concurrent systems to offer support for mutual-exclusion and synchronisation. 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.
16
17\subsection{Mutual-Exclusion}
18As mentionned 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 mutual-exclusion methods, 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 organising for multiple locks to be used while preventing deadlocks. Easing composability is another feature higher-level mutual-exclusion mechanisms often offer.
19
20\subsection{Synchronization}
21As for mutual-exclusion, low-level synchronisation primitives often offer good performance and good flexibility at the cost of ease of use. Again, higher-level mechanism often simplify usage by adding better coupling between synchronization and data, e.g.: message passing, or offering simple solution to otherwise involved challenges. An example is barging. As mentioned above, synchronization can be expressed as guaranteeing that event \textit{X} always happens before \textit{Y}. Most of the time, synchronisation happens around a critical section, where threads must acquire critical sections 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 called 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}. 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 status flags and other flag variables to detect barging threads are said to be using barging avoidance while algorithms that baton-passing locks between threads instead of releasing the locks are said to be using barging prevention.
22
23% ======================================================================
24% ======================================================================
25\section{Monitors}
26% ======================================================================
27% ======================================================================
28A monitor is a set of routines that ensure mutual exclusion when accessing shared state. 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 requirements is the ability to declare a handle to a shared object and a set of routines that act on it :
29\begin{cfacode}
30typedef /*some monitor type*/ monitor;
31int f(monitor & m);
32
33int main() {
34        monitor m;  //Handle m
35        f(m);       //Routine using handle
36}
37\end{cfacode}
38
39% ======================================================================
40% ======================================================================
41\subsection{Call semantics} \label{call}
42% ======================================================================
43% ======================================================================
44The 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 implicitly non-copyable objects.
45
46Another 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. Pass through can occur for generic helper routines (\code{swap}, \code{sort}, etc.) or specific helper routines like the following to implement an atomic counter :
47
48\begin{cfacode}
49monitor counter_t { /*...see section $\ref{data}$...*/ };
50
51void ?{}(counter_t & nomutex this); //constructor
52size_t ++?(counter_t & mutex this); //increment
53
54//need for mutex is platform dependent
55void ?{}(size_t * this, counter_t & mutex cnt); //conversion
56\end{cfacode}
57This counter is used as follows:
58\begin{center}
59\begin{tabular}{c @{\hskip 0.35in} c @{\hskip 0.35in} c}
60\begin{cfacode}
61//shared counter
62counter_t cnt1, cnt2;
63
64//multiple threads access counter
65thread 1 : cnt1++; cnt2++;
66thread 2 : cnt1++; cnt2++;
67thread 3 : cnt1++; cnt2++;
68        ...
69thread N : cnt1++; cnt2++;
70\end{cfacode}
71\end{tabular}
72\end{center}
73Notice how the counter is used without any explicit synchronisation and yet supports thread-safe semantics for both reading and writting.
74
75Here, 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 constructed should never be shared and therefore does not require mutual exclusion. 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.
76
77For maximum usability, monitors use \gls{multi-acq} semantics, which means a single thread can acquire multiple times the same monitor without deadlock. For example, figure \ref{fig:search} uses recursion and \gls{multi-acq} to print values inside a binary tree.
78\begin{figure}
79\label{fig:search}
80\begin{cfacode}
81monitor printer { ... };
82struct tree {
83        tree * left, right;
84        char * value;
85};
86void print(printer & mutex p, char * v);
87
88void print(printer & mutex p, tree * t) {
89        print(p, t->value);
90        print(p, t->left );
91        print(p, t->right);
92}
93\end{cfacode}
94\caption{Recursive printing algorithm using \gls{multi-acq}.}
95\end{figure}
96
97Having both \code{mutex} and \code{nomutex} keywords is redundant based on the meaning of a routine having neither of these keywords. For example, given a routine without qualifiers \code{void foo(counter_t & this)}, then it is reasonable that it should default to the safest option \code{mutex}, whereas assuming \code{nomutex} is unsafe and may cause subtle errors. In fact, \code{nomutex} is the "normal" parameter behaviour, with the \code{nomutex} keyword effectively stating explicitly that "this routine is not special". Another alternative is making exactly one of these keywords mandatory, which would provide 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}.
98
99The next semantic decision is to establish when \code{mutex} may be used as a type qualifier. Consider the following declarations:
100\begin{cfacode}
101int f1(monitor & mutex m);
102int f2(const monitor & mutex m);
103int f3(monitor ** mutex m);
104int f4(monitor * mutex m []);
105int f5(graph(monitor*) & mutex m);
106\end{cfacode}
107The problem is to indentify 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 indentify 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 be acquired, passing an array to this routine would be type safe and yet result in undefined behavior 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:
108\begin{cfacode}
109int f1(monitor & mutex m);   //Okay : recommanded case
110int f2(monitor * mutex m);   //Okay : could be an array but probably not
111int f3(monitor mutex m []);  //Not Okay : Array of unkown length
112int f4(monitor ** mutex m);  //Not Okay : Could be an array
113int f5(monitor * mutex m []); //Not Okay : Array of unkown length
114\end{cfacode}
115Note that not all array functions are actually distinct in the type system sense. However, even the code generation could tell the difference, the extra information is still not sufficient to extend meaningfully the monitor call semantic.
116
117Unlike object-oriented monitors, where calling a mutex member \emph{implicitly} acquires mutual-exclusion often receives an object, \CFA uses an explicit mechanism to acquire mutual-exclusion. A consequence of this approach is that it extends naturally to multi-monitor calls.
118\begin{cfacode}
119int f(MonitorA & mutex a, MonitorB & mutex b);
120
121MonitorA a;
122MonitorB b;
123f(a,b);
124\end{cfacode}
125The capacity to acquire multiple locks before entering a critical section is called \emph{\gls{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 aquisition is consistent across calls to routines using the same monitors as arguments. However, since \CFA monitors use \gls{multi-acq} locks, users can effectively force the acquiring order. For example, notice which routines use \code{mutex}/\code{nomutex} and how this affects aquiring order:
126\begin{cfacode}
127void foo(A & mutex a, B & mutex b) { //acquire a & b
128        ...
129}
130
131void bar(A & mutex a, B & /*nomutex*/ b) { //acquire a
132        ... foo(a, b); ... //acquire b
133}
134
135void baz(A & /*nomutex*/ a, B & mutex b) { //acquire b
136        ... foo(a, b); ... //acquire a
137}
138\end{cfacode}
139The \gls{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.
140
141However, such use leads to the lock acquiring order problem. 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 mistake means that calling these routines concurrently may lead to deadlock and is therefore undefined behavior. As shown on several occasion\cit, solving this problem requires:
142\begin{enumerate}
143        \item Dynamically tracking of the monitor-call order.
144        \item Implement rollback semantics.
145\end{enumerate}
146While the first requirement is already a significant constraint on the system, implementing a general rollback semantics in a C-like language is prohibitively complex \cit. In \CFA, users simply need to be carefull when acquiring multiple monitors at the same time or only use \gls{bulk-acq} of all the monitors.
147
148\Gls{multi-acq} and \gls{bulk-acq} can be used together in interesting ways, for example:
149\begin{cfacode}
150monitor bank { ... };
151
152void deposit( bank & mutex b, int deposit );
153
154void transfer( bank & mutex mybank, bank & mutex yourbank, int me2you) {
155        deposit( mybank, -me2you );
156        deposit( yourbank, me2you );
157}
158\end{cfacode}
159This example shows a trivial solution to the bank account transfer problem\cit. Without \gls{multi-acq} and \gls{bulk-acq}, the solution to this problem is much more involved and requires carefull engineering.
160
161\subsubsection{\code{mutex} statement} \label{mutex-stmt}
162
163The call semantics discussed aboved have one software engineering issue, only a named routine can acquire the mutual-exclusion of a set of monitor. \CFA offers the \code{mutex} statement to workaround the need for unnecessary names, avoiding a major software engineering problem\cit. Listing \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.
164
165\begin{figure}
166\begin{center}
167\begin{tabular}{|c|c|}
168function call & \code{mutex} statement \\
169\hline
170\begin{cfacode}[tabsize=3]
171monitor M {};
172void foo( M & mutex m ) {
173        //critical section
174}
175
176void bar( M & m ) {
177        foo( m );
178}
179\end{cfacode}&\begin{cfacode}[tabsize=3]
180monitor M {};
181void bar( M & m ) {
182        mutex(m) {
183                //critical section
184        }
185}
186
187
188\end{cfacode}
189\end{tabular}
190\end{center}
191\caption{Regular call semantics vs. \code{mutex} statement}
192\label{lst:mutex-stmt}
193\end{figure}
194
195% ======================================================================
196% ======================================================================
197\subsection{Data semantics} \label{data}
198% ======================================================================
199% ======================================================================
200Once 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 showed in section \ref{call}:
201\begin{cfacode}
202monitor counter_t {
203        int value;
204};
205
206void ?{}(counter_t & this) {
207        this.cnt = 0;
208}
209
210int ?++(counter_t & mutex this) {
211        return ++this.value;
212}
213
214//need for mutex is platform dependent here
215void ?{}(int * this, counter_t & mutex cnt) {
216        *this = (int)cnt;
217}
218\end{cfacode}
219
220
221% ======================================================================
222% ======================================================================
223\section{Internal scheduling} \label{intsched}
224% ======================================================================
225% ======================================================================
226In addition to mutual exclusion, the monitors at the core of \CFA's concurrency can also be used to achieve synchronisation. With monitors, this capability is generally achieved with internal or external scheduling as in\cit. Since internal scheduling within a single monitor is mostly a solved problem, this thesis concentrates on extending internal scheduling to multiple monitors. Indeed, like the \gls{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.
227
228First, here is a simple example of such a technique:
229
230\begin{cfacode}
231monitor A {
232        condition e;
233}
234
235void foo(A & mutex a) {
236        ...
237        //Wait for cooperation from bar()
238        wait(a.e);
239        ...
240}
241
242void bar(A & mutex a) {
243        //Provide cooperation for foo()
244        ...
245        //Unblock foo
246        signal(a.e);
247}
248\end{cfacode}
249
250There are two details to note here. First, the \code{signal} is a delayed operation, it only unblocks the waiting thread when it reaches the end of the critical section. This semantic is needed to respect mutual-exclusion. Second, in \CFA, 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, effectively ensuring a basic ordering.
251
252An important aspect of the implementation is that \CFA does not allow barging, which means that once function \code{bar} releases the monitor, foo is guaranteed to resume immediately after (unless some other thread waited on the same condition). This guarantees offers the benefit of not having to loop arount waits in order to guarantee that a condition is still 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 of \CFA concurrency.
253
254% ======================================================================
255% ======================================================================
256\subsection{Internal Scheduling - multi monitor}
257% ======================================================================
258% ======================================================================
259It is easier to understand the problem of multi-monitor scheduling using a series of pseudo-code. 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 a single condition as paremeter and waits on the monitors associated with the condition.
260
261\begin{multicols}{2}
262thread 1
263\begin{pseudo}
264acquire A
265        wait A
266release A
267\end{pseudo}
268
269\columnbreak
270
271thread 2
272\begin{pseudo}
273acquire A
274        signal A
275release A
276\end{pseudo}
277\end{multicols}
278The example shows the simple case of having two threads (one for each column) and a single monitor A. 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.
279
280A direct extension of the previous example is a \gls{bulk-acq} version:
281
282\begin{multicols}{2}
283\begin{pseudo}
284acquire A & B
285        wait A & B
286release A & B
287\end{pseudo}
288
289\columnbreak
290
291\begin{pseudo}
292acquire A & B
293        signal A & B
294release A & B
295\end{pseudo}
296\end{multicols}
297This version uses \gls{bulk-acq} (denoted using the \& 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 more monitors. On the implementation side, handling multiple monitors does add a degree of complexity as the next few examples demonstrate.
298
299While 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\cit, which occurs when a \code{wait} is made on a thread that holds more than one monitor. For example, the following pseudo-code will run into the nested monitor problem :
300\begin{multicols}{2}
301\begin{pseudo}
302acquire A
303        acquire B
304                wait B
305        release B
306release A
307\end{pseudo}
308
309\columnbreak
310
311\begin{pseudo}
312acquire A
313        acquire B
314                signal B
315        release B
316release A
317\end{pseudo}
318\end{multicols}
319However, 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.
320
321\begin{multicols}{2}
322\begin{pseudo}
323acquire A
324        acquire B
325                wait B
326        release B
327release A
328\end{pseudo}
329
330\columnbreak
331
332\begin{pseudo}
333
334acquire B
335        signal B
336release B
337
338\end{pseudo}
339\end{multicols}
340
341Listing \ref{lst:int-bulk-pseudo} shows an example where \gls{bulk-acq} adds a significant layer of complexity to the internal signalling semantics. Listing \ref{lst:int-bulk-cfa} shows the corresponding \CFA code which implements the pseudo-code in listing \ref{lst:int-bulk-pseudo}. Note that listing \ref{lst:int-bulk-cfa} uses non-\code{mutex} parameter to introduce monitor \code{b} into context. However, for the purpose of translating the given pseudo-code into \CFA-code any method of introducing new monitors into context, other than a \code{mutex} parameter, is acceptable, e.g. global variables, pointer parameters or using locals with the \code{mutex}-statement.
342
343\begin{figure}[!b]
344\begin{multicols}{2}
345Waiting thread
346\begin{pseudo}[numbers=left]
347acquire A
348        //Code Section 1
349        acquire A & B
350                //Code Section 2
351                wait A & B
352                //Code Section 3
353        release A & B
354        //Code Section 4
355release A
356\end{pseudo}
357
358\columnbreak
359
360Signalling thread
361\begin{pseudo}[numbers=left, firstnumber=10]
362acquire A
363        //Code Section 5
364        acquire A & B
365                //Code Section 6
366                signal A & B
367                //Code Section 7
368        release A & B
369        //Code Section 8
370release A
371\end{pseudo}
372\end{multicols}
373\caption{Internal scheduling with \gls{bulk-acq}}
374\label{lst:int-bulk-pseudo}
375\end{figure}
376
377\begin{figure}[!b]
378\begin{multicols}{2}
379Waiting thread
380\begin{cfacode}
381monitor A;
382monitor B;
383extern condition c;
384void foo(A & mutex a, B & b) {
385        //Code Section 1
386        mutex(a, b) {
387                //Code Section 2
388                wait(c);
389                //Code Section 3
390        }
391        //Code Section 4
392}
393\end{cfacode}
394
395\columnbreak
396
397Signalling thread
398\begin{cfacode}
399monitor A;
400monitor B;
401extern condition c;
402void foo(A & mutex a, B & b) {
403        //Code Section 5
404        mutex(a, b) {
405                //Code Section 6
406                signal(c);
407                //Code Section 7
408        }
409        //Code Section 8
410}
411\end{cfacode}
412\end{multicols}
413\caption{Equivalent \CFA code for listing \ref{lst:int-bulk-pseudo}}
414\label{lst:int-bulk-cfa}
415\end{figure}
416
417It is particularly important to pay attention to code sections 4 and 8, which are where the existing semantics of internal scheduling need to be extended for multiple monitors. The root of the problem is that \gls{bulk-acq} is used in a context where one of the monitors is already acquired and 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 A \& B" (line 16), it must actually transfer ownership of monitor B to the waiting thread. This ownership trasnfer is required in order to prevent barging. Since the signalling thread still needs monitor A, simply waking up the waiting thread is not an option because it would violate mutual exclusion. There are three options.
418
419\subsubsection{Delaying signals}
420The first more obvious solution to solve 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 the correct time to transfer ownership 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 mutiple objects to a single group of objects, effectively making the existing single monitor semantic viable by simply changing monitors to monitor groups.
421\begin{multicols}{2}
422Waiter
423\begin{pseudo}[numbers=left]
424acquire A
425        acquire A & B
426                wait A & B
427        release A & B
428release A
429\end{pseudo}
430
431\columnbreak
432
433Signaller
434\begin{pseudo}[numbers=left, firstnumber=6]
435acquire A
436        acquire A & B
437                signal A & B
438        release A & B
439        //Secretly keep B here
440release A
441//Wakeup waiter and transfer A & B
442\end{pseudo}
443\end{multicols}
444However, this solution can become much more complicated depending on what is executed while secretly holding B (at line 10). Indeed, nothing prevents signalling monitor A on a different condition variable:
445\begin{multicols}{2}
446Thread 1
447\begin{pseudo}[numbers=left, firstnumber=1]
448acquire A
449        acquire A & B
450                wait A & B
451        release A & B
452release A
453\end{pseudo}
454
455Thread 2
456\begin{pseudo}[numbers=left, firstnumber=6]
457acquire A
458        wait A
459release A
460\end{pseudo}
461
462\columnbreak
463
464Thread 3
465\begin{pseudo}[numbers=left, firstnumber=9]
466acquire A
467        acquire A & B
468                signal A & B
469        release A & B
470        //Secretly keep B here
471        signal A
472release A
473//Wakeup thread 1 or 2?
474//Who wakes up the other thread?
475\end{pseudo}
476\end{multicols}
477
478The goal in this solution is to avoid the need to transfer ownership of a subset of the condition monitors. However, this goal is unreacheable in the previous example. Depending on the order of signals (line 12 and 15) two cases can happen.
479
480\paragraph{Case 1: thread 1 goes first.} In this case, the problem is that monitor A needs to be passed to thread 2 when thread 1 is done with it.
481\paragraph{Case 2: thread 2 goes first.} In this case, the problem is that monitor B needs to be passed to thread 1, which can be done directly or using thread 2 as an intermediate.
482\\
483
484Note 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 15 before line 11 and get the reverse effect.
485
486In 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 monitors cannot be handled as a single homogenous group and therefore invalidates the main benefit of this approach.
487
488\subsubsection{Dependency graphs}
489In the Listing 1 pseudo-code, there is a solution which statisfies both barging prevention and mutual exclusion. If ownership of both monitors is transferred to the waiter when the signaller releases A and then the waiter transfers back ownership of A when it releases it, then the problem is solved. Dynamically finding the correct order is therefore the second possible solution. The problem it encounters is that it effectively boils down to resolving a dependency graph of ownership requirements. Here even the simplest of code snippets requires two transfers and it seems to increase in a manner closer to polynomial. For example, the following code, which is just a direct extension to three monitors, requires at least three ownership transfer and has multiple solutions:
490
491\begin{multicols}{2}
492\begin{pseudo}
493acquire A
494        acquire B
495                acquire C
496                        wait A & B & C
497                release C
498        release B
499release A
500\end{pseudo}
501
502\columnbreak
503
504\begin{pseudo}
505acquire A
506        acquire B
507                acquire C
508                        signal A & B & C
509                release C
510        release B
511release A
512\end{pseudo}
513\end{multicols}
514
515\begin{figure}
516\begin{multicols}{3}
517Thread $\alpha$
518\begin{pseudo}[numbers=left, firstnumber=1]
519acquire A
520        acquire A & B
521                wait A & B
522        release A & B
523release A
524\end{pseudo}
525
526\columnbreak
527
528Thread $\gamma$
529\begin{pseudo}[numbers=left, firstnumber=1]
530acquire A
531        acquire A & B
532                signal A & B
533        release A & B
534        signal A
535release A
536\end{pseudo}
537
538\columnbreak
539
540Thread $\beta$
541\begin{pseudo}[numbers=left, firstnumber=1]
542acquire A
543        wait A
544release A
545\end{pseudo}
546
547\end{multicols}
548\caption{Dependency graph}
549\label{lst:dependency}
550\end{figure}
551
552\begin{figure}
553\begin{center}
554\input{dependency}
555\end{center}
556\label{fig:dependency}
557\caption{Dependency graph of the statements in listing \ref{lst:dependency}}
558\end{figure}
559
560Listing \ref{lst:dependency} is the three thread example rewritten for dependency graphs as well as the corresponding dependency graph. 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. The extra challenge is that this dependency graph is effectively post-mortem, but the run time system needs to be able to build and solve these graphs as the dependency unfolds. Resolving dependency graph being a complex and expensive endeavour, this solution is not the preffered one.
561
562\subsubsection{Partial signalling} \label{partial-sig}
563Finally, the solution that is chosen for \CFA is to use partial signalling. Consider the following case:
564
565\begin{multicols}{2}
566\begin{pseudo}[numbers=left]
567acquire A
568        acquire A & B
569                wait A & B
570        release A & B
571release A
572\end{pseudo}
573
574\columnbreak
575
576\begin{pseudo}[numbers=left, firstnumber=6]
577acquire A
578        acquire A & B
579                signal A & B
580        release A & B
581        //... More code
582release A
583\end{pseudo}
584\end{multicols}
585The partial signalling solution transfers ownership of monitor B at lines 10 but does not wake the waiting thread since it is still using monitor A. Only when it reaches line 11 does it actually wakeup 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 release and conditionally waking threads if all conditions are met. This solution has a much simpler implementation than a dependency graph solving algorithm which is why it was chosen.
586
587% ======================================================================
588% ======================================================================
589\subsection{Signalling: Now or Later}
590% ======================================================================
591% ======================================================================
592An 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\footnote{name to be discussed}.
593
594The example in listing \ref{lst:datingservice} highlights the difference in behaviour. As mentioned, \code{signal} only transfers ownership once the current critical section exits, this behaviour cause the need for additional synchronisation when a two-way handshake is needed. To avoid this extraneous synchronisation, the \code{condition} type offers the \code{signal_block} routine which handle two-way handshakes as shown in the example. This 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 frond-end and the back-end of the call to \code{signal_block}, meaning no other thread can acquire the monitor neither before nor after the call.
595\begin{figure}
596\begin{tabular}{|c|c|}
597\code{signal} & \code{signal_block} \\
598\hline
599\begin{cfacode}[tabsize=3]
600monitor DatingService
601{
602        //compatibility codes
603        enum{ CCodes = 20 };
604
605        int girlPhoneNo
606        int boyPhoneNo;
607};
608
609condition girls[CCodes];
610condition boys [CCodes];
611condition exchange;
612
613int girl(int phoneNo, int ccode)
614{
615        //no compatible boy ?
616        if(empty(boys[ccode]))
617        {
618                //wait for boy
619                wait(girls[ccode]);
620
621                //make phone number available
622                girlPhoneNo = phoneNo;
623
624                //wake boy fron chair
625                signal(exchange);
626        }
627        else
628        {
629                //make phone number available
630                girlPhoneNo = phoneNo;
631
632                //wake boy
633                signal(boys[ccode]);
634
635                //sit in chair
636                wait(exchange);
637        }
638        return boyPhoneNo;
639}
640
641int boy(int phoneNo, int ccode)
642{
643        //same as above
644        //with boy/girl interchanged
645}
646\end{cfacode}&\begin{cfacode}[tabsize=3]
647monitor DatingService
648{
649        //compatibility codes
650        enum{ CCodes = 20 };
651
652        int girlPhoneNo;
653        int boyPhoneNo;
654};
655
656condition girls[CCodes];
657condition boys [CCodes];
658//exchange is not needed
659
660int girl(int phoneNo, int ccode)
661{
662        //no compatible boy ?
663        if(empty(boys[ccode]))
664        {
665                //wait for boy
666                wait(girls[ccode]);
667
668                //make phone number available
669                girlPhoneNo = phoneNo;
670
671                //wake boy fron chair
672                signal(exchange);
673        }
674        else
675        {
676                //make phone number available
677                girlPhoneNo = phoneNo;
678
679                //wake boy
680                signal_block(boys[ccode]);
681
682                //second handshake unnecessary
683
684        }
685        return boyPhoneNo;
686}
687
688int boy(int phoneNo, int ccode)
689{
690        //same as above
691        //with boy/girl interchanged
692}
693\end{cfacode}
694\end{tabular}
695\caption{Dating service example using \code{signal} and \code{signal_block}. }
696\label{lst:datingservice}
697\end{figure}
698
699% ======================================================================
700% ======================================================================
701\section{External scheduling} \label{extsched}
702% ======================================================================
703% ======================================================================
704An alternative to internal scheduling is to use external scheduling.
705\begin{center}
706\begin{tabular}{|c|c|}
707Internal Scheduling & External Scheduling \\
708\hline
709\begin{ucppcode}
710_Monitor Semaphore {
711        condition c;
712        bool inUse;
713public:
714        void P() {
715                if(inUse) wait(c);
716                inUse = true;
717        }
718        void V() {
719                inUse = false;
720                signal(c);
721        }
722}
723\end{ucppcode}&\begin{ucppcode}
724_Monitor Semaphore {
725
726        bool inUse;
727public:
728        void P() {
729                if(inUse) _Accept(V);
730                inUse = true;
731        }
732        void V() {
733                inUse = false;
734
735        }
736}
737\end{ucppcode}
738\end{tabular}
739\end{center}
740This method is more constrained and explicit, which helps users tone down the undeterministic nature of concurrency. Indeed, as the following examples demonstrates, external scheduling allows users to wait for events from other threads without the concern of unrelated events occuring. External scheduling can generally be done either in terms of control flow (e.g., \uC with \code{_Accept}) or in terms of data (e.g. Go with channels). Of course, both of these paradigms have their own strenghts and weaknesses but for this project control-flow semantics were 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 multi-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 \acrshort{api}s.
741
742In the case of internal scheduling, the call to \code{wait} only guarantees that \code{V} is the last routine to access the monitor. This entails that a third routine, say \code{isInUse()}, may have acquired mutual exclusion several times while routine \code{P} was waiting. On the other hand, external scheduling guarantees that while routine \code{P} was waiting, no routine other than \code{V} could acquire the monitor.
743
744% ======================================================================
745% ======================================================================
746\subsection{Loose object definitions}
747% ======================================================================
748% ======================================================================
749In \uC, monitor declarations include an exhaustive list of monitor operations. Since \CFA is not object oriented it becomes both more difficult to implement but also less clear for the user:
750
751\begin{cfacode}
752monitor A {};
753
754void f(A & mutex a);
755void g(A & mutex a) {
756        waitfor(f); //Obvious which f() to wait for
757}
758
759void f(A & mutex a, int); //New different F added in scope
760void h(A & mutex a) {
761        waitfor(f); //Less obvious which f() to wait for
762}
763\end{cfacode}
764
765Furthermore, external scheduling is an example where implementation constraints become visible from the interface. Indeed, since there is no hard limit to the number of threads trying to acquire a monitor concurrently, performance is a significant concern. Here is the pseudo code for the entering phase of a monitor:
766
767\begin{center}
768\begin{tabular}{l}
769\begin{pseudo}
770        if monitor is free
771                enter
772        elif already own the monitor
773                continue
774        elif monitor accepts me
775                enter
776        else
777                block
778\end{pseudo}
779\end{tabular}
780\end{center}
781
782For the first two conditions, it is easy to implement a check that can evaluate the condition in a few instruction. 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 the following figure:
783
784\begin{center}
785{\resizebox{0.4\textwidth}{!}{\input{monitor}}}
786\end{center}
787
788There are other alternatives to these pictures, but in the case of this picture, implementing a fast accept check is relatively easy. Indeed simply updating a bitmask when the acceptor queue changes is enough to have a check that executes in a single instruction, even with a fairly large number (e.g. 128) of mutex members. This technique cannot be used in \CFA because it relies on the fact that the monitor type declares all the acceptable routines. For OO languages this does not compromise much since monitors already have an exhaustive list of member routines. However, for \CFA this is not the case; routines can be added to a type anywhere after its declaration. Its important to note that the bitmask approach does not actually require an exhaustive list of routines, but it requires a dense unique ordering of routines with an upper-bound and that ordering must be consistent across translation units.
789The alternative is to have a picture like this one:
790
791\begin{center}
792{\resizebox{0.4\textwidth}{!}{\input{ext_monitor}}}
793\end{center}
794
795Not storing the mask inside the monitor 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 function-pointers would solve the issue of uniquely identifying acceptable routines. However, the single instruction bitmask compare has been replaced by dereferencing a pointer followed by a linear search. Furthermore, supporting nested external scheduling may now require additionnal searches on calls to waitfor to check if a routine is already queued in.
796
797Note that in the second picture, tasks need to always keep track of through which routine they are attempting to acquire the monitor and the routine mask needs to have both a function pointer and a set of monitors, as will be discussed in the next section. These details where omitted from the picture for the sake of simplifying the representation.
798
799At this point we must make a decision 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 prohibitively hard to write. This decision is based on the assumption that writing fast but inflexible locks is closer to a solved problems than writing locks that are as flexible as external scheduling in \CFA.
800
801% ======================================================================
802% ======================================================================
803\subsection{Multi-monitor scheduling}
804% ======================================================================
805% ======================================================================
806
807External scheduling, like internal scheduling, becomes significantly more complex when introducing multi-monitor syntax. Even in the simplest possible case, some new semantics need to be established:
808\begin{cfacode}
809monitor M {};
810
811void f(M & mutex a);
812
813void g(M & mutex a, M & mutex b) {
814        waitfor(f); //ambiguous, keep a pass b or other way around?
815}
816\end{cfacode}
817
818The obvious solution is to specify the correct monitor as follows:
819
820\begin{cfacode}
821monitor M {};
822
823void f(M & mutex a);
824
825void g(M & mutex a, M & mutex b) {
826        waitfor( f, b );
827}
828\end{cfacode}
829
830This syntax is unambiguous. Both locks are acquired and kept. 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 behavior can be extended to multi-monitor waitfor statement as follows.
831
832\begin{cfacode}
833monitor M {};
834
835void f(M & mutex a, M & mutex b);
836
837void g(M & mutex a, M & mutex b) {
838        waitfor( f, a, b);
839}
840\end{cfacode}
841
842Note 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.
843
844An important behavior to note is that what happens when a set of monitors only match partially :
845
846\begin{cfacode}
847mutex struct A {};
848
849mutex struct B {};
850
851void g(A & mutex a, B & mutex b) {
852        waitfor(f, a, b);
853}
854
855A a1, a2;
856B b;
857
858void foo() {
859        g(a1, b); //block on accept
860}
861
862void bar() {
863        f(a2, b); //fufill cooperation
864}
865\end{cfacode}
866
867While 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 wake-up the waiting thread. It is also important to note that in the case of external scheduling, as for routine calls, the order of parameters is important; \code{waitfor(f,a,b)} and \code{waitfor(f,b,a)} are to distinct waiting condition.
868
869% ======================================================================
870% ======================================================================
871\subsection{\code{waitfor} semantics}
872% ======================================================================
873% ======================================================================
874
875Syntactically, the \code{waitfor} statement takes a function identifier and a set of monitors. While the set of monitors can be any list of expression, the function name is more restricted. This is because the compiler validates at compile time the validity of the waitfor statement. It checks that the set of monitor passed in matches the requirements for a function call. Listing \ref{lst:waitfor} shows various usage 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.
876\begin{figure}
877\begin{cfacode}
878monitor A{};
879monitor B{};
880
881void f1( A & mutex );
882void f2( A & mutex, B & mutex );
883void f3( A & mutex, int );
884void f4( A & mutex, int );
885void f4( A & mutex, double );
886
887void foo( A & mutex a1, A & mutex a2, B & mutex b1, B & b2 ) {
888        A * ap = & a1;
889        void (*fp)( A & mutex ) = f1;
890
891        waitfor(f1, a1);     //Correct : 1 monitor case
892        waitfor(f2, a1, b1); //Correct : 2 monitor case
893        waitfor(f3, a1);     //Correct : non-mutex arguments are ignored
894        waitfor(f1, *ap);    //Correct : expression as argument
895
896        waitfor(f1, a1, b1); //Incorrect : Too many mutex arguments
897        waitfor(f2, a1);     //Incorrect : Too few mutex arguments
898        waitfor(f2, a1, a2); //Incorrect : Mutex arguments don't match
899        waitfor(f1, 1);      //Incorrect : 1 not a mutex argument
900        waitfor(f4, a1);     //Incorrect : f9 not a function
901        waitfor(*fp, a1 );   //Incorrect : fp not a identifier
902        waitfor(f4, a1);     //Incorrect : f4 ambiguous
903
904        waitfor(f2, a1, b2); //Undefined Behaviour : b2 may not acquired
905}
906\end{cfacode}
907\caption{Various correct and incorrect uses of the waitfor statement}
908\label{lst:waitfor}
909\end{figure}
910
911Finally, for added flexibility, \CFA supports constructing complex waitfor mask using the \code{or}, \code{timeout} and \code{else}. Indeed, multiple \code{waitfor} can be chained together using \code{or}; this chain will form a single statement which will baton-pass to any one function that fits one of the function+monitor set which was passed in. To eanble users to tell which was the accepted function, \code{waitfor}s are followed by a statement (including the null statement \code{;}) or a compound statement. When multiple \code{waitfor} are chained together, only the statement corresponding to the accepted function is executed. 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, that is only check of a matching function already arrived and return immediately otherwise. Any and all of these clauses can be preceded by a \code{when} condition to dynamically construct the mask based on some current state. Listing \ref{lst:waitfor2}, demonstrates several complex masks and some incorrect ones.
912
913\begin{figure}
914\begin{cfacode}
915monitor A{};
916
917void f1( A & mutex );
918void f2( A & mutex );
919
920void foo( A & mutex a, bool b, int t ) {
921        //Correct : blocking case
922        waitfor(f1, a);
923
924        //Correct : block with statement
925        waitfor(f1, a) {
926                sout | "f1" | endl;
927        }
928
929        //Correct : block waiting for f1 or f2
930        waitfor(f1, a) {
931                sout | "f1" | endl;
932        } or waitfor(f2, a) {
933                sout | "f2" | endl;
934        }
935
936        //Correct : non-blocking case
937        waitfor(f1, a); or else;
938
939        //Correct : non-blocking case
940        waitfor(f1, a) {
941                sout | "blocked" | endl;
942        } or else {
943                sout | "didn't block" | endl;
944        }
945
946        //Correct : block at most 10 seconds
947        waitfor(f1, a) {
948                sout | "blocked" | endl;
949        } or timeout( 10`s) {
950                sout | "didn't block" | endl;
951        }
952
953        //Correct : block only if b == true
954        //if b == false, don't even make the call
955        when(b) waitfor(f1, a);
956
957        //Correct : block only if b == true
958        //if b == false, make non-blocking call
959        waitfor(f1, a); or when(!b) else;
960
961        //Correct : block only of t > 1
962        waitfor(f1, a); or when(t > 1) timeout(t); or else;
963
964        //Incorrect : timeout clause is dead code
965        waitfor(f1, a); or timeout(t); or else;
966
967        //Incorrect : order must be
968        //waitfor [or waitfor... [or timeout] [or else]]
969        timeout(t); or waitfor(f1, a); or else;
970}
971\end{cfacode}
972\caption{Various correct and incorrect uses of the or, else, and timeout clause around a waitfor statement}
973\label{lst:waitfor2}
974\end{figure}
975
976% ======================================================================
977% ======================================================================
978\subsection{Waiting for the destructor}
979% ======================================================================
980% ======================================================================
981An important exception for the \code{waitfor} statement is destructor semantics. Indeed, the \code{waitfor} statement can accept any \code{mutex} routine, which counts the destructor. 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 to fix this hole in the semantics would be disallowing \code{waitfor} on 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.
982\begin{figure}
983\begin{cfacode}
984monitor Executer {};
985struct  Action;
986
987void ^?{}   (Executer & mutex this);
988void execute(Executer & mutex this, const Action & );
989void run    (Executer & mutex this) {
990        while(true) {
991                   waitfor(execute, this);
992                or waitfor(^?{}   , this) {
993                        break;
994                }
995        }
996}
997\end{cfacode}
998\caption{Example of an executor which executes action in series until the destructor is called.}
999\label{lst:dtor-order}
1000\end{figure}
1001For 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.
Note: See TracBrowser for help on using the repository browser.