[concurrency-interest] Concurrency-interest Digest, Vol 87, Issue 27

Khilan Gudka khilan at doc.ic.ac.uk
Mon Apr 16 15:22:10 EDT 2012


Hi Rohit

This sounds very interesting.

A few questions though...

What happens if you have already acquired locks further up and have
performed side effects since then? Also, what if you are involved in a
mutual deadlock whereby side effects have occurred in two or more processes
so you cannot simply rollback the locks?

In our lock inference work, we bunch all the locks together so if a
deadlock occurs, like you say, you can just release already acquired locks
and try again. We don't do cycle detection though because of the potential
overheads. What kind of overheads do you get? Is it proportional to the
number of threads?

Thanks

--
Khilan Gudka
PhD Student
Department of Computing
Imperial College London
http://www.doc.ic.ac.uk/~khilan/



On 16 April 2012 19:36, Rohit Kumar <rohitk242 at yahoo.co.in> wrote:

> Hans,Nathan,
>
> I am not aware of anything like Gadara yet. My implementation assumes that
> there are no synchnonized keywords (implicit locks) in your program. It
> requires you to use explicit locks (e.g. Lock class in java concurrent
> api). There is no background thread running here. The deadlock-check
> happens at each lock acquisition request; it runs an algorithms internally
> to see if there are any cycles formed in the lock-acquisition graph. If
> yes, it doesn't let you acquire the lock but throws an exception. You can
> catch this exception and take necessary steps to re-acquire all the locks
> once again or whatever you want. The program is written in java and it is
> hardly 270 lines of code.
>
> Finally where do I need to present this writeup ?
>
> Thanks & Regards,
> Rohit Kumar
>
>   *From:* "concurrency-interest-request at cs.oswego.edu" <
> concurrency-interest-request at cs.oswego.edu>
> *To:* concurrency-interest at cs.oswego.edu
> *Sent:* Monday, 16 April 2012 10:50 PM
> *Subject:* Concurrency-interest Digest, Vol 87, Issue 27
>
> Send Concurrency-interest mailing list submissions to
>     concurrency-interest at cs.oswego.edu
>
> To subscribe or unsubscribe via the World Wide Web, visit
>     http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> or, via email, send a message with subject or body 'help' to
>     concurrency-interest-request at cs.oswego.edu
>
> You can reach the person managing the list at
>     concurrency-interest-owner at cs.oswego.edu
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Concurrency-interest digest..."
>
>
> Today's Topics:
>
>   1. Re: CountedCompleters (Wolfgang Baltes)
>   2. Re: Java Deadlocks prevented (Boehm, Hans)
>   3. Re: Java Deadlocks prevented (Nathan Reynolds)
>   4. Re: Concurrency-interest Digest, Vol 87,    Issue 26 (Java
>       deadlock prevented) (Rohit Kumar)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 16 Apr 2012 18:01:50 +0200
> From: Wolfgang Baltes <wolfgang.baltes at laposte.net>
> To: concurrency-interest at cs.oswego.edu
> Subject: Re: [concurrency-interest] CountedCompleters
> Message-ID: <4F8C426E.6090704 at laposte.net>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> I apologize for a mistake in the last paragraph of my memo: using
> quietlyJoinUnforked() in the SDK7 sample code for MergeSort does have a
> non-negligible performance impact (not no impact as stated). There is
> better performance in case of recursions which produce many tasks that
> are not explicitly forked, and it reduces the number of extra threads
> significantly, allowing larger problems to be solved with smaller memory
> footprint.
>
> Wolfgang.
>
> On 2012-04-16 16:48, Wolfgang Baltes wrote:
> > Thanks, Doug, for a this addition to the FJ framework. I think that
> > CountedCompleter will address the needs of an entire class of
> > applications in an efficient and simple to use manner.
> >
> > I used the code and noticed that method doJoin() has become more
> > effective in avoiding blocking threads, and as a result fewer extra
> > threads are created. I found the performance, compared to
> > RecursiveAction, to be equal or insignificantly different. This
> > reduces the problem described in item 3 below.
> >
> > However, at the same time, CountedCompleter does not fully satisfy the
> > needs for a class of problems I work on. To this end, here are a few
> > enhancements I would like to suggest:
> >
> > 1: Symmetrically to onCompletion(), provide
> > onExceptionalCompletion(Throwable). This allows filtering exception
> > propagation. There are cases where the propagation of the exception is
> > desired, and others where a corrective action is taken instead, such
> > as a retry.
> >
> > 2: As a further enhancement to 1: enable any Throwable, including
> > checked exceptions. This allows the use of a CountedCompleter as a
> > CompletionHandler for asynchronous IO operations or as a wrapper for
> > MethodHandles (which throw Throwable) without adding extra logic to
> > capture and convert an IO exception. I read the documentation which
> > explains why this is currently limited to unchecked exceptions. While
> > I can agree with this in general, I feel the argument is weak for
> > CountedCompleter if it is there to support asynchronous tasks/events.
> > (May I add that using this type of framework is not for the
> > faint-hearted anyway!?)
> >
> > 3: Provide a method to join a task that is not forked and/or not
> > completable, while minimizing worker thread blocking. For example,
> > CountedCompleter allows creating chains of dependent tasks. Unless the
> > ultimate task (the last in the chain) is forked/exists on the task
> > stack AND can complete because all dependencies are resolved, joining
> > it will block the worker thread. I noticed (and my testing is limited
> > to a few test cases and therefore not representative) the blocking and
> > the creation of other worker threads, ultimately running out of memory
> > or reaching the thread count limit. If this task is not forked, then
> > join()/quietlyJoin() will block the worker thread. The following code
> > is my (inexpert) attempt to provide a remedy. It is based on the
> > assumption that a task that depends on others for completion is not
> > forked until all dependencies are resolved. For example, a
> > CountedCompleter implementing CompletionHandler would fork itself
> > ("implicit fork") when the IO operation is done. This works very well
> > in my test cases, but at this time I would not claim it to be
> > universally applicable or error free. It is shown here more to
> > demonstrate the attempt rather than as a reference implementation.
> > With access to private data structures, this can be done more
> > elegantly and more reliably.
> >
> >        static final int RETRIES = 16;
> >        static final long WAIT_TIMEOUT = 1_000;    // Timeout in
> > microseconds.
> >
> >        public final void quietlyJoinUnforked() {
> >            this.doJoinUnforked(false);
> >        }
> >
> >        public final void quietlyJoinUnforkedInterruptibly()
> >        throws InterruptedException {
> >            if (this.doJoinUnforked(true)) {
> >                throw new InterruptedException();
> >            }
> >        }
> >
> >        public final boolean doJoinUnforked(final boolean
> > interruptibly) {
> >            int retries = RETRIES;
> >            boolean wasInterrupted = false;
> >            while (!this.isDone()) {
> >                ForkJoinTask<?> t;
> >                if ((t = pollTask()) != null) {
> >                    t.quietlyInvoke();
> >                    if (t == this) {
> >                        break;
> >                    }
> >                }
> >                else {
> >                    if (retries-- > 0) {
> >                        Thread.yield();
> >                        continue;
> >                    }
> >                    wasInterrupted = Thread.interrupted();
> >                    try {
> >                        // get(...) is used as a timed join(). It is
> > assumed that
> >                        // other code will perform get() on this task
> > to retrieve
> >                        // the task's result or exception.
> >                        this.get(WAIT_TIMEOUT, TimeUnit.MICROSECONDS);
> >                        break;
> >                    }
> >                    catch (final InterruptedException consumed) {
> >                        if (!interruptibly) {
> >                            wasInterrupted = true;
> >                            continue;
> >                        }
> >                        return true;
> >                    }
> >                    catch (final ExecutionException ignored) {
> >                        // See comment on get() above.
> >                        break;
> >                    }
> >                    catch (final TimeoutException ignored) {
> >                        continue;
> >                    }
> >                }
> >                retries = RETRIES;
> >            }
> >            if (wasInterrupted && !interruptibly) {
> >                Thread.currentThread().interrupt();
> >                return false;
> >            }
> >            return wasInterrupted;
> >        }
> >
> > As already mentioned this works quite well in a number of cases. For
> > example, adding this method to the example MergeSort code and calling
> > quietlyJoinUnforked(), results in the same overall performance,
> > reduces the number of extra blocked worker threads to 1 if any
> > (instead of up to 8 for the unmodified code; on a PC with 4
> > hyper-threading cores/8 threads), and allows for some extra
> > (recreational?) freedom in joining the right and left sub-tasks in any
> > order. It works in cases where no sub-task is forked explicitly. I
> > observed that worker thread blocking only occurs towards the end of a
> > large recursion, suggesting that worker threads only block - as
> > intended - when there is no other work available (sometimes while
> > implicit forking has not yet happened).
> >
> > Wolfgang.
> >
> >
> >
> > On 2012-04-09 16:16, Doug Lea wrote:
> >>
> >> After sitting on multiple variations for months, I committed
> >> CountedCompleter, a completion-based flavor of ForkJoinTask.
> >>
> >> As mentioned a few times over the past year, the main motivation
> >> is to better support tasks that perform IO or other base
> >> actions that may (or may not) take a lot of time to execute.
> >> As is the case with JDK7 async IO and other completion-based
> >> frameworks, the most common path to efficiency is for such tasks
> >> to arrange continuation actions that occur upon their completion.
> >> The main twist for CountedCompleters is that continuations
> >> might be dependent on multiple actions, not just one. (Or in
> >> other words, the continuations must be preceded by a specialized,
> >> "bottom-up" form of join.)
> >>
> >> The CountedCompleter abstract class provides a minimal basis
> >> for these kinds of tasks. While some of the mechanics are
> >> reminiscent of other FJ-like frameworks such as Intel TBB,
> >> CountedCompleters are designed to fit smoothly with other
> >> kinds of ForkJoinTasks (like RecursiveActions), and so still
> >> allow people to use the more pleasant Future-style conventions
> >> rather than count-based bottom-up joining unless they need them.
> >> At the same time, the CountedCompleter class exposes enough
> >> mechanics to allow all sorts of tweaks that people can use
> >> to improve performance.
> >> In particular, in addition to usually being the best way to deal
> >> with IO etc bound tasks, CountedCompleters sometimes fare better
> >> than RecursiveActions in programs that entail lots of garbage
> >> collection because GC can have similar impact on task variability.
> >>
> >> Even though targeted for JDK8, versions of CountedCompleter
> >> appear in the jsr166y and main repositories, not jsr166e. This is
> >> because they require a non-public hook into modified ForkJoinTask
> >> exception handling mechanics in order to properly propagate
> >> exceptional completions. For sources, docs, and jar files, see
> >> the usual links at
> >> http://gee.cs.oswego.edu/dl/concurrency-interest/index.html
> >>
> >> The API docs include more details and some examples:
> >>
> http://gee.cs.oswego.edu/dl/jsr166/dist/docs/java/util/concurrent/CountedCompleter.html
> >>
> >>
> >> I also added a few (with more to come) test/demo programs that
> >> illustrate
> >> other usages. See CCBoxedLongSort and CCJacobi in
> >> http://gee.cs.oswego.edu/cgi-bin/viewcvs.cgi/jsr166/src/test/loops/
> >>
> >> Please try these out. As always, comments and suggestions
> >> (hopefully based on usage experience) would be welcome.
> >>
> >> -Doug
> >>
> >>
> >> _______________________________________________
> >> Concurrency-interest mailing list
> >> Concurrency-interest at cs.oswego.edu
> >> http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> >>
> > _______________________________________________
> > Concurrency-interest mailing list
> > Concurrency-interest at cs.oswego.edu
> > http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> >
>
>
> ------------------------------
>
> Message: 2
> Date: Mon, 16 Apr 2012 16:38:00 +0000
> From: "Boehm, Hans" <hans.boehm at hp.com>
> To: Andrew Haley <aph at redhat.com>,
>     "concurrency-interest at cs.oswego.edu"
>     <concurrency-interest at cs.oswego.edu>
> Subject: Re: [concurrency-interest] Java Deadlocks prevented
> Message-ID:
>     <A3E67C2071F49C4CBC4F17E6D77CDDD235543F5B at G4W3299.americas.hpqcorp.net
> >
>
> Content-Type: text/plain; charset="us-ascii"
>
> Important questions to consider when you write it up:
>
> How does it compare to something like Gadara that uses whole program
> analysis and scheduling to avoid lock-based deadlocks?  (Hopefully it
> doesn't need whole program analysis.)  Other deadlock avoidance schemes?
>
> Once you detect a potential deadlock, how do you recover?  Or can you
> always schedule so that the possibility doesn't arise?
>
> Hans
>
> > -----Original Message-----
> > From: concurrency-interest-bounces at cs.oswego.edu [mailto:concurrency-
> > interest-bounces at cs.oswego.edu] On Behalf Of Andrew Haley
> > Sent: Monday, April 16, 2012 4:34 AM
> > To: concurrency-interest at cs.oswego.edu
> > Subject: Re: [concurrency-interest] Java Deadlocks prevented
> >
> > On 04/16/2012 12:06 PM, Rohit Kumar wrote:
> >
> > > I have found a way of preventing deadlocks in java. The
> > > methodology(which is code) completely prevents the deadlock from
> > > occuring by detecting it in advance. This can be used across the
> > > systems seamlessly.
> > >
> > > Kindly let me know what I need to do next. I want this to be part of
> > > next jdk release. I am writing this email as I have no idea what I
> > > need to do next to bring it into limelight.
> >
> > Write it up, maybe present it to a conference, and wait for feedback.
> > That's how it always works.  If your idea really works, people will
> > use it.
> >
> > Andrew.
> > _______________________________________________
> > Concurrency-interest mailing list
> > Concurrency-interest at cs.oswego.edu
> > http://cs.oswego.edu/mailman/listinfo/concurrency-interest
>
>
>
> ------------------------------
>
> Message: 3
> Date: Mon, 16 Apr 2012 10:11:00 -0700
> From: Nathan Reynolds <nathan.reynolds at oracle.com>
> To: "Boehm, Hans" <hans.boehm at hp.com>
> Cc: "concurrency-interest at cs.oswego.edu"
>     <concurrency-interest at cs.oswego.edu>
> Subject: Re: [concurrency-interest] Java Deadlocks prevented
> Message-ID: <4F8C52A4.70002 at oracle.com>
> Content-Type: text/plain; charset="iso-8859-1"; Format="flowed"
>
> Deadlock prevention is very valuable.  It means deadlock prone code
> won't bring down a production server and cost the company millions in
> down time.  It means consumers won't kill the process and request a refund.
>
> How much does deadlock prevention cost?  Is the cost on the thread that
> acquires locks or is it in a background thread?
>
> Each time processors or systems increase the number of cores, I find we
> have to do a round of lock contention fixing.  I have only seen 1 lock
> at a time be the bottleneck in the system.  Does deadlock prevention
> increase the critical region of locks?  If so, this will definitely
> reduce the scalability of the system if it impacts the 1 bottlenecking
> lock.
>
> Lock performance is a very important consideration.  Locks have evolved
> from fat locks (i.e trips into the OS kernel) to thin locks (i.e. spin
> and CAS in user mode) to biased/lazy locks (i.e. no CAS and an
> indefinite lock owner).  All of this was done to reduce the performance
> overhead of locks.  How does deadlock prevention impact the performance
> of biased, thin and fat locks?  I am not as concerned about fat lock
> performance since most of the time the thread is going to block.
>
> Nathan Reynolds
> <http://psr.us.oracle.com/wiki/index.php/User:Nathan_Reynolds> |
> Consulting Member of Technical Staff | 602.333.9091
> Oracle PSR Engineering <http://psr.us.oracle.com/> | Server Technology
>
> On 4/16/2012 9:38 AM, Boehm, Hans wrote:
> > Important questions to consider when you write it up:
> >
> > How does it compare to something like Gadara that uses whole program
> analysis and scheduling to avoid lock-based deadlocks?  (Hopefully it
> doesn't need whole program analysis.)  Other deadlock avoidance schemes?
> >
> > Once you detect a potential deadlock, how do you recover?  Or can you
> always schedule so that the possibility doesn't arise?
> >
> > Hans
> >
> >> -----Original Message-----
> >> From: concurrency-interest-bounces at cs.oswego.edu [mailto:concurrency-
> >> interest-bounces at cs.oswego.edu] On Behalf Of Andrew Haley
> >> Sent: Monday, April 16, 2012 4:34 AM
> >> To: concurrency-interest at cs.oswego.edu
> >> Subject: Re: [concurrency-interest] Java Deadlocks prevented
> >>
> >> On 04/16/2012 12:06 PM, Rohit Kumar wrote:
> >>
> >>> I have found a way of preventing deadlocks in java. The
> >>> methodology(which is code) completely prevents the deadlock from
> >>> occuring by detecting it in advance. This can be used across the
> >>> systems seamlessly.
> >>>
> >>> Kindly let me know what I need to do next. I want this to be part of
> >>> next jdk release. I am writing this email as I have no idea what I
> >>> need to do next to bring it into limelight.
> >> Write it up, maybe present it to a conference, and wait for feedback.
> >> That's how it always works.  If your idea really works, people will
> >> use it.
> >>
> >> Andrew.
> >> _______________________________________________
> >> Concurrency-interest mailing list
> >> Concurrency-interest at cs.oswego.edu
> >> http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> > _______________________________________________
> > Concurrency-interest mailing list
> > Concurrency-interest at cs.oswego.edu
> > http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <
> http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20120416/7ea89bb3/attachment-0001.html
> >
>
> ------------------------------
>
> Message: 4
> Date: Tue, 17 Apr 2012 01:20:15 +0800 (SGT)
> From: Rohit Kumar <rohitk242 at yahoo.co.in>
> To: "concurrency-interest at cs.oswego.edu"
>     <concurrency-interest at cs.oswego.edu>
> Subject: Re: [concurrency-interest] Concurrency-interest Digest, Vol
>     87,    Issue 26 (Java deadlock prevented)
> Message-ID:
>     <1334596815.27848.YahooMailNeo at web193403.mail.sg3.yahoo.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Andrew:
> ?
> Thanks for the reply. Can you kindly let me know where do I need to
> present it ? Where will the conference be held ? Can I present it online or
> I have to come in person ?
> ?
> Waiting for your early reply once again.
> ?
> Thanks & Regards,
> Rohit Kumar
>
> From: "concurrency-interest-request at cs.oswego.edu" <
> concurrency-interest-request at cs.oswego.edu>
> To: concurrency-interest at cs.oswego.edu
> Sent: Monday, 16 April 2012 9:30 PM
> Subject: Concurrency-interest Digest, Vol 87, Issue 26
>
> Send Concurrency-interest mailing list submissions to
> ??? concurrency-interest at cs.oswego.edu
>
> To subscribe or unsubscribe via the World Wide Web, visit
> ??? http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> or, via email, send a message with subject or body 'help' to
> ??? concurrency-interest-request at cs.oswego.edu
>
> You can reach the person managing the list at
> ??? concurrency-interest-owner at cs.oswego.edu
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Concurrency-interest digest..."
>
>
> Today's Topics:
>
> ? 1. (no subject) (Rohit Kumar)
> ? 2. Java Deadlocks prevented (Rohit Kumar)
> ? 3. Re: Java Deadlocks prevented (Andrew Haley)
> ? 4. Re: CountedCompleters (Wolfgang Baltes)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Mon, 16 Apr 2012 19:03:59 +0800 (SGT)
> From: Rohit Kumar <rohitk242 at yahoo.co.in>
> To: "concurrency-interest at cs.oswego.edu"
> ??? <concurrency-interest at cs.oswego.edu>
> Cc: "concurrency-interest-owner at cs.oswego.edu"
> ??? <concurrency-interest-owner at cs.oswego.edu>
> Subject: [concurrency-interest] (no subject)
> Message-ID:
> ??? <1334574239.81244.YahooMailNeo at web193402.mail.sg3.yahoo.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Hi All,
> ?
> I have found a way of preventing deadlocks in java. The methodology(which
> is code) completely prevents the deadlock from accuring by detecting it in
> advance. This can be used across the systems seamlessly.
> ?
> Kindly let me know what I need to do next. I want this to be part of next
> jdk release.
> ?
> Thanks & Regards,
> Rohit Kumar
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <
> http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20120416/a6588341/attachment-0001.html
> >
>
> ------------------------------
>
> Message: 2
> Date: Mon, 16 Apr 2012 19:06:22 +0800 (SGT)
> From: Rohit Kumar <rohitk242 at yahoo.co.in>
> To: "concurrency-interest at cs.oswego.edu"
> ??? <concurrency-interest at cs.oswego.edu>
> Cc: "concurrency-interest-owner at cs.oswego.edu"
> ??? <concurrency-interest-owner at cs.oswego.edu>
> Subject: [concurrency-interest] Java Deadlocks prevented
> Message-ID:
> ??? <1334574382.90584.YahooMailNeo at web193403.mail.sg3.yahoo.com>
> Content-Type: text/plain; charset="utf-8"
>
>
>
> Hi All,
>
> I have found a way of preventing deadlocks in java. The methodology(which
> is code) completely prevents the deadlock from occuring by detecting it in
> advance. This can be used across the systems seamlessly.
>
> Kindly let me know what I need to do next. I want this to be part of next
> jdk release. I am writing this email as I have no idea what I need to do
> next to bring it into limelight.
>
> Thanks & Regards,
> Rohit Kumar
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <
> http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20120416/064a4873/attachment-0001.html
> >
>
> ------------------------------
>
> Message: 3
> Date: Mon, 16 Apr 2012 12:33:45 +0100
> From: Andrew Haley <aph at redhat.com>
> To: concurrency-interest at cs.oswego.edu
> Subject: Re: [concurrency-interest] Java Deadlocks prevented
> Message-ID: <4F8C0399.8040301 at redhat.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> On 04/16/2012 12:06 PM, Rohit Kumar wrote:
>
> > I have found a way of preventing deadlocks in java. The
> > methodology(which is code) completely prevents the deadlock from
> > occuring by detecting it in advance. This can be used across the
> > systems seamlessly.
> >
> > Kindly let me know what I need to do next. I want this to be part of
> > next jdk release. I am writing this email as I have no idea what I
> > need to do next to bring it into limelight.
>
> Write it up, maybe present it to a conference, and wait for feedback.
> That's how it always works.? If your idea really works, people will
> use it.
>
> Andrew.
>
>
> ------------------------------
>
> Message: 4
> Date: Mon, 16 Apr 2012 16:48:31 +0200
> From: Wolfgang Baltes <wolfgang.baltes at laposte.net>
> To: "Concurrency-interest at cs.oswego.edu"
> ??? <Concurrency-interest at cs.oswego.edu>
> Subject: Re: [concurrency-interest] CountedCompleters
> Message-ID: <4F8C313F.2080308 at laposte.net>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Thanks, Doug, for a this addition to the FJ framework. I think that
> CountedCompleter will address the needs of an entire class of
> applications in an efficient and simple to use manner.
>
> I used the code and noticed that method doJoin() has become more
> effective in avoiding blocking threads, and as a result fewer extra
> threads are created. I found the performance, compared to
> RecursiveAction, to be equal or insignificantly different. This reduces
> the problem described in item 3 below.
>
> However, at the same time, CountedCompleter does not fully satisfy the
> needs for a class of problems I work on. To this end, here are a few
> enhancements I would like to suggest:
>
> 1: Symmetrically to onCompletion(), provide
> onExceptionalCompletion(Throwable). This allows filtering exception
> propagation. There are cases where the propagation of the exception is
> desired, and others where a corrective action is taken instead, such as
> a retry.
>
> 2: As a further enhancement to 1: enable any Throwable, including
> checked exceptions. This allows the use of a CountedCompleter as a
> CompletionHandler for asynchronous IO operations or as a wrapper for
> MethodHandles (which throw Throwable) without adding extra logic to
> capture and convert an IO exception. I read the documentation which
> explains why this is currently limited to unchecked exceptions. While I
> can agree with this in general, I feel the argument is weak for
> CountedCompleter if it is there to support asynchronous tasks/events.
> (May I add that using this type of framework is not for the
> faint-hearted anyway!?)
>
> 3: Provide a method to join a task that is not forked and/or not
> completable, while minimizing worker thread blocking. For example,
> CountedCompleter allows creating chains of dependent tasks. Unless the
> ultimate task (the last in the chain) is forked/exists on the task stack
> AND can complete because all dependencies are resolved, joining it will
> block the worker thread. I noticed (and my testing is limited to a few
> test cases and therefore not representative) the blocking and the
> creation of other worker threads, ultimately running out of memory or
> reaching the thread count limit. If this task is not forked, then
> join()/quietlyJoin() will block the worker thread. The following code is
> my (inexpert) attempt to provide a remedy. It is based on the assumption
> that a task that depends on others for completion is not forked until
> all dependencies are resolved. For example, a CountedCompleter
> implementing CompletionHandler would fork itself ("implicit fork") when
> the IO operation is done. This works very well in my test cases, but at
> this time I would not claim it to be universally applicable or error
> free. It is shown here more to demonstrate the attempt rather than as a
> reference implementation. With access to private data structures, this
> can be done more elegantly and more reliably.
>
> ? ? ? ? static final int RETRIES = 16;
> ? ? ? ? static final long WAIT_TIMEOUT = 1_000;? ? // Timeout in
> microseconds.
>
> ? ? ? ? public final void quietlyJoinUnforked() {
> ? ? ? ? ? ? this.doJoinUnforked(false);
> ? ? ? ? }
>
> ? ? ? ? public final void quietlyJoinUnforkedInterruptibly()
> ? ? ? ? throws InterruptedException {
> ? ? ? ? ? ? if (this.doJoinUnforked(true)) {
> ? ? ? ? ? ? ? ? throw new InterruptedException();
> ? ? ? ? ? ? }
> ? ? ? ? }
>
> ? ? ? ? public final boolean doJoinUnforked(final boolean interruptibly) {
> ? ? ? ? ? ? int retries = RETRIES;
> ? ? ? ? ? ? boolean wasInterrupted = false;
> ? ? ? ? ? ? while (!this.isDone()) {
> ? ? ? ? ? ? ? ? ForkJoinTask<?> t;
> ? ? ? ? ? ? ? ? if ((t = pollTask()) != null) {
> ? ? ? ? ? ? ? ? ? ? t.quietlyInvoke();
> ? ? ? ? ? ? ? ? ? ? if (t == this) {
> ? ? ? ? ? ? ? ? ? ? ? ? break;
> ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? else {
> ? ? ? ? ? ? ? ? ? ? if (retries-- > 0) {
> ? ? ? ? ? ? ? ? ? ? ? ? Thread.yield();
> ? ? ? ? ? ? ? ? ? ? ? ? continue;
> ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? ? ? wasInterrupted = Thread.interrupted();
> ? ? ? ? ? ? ? ? ? ? try {
> ? ? ? ? ? ? ? ? ? ? ? ? // get(...) is used as a timed join(). It is
> assumed that
> ? ? ? ? ? ? ? ? ? ? ? ? // other code will perform get() on this task
> to retrieve
> ? ? ? ? ? ? ? ? ? ? ? ? // the task's result or exception.
> ? ? ? ? ? ? ? ? ? ? ? ? this.get(WAIT_TIMEOUT, TimeUnit.MICROSECONDS);
> ? ? ? ? ? ? ? ? ? ? ? ? break;
> ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? ? ? catch (final InterruptedException consumed) {
> ? ? ? ? ? ? ? ? ? ? ? ? if (!interruptibly) {
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? wasInterrupted = true;
> ? ? ? ? ? ? ? ? ? ? ? ? ? ? continue;
> ? ? ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? ? ? ? ? return true;
> ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? ? ? catch (final ExecutionException ignored) {
> ? ? ? ? ? ? ? ? ? ? ? ? // See comment on get() above.
> ? ? ? ? ? ? ? ? ? ? ? ? break;
> ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? ? ? catch (final TimeoutException ignored) {
> ? ? ? ? ? ? ? ? ? ? ? ? continue;
> ? ? ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? }
> ? ? ? ? ? ? ? ? retries = RETRIES;
> ? ? ? ? ? ? }
> ? ? ? ? ? ? if (wasInterrupted && !interruptibly) {
> ? ? ? ? ? ? ? ? Thread.currentThread().interrupt();
> ? ? ? ? ? ? ? ? return false;
> ? ? ? ? ? ? }
> ? ? ? ? ? ? return wasInterrupted;
> ? ? ? ? }
>
> As already mentioned this works quite well in a number of cases. For
> example, adding this method to the example MergeSort code and calling
> quietlyJoinUnforked(), results in the same overall performance, reduces
> the number of extra blocked worker threads to 1 if any (instead of up to
> 8 for the unmodified code; on a PC with 4 hyper-threading cores/8
> threads), and allows for some extra (recreational?) freedom in joining
> the right and left sub-tasks in any order. It works in cases where no
> sub-task is forked explicitly. I observed that worker thread blocking
> only occurs towards the end of a large recursion, suggesting that worker
> threads only block - as intended - when there is no other work available
> (sometimes while implicit forking has not yet happened).
>
> Wolfgang.
>
>
>
> On 2012-04-09 16:16, Doug Lea wrote:
> >
> > After sitting on multiple variations for months, I committed
> > CountedCompleter, a completion-based flavor of ForkJoinTask.
> >
> > As mentioned a few times over the past year, the main motivation
> > is to better support tasks that perform IO or other base
> > actions that may (or may not) take a lot of time to execute.
> > As is the case with JDK7 async IO and other completion-based
> > frameworks, the most common path to efficiency is for such tasks
> > to arrange continuation actions that occur upon their completion.
> > The main twist for CountedCompleters is that continuations
> > might be dependent on multiple actions, not just one. (Or in
> > other words, the continuations must be preceded by a specialized,
> > "bottom-up" form of join.)
> >
> > The CountedCompleter abstract class provides a minimal basis
> > for these kinds of tasks. While some of the mechanics are
> > reminiscent of other FJ-like frameworks such as Intel TBB,
> > CountedCompleters are designed to fit smoothly with other
> > kinds of ForkJoinTasks (like RecursiveActions), and so still
> > allow people to use the more pleasant Future-style conventions
> > rather than count-based bottom-up joining unless they need them.
> > At the same time, the CountedCompleter class exposes enough
> > mechanics to allow all sorts of tweaks that people can use
> > to improve performance.
> > In particular, in addition to usually being the best way to deal
> > with IO etc bound tasks, CountedCompleters sometimes fare better
> > than RecursiveActions in programs that entail lots of garbage
> > collection because GC can have similar impact on task variability.
> >
> > Even though targeted for JDK8, versions of CountedCompleter
> > appear in the jsr166y and main repositories, not jsr166e. This is
> > because they require a non-public hook into modified ForkJoinTask
> > exception handling mechanics in order to properly propagate
> > exceptional completions. For sources, docs, and jar files, see
> > the usual links at
> > http://gee.cs.oswego.edu/dl/concurrency-interest/index.html
> >
> > The API docs include more details and some examples:
> >
> http://gee.cs.oswego.edu/dl/jsr166/dist/docs/java/util/concurrent/CountedCompleter.html
> >
> >
> > I also added a few (with more to come) test/demo programs that illustrate
> > other usages. See CCBoxedLongSort and CCJacobi in
> > http://gee.cs.oswego.edu/cgi-bin/viewcvs.cgi/jsr166/src/test/loops/
> >
> > Please try these out. As always, comments and suggestions
> > (hopefully based on usage experience) would be welcome.
> >
> > -Doug
> >
> >
> > _______________________________________________
> > Concurrency-interest mailing list
> > Concurrency-interest at cs.oswego.edu
> > http://cs.oswego.edu/mailman/listinfo/concurrency-interest
> >
>
>
> ------------------------------
>
> _______________________________________________
> Concurrency-interest mailing list
> Concurrency-interest at cs.oswego.edu
> http://cs.oswego.edu/mailman/listinfo/concurrency-interest
>
>
> End of Concurrency-interest Digest, Vol 87, Issue 26
> ****************************************************
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <
> http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20120417/75746dcb/attachment.html
> >
>
> ------------------------------
>
> _______________________________________________
> Concurrency-interest mailing list
> Concurrency-interest at cs.oswego.edu
> http://cs.oswego.edu/mailman/listinfo/concurrency-interest
>
>
> End of Concurrency-interest Digest, Vol 87, Issue 27
> ****************************************************
>
>
>
> _______________________________________________
> Concurrency-interest mailing list
> Concurrency-interest at cs.oswego.edu
> http://cs.oswego.edu/mailman/listinfo/concurrency-interest
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20120416/938491d7/attachment-0001.html>


More information about the Concurrency-interest mailing list