Ignore:
Timestamp:
Sep 19, 2017, 1:22:55 PM (4 years ago)
Author:
Rob Schluntz <rschlunt@…>
Branches:
aaron-thesis, arm-eh, cleanup-dtors, deferred_resn, demangler, jacob/cs343-translation, jenkins-sandbox, master, new-ast, new-ast-unique-expr, new-env, no_list, persistent-indexer, resolv-new, with_gc
Children:
d22e90f
Parents:
d48e529 (diff), e06be49 (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 branch 'master' of plg.uwaterloo.ca:/u/cforall/software/cfa/cfa-cc

File:
1 edited

Legend:

Unmodified
Added
Removed
  • doc/proposals/concurrency/text/concurrency.tex

    rd48e529 ra9a4771  
    300300% ======================================================================
    301301% ======================================================================
    302 It 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.
     302It 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.
    303303
    304304\begin{multicols}{2}
     
    397397\end{center}
    398398
    399 It is particularly important to pay attention to code sections 8 and 3, which are where the existing semantics of internal scheduling need to be extended for multiple monitors. The root of the problem is that \gls{group-acquire} 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 the monitor A, simply waking up the waiting thread is not an option because it would violate mutual exclusion. There are three options:
     399It is particularly important to pay attention to code sections 8 and 4, which are where the existing semantics of internal scheduling need to be extended for multiple monitors. The root of the problem is that \gls{group-acquire} 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 the monitor A, simply waking up the waiting thread is not an option because it would violate mutual exclusion. There are three options:
    400400
    401401\subsubsection{Delaying signals}
     
    467467Note 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.
    468468
    469 In both cases however, 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.
     469In 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.
    470470
    471471\subsubsection{Dependency graphs}
     
    497497Resolving dependency graph being a complex and expensive endeavour, this solution is not the preffered one.
    498498
    499 \subsubsection{Partial signalling}
     499\subsubsection{Partial signalling} \label{partial-sig}
    500500Finally, the solution that is chosen for \CFA is to use partial signalling. Consider the following case:
    501501
     
    605605% ======================================================================
    606606% ======================================================================
    607 \subsection{Internal scheduling: Implementation} \label{insched-impl}
    608 % ======================================================================
    609 % ======================================================================
    610 \TODO
    611 
     607\subsection{Internal scheduling: Implementation} \label{inschedimpl}
     608% ======================================================================
     609% ======================================================================
     610There are several challenges specific to \CFA when implementing internal scheduling. These challenges are direct results of \gls{group-acquire} and loose object definitions. These two constraints are to root cause of most design decisions in the implementation of internal scheduling. Furthermore, to avoid the head-aches of dynamically allocating memory in a concurrent environment, the internal-scheduling design is entirely free of mallocs and other dynamic memory allocation scheme. This is to avoid the chicken and egg problem 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.
     611
     612The main memory concern for concurrency is queues. All blocking operations are made by parking threads onto queues. These queues need to be intrinsic\cit to avoid the need memory allocation. This entails that all the fields needed to keep track of all needed information. Since internal scheduling can use an unbound amount of memory (depending on \gls{group-acquire}) statically defining information information in the intrusive fields of threads is insufficient. The only variable sized container that does not require memory allocation is the callstack, which is heavily used in the implementation of internal scheduling. Particularly the GCC extension variable length arrays which is used extensively.
     613
     614Since stack allocation is based around scope, 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. In the case of external scheduling, the threads and the condition both allow a fixed amount of memory to be stored, while mutex-routines and the actual blocking call allow for an unbound amount (though adding too much to the mutex routine stack size can become expansive faster).
     615
     616The following figure is the traditionnal illustration of a monitor :
     617
     618\begin{center}
     619{\resizebox{0.4\textwidth}{!}{\input{monitor}}}
     620\end{center}
     621
     622For \CFA, the previous picture does not have support for blocking multiple monitors on a single condition. To support \gls{group-acquire} two changes to this picture are required. First, it doesn't make sense to tie the condition to a single monitor since blocking two monitors as one would require arbitrarily picking a monitor to hold the condition. Secondly, the object waiting on the conditions and AS-stack cannot simply contain the waiting thread since a single thread can potentially wait on multiple monitors. As mentionned in section \ref{inschedimpl}, the handling in multiple monitors is done by partially passing, which entails that each concerned monitor needs to have a node object. However, for waiting on the condition, since all threads need to wait together, a single object needs to be queued in the condition. Moving out the condition and updating the node types yields :
     623
     624\begin{center}
     625{\resizebox{0.8\textwidth}{!}{\input{int_monitor}}}
     626\end{center}
     627
     628\newpage
     629
     630This picture and the proper entry and leave algorithms is the fundamental implementation of internal scheduling.
     631
     632\begin{multicols}{2}
     633Entry
     634\begin{pseudo}[numbers=left]
     635if monitor is free
     636        enter
     637elif I already own the monitor
     638        continue
     639else
     640        block
     641increment recursion
     642
     643\end{pseudo}
     644\columnbreak
     645Exit
     646\begin{pseudo}[numbers=left, firstnumber=8]
     647decrement recursion
     648if recursion == 0
     649        if signal_stack not empty
     650                set_owner to thread
     651                if all monitors ready
     652                        wake-up thread
     653
     654        if entry queue not empty
     655                wake-up thread
     656\end{pseudo}
     657\end{multicols}
     658
     659Some important things to notice about the exit routine. The solution discussed in \ref{inschedimpl} can be seen on line 11 of the previous pseudo code. Basically, the solution boils down to having a seperate 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 trasnferred ownership. This solution is safe as well as preventing any potential barging.
    612660
    613661% ======================================================================
     
    644692                inUse = true;
    645693        }
    646         void g() {
     694        void V() {
    647695                inUse = false;
    648696
     
    652700\end{tabular}
    653701\end{center}
    654 This method is more constrained and explicit, which may help 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) or in terms of data (e.g. Go). Of course, both of these paradigms have their own strenghts and weaknesses but for this project control flow semantics where 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 following example shows a simple use \code{accept} versus \code{wait}/\code{signal} and its advantages.
    655 
    656 In the case of internal scheduling, the call to \code{wait} only guarantees that \code{g} is the last routine to access the monitor. This entails that the routine \code{f} may have acquired mutual exclusion several times while routine \code{h} was waiting. On the other hand, external scheduling guarantees that while routine \code{h} was waiting, no routine other than \code{g} could acquire the monitor.
     702This method is more constrained and explicit, which may help 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) or in terms of data (e.g. Go). 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} as the core external scheduling keyword, \CFA uses \code{waitfor} to prevent name collisions with existing socket APIs.
     703
     704In 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 the routine \code{V} 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.
    657705
    658706% ======================================================================
     
    667715
    668716        void f(A & mutex a);
    669         void g(A & mutex a) { accept(f); }
    670 \end{cfacode}
    671 
    672 However, 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:
     717        void f(int a, float b);
     718        void g(A & mutex a) {
     719                waitfor(f); // Less obvious which f() to wait for
     720        }
     721\end{cfacode}
     722
     723Furthermore, 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:
    673724
    674725\begin{center}
     
    677728        if monitor is free
    678729                enter
     730        elif I already own the monitor
     731                continue
    679732        elif monitor accepts me
    680733                enter
     
    685738\end{center}
    686739
    687 For the \pscode{monitor is free} condition 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:
     740For the fist 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:
    688741
    689742\begin{center}
     
    691744\end{center}
    692745
    693 There 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. However, this relies on the fact that all the acceptable routines are declared with the monitor type. 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.
     746There 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.
    694747The alternative would be to have a picture more like this one:
    695748
     
    698751\end{center}
    699752
    700 Not storing the queues inside the monitor means that the storage can vary between routines, 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 accept to check if a routine is already queued in.
    701 
    702 At 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.
    703 
    704 In either cases here are a few alternatives for the different syntaxes this syntax : \\
    705 \begin{center}
    706 {\renewcommand{\arraystretch}{1.5}
    707 \begin{tabular}[t]{l @{\hskip 0.35in} l}
    708 \hline
    709 \multicolumn{2}{ c }{\code{accept} on type}\\
    710 \hline
    711 Alternative 1 & Alternative 2 \\
    712 \begin{lstlisting}
    713 mutex struct A
    714 accept( void f(A & mutex a) )
    715 {};
    716 \end{lstlisting} &\begin{lstlisting}
    717 mutex struct A {}
    718 accept( void f(A & mutex a) );
    719 
    720 \end{lstlisting} \\
    721 Alternative 3 & Alternative 4 \\
    722 \begin{lstlisting}
    723 mutex struct A {
    724         accept( void f(A & mutex a) )
    725 };
    726 
    727 \end{lstlisting} &\begin{lstlisting}
    728 mutex struct A {
    729         accept :
    730                 void f(A & mutex a) );
    731 };
    732 \end{lstlisting}\\
    733 \hline
    734 \multicolumn{2}{ c }{\code{accept} on routine}\\
    735 \hline
    736 \begin{lstlisting}
    737 mutex struct A {};
    738 
    739 void f(A & mutex a)
    740 
    741 accept( void f(A & mutex a) )
    742 void g(A & mutex a) {
    743         /*...*/
    744 }
    745 \end{lstlisting}&\\
    746 \end{tabular}
    747 }
    748 \end{center}
    749 
    750 Another aspect to consider is what happens if multiple overloads of the same routine are used. For the time being it is assumed that multiple overloads of the same routine should be scheduled regardless of the overload used. However, this could easily be extended in the future.
     753Not storing the queues inside the monitor means that the storage can vary between routines, 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.
     754
     755At 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 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.
     756
     757Another aspect to consider is what happens if multiple overloads of the same routine are used. For the time being it is assumed that multiple overloads of the same routine are considered as distinct routines. However, this could easily be extended in the future.
    751758
    752759% ======================================================================
     
    758765External scheduling, like internal scheduling, becomes orders of magnitude more complex when we start introducing multi-monitor syntax. Even in the simplest possible case some new semantics need to be established:
    759766\begin{cfacode}
    760         accept( void f(mutex struct A & mutex this))
    761767        mutex struct A {};
    762768
     
    764770
    765771        void g(A & mutex a, B & mutex b) {
    766                 accept(f); //ambiguous, which monitor
     772                waitfor(f); //ambiguous, which monitor
    767773        }
    768774\end{cfacode}
     
    771777
    772778\begin{cfacode}
    773         accept( void f(mutex struct A & mutex this))
    774779        mutex struct A {};
    775780
     
    777782
    778783        void g(A & mutex a, B & mutex b) {
    779                 accept( b, f );
    780         }
    781 \end{cfacode}
    782 
    783 This is unambiguous. Both locks will be acquired and kept, when routine \code{f} is called the lock for monitor \code{a} will be temporarily transferred from \code{g} to \code{f} (while \code{g} still holds lock \code{b}). This behavior can be extended to multi-monitor accept statment as follows.
    784 
    785 \begin{cfacode}
    786         accept( void f(mutex struct A & mutex, mutex struct A & mutex))
     784                waitfor( f, b );
     785        }
     786\end{cfacode}
     787
     788This is unambiguous. Both locks will be acquired and kept, when routine \code{f} is called the lock for monitor \code{b} will be 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 statment as follows.
     789
     790\begin{cfacode}
    787791        mutex struct A {};
    788792
     
    790794
    791795        void g(A & mutex a, B & mutex b) {
    792                 accept( b, a, f);
    793         }
    794 \end{cfacode}
    795 
    796 Note that the set of monitors passed to the \code{accept} statement must be entirely contained in the set of monitor already acquired in the routine. \code{accept} used in any other context is Undefined Behaviour.
     796                waitfor( f, a, b);
     797        }
     798\end{cfacode}
     799
     800Note that the set of monitors passed to the \code{waitfor} statement must be entirely contained in the set of monitor already acquired in the routine. \code{waitfor} used in any other context is Undefined Behaviour.
     801
     802An important behavior to note is that what happens when set of monitors only match partially :
     803
     804\begin{cfacode}
     805        mutex struct A {};
     806
     807        mutex struct B {};
     808
     809        void g(A & mutex a, B & mutex b) {
     810                waitfor(f, a, b);
     811        }
     812
     813        A a1, a2;
     814        B b;
     815
     816        void foo() {
     817                g(a1, b);
     818        }
     819
     820        void bar() {
     821                f(a2, b);
     822        }
     823\end{cfacode}
     824
     825While the equivalent can happen when using internal scheduling, the fact that conditions are branded on first use means that users have to use two different condition variables. In both cases, partially matching monitor sets will 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.
    797826
    798827% ======================================================================
Note: See TracChangeset for help on using the changeset viewer.