A {@code ForkJoinPool} differs from other kinds of {@link * ExecutorService} mainly by virtue of employing * work-stealing: all threads in the pool attempt to find and * execute tasks submitted to the pool and/or created by other active * tasks (eventually blocking waiting for work if none exist). This * enables efficient processing when most tasks spawn other subtasks * (as do most {@code ForkJoinTask}s), as well as when many small * tasks are submitted to the pool from external clients. Especially * when setting asyncMode to true in constructors, {@code * ForkJoinPool}s may also be appropriate for use with event-style * tasks that are never joined. All worker threads are initialized * with {@link Thread#isDaemon} set {@code true}. * *
A static {@link #commonPool()} is available and appropriate for * most applications. The common pool is used by any ForkJoinTask that * is not explicitly submitted to a specified pool. Using the common * pool normally reduces resource usage (its threads are slowly * reclaimed during periods of non-use, and reinstated upon subsequent * use). * *
For applications that require separate or custom pools, a {@code * ForkJoinPool} may be constructed with a given target parallelism * level; by default, equal to the number of available processors. * The pool attempts to maintain enough active (or available) threads * by dynamically adding, suspending, or resuming internal worker * threads, even if some tasks are stalled waiting to join others. * However, no such adjustments are guaranteed in the face of blocked * I/O or other unmanaged synchronization. The nested {@link * ManagedBlocker} interface enables extension of the kinds of * synchronization accommodated. The default policies may be * overridden using a constructor with parameters corresponding to * those documented in class {@link ThreadPoolExecutor}. * *
In addition to execution and lifecycle control methods, this * class provides status check methods (for example * {@link #getStealCount}) that are intended to aid in developing, * tuning, and monitoring fork/join applications. Also, method * {@link #toString} returns indications of pool state in a * convenient form for informal monitoring. * *
As is the case with other ExecutorServices, there are three * main task execution methods summarized in the following table. * These are designed to be used primarily by clients not already * engaged in fork/join computations in the current pool. The main * forms of these methods accept instances of {@code ForkJoinTask}, * but overloaded forms also allow mixed execution of plain {@code * Runnable}- or {@code Callable}- based activities as well. However, * tasks that are already executing in a pool should normally instead * use the within-computation forms listed in the table unless using * async event-style tasks that are not usually joined, in which case * there is little difference among choice of methods. * *
| * | Call from non-fork/join clients | *Call from within fork/join computations | *
| Arrange async execution | *{@link #execute(ForkJoinTask)} | *{@link ForkJoinTask#fork} | *
| Await and obtain result | *{@link #invoke(ForkJoinTask)} | *{@link ForkJoinTask#invoke} | *
| Arrange exec and obtain Future | *{@link #submit(ForkJoinTask)} | *{@link ForkJoinTask#fork} (ForkJoinTasks are Futures) | *
The parameters used to construct the common pool may be controlled by * setting the following {@linkplain System#getProperty system properties}: *
Implementation notes: This implementation restricts the * maximum number of running threads to 32767. Attempts to create * pools with greater than the maximum number result in * {@code IllegalArgumentException}. * *
This implementation rejects submitted tasks (that is, by throwing
* {@link RejectedExecutionException}) only when the pool is shut down
* or internal resources have been exhausted.
*
* @since 1.7
* @author Doug Lea
*/
public class ForkJoinPool extends AbstractExecutorService {
/*
* Implementation Overview
*
* This class and its nested classes provide the main
* functionality and control for a set of worker threads:
* Submissions from non-FJ threads enter into submission queues.
* Workers take these tasks and typically split them into subtasks
* that may be stolen by other workers. Preference rules give
* first priority to processing tasks from their own queues (LIFO
* or FIFO, depending on mode), then to randomized FIFO steals of
* tasks in other queues. This framework began as vehicle for
* supporting tree-structured parallelism using work-stealing.
* Over time, its scalability advantages led to extensions and
* changes to better support more diverse usage contexts. Because
* most internal methods and nested classes are interrelated,
* their main rationale and descriptions are presented here;
* individual methods and nested classes contain only brief
* comments about details.
*
* WorkQueues
* ==========
*
* Most operations occur within work-stealing queues (in nested
* class WorkQueue). These are special forms of Deques that
* support only three of the four possible end-operations -- push,
* pop, and poll (aka steal), under the further constraints that
* push and pop are called only from the owning thread (or, as
* extended here, under a lock), while poll may be called from
* other threads. (If you are unfamiliar with them, you probably
* want to read Herlihy and Shavit's book "The Art of
* Multiprocessor programming", chapter 16 describing these in
* more detail before proceeding.) The main work-stealing queue
* design is roughly similar to those in the papers "Dynamic
* Circular Work-Stealing Deque" by Chase and Lev, SPAA 2005
* (http://research.sun.com/scalable/pubs/index.html) and
* "Idempotent work stealing" by Michael, Saraswat, and Vechev,
* PPoPP 2009 (http://portal.acm.org/citation.cfm?id=1504186).
* The main differences ultimately stem from GC requirements that
* we null out taken slots as soon as we can, to maintain as small
* a footprint as possible even in programs generating huge
* numbers of tasks. To accomplish this, we shift the CAS
* arbitrating pop vs poll (steal) from being on the indices
* ("base" and "top") to the slots themselves.
*
* Adding tasks then takes the form of a classic array push(task)
* in a circular buffer:
* q.array[q.top++ % length] = task;
*
* (The actual code needs to null-check and size-check the array,
* uses masking, not mod, for indexing a power-of-two-sized array,
* properly fences accesses, and possibly signals waiting workers
* to start scanning -- see below.) Both a successful pop and
* poll mainly entail a CAS of a slot from non-null to null.
*
* The pop operation (always performed by owner) is:
* if ((the task at top slot is not null) and
* (CAS slot to null))
* decrement top and return task;
*
* And the poll operation (usually by a stealer) is
* if ((the task at base slot is not null) and
* (CAS slot to null))
* increment base and return task;
*
* There are several variants of each of these. In particular,
* almost all uses of poll occur within scan operations that also
* interleave contention tracking (with associated code sprawl.)
*
* Memory ordering. See "Correct and Efficient Work-Stealing for
* Weak Memory Models" by Le, Pop, Cohen, and Nardelli, PPoPP 2013
* (http://www.di.ens.fr/~zappa/readings/ppopp13.pdf) for an
* analysis of memory ordering requirements in work-stealing
* algorithms similar to (but different than) the one used here.
* Extracting tasks in array slots via (fully fenced) CAS provides
* primary synchronization. The base and top indices imprecisely
* guide where to extract from. We do not always require strict
* orderings of array and index updates, so sometimes let them be
* subject to compiler and processor reorderings. However, the
* volatile "base" index also serves as a basis for memory
* ordering: Slot accesses are preceded by a read of base,
* ensuring happens-before ordering with respect to stealers (so
* the slots themselves can be read via plain array reads.) The
* only other memory orderings relied on are maintained in the
* course of signalling and activation (see below). A check that
* base == top indicates (momentary) emptiness, but otherwise may
* err on the side of possibly making the queue appear nonempty
* when a push, pop, or poll have not fully committed, or making
* it appear empty when an update of top has not yet been visibly
* written. (Method isEmpty() checks the case of a partially
* completed removal of the last element.) Because of this, the
* poll operation, considered individually, is not wait-free. One
* thief cannot successfully continue until another in-progress
* one (or, if previously empty, a push) visibly completes.
* However, in the aggregate, we ensure at least probabilistic
* non-blockingness. If an attempted steal fails, a scanning
* thief chooses a different random victim target to try next. So,
* in order for one thief to progress, it suffices for any
* in-progress poll or new push on any empty queue to
* complete.
*
* This approach also enables support of a user mode in which
* local task processing is in FIFO, not LIFO order, simply by
* using poll rather than pop. This can be useful in
* message-passing frameworks in which tasks are never joined.
*
* WorkQueues are also used in a similar way for tasks submitted
* to the pool. We cannot mix these tasks in the same queues used
* by workers. Instead, we randomly associate submission queues
* with submitting threads, using a form of hashing. The
* ThreadLocalRandom probe value serves as a hash code for
* choosing existing queues, and may be randomly repositioned upon
* contention with other submitters. In essence, submitters act
* like workers except that they are restricted to executing local
* tasks that they submitted. Insertion of tasks in shared mode
* requires a lock but we use only a simple spinlock (using field
* phase), because submitters encountering a busy queue move to a
* different position to use or create other queues -- they block
* only when creating and registering new queues. Because it is
* used only as a spinlock, unlocking requires only a "releasing"
* store (using setRelease).
*
* Management
* ==========
*
* The main throughput advantages of work-stealing stem from
* decentralized control -- workers mostly take tasks from
* themselves or each other, at rates that can exceed a billion
* per second. The pool itself creates, activates (enables
* scanning for and running tasks), deactivates, blocks, and
* terminates threads, all with minimal central information.
* There are only a few properties that we can globally track or
* maintain, so we pack them into a small number of variables,
* often maintaining atomicity without blocking or locking.
* Nearly all essentially atomic control state is held in a few
* volatile variables that are by far most often read (not
* written) as status and consistency checks. We pack as much
* information into them as we can.
*
* Field "ctl" contains 64 bits holding information needed to
* atomically decide to add, enqueue (on an event queue), and
* dequeue (and release)-activate workers. To enable this
* packing, we restrict maximum parallelism to (1<<15)-1 (which is
* far in excess of normal operating range) to allow ids, counts,
* and their negations (used for thresholding) to fit into 16bit
* subfields.
*
* Field "mode" holds configuration parameters as well as lifetime
* status, atomically and monotonically setting SHUTDOWN, STOP,
* and finally TERMINATED bits.
*
* Field "workQueues" holds references to WorkQueues. It is
* updated (only during worker creation and termination) under
* lock (using field workerNamePrefix as lock), but is otherwise
* concurrently readable, and accessed directly. We also ensure
* that uses of the array reference itself never become too stale
* in case of resizing. To simplify index-based operations, the
* array size is always a power of two, and all readers must
* tolerate null slots. Worker queues are at odd indices. Shared
* (submission) queues are at even indices, up to a maximum of 64
* slots, to limit growth even if array needs to expand to add
* more workers. Grouping them together in this way simplifies and
* speeds up task scanning.
*
* All worker thread creation is on-demand, triggered by task
* submissions, replacement of terminated workers, and/or
* compensation for blocked workers. However, all other support
* code is set up to work with other policies. To ensure that we
* do not hold on to worker references that would prevent GC, all
* accesses to workQueues are via indices into the workQueues
* array (which is one source of some of the messy code
* constructions here). In essence, the workQueues array serves as
* a weak reference mechanism. Thus for example the stack top
* subfield of ctl stores indices, not references.
*
* Queuing Idle Workers. Unlike HPC work-stealing frameworks, we
* cannot let workers spin indefinitely scanning for tasks when
* none can be found immediately, and we cannot start/resume
* workers unless there appear to be tasks available. On the
* other hand, we must quickly prod them into action when new
* tasks are submitted or generated. In many usages, ramp-up time
* is the main limiting factor in overall performance, which is
* compounded at program start-up by JIT compilation and
* allocation. So we streamline this as much as possible.
*
* The "ctl" field atomically maintains total worker and
* "released" worker counts, plus the head of the available worker
* queue (actually stack, represented by the lower 32bit subfield
* of ctl). Released workers are those known to be scanning for
* and/or running tasks. Unreleased ("available") workers are
* recorded in the ctl stack. These workers are made available for
* signalling by enqueuing in ctl (see method runWorker). The
* "queue" is a form of Treiber stack. This is ideal for
* activating threads in most-recently used order, and improves
* performance and locality, outweighing the disadvantages of
* being prone to contention and inability to release a worker
* unless it is topmost on stack. To avoid missed signal problems
* inherent in any wait/signal design, available workers rescan
* for (and if found run) tasks after enqueuing. Normally their
* release status will be updated while doing so, but the released
* worker ctl count may underestimate the number of active
* threads. (However, it is still possible to determine quiescence
* via a validation traversal -- see isQuiescent). After an
* unsuccessful rescan, available workers are blocked until
* signalled (see signalWork). The top stack state holds the
* value of the "phase" field of the worker: its index and status,
* plus a version counter that, in addition to the count subfields
* (also serving as version stamps) provide protection against
* Treiber stack ABA effects.
*
* Creating workers. To create a worker, we pre-increment counts
* (serving as a reservation), and attempt to construct a
* ForkJoinWorkerThread via its factory. Upon construction, the
* new thread invokes registerWorker, where it constructs a
* WorkQueue and is assigned an index in the workQueues array
* (expanding the array if necessary). The thread is then started.
* Upon any exception across these steps, or null return from
* factory, deregisterWorker adjusts counts and records
* accordingly. If a null return, the pool continues running with
* fewer than the target number workers. If exceptional, the
* exception is propagated, generally to some external caller.
* Worker index assignment avoids the bias in scanning that would
* occur if entries were sequentially packed starting at the front
* of the workQueues array. We treat the array as a simple
* power-of-two hash table, expanding as needed. The seedIndex
* increment ensures no collisions until a resize is needed or a
* worker is deregistered and replaced, and thereafter keeps
* probability of collision low. We cannot use
* ThreadLocalRandom.getProbe() for similar purposes here because
* the thread has not started yet, but do so for creating
* submission queues for existing external threads (see
* externalPush).
*
* WorkQueue field "phase" is used by both workers and the pool to
* manage and track whether a worker is UNSIGNALLED (possibly
* blocked waiting for a signal). When a worker is enqueued its
* phase field is set. Note that phase field updates lag queue CAS
* releases so usage requires care -- seeing a negative phase does
* not guarantee that the worker is available. When queued, the
* lower 16 bits of scanState must hold its pool index. So we
* place the index there upon initialization (see registerWorker)
* and otherwise keep it there or restore it when necessary.
*
* The ctl field also serves as the basis for memory
* synchronization surrounding activation. This uses a more
* efficient version of a Dekker-like rule that task producers and
* consumers sync with each other by both writing/CASing ctl (even
* if to its current value). This would be extremely costly. So
* we relax it in several ways: (1) Producers only signal when
* their queue is empty. Other workers propagate this signal (in
* method scan) when they find tasks; to further reduce flailing,
* each worker signals only one other per activation. (2) Workers
* only enqueue after scanning (see below) and not finding any
* tasks. (3) Rather than CASing ctl to its current value in the
* common case where no action is required, we reduce write
* contention by equivalently prefacing signalWork when called by
* an external task producer using a memory access with
* full-volatile semantics or a "fullFence".
*
* Almost always, too many signals are issued. A task producer
* cannot in general tell if some existing worker is in the midst
* of finishing one task (or already scanning) and ready to take
* another without being signalled. So the producer might instead
* activate a different worker that does not find any work, and
* then inactivates. This scarcely matters in steady-state
* computations involving all workers, but can create contention
* and bookkeeping bottlenecks during ramp-up, ramp-down, and small
* computations involving only a few workers.
*
* Scanning. Method runWorker performs top-level scanning for
* tasks. Each scan traverses and tries to poll from each queue
* starting at a random index and circularly stepping. Scans are
* not performed in ideal random permutation order, to reduce
* cacheline contention. The pseudorandom generator need not have
* high-quality statistical properties in the long term, but just
* within computations; We use Marsaglia XorShifts (often via
* ThreadLocalRandom.nextSecondarySeed), which are cheap and
* suffice. Scanning also employs contention reduction: When
* scanning workers fail to extract an apparently existing task,
* they soon restart at a different pseudorandom index. This
* improves throughput when many threads are trying to take tasks
* from few queues, which can be common in some usages. Scans do
* not otherwise explicitly take into account core affinities,
* loads, cache localities, etc, However, they do exploit temporal
* locality (which usually approximates these) by preferring to
* re-poll (at most #workers times) from the same queue after a
* successful poll before trying others.
*
* Trimming workers. To release resources after periods of lack of
* use, a worker starting to wait when the pool is quiescent will
* time out and terminate (see method scan) if the pool has
* remained quiescent for period given by field keepAlive.
*
* Shutdown and Termination. A call to shutdownNow invokes
* tryTerminate to atomically set a runState bit. The calling
* thread, as well as every other worker thereafter terminating,
* helps terminate others by cancelling their unprocessed tasks,
* and waking them up, doing so repeatedly until stable. Calls to
* non-abrupt shutdown() preface this by checking whether
* termination should commence by sweeping through queues (until
* stable) to ensure lack of in-flight submissions and workers
* about to process them before triggering the "STOP" phase of
* termination.
*
* Joining Tasks
* =============
*
* Any of several actions may be taken when one worker is waiting
* to join a task stolen (or always held) by another. Because we
* are multiplexing many tasks on to a pool of workers, we can't
* always just let them block (as in Thread.join). We also cannot
* just reassign the joiner's run-time stack with another and
* replace it later, which would be a form of "continuation", that
* even if possible is not necessarily a good idea since we may
* need both an unblocked task and its continuation to progress.
* Instead we combine two tactics:
*
* Helping: Arranging for the joiner to execute some task that it
* would be running if the steal had not occurred.
*
* Compensating: Unless there are already enough live threads,
* method tryCompensate() may create or re-activate a spare
* thread to compensate for blocked joiners until they unblock.
*
* A third form (implemented in tryRemoveAndExec) amounts to
* helping a hypothetical compensator: If we can readily tell that
* a possible action of a compensator is to steal and execute the
* task being joined, the joining thread can do so directly,
* without the need for a compensation thread.
*
* The ManagedBlocker extension API can't use helping so relies
* only on compensation in method awaitBlocker.
*
* The algorithm in awaitJoin entails a form of "linear helping".
* Each worker records (in field source) the id of the queue from
* which it last stole a task. The scan in method awaitJoin uses
* these markers to try to find a worker to help (i.e., steal back
* a task from and execute it) that could hasten completion of the
* actively joined task. Thus, the joiner executes a task that
* would be on its own local deque if the to-be-joined task had
* not been stolen. This is a conservative variant of the approach
* described in Wagner & Calder "Leapfrogging: a portable
* technique for implementing efficient futures" SIGPLAN Notices,
* 1993 (http://portal.acm.org/citation.cfm?id=155354). It differs
* mainly in that we only record queue ids, not full dependency
* links. This requires a linear scan of the workQueues array to
* locate stealers, but isolates cost to when it is needed, rather
* than adding to per-task overhead. Searches can fail to locate
* stealers GC stalls and the like delay recording sources.
* Further, even when accurately identified, stealers might not
* ever produce a task that the joiner can in turn help with. So,
* compensation is tried upon failure to find tasks to run.
*
* Compensation does not by default aim to keep exactly the target
* parallelism number of unblocked threads running at any given
* time. Some previous versions of this class employed immediate
* compensations for any blocked join. However, in practice, the
* vast majority of blockages are transient byproducts of GC and
* other JVM or OS activities that are made worse by replacement.
* Rather than impose arbitrary policies, we allow users to
* override the default of only adding threads upon apparent
* starvation. The compensation mechanism may also be bounded.
* Bounds for the commonPool (see COMMON_MAX_SPARES) better enable
* JVMs to cope with programming errors and abuse before running
* out of resources to do so.
*
* Common Pool
* ===========
*
* The static common pool always exists after static
* initialization. Since it (or any other created pool) need
* never be used, we minimize initial construction overhead and
* footprint to the setup of about a dozen fields.
*
* When external threads submit to the common pool, they can
* perform subtask processing (see externalHelpComplete and
* related methods) upon joins. This caller-helps policy makes it
* sensible to set common pool parallelism level to one (or more)
* less than the total number of available cores, or even zero for
* pure caller-runs. We do not need to record whether external
* submissions are to the common pool -- if not, external help
* methods return quickly. These submitters would otherwise be
* blocked waiting for completion, so the extra effort (with
* liberally sprinkled task status checks) in inapplicable cases
* amounts to an odd form of limited spin-wait before blocking in
* ForkJoinTask.join.
*
* As a more appropriate default in managed environments, unless
* overridden by system properties, we use workers of subclass
* InnocuousForkJoinWorkerThread when there is a SecurityManager
* present. These workers have no permissions set, do not belong
* to any user-defined ThreadGroup, and erase all ThreadLocals
* after executing any top-level task (see
* WorkQueue.afterTopLevelExec). The associated mechanics (mainly
* in ForkJoinWorkerThread) may be JVM-dependent and must access
* particular Thread class fields to achieve this effect.
*
* Style notes
* ===========
*
* Memory ordering relies mainly on VarHandles. This can be
* awkward and ugly, but also reflects the need to control
* outcomes across the unusual cases that arise in very racy code
* with very few invariants. All fields are read into locals
* before use, and null-checked if they are references. This is
* usually done in a "C"-like style of listing declarations at the
* heads of methods or blocks, and using inline assignments on
* first encounter. Nearly all explicit checks lead to
* bypass/return, not exception throws, because they may
* legitimately arise due to cancellation/revocation during
* shutdown.
*
* There is a lot of representation-level coupling among classes
* ForkJoinPool, ForkJoinWorkerThread, and ForkJoinTask. The
* fields of WorkQueue maintain data structures managed by
* ForkJoinPool, so are directly accessed. There is little point
* trying to reduce this, since any associated future changes in
* representations will need to be accompanied by algorithmic
* changes anyway. Several methods intrinsically sprawl because
* they must accumulate sets of consistent reads of fields held in
* local variables. There are also other coding oddities
* (including several unnecessary-looking hoisted null checks)
* that help some methods perform reasonably even when interpreted
* (not compiled).
*
* The order of declarations in this file is (with a few exceptions):
* (1) Static utility functions
* (2) Nested (static) classes
* (3) Static fields
* (4) Fields, along with constants used when unpacking some of them
* (5) Internal control methods
* (6) Callbacks and other support for ForkJoinTask methods
* (7) Exported methods
* (8) Static block initializing statics in minimally dependent order
*/
// Static utilities
/**
* If there is a security manager, makes sure caller has
* permission to modify threads.
*/
private static void checkPermission() {
SecurityManager security = System.getSecurityManager();
if (security != null)
security.checkPermission(modifyThreadPermission);
}
// Nested classes
/**
* Factory for creating new {@link ForkJoinWorkerThread}s.
* A {@code ForkJoinWorkerThreadFactory} must be defined and used
* for {@code ForkJoinWorkerThread} subclasses that extend base
* functionality or initialize threads with different contexts.
*/
public static interface ForkJoinWorkerThreadFactory {
/**
* Returns a new worker thread operating in the given pool.
* Returning null or throwing an exception may result in tasks
* never being executed. If this method throws an exception,
* it is relayed to the caller of the method (for example
* {@code execute}) causing attempted thread creation. If this
* method returns null or throws an exception, it is not
* retried until the next attempted creation (for example
* another call to {@code execute}).
*
* @param pool the pool this thread works in
* @return the new worker thread, or {@code null} if the request
* to create a thread is rejected
* @throws NullPointerException if the pool is null
*/
public ForkJoinWorkerThread newThread(ForkJoinPool pool);
}
static AccessControlContext contextWithPermissions(Permission ... perms) {
Permissions permissions = new Permissions();
for (Permission perm : perms)
permissions.add(perm);
return new AccessControlContext(
new ProtectionDomain[] { new ProtectionDomain(null, permissions) });
}
/**
* Default ForkJoinWorkerThreadFactory implementation; creates a
* new ForkJoinWorkerThread using the system class loader as the
* thread context class loader.
*/
private static final class DefaultForkJoinWorkerThreadFactory
implements ForkJoinWorkerThreadFactory {
private static final AccessControlContext ACC = contextWithPermissions(
new RuntimePermission("getClassLoader"),
new RuntimePermission("setContextClassLoader"));
public final ForkJoinWorkerThread newThread(ForkJoinPool pool) {
return AccessController.doPrivileged(
new PrivilegedAction<>() {
public ForkJoinWorkerThread run() {
return new ForkJoinWorkerThread(
pool, ClassLoader.getSystemClassLoader()); }},
ACC);
}
}
// Constants shared across ForkJoinPool and WorkQueue
// Bounds
static final int SWIDTH = 16; // width of short
static final int SMASK = 0xffff; // short bits == max index
static final int MAX_CAP = 0x7fff; // max #workers - 1
static final int SQMASK = 0x007e; // max 64 (even) slots
// Masks and units for WorkQueue.phase and ctl sp subfield
static final int UNSIGNALLED = 1 << 31; // must be negative
static final int SS_SEQ = 1 << 16; // version count
static final int QLOCK = 1; // must be 1
// Mode bits and sentinels, some also used in WorkQueue id and.source fields
static final int OWNED = 1; // queue has owner thread
static final int FIFO = 1 << 16; // fifo queue or access mode
static final int SHUTDOWN = 1 << 18;
static final int TERMINATED = 1 << 19;
static final int STOP = 1 << 31; // must be negative
static final int QUIET = 1 << 30; // not scanning or working
static final int DORMANT = QUIET | UNSIGNALLED;
/**
* The maximum number of local polls from the same queue before
* checking others. This is a safeguard against infinitely unfair
* looping under unbounded user task recursion, and must be larger
* than plausible cases of intentional bounded task recursion.
*/
static final int POLL_LIMIT = 1 << 10;
/**
* Queues supporting work-stealing as well as external task
* submission. See above for descriptions and algorithms.
* Performance on most platforms is very sensitive to placement of
* instances of both WorkQueues and their arrays -- we absolutely
* do not want multiple WorkQueue instances or multiple queue
* arrays sharing cache lines. The @Contended annotation alerts
* JVMs to try to keep instances apart.
*/
@jdk.internal.vm.annotation.Contended
static final class WorkQueue {
/**
* Capacity of work-stealing queue array upon initialization.
* Must be a power of two; at least 4, but should be larger to
* reduce or eliminate cacheline sharing among queues.
* Currently, it is much larger, as a partial workaround for
* the fact that JVMs often place arrays in locations that
* share GC bookkeeping (especially cardmarks) such that
* per-write accesses encounter serious memory contention.
*/
static final int INITIAL_QUEUE_CAPACITY = 1 << 13;
/**
* Maximum size for queue arrays. Must be a power of two less
* than or equal to 1 << (31 - width of array entry) to ensure
* lack of wraparound of index calculations, but defined to a
* value a bit less than this to help users trap runaway
* programs before saturating systems.
*/
static final int MAXIMUM_QUEUE_CAPACITY = 1 << 26; // 64M
// Instance fields
volatile int phase; // versioned, negative: queued, 1: locked
int stackPred; // pool stack (ctl) predecessor link
int nsteals; // number of steals
int id; // index, mode, tag
volatile int source; // source queue id, or sentinel
volatile int base; // index of next slot for poll
int top; // index of next slot for push
ForkJoinTask[] array; // the elements (initially unallocated)
final ForkJoinPool pool; // the containing pool (may be null)
final ForkJoinWorkerThread owner; // owning thread or null if shared
WorkQueue(ForkJoinPool pool, ForkJoinWorkerThread owner) {
this.pool = pool;
this.owner = owner;
// Place indices in the center of array (that is not yet allocated)
base = top = INITIAL_QUEUE_CAPACITY >>> 1;
}
/**
* Returns an exportable index (used by ForkJoinWorkerThread).
*/
final int getPoolIndex() {
return (id & 0xffff) >>> 1; // ignore odd/even tag bit
}
/**
* Returns the approximate number of tasks in the queue.
*/
final int queueSize() {
int n = base - top; // read base first
return (n >= 0) ? 0 : -n; // ignore transient negative
}
/**
* Provides a more accurate estimate of whether this queue has
* any tasks than does queueSize, by checking whether a
* near-empty queue has at least one unclaimed task.
*/
final boolean isEmpty() {
ForkJoinTask[] a; int n, al, b;
return ((n = (b = base) - top) >= 0 || // possibly one task
(n == -1 && ((a = array) == null ||
(al = a.length) == 0 ||
a[(al - 1) & b] == null)));
}
/**
* Pushes a task. Call only by owner in unshared queues.
*
* @param task the task. Caller must ensure non-null.
* @throws RejectedExecutionException if array cannot be resized
*/
final void push(ForkJoinTask task) {
int s = top; ForkJoinTask[] a; int al, d;
if ((a = array) != null && (al = a.length) > 0) {
int index = (al - 1) & s;
ForkJoinPool p = pool;
top = s + 1;
QA.setRelease(a, index, task);
if ((d = base - s) == 0 && p != null) {
VarHandle.fullFence();
p.signalWork();
}
else if (d + al == 1)
growArray();
}
}
/**
* Initializes or doubles the capacity of array. Call either
* by owner or with lock held -- it is OK for base, but not
* top, to move while resizings are in progress.
*/
final ForkJoinTask[] growArray() {
ForkJoinTask[] oldA = array;
int oldSize = oldA != null ? oldA.length : 0;
int size = oldSize > 0 ? oldSize << 1 : INITIAL_QUEUE_CAPACITY;
if (size < INITIAL_QUEUE_CAPACITY || size > MAXIMUM_QUEUE_CAPACITY)
throw new RejectedExecutionException("Queue capacity exceeded");
int oldMask, t, b;
ForkJoinTask[] a = array = new ForkJoinTask[size];
if (oldA != null && (oldMask = oldSize - 1) > 0 &&
(t = top) - (b = base) > 0) {
int mask = size - 1;
do { // emulate poll from old array, push to new array
int index = b & oldMask;
ForkJoinTask x = (ForkJoinTask)
QA.getAcquire(oldA, index);
if (x != null &&
QA.compareAndSet(oldA, index, x, null))
a[b & mask] = x;
} while (++b != t);
VarHandle.releaseFence();
}
return a;
}
/**
* Takes next task, if one exists, in LIFO order. Call only
* by owner in unshared queues.
*/
final ForkJoinTask pop() {
int b = base, s = top, al, i; ForkJoinTask[] a;
if ((a = array) != null && b != s && (al = a.length) > 0) {
int index = (al - 1) & --s;
ForkJoinTask t = (ForkJoinTask)
QA.get(a, index);
if (t != null &&
QA.compareAndSet(a, index, t, null)) {
top = s;
VarHandle.releaseFence();
return t;
}
}
return null;
}
/**
* Takes next task, if one exists, in FIFO order.
*/
final ForkJoinTask poll() {
for (;;) {
int b = base, s = top, d, al; ForkJoinTask[] a;
if ((a = array) != null && (d = b - s) < 0 &&
(al = a.length) > 0) {
int index = (al - 1) & b;
ForkJoinTask t = (ForkJoinTask)
QA.getAcquire(a, index);
if (b++ == base) {
if (t != null) {
if (QA.compareAndSet(a, index, t, null)) {
base = b;
return t;
}
}
else if (d == -1)
break; // now empty
}
}
else
break;
}
return null;
}
/**
* Takes next task, if one exists, in order specified by mode.
*/
final ForkJoinTask nextLocalTask() {
return ((id & FIFO) != 0) ? poll() : pop();
}
/**
* Returns next task, if one exists, in order specified by mode.
*/
final ForkJoinTask peek() {
int al; ForkJoinTask[] a;
return ((a = array) != null && (al = a.length) > 0) ?
a[(al - 1) &
((id & FIFO) != 0 ? base : top - 1)] : null;
}
/**
* Pops the given task only if it is at the current top.
*/
final boolean tryUnpush(ForkJoinTask task) {
int b = base, s = top, al; ForkJoinTask[] a;
if ((a = array) != null && b != s && (al = a.length) > 0) {
int index = (al - 1) & --s;
if (QA.compareAndSet(a, index, task, null)) {
top = s;
VarHandle.releaseFence();
return true;
}
}
return false;
}
/**
* Removes and cancels all known tasks, ignoring any exceptions.
*/
final void cancelAll() {
for (ForkJoinTask t; (t = poll()) != null; )
ForkJoinTask.cancelIgnoringExceptions(t);
}
// Specialized execution methods
/**
* Pops and executes up to limit consecutive tasks or until empty.
*
* @param limit max runs, or zero for no limit
*/
final void localPopAndExec(int limit) {
for (;;) {
int b = base, s = top, al; ForkJoinTask[] a;
if ((a = array) != null && b != s && (al = a.length) > 0) {
int index = (al - 1) & --s;
ForkJoinTask t = (ForkJoinTask)
QA.getAndSet(a, index, null);
if (t != null) {
top = s;
VarHandle.releaseFence();
t.doExec();
if (limit != 0 && --limit == 0)
break;
}
else
break;
}
else
break;
}
}
/**
* Polls and executes up to limit consecutive tasks or until empty.
*
* @param limit, or zero for no limit
*/
final void localPollAndExec(int limit) {
for (int polls = 0;;) {
int b = base, s = top, d, al; ForkJoinTask[] a;
if ((a = array) != null && (d = b - s) < 0 &&
(al = a.length) > 0) {
int index = (al - 1) & b++;
ForkJoinTask t = (ForkJoinTask)
QA.getAndSet(a, index, null);
if (t != null) {
base = b;
t.doExec();
if (limit != 0 && ++polls == limit)
break;
}
else if (d == -1)
break; // now empty
else
polls = 0; // stolen; reset
}
else
break;
}
}
/**
* If present, removes task from queue and executes it.
*/
final void tryRemoveAndExec(ForkJoinTask task) {
ForkJoinTask[] wa; int s, wal;
if (base - (s = top) < 0 && // traverse from top
(wa = array) != null && (wal = wa.length) > 0) {
for (int m = wal - 1, ns = s - 1, i = ns; ; --i) {
int index = i & m;
ForkJoinTask t = (ForkJoinTask)
QA.get(wa, index);
if (t == null)
break;
else if (t == task) {
if (QA.compareAndSet(wa, index, t, null)) {
top = ns; // safely shift down
for (int j = i; j != ns; ++j) {
ForkJoinTask f;
int pindex = (j + 1) & m;
f = (ForkJoinTask)QA.get(wa, pindex);
QA.setVolatile(wa, pindex, null);
int jindex = j & m;
QA.setRelease(wa, jindex, f);
}
VarHandle.releaseFence();
t.doExec();
}
break;
}
}
}
}
/**
* Tries to steal and run tasks within the target's
* computation until done, not found, or limit exceeded.
*
* @param task root of CountedCompleter computation
* @param limit max runs, or zero for no limit
* @return task status on exit
*/
final int localHelpCC(CountedCompleter task, int limit) {
int status = 0;
if (task != null && (status = task.status) >= 0) {
for (;;) {
boolean help = false;
int b = base, s = top, al; ForkJoinTask[] a;
if ((a = array) != null && b != s && (al = a.length) > 0) {
int index = (al - 1) & (s - 1);
ForkJoinTask o = (ForkJoinTask)
QA.get(a, index);
if (o instanceof CountedCompleter) {
CountedCompleter t = (CountedCompleter)o;
for (CountedCompleter f = t;;) {
if (f != task) {
if ((f = f.completer) == null) // try parent
break;
}
else {
if (QA.compareAndSet(a, index, t, null)) {
top = s - 1;
VarHandle.releaseFence();
t.doExec();
help = true;
}
break;
}
}
}
}
if ((status = task.status) < 0 || !help ||
(limit != 0 && --limit == 0))
break;
}
}
return status;
}
// Operations on shared queues
/**
* Tries to lock shared queue by CASing phase field.
*/
final boolean tryLockSharedQueue() {
return PHASE.compareAndSet(this, 0, QLOCK);
}
/**
* Shared version of tryUnpush.
*/
final boolean trySharedUnpush(ForkJoinTask task) {
boolean popped = false;
int s = top - 1, al; ForkJoinTask[] a;
if ((a = array) != null && (al = a.length) > 0) {
int index = (al - 1) & s;
ForkJoinTask t = (ForkJoinTask) QA.get(a, index);
if (t == task &&
PHASE.compareAndSet(this, 0, QLOCK)) {
if (top == s + 1 && array == a &&
QA.compareAndSet(a, index, task, null)) {
popped = true;
top = s;
}
PHASE.setRelease(this, 0);
}
}
return popped;
}
/**
* Shared version of localHelpCC.
*/
final int sharedHelpCC(CountedCompleter task, int limit) {
int status = 0;
if (task != null && (status = task.status) >= 0) {
for (;;) {
boolean help = false;
int b = base, s = top, al; ForkJoinTask[] a;
if ((a = array) != null && b != s && (al = a.length) > 0) {
int index = (al - 1) & (s - 1);
ForkJoinTask o = (ForkJoinTask)
QA.get(a, index);
if (o instanceof CountedCompleter) {
CountedCompleter t = (CountedCompleter)o;
for (CountedCompleter f = t;;) {
if (f != task) {
if ((f = f.completer) == null)
break;
}
else {
if (PHASE.compareAndSet(this, 0, QLOCK)) {
if (top == s && array == a &&
QA.compareAndSet(a, index, t, null)) {
help = true;
top = s - 1;
}
PHASE.setRelease(this, 0);
if (help)
t.doExec();
}
break;
}
}
}
}
if ((status = task.status) < 0 || !help ||
(limit != 0 && --limit == 0))
break;
}
}
return status;
}
/**
* Returns true if owned and not known to be blocked.
*/
final boolean isApparentlyUnblocked() {
Thread wt; Thread.State s;
return ((wt = owner) != null &&
(s = wt.getState()) != Thread.State.BLOCKED &&
s != Thread.State.WAITING &&
s != Thread.State.TIMED_WAITING);
}
// VarHandle mechanics.
private static final VarHandle PHASE;
static {
try {
MethodHandles.Lookup l = MethodHandles.lookup();
PHASE = l.findVarHandle(WorkQueue.class, "phase", int.class);
} catch (ReflectiveOperationException e) {
throw new Error(e);
}
}
}
// static fields (initialized in static initializer below)
/**
* Creates a new ForkJoinWorkerThread. This factory is used unless
* overridden in ForkJoinPool constructors.
*/
public static final ForkJoinWorkerThreadFactory
defaultForkJoinWorkerThreadFactory;
/**
* Permission required for callers of methods that may start or
* kill threads.
*/
static final RuntimePermission modifyThreadPermission;
/**
* Common (static) pool. Non-null for public use unless a static
* construction exception, but internal usages null-check on use
* to paranoically avoid potential initialization circularities
* as well as to simplify generated code.
*/
static final ForkJoinPool common;
/**
* Common pool parallelism. To allow simpler use and management
* when common pool threads are disabled, we allow the underlying
* common.parallelism field to be zero, but in that case still report
* parallelism as 1 to reflect resulting caller-runs mechanics.
*/
static final int COMMON_PARALLELISM;
/**
* Limit on spare thread construction in tryCompensate.
*/
private static final int COMMON_MAX_SPARES;
/**
* Sequence number for creating workerNamePrefix.
*/
private static int poolNumberSequence;
/**
* Returns the next sequence number. We don't expect this to
* ever contend, so use simple builtin sync.
*/
private static final synchronized int nextPoolId() {
return ++poolNumberSequence;
}
// static configuration constants
/**
* Default idle timeout value (in milliseconds) for the thread
* triggering quiescence to park waiting for new work
*/
private static final long DEFAULT_KEEPALIVE = 60_000L;
/**
* Undershoot tolerance for idle timeouts
*/
private static final long TIMEOUT_SLOP = 20L;
/**
* The default value for COMMON_MAX_SPARES. Overridable using the
* "java.util.concurrent.ForkJoinPool.common.maximumSpares" system
* property. The default value is far in excess of normal
* requirements, but also far short of MAX_CAP and typical OS
* thread limits, so allows JVMs to catch misuse/abuse before
* running out of resources needed to do so.
*/
private static final int DEFAULT_COMMON_MAX_SPARES = 256;
/**
* Increment for seed generators. See class ThreadLocal for
* explanation.
*/
private static final int SEED_INCREMENT = 0x9e3779b9;
/*
* Bits and masks for field ctl, packed with 4 16 bit subfields:
* RC: Number of released (unqueued) workers minus target parallelism
* TC: Number of total workers minus target parallelism
* SS: version count and status of top waiting thread
* ID: poolIndex of top of Treiber stack of waiters
*
* When convenient, we can extract the lower 32 stack top bits
* (including version bits) as sp=(int)ctl. The offsets of counts
* by the target parallelism and the positionings of fields makes
* it possible to perform the most common checks via sign tests of
* fields: When ac is negative, there are not enough unqueued
* workers, when tc is negative, there are not enough total
* workers. When sp is non-zero, there are waiting workers. To
* deal with possibly negative fields, we use casts in and out of
* "short" and/or signed shifts to maintain signedness.
*
* Because it occupies uppermost bits, we can add one release count
* using getAndAddLong of RC_UNIT, rather than CAS, when returning
* from a blocked join. Other updates entail multiple subfields
* and masking, requiring CAS.
*
* The limits packed in field "bounds" are also offset by the
* parallelism level to make them comparable to the ctl rc and tc
* fields.
*/
// Lower and upper word masks
private static final long SP_MASK = 0xffffffffL;
private static final long UC_MASK = ~SP_MASK;
// Release counts
private static final int RC_SHIFT = 48;
private static final long RC_UNIT = 0x0001L << RC_SHIFT;
private static final long RC_MASK = 0xffffL << RC_SHIFT;
// Total counts
private static final int TC_SHIFT = 32;
private static final long TC_UNIT = 0x0001L << TC_SHIFT;
private static final long TC_MASK = 0xffffL << TC_SHIFT;
private static final long ADD_WORKER = 0x0001L << (TC_SHIFT + 15); // sign
// Instance fields
volatile long stealCount; // collects worker nsteals
final long keepAlive; // milliseconds before dropping if idle
int indexSeed; // next worker index
final int bounds; // min, max threads packed as shorts
volatile int mode; // parallelism, runstate, queue mode
WorkQueue[] workQueues; // main registry
final String workerNamePrefix; // for worker thread string; sync lock
final ForkJoinWorkerThreadFactory factory;
final UncaughtExceptionHandler ueh; // per-worker UEH
final Predicate saturate;
@jdk.internal.vm.annotation.Contended("fjpctl") // segregate
volatile long ctl; // main pool control
// Creating, registering and deregistering workers
/**
* Tries to construct and start one worker. Assumes that total
* count has already been incremented as a reservation. Invokes
* deregisterWorker on any failure.
*
* @return true if successful
*/
private boolean createWorker() {
ForkJoinWorkerThreadFactory fac = factory;
Throwable ex = null;
ForkJoinWorkerThread wt = null;
try {
if (fac != null && (wt = fac.newThread(this)) != null) {
wt.start();
return true;
}
} catch (Throwable rex) {
ex = rex;
}
deregisterWorker(wt, ex);
return false;
}
/**
* Tries to add one worker, incrementing ctl counts before doing
* so, relying on createWorker to back out on failure.
*
* @param c incoming ctl value, with total count negative and no
* idle workers. On CAS failure, c is refreshed and retried if
* this holds (otherwise, a new worker is not needed).
*/
private void tryAddWorker(long c) {
do {
long nc = ((RC_MASK & (c + RC_UNIT)) |
(TC_MASK & (c + TC_UNIT)));
if (ctl == c && CTL.compareAndSet(this, c, nc)) {
createWorker();
break;
}
} while (((c = ctl) & ADD_WORKER) != 0L && (int)c == 0);
}
/**
* Callback from ForkJoinWorkerThread constructor to establish and
* record its WorkQueue.
*
* @param wt the worker thread
* @return the worker's queue
*/
final WorkQueue registerWorker(ForkJoinWorkerThread wt) {
UncaughtExceptionHandler handler;
wt.setDaemon(true); // configure thread
if ((handler = ueh) != null)
wt.setUncaughtExceptionHandler(handler);
WorkQueue w = new WorkQueue(this, wt);
int tid = 0; // for thread name
int fifo = mode & FIFO;
String prefix = workerNamePrefix;
if (prefix != null) {
synchronized (prefix) {
WorkQueue[] ws = workQueues; int n;
int s = indexSeed += SEED_INCREMENT;
if (ws != null && (n = ws.length) > 1) {
int m = n - 1;
tid = s & m;
int i = m & ((s << 1) | 1); // odd-numbered indices
for (int probes = n >>> 1;;) { // find empty slot
WorkQueue q;
if ((q = ws[i]) == null || q.phase == QUIET)
break;
else if (--probes == 0) {
i = n | 1; // resize below
break;
}
else
i = (i + 2) & m;
}
int id = i | fifo | (s & ~(SMASK | FIFO | DORMANT));
w.phase = w.id = id; // now publishable
if (i < n)
ws[i] = w;
else { // expand array
int an = n << 1;
WorkQueue[] as = new WorkQueue[an];
as[i] = w;
int am = an - 1;
for (int j = 0; j < n; ++j) {
WorkQueue v; // copy external queue
if ((v = ws[j]) != null) // position may change
as[v.id & am & SQMASK] = v;
if (++j >= n)
break;
as[j] = ws[j]; // copy worker
}
workQueues = as;
}
}
}
wt.setName(prefix.concat(Integer.toString(tid)));
}
return w;
}
/**
* Final callback from terminating worker, as well as upon failure
* to construct or start a worker. Removes record of worker from
* array, and adjusts counts. If pool is shutting down, tries to
* complete termination.
*
* @param wt the worker thread, or null if construction failed
* @param ex the exception causing failure, or null if none
*/
final void deregisterWorker(ForkJoinWorkerThread wt, Throwable ex) {
WorkQueue w = null;
int phase = 0;
if (wt != null && (w = wt.workQueue) != null) {
Object lock = workerNamePrefix;
long ns = (long)w.nsteals & 0xffffffffL;
int idx = w.id & SMASK;
if (lock != null) {
WorkQueue[] ws; // remove index from array
synchronized (lock) {
if ((ws = workQueues) != null && ws.length > idx &&
ws[idx] == w)
ws[idx] = null;
stealCount += ns;
}
}
phase = w.phase;
}
if (phase != QUIET) { // else pre-adjusted
long c; // decrement counts
do {} while (!CTL.weakCompareAndSet
(this, c = ctl, ((RC_MASK & (c - RC_UNIT)) |
(TC_MASK & (c - TC_UNIT)) |
(SP_MASK & c))));
}
if (w != null)
w.cancelAll(); // cancel remaining tasks
if (!tryTerminate(false, false) && // possibly replace worker
w != null && w.array != null) // avoid repeated failures
signalWork();
if (ex == null) // help clean on way out
ForkJoinTask.helpExpungeStaleExceptions();
else // rethrow
ForkJoinTask.rethrow(ex);
}
/**
* Tries to create or release a worker if too few are running.
*/
final void signalWork() {
for (;;) {
long c; int sp; WorkQueue[] ws; int i; WorkQueue v;
if ((c = ctl) >= 0L) // enough workers
break;
else if ((sp = (int)c) == 0) { // no idle workers
if ((c & ADD_WORKER) != 0L) // too few workers
tryAddWorker(c);
break;
}
else if ((ws = workQueues) == null)
break; // unstarted/terminated
else if (ws.length <= (i = sp & SMASK))
break; // terminated
else if ((v = ws[i]) == null)
break; // terminating
else {
int np = sp & ~UNSIGNALLED;
int vp = v.phase;
long nc = (v.stackPred & SP_MASK) | (UC_MASK & (c + RC_UNIT));
Thread vt = v.owner;
if (sp == vp && CTL.compareAndSet(this, c, nc)) {
v.phase = np;
if (v.source < 0)
LockSupport.unpark(vt);
break;
}
}
}
}
/**
* Tries to decrement counts (sometimes implicitly) and possibly
* arrange for a compensating worker in preparation for blocking:
* If not all core workers yet exist, creates one, else if any are
* unreleased (possibly including caller) releases one, else if
* fewer than the minimum allowed number of workers running,
* checks to see that they are all active, and if so creates an
* extra worker unless over maximum limit and policy is to
* saturate. Most of these steps can fail due to interference, in
* which case 0 is returned so caller will retry. A negative
* return value indicates that the caller doesn't need to
* re-adjust counts when later unblocked.
*
* @return 1: block then adjust, -1: block without adjust, 0 : retry
*/
private int tryCompensate(WorkQueue w) {
int t, n, sp;
long c = ctl;
WorkQueue[] ws = workQueues;
if ((t = (short)(c >>> TC_SHIFT)) >= 0) {
if (ws == null || (n = ws.length) <= 0 || w == null)
return 0; // disabled
else if ((sp = (int)c) != 0) { // replace or release
WorkQueue v = ws[sp & (n - 1)];
int wp = w.phase;
long uc = UC_MASK & ((wp < 0) ? c + RC_UNIT : c);
int np = sp & ~UNSIGNALLED;
if (v != null) {
int vp = v.phase;
Thread vt = v.owner;
long nc = ((long)v.stackPred & SP_MASK) | uc;
if (vp == sp && CTL.compareAndSet(this, c, nc)) {
v.phase = np;
if (v.source < 0)
LockSupport.unpark(vt);
return (wp < 0) ? -1 : 1;
}
}
return 0;
}
else if ((int)(c >> RC_SHIFT) - // reduce parallelism
(short)(bounds & SMASK) > 0) {
long nc = ((RC_MASK & (c - RC_UNIT)) | (~RC_MASK & c));
return CTL.compareAndSet(this, c, nc) ? 1 : 0;
}
else { // validate
int md = mode, pc = md & SMASK, tc = pc + t, bc = 0;
boolean unstable = false;
for (int i = 1; i < n; i += 2) {
WorkQueue q; Thread wt; Thread.State ts;
if ((q = ws[i]) != null) {
if (q.source == 0) {
unstable = true;
break;
}
else {
--tc;
if ((wt = q.owner) != null &&
((ts = wt.getState()) == Thread.State.BLOCKED ||
ts == Thread.State.WAITING))
++bc; // worker is blocking
}
}
}
if (unstable || tc != 0 || ctl != c)
return 0; // inconsistent
else if (t + pc >= MAX_CAP || t >= (bounds >>> SWIDTH)) {
Predicate sat;
if ((sat = saturate) != null && sat.test(this))
return -1;
else if (bc < pc) { // lagging
Thread.yield(); // for retry spins
return 0;
}
else
throw new RejectedExecutionException(
"Thread limit exceeded replacing blocked worker");
}
}
}
long nc = ((c + TC_UNIT) & TC_MASK) | (c & ~TC_MASK); // expand pool
return CTL.compareAndSet(this, c, nc) && createWorker() ? 1 : 0;
}
/**
* Top-level runloop for workers, called by ForkJoinWorkerThread.run.
* See above for explanation.
*/
final void runWorker(WorkQueue w) {
WorkQueue[] ws;
w.growArray(); // allocate queue
int r = w.id ^ ThreadLocalRandom.nextSecondarySeed();
if (r == 0) // initial nonzero seed
r = 1;
int lastSignalId = 0; // avoid unneeded signals
while ((ws = workQueues) != null) {
boolean nonempty = false; // scan
for (int n = ws.length, j = n, m = n - 1; j > 0; --j) {
WorkQueue q; int i, b, al; ForkJoinTask[] a;
if ((i = r & m) >= 0 && i < n && // always true
(q = ws[i]) != null && (b = q.base) - q.top < 0 &&
(a = q.array) != null && (al = a.length) > 0) {
int qid = q.id; // (never zero)
int index = (al - 1) & b;
ForkJoinTask t = (ForkJoinTask)
QA.getAcquire(a, index);
if (t != null && b++ == q.base &&
QA.compareAndSet(a, index, t, null)) {
if ((q.base = b) - q.top < 0 && qid != lastSignalId)
signalWork(); // propagate signal
w.source = lastSignalId = qid;
t.doExec();
if ((w.id & FIFO) != 0) // run remaining locals
w.localPollAndExec(POLL_LIMIT);
else
w.localPopAndExec(POLL_LIMIT);
ForkJoinWorkerThread thread = w.owner;
++w.nsteals;
w.source = 0; // now idle
if (thread != null)
thread.afterTopLevelExec();
}
nonempty = true;
}
else if (nonempty)
break;
else
++r;
}
if (nonempty) { // move (xorshift)
r ^= r << 13; r ^= r >>> 17; r ^= r << 5;
}
else {
int phase;
lastSignalId = 0; // clear for next scan
if ((phase = w.phase) >= 0) { // enqueue
int np = w.phase = (phase + SS_SEQ) | UNSIGNALLED;
long c, nc;
do {
w.stackPred = (int)(c = ctl);
nc = ((c - RC_UNIT) & UC_MASK) | (SP_MASK & np);
} while (!CTL.weakCompareAndSet(this, c, nc));
}
else { // already queued
int pred = w.stackPred;
w.source = DORMANT; // enable signal
for (int steps = 0;;) {
int md, rc; long c;
if (w.phase >= 0) {
w.source = 0;
break;
}
else if ((md = mode) < 0) // shutting down
return;
else if ((rc = ((md & SMASK) + // possibly quiescent
(int)((c = ctl) >> RC_SHIFT))) <= 0 &&
(md & SHUTDOWN) != 0 &&
tryTerminate(false, false))
return; // help terminate
else if ((++steps & 1) == 0)
Thread.interrupted(); // clear between parks
else if (rc <= 0 && pred != 0 && phase == (int)c) {
long d = keepAlive + System.currentTimeMillis();
LockSupport.parkUntil(this, d);
if (ctl == c &&
d - System.currentTimeMillis() <= TIMEOUT_SLOP) {
long nc = ((UC_MASK & (c - TC_UNIT)) |
(SP_MASK & pred));
if (CTL.compareAndSet(this, c, nc)) {
w.phase = QUIET;
return; // drop on timeout
}
}
}
else
LockSupport.park(this);
}
}
}
}
}
/**
* Helps and/or blocks until the given task is done or timeout.
* First tries locally helping, then scans other queues for a task
* produced by one of w's stealers; compensating and blocking if
* none are found (rescanning if tryCompensate fails).
*
* @param w caller
* @param task the task
* @param deadline for timed waits, if nonzero
* @return task status on exit
*/
final int awaitJoin(WorkQueue w, ForkJoinTask task, long deadline) {
int s = 0;
if (w != null && task != null &&
(!(task instanceof CountedCompleter) ||
(s = w.localHelpCC((CountedCompleter)task, 0)) >= 0)) {
w.tryRemoveAndExec(task);
int src = w.source, id = w.id;
s = task.status;
while (s >= 0) {
WorkQueue[] ws;
boolean nonempty = false;
int r = ThreadLocalRandom.nextSecondarySeed() | 1; // odd indices
if ((ws = workQueues) != null) { // scan for matching id
for (int n = ws.length, m = n - 1, j = -n; j < n; j += 2) {
WorkQueue q; int i, b, al; ForkJoinTask[] a;
if ((i = (r + j) & m) >= 0 && i < n &&
(q = ws[i]) != null && q.source == id &&
(b = q.base) - q.top < 0 &&
(a = q.array) != null && (al = a.length) > 0) {
int qid = q.id;
int index = (al - 1) & b;
ForkJoinTask t = (ForkJoinTask)
QA.getAcquire(a, index);
if (t != null && b++ == q.base && id == q.source &&
QA.compareAndSet(a, index, t, null)) {
q.base = b;
w.source = qid;
t.doExec();
w.source = src;
}
nonempty = true;
break;
}
}
}
if ((s = task.status) < 0)
break;
else if (!nonempty) {
long ms, ns; int block;
if (deadline == 0L)
ms = 0L; // untimed
else if ((ns = deadline - System.nanoTime()) <= 0L)
break; // timeout
else if ((ms = TimeUnit.NANOSECONDS.toMillis(ns)) <= 0L)
ms = 1L; // avoid 0 for timed wait
if ((block = tryCompensate(w)) != 0) {
task.internalWait(ms);
CTL.getAndAdd(this, (block > 0) ? RC_UNIT : 0L);
}
s = task.status;
}
}
}
return s;
}
/**
* Runs tasks until {@code isQuiescent()}. Rather than blocking
* when tasks cannot be found, rescans until all others cannot
* find tasks either.
*/
final void helpQuiescePool(WorkQueue w) {
int prevSrc = w.source, fifo = w.id & FIFO;
for (int source = prevSrc, released = -1;;) { // -1 until known
WorkQueue[] ws;
if (fifo != 0)
w.localPollAndExec(0);
else
w.localPopAndExec(0);
if (released == -1 && w.phase >= 0)
released = 1;
boolean quiet = true, empty = true;
int r = ThreadLocalRandom.nextSecondarySeed();
if ((ws = workQueues) != null) {
for (int n = ws.length, j = n, m = n - 1; j > 0; --j) {
WorkQueue q; int i, b, al; ForkJoinTask[] a;
if ((i = (r - j) & m) >= 0 && i < n && (q = ws[i]) != null) {
if ((b = q.base) - q.top < 0 &&
(a = q.array) != null && (al = a.length) > 0) {
int qid = q.id;
if (released == 0) { // increment
released = 1;
CTL.getAndAdd(this, RC_UNIT);
}
int index = (al - 1) & b;
ForkJoinTask t = (ForkJoinTask)
QA.getAcquire(a, index);
if (t != null && b++ == q.base &&
QA.compareAndSet(a, index, t, null)) {
q.base = b;
w.source = source = q.id;
t.doExec();
w.source = source = prevSrc;
}
quiet = empty = false;
break;
}
else if ((q.source & QUIET) == 0)
quiet = false;
}
}
}
if (quiet) {
if (released == 0)
CTL.getAndAdd(this, RC_UNIT);
w.source = prevSrc;
break;
}
else if (empty) {
if (source != QUIET)
w.source = source = QUIET;
if (released == 1) { // decrement
released = 0;
CTL.getAndAdd(this, RC_MASK & -RC_UNIT);
}
}
}
}
/**
* Scans for and returns a polled task, if available.
* Used only for untracked polls.
*
* @param submissionsOnly if true, only scan submission queues
*/
private ForkJoinTask pollScan(boolean submissionsOnly) {
WorkQueue[] ws; int n;
rescan: while ((mode & STOP) == 0 && (ws = workQueues) != null &&
(n = ws.length) > 0) {
int m = n - 1;
int r = ThreadLocalRandom.nextSecondarySeed();
int h = r >>> 16;
int origin, step;
if (submissionsOnly) {
origin = (r & ~1) & m; // even indices and steps
step = (h & ~1) | 2;
}
else {
origin = r & m;
step = h | 1;
}
for (int k = origin, oldSum = 0, checkSum = 0;;) {
WorkQueue q; int b, al; ForkJoinTask[] a;
if ((q = ws[k]) != null) {
checkSum += b = q.base;
if (b - q.top < 0 &&
(a = q.array) != null && (al = a.length) > 0) {
int index = (al - 1) & b;
ForkJoinTask t = (ForkJoinTask)
QA.getAcquire(a, index);
if (t != null && b++ == q.base &&
QA.compareAndSet(a, index, t, null)) {
q.base = b;
return t;
}
else
break; // restart
}
}
if ((k = (k + step) & m) == origin) {
if (oldSum == (oldSum = checkSum))
break rescan;
checkSum = 0;
}
}
}
return null;
}
/**
* Gets and removes a local or stolen task for the given worker.
*
* @return a task, if available
*/
final ForkJoinTask nextTaskFor(WorkQueue w) {
ForkJoinTask t;
if (w != null &&
(t = (w.id & FIFO) != 0 ? w.poll() : w.pop()) != null)
return t;
else
return pollScan(false);
}
// External operations
/**
* Adds the given task to a submission queue at submitter's
* current queue, creating one if null or contended.
*
* @param task the task. Caller must ensure non-null.
*/
final void externalPush(ForkJoinTask task) {
int r; // initialize caller's probe
if ((r = ThreadLocalRandom.getProbe()) == 0) {
ThreadLocalRandom.localInit();
r = ThreadLocalRandom.getProbe();
}
for (;;) {
int md = mode, n;
WorkQueue[] ws = workQueues;
if ((md & SHUTDOWN) != 0 || ws == null || (n = ws.length) <= 0)
throw new RejectedExecutionException();
else {
WorkQueue q;
boolean push = false, grow = false;
if ((q = ws[(n - 1) & r & SQMASK]) == null) {
Object lock = workerNamePrefix;
int qid = (r | QUIET) & ~(FIFO | OWNED);
q = new WorkQueue(this, null);
q.id = qid;
q.source = QUIET;
q.phase = QLOCK; // lock queue
if (lock != null) {
synchronized (lock) { // lock pool to install
int i;
if ((ws = workQueues) != null &&
(n = ws.length) > 0 &&
ws[i = qid & (n - 1) & SQMASK] == null) {
ws[i] = q;
push = grow = true;
}
}
}
}
else if (q.tryLockSharedQueue()) {
int b = q.base, s = q.top, al, d; ForkJoinTask[] a;
if ((a = q.array) != null && (al = a.length) > 0 &&
al - 1 + (d = b - s) > 0) {
a[(al - 1) & s] = task;
q.top = s + 1; // relaxed writes OK here
q.phase = 0;
if (d < 0 && q.base - s < -1)
break; // no signal needed
}
else
grow = true;
push = true;
}
if (push) {
if (grow) {
try {
q.growArray();
int s = q.top, al; ForkJoinTask[] a;
if ((a = q.array) != null && (al = a.length) > 0) {
a[(al - 1) & s] = task;
q.top = s + 1;
}
} finally {
q.phase = 0;
}
}
signalWork();
break;
}
else // move if busy
r = ThreadLocalRandom.advanceProbe(r);
}
}
}
/**
* Pushes a possibly-external submission.
*/
private A {@code ManagedBlocker} provides two methods. Method
* {@link #isReleasable} must return {@code true} if blocking is
* not necessary. Method {@link #block} blocks the current thread
* if necessary (perhaps internally invoking {@code isReleasable}
* before actually blocking). These actions are performed by any
* thread invoking {@link ForkJoinPool#managedBlock(ManagedBlocker)}.
* The unusual methods in this API accommodate synchronizers that
* may, but don't usually, block for long periods. Similarly, they
* allow more efficient internal handling of cases in which
* additional workers may be, but usually are not, needed to
* ensure sufficient parallelism. Toward this end,
* implementations of method {@code isReleasable} must be amenable
* to repeated invocation.
*
* For example, here is a ManagedBlocker based on a
* ReentrantLock:
* Here is a class that possibly blocks waiting for an
* item on a given queue:
* This method repeatedly calls {@code blocker.isReleasable()} and
* {@code blocker.block()} until either method returns {@code true}.
* Every call to {@code blocker.block()} is preceded by a call to
* {@code blocker.isReleasable()} that returned {@code false}.
*
* If not running in a ForkJoinPool, this method is
* behaviorally equivalent to
* {@code
* class ManagedLocker implements ManagedBlocker {
* final ReentrantLock lock;
* boolean hasLock = false;
* ManagedLocker(ReentrantLock lock) { this.lock = lock; }
* public boolean block() {
* if (!hasLock)
* lock.lock();
* return true;
* }
* public boolean isReleasable() {
* return hasLock || (hasLock = lock.tryLock());
* }
* }}
*
* {@code
* class QueueTaker
*/
public static interface ManagedBlocker {
/**
* Possibly blocks the current thread, for example waiting for
* a lock or condition.
*
* @return {@code true} if no additional blocking is necessary
* (i.e., if isReleasable would return true)
* @throws InterruptedException if interrupted while waiting
* (the method is not required to do so, but is allowed to)
*/
boolean block() throws InterruptedException;
/**
* Returns {@code true} if blocking is unnecessary.
* @return {@code true} if blocking is unnecessary
*/
boolean isReleasable();
}
/**
* Runs the given possibly blocking task. When {@linkplain
* ForkJoinTask#inForkJoinPool() running in a ForkJoinPool}, this
* method possibly arranges for a spare thread to be activated if
* necessary to ensure sufficient parallelism while the current
* thread is blocked in {@link ManagedBlocker#block blocker.block()}.
*
* {@code
* while (!blocker.isReleasable())
* if (blocker.block())
* break;}
*
* If running in a ForkJoinPool, the pool may first be expanded to
* ensure sufficient parallelism available during the call to
* {@code blocker.block()}.
*
* @param blocker the blocker task
* @throws InterruptedException if {@code blocker.block()} did so
*/
public static void managedBlock(ManagedBlocker blocker)
throws InterruptedException {
ForkJoinPool p;
ForkJoinWorkerThread wt;
WorkQueue w;
Thread t = Thread.currentThread();
if ((t instanceof ForkJoinWorkerThread) &&
(p = (wt = (ForkJoinWorkerThread)t).pool) != null &&
(w = wt.workQueue) != null) {
int block;
while (!blocker.isReleasable()) {
if ((block = p.tryCompensate(w)) != 0) {
try {
do {} while (!blocker.isReleasable() &&
!blocker.block());
} finally {
CTL.getAndAdd(p, (block > 0) ? RC_UNIT : 0L);
}
break;
}
}
}
else {
do {} while (!blocker.isReleasable() &&
!blocker.block());
}
}
/**
* If the given executor is a ForkJoinPool, poll and execute
* AsynchronousCompletionTasks from worker's queue until none are
* available or blocker is released.
*/
static void helpAsyncBlocker(Executor e, ManagedBlocker blocker) {
if (blocker != null && (e instanceof ForkJoinPool)) {
WorkQueue w; ForkJoinWorkerThread wt; WorkQueue[] ws; int r, n;
ForkJoinPool p = (ForkJoinPool)e;
Thread thread = Thread.currentThread();
if (thread instanceof ForkJoinWorkerThread &&
(wt = (ForkJoinWorkerThread)thread).pool == p)
w = wt.workQueue;
else if ((r = ThreadLocalRandom.getProbe()) != 0 &&
(ws = p.workQueues) != null && (n = ws.length) > 0)
w = ws[(n - 1) & r & SQMASK];
else
w = null;
if (w != null) {
for (;;) {
int b = w.base, s = w.top, d, al; ForkJoinTask[] a;
if ((a = w.array) != null && (d = b - s) < 0 &&
(al = a.length) > 0) {
int index = (al - 1) & b;
ForkJoinTask t = (ForkJoinTask)
QA.getAcquire(a, index);
if (blocker.isReleasable())
break;
else if (b++ == w.base) {
if (t == null) {
if (d == -1)
break;
}
else if (!(t instanceof CompletableFuture.
AsynchronousCompletionTask))
break;
else if (QA.compareAndSet(a, index, t, null)) {
w.base = b;
t.doExec();
}
}
}
else
break;
}
}
}
}
// AbstractExecutorService overrides. These rely on undocumented
// fact that ForkJoinTask.adapt returns ForkJoinTasks that also
// implement RunnableFuture.
protected