1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
//! A concurrent work-stealing deque.
//!
//! This data structure is most commonly used in schedulers. The typical setup involves a number of
//! threads where each thread has its own deque containing tasks. A thread may push tasks into its
//! deque as well as pop tasks from it. Once it runs out of tasks, it may steal some from other
//! threads to help complete tasks more quickly. Therefore, work-stealing deques supports three
//! essential operations: *push*, *pop*, and *steal*.
//!
//! # Types of deques
//!
//! There are two types of deques, differing only in which order tasks get pushed and popped. The
//! two task ordering strategies are:
//!
//! * First-in first-out (FIFO)
//! * Last-in first-out (LIFO)
//!
//! A deque is a buffer with two ends, front and back. In a FIFO deque, tasks are pushed into the
//! back, popped from the front, and stolen from the front. However, in a LIFO deque, tasks are
//! popped from the back instead - that is the only difference.
//!
//! # Workers and stealers
//!
//! There are two functions that construct a deque: [`fifo`] and [`lifo`]. These functions return a
//! [`Worker`] and a [`Stealer`]. The thread which owns the deque is usually called *worker*, while
//! all other threads are *stealers*.
//!
//! [`Worker`] is able to push and pop tasks. It cannot be shared among multiple threads - only
//! one thread owns it.
//!
//! [`Stealer`] can only steal tasks. It can be shared among multiple threads by reference or by
//! cloning. Cloning a [`Stealer`] simply creates another one associated with the same deque.
//!
//! # Examples
//!
//! ```
//! use crossbeam_deque::{self as deque, Pop, Steal};
//! use std::thread;
//!
//! // Create a LIFO deque.
//! let (w, s) = deque::lifo();
//!
//! // Push several elements into the back.
//! w.push(1);
//! w.push(2);
//! w.push(3);
//!
//! // This is a LIFO deque, which means an element is popped from the back.
//! // If it was a FIFO deque, `w.pop()` would return `Some(1)`.
//! assert_eq!(w.pop(), Pop::Data(3));
//!
//! // Create a stealer thread.
//! thread::spawn(move || {
//!     assert_eq!(s.steal(), Steal::Data(1));
//!     assert_eq!(s.steal(), Steal::Data(2));
//! }).join().unwrap();
//! ```
//!
//! [`Worker`]: struct.Worker.html
//! [`Stealer`]: struct.Stealer.html
//! [`fifo`]: fn.fifo.html
//! [`lifo`]: fn.lifo.html

#![warn(missing_docs)]
#![warn(missing_debug_implementations)]

extern crate crossbeam_epoch as epoch;
extern crate crossbeam_utils as utils;

use std::cell::Cell;
use std::cmp;
use std::fmt;
use std::marker::PhantomData;
use std::mem;
use std::ptr;
use std::sync::atomic::{self, AtomicIsize, Ordering};
use std::sync::Arc;

use epoch::{Atomic, Owned};
use utils::CachePadded;

/// Minimum buffer capacity for a deque.
const MIN_CAP: usize = 32;

/// Maximum number of additional elements that can be stolen in `steal_many`.
const MAX_BATCH: usize = 128;

/// If a buffer of at least this size is retired, thread-local garbage is flushed so that it gets
/// deallocated as soon as possible.
const FLUSH_THRESHOLD_BYTES: usize = 1 << 10;

/// Creates a work-stealing deque with the first-in first-out strategy.
///
/// Elements are pushed into the back, popped from the front, and stolen from the front. In other
/// words, the worker side behaves as a FIFO queue.
///
/// # Examples
///
/// ```
/// use crossbeam_deque::{self as deque, Pop, Steal};
///
/// let (w, s) = deque::fifo::<i32>();
/// w.push(1);
/// w.push(2);
/// w.push(3);
///
/// assert_eq!(s.steal(), Steal::Data(1));
/// assert_eq!(w.pop(), Pop::Data(2));
/// assert_eq!(w.pop(), Pop::Data(3));
/// ```
pub fn fifo<T>() -> (Worker<T>, Stealer<T>) {
    let buffer = Buffer::alloc(MIN_CAP);

    let inner = Arc::new(CachePadded::new(Inner {
        front: AtomicIsize::new(0),
        back: AtomicIsize::new(0),
        buffer: Atomic::new(buffer),
    }));

    let w = Worker {
        inner: inner.clone(),
        cached_buffer: Cell::new(buffer),
        flavor: Flavor::Fifo,
        _marker: PhantomData,
    };
    let s = Stealer {
        inner,
        flavor: Flavor::Fifo,
    };
    (w, s)
}

/// Creates a work-stealing deque with the last-in first-out strategy.
///
/// Elements are pushed into the back, popped from the back, and stolen from the front. In other
/// words, the worker side behaves as a LIFO stack.
///
/// # Examples
///
/// ```
/// use crossbeam_deque::{self as deque, Pop, Steal};
///
/// let (w, s) = deque::lifo::<i32>();
/// w.push(1);
/// w.push(2);
/// w.push(3);
///
/// assert_eq!(s.steal(), Steal::Data(1));
/// assert_eq!(w.pop(), Pop::Data(3));
/// assert_eq!(w.pop(), Pop::Data(2));
/// ```
pub fn lifo<T>() -> (Worker<T>, Stealer<T>) {
    let buffer = Buffer::alloc(MIN_CAP);

    let inner = Arc::new(CachePadded::new(Inner {
        front: AtomicIsize::new(0),
        back: AtomicIsize::new(0),
        buffer: Atomic::new(buffer),
    }));

    let w = Worker {
        inner: inner.clone(),
        cached_buffer: Cell::new(buffer),
        flavor: Flavor::Lifo,
        _marker: PhantomData,
    };
    let s = Stealer {
        inner,
        flavor: Flavor::Lifo,
    };
    (w, s)
}

/// A buffer that holds elements in a deque.
///
/// This is just a pointer to the buffer and its length - dropping an instance of this struct will
/// *not* deallocate the buffer.
struct Buffer<T> {
    /// Pointer to the allocated memory.
    ptr: *mut T,

    /// Capacity of the buffer. Always a power of two.
    cap: usize,
}

unsafe impl<T> Send for Buffer<T> {}

impl<T> Buffer<T> {
    /// Allocates a new buffer with the specified capacity.
    fn alloc(cap: usize) -> Self {
        debug_assert_eq!(cap, cap.next_power_of_two());

        let mut v = Vec::with_capacity(cap);
        let ptr = v.as_mut_ptr();
        mem::forget(v);

        Buffer { ptr, cap }
    }

    /// Deallocates the buffer.
    unsafe fn dealloc(self) {
        drop(Vec::from_raw_parts(self.ptr, 0, self.cap));
    }

    /// Returns a pointer to the element at the specified `index`.
    unsafe fn at(&self, index: isize) -> *mut T {
        // `self.cap` is always a power of two.
        self.ptr.offset(index & (self.cap - 1) as isize)
    }

    /// Writes `value` into the specified `index`.
    ///
    /// Using this concurrently with another `read` or `write` is technically
    /// speaking UB due to data races.  We should be using relaxed accesses, but
    /// that would cost too much performance.  Hence, as a HACK, we use volatile
    /// accesses instead.  Experimental evidence shows that this works.
    unsafe fn write(&self, index: isize, value: T) {
        ptr::write_volatile(self.at(index), value)
    }

    /// Reads a value from the specified `index`.
    ///
    /// Using this concurrently with a `write` is technically speaking UB due to
    /// data races.  We should be using relaxed accesses, but that would cost
    /// too much performance.  Hence, as a HACK, we use volatile accesses
    /// instead.  Experimental evidence shows that this works.
    unsafe fn read(&self, index: isize) -> T {
        ptr::read_volatile(self.at(index))
    }
}

impl<T> Clone for Buffer<T> {
    fn clone(&self) -> Buffer<T> {
        Buffer {
            ptr: self.ptr,
            cap: self.cap,
        }
    }
}

impl<T> Copy for Buffer<T> {}

/// Possible outcomes of a pop operation.
#[must_use]
#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Copy, Clone)]
pub enum Pop<T> {
    /// The deque was empty at the time of popping.
    Empty,

    /// Some data has been successfully popped.
    Data(T),

    /// Lost the race for popping data to another concurrent steal operation. Try again.
    Retry,
}

/// Possible outcomes of a steal operation.
#[must_use]
#[derive(Debug, PartialEq, Eq, PartialOrd, Ord, Copy, Clone)]
pub enum Steal<T> {
    /// The deque was empty at the time of stealing.
    Empty,

    /// Some data has been successfully stolen.
    Data(T),

    /// Lost the race for stealing data to another concurrent steal or pop operation. Try again.
    Retry,
}

/// Internal data that is shared between the worker and stealers.
///
/// The implementation is based on the following work:
///
/// 1. [Chase and Lev. Dynamic circular work-stealing deque. SPAA 2005.][chase-lev]
/// 2. [Le, Pop, Cohen, and Nardelli. Correct and efficient work-stealing for weak memory models.
///    PPoPP 2013.][weak-mem]
/// 3. [Norris and Demsky. CDSchecker: checking concurrent data structures written with C/C++
///    atomics. OOPSLA 2013.][checker]
///
/// [chase-lev]: https://dl.acm.org/citation.cfm?id=1073974
/// [weak-mem]: https://dl.acm.org/citation.cfm?id=2442524
/// [checker]: https://dl.acm.org/citation.cfm?id=2509514
struct Inner<T> {
    /// The front index.
    front: AtomicIsize,

    /// The back index.
    back: AtomicIsize,

    /// The underlying buffer.
    buffer: Atomic<Buffer<T>>,
}

impl<T> Drop for Inner<T> {
    fn drop(&mut self) {
        // Load the back index, front index, and buffer.
        let b = self.back.load(Ordering::Relaxed);
        let f = self.front.load(Ordering::Relaxed);

        unsafe {
            let buffer = self.buffer.load(Ordering::Relaxed, epoch::unprotected());

            // Go through the buffer from front to back and drop all elements in the deque.
            let mut i = f;
            while i != b {
                ptr::drop_in_place(buffer.deref().at(i));
                i = i.wrapping_add(1);
            }

            // Free the memory allocated by the buffer.
            buffer.into_owned().into_box().dealloc();
        }
    }
}

/// The flavor of a deque: FIFO or LIFO.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
enum Flavor {
    /// The first-in first-out flavor.
    Fifo,

    /// The last-in first-out flavor.
    Lifo,
}

/// The worker side of a deque.
///
/// Workers push elements into the back and pop elements depending on the strategy:
///
/// * In FIFO deques, elements are popped from the front.
/// * In LIFO deques, elements are popped from the back.
///
/// A deque has only one worker. Workers are not intended to be shared among multiple threads.
pub struct Worker<T> {
    /// A reference to the inner representation of the deque.
    inner: Arc<CachePadded<Inner<T>>>,

    /// A copy of `inner.buffer` for quick access.
    cached_buffer: Cell<Buffer<T>>,

    /// The flavor of the deque.
    flavor: Flavor,

    /// Indicates that the worker cannot be shared among threads.
    _marker: PhantomData<*mut ()>, // !Send + !Sync
}

unsafe impl<T: Send> Send for Worker<T> {}

impl<T> Worker<T> {
    /// Resizes the internal buffer to the new capacity of `new_cap`.
    #[cold]
    unsafe fn resize(&self, new_cap: usize) {
        // Load the back index, front index, and buffer.
        let b = self.inner.back.load(Ordering::Relaxed);
        let f = self.inner.front.load(Ordering::Relaxed);
        let buffer = self.cached_buffer.get();

        // Allocate a new buffer.
        let new = Buffer::alloc(new_cap);
        self.cached_buffer.set(new);

        // Copy data from the old buffer to the new one.
        let mut i = f;
        while i != b {
            ptr::copy_nonoverlapping(buffer.at(i), new.at(i), 1);
            i = i.wrapping_add(1);
        }

        let guard = &epoch::pin();

        // Replace the old buffer with the new one.
        let old =
            self.inner
                .buffer
                .swap(Owned::new(new).into_shared(guard), Ordering::Release, guard);

        // Destroy the old buffer later.
        guard.defer_unchecked(move || old.into_owned().into_box().dealloc());

        // If the buffer is very large, then flush the thread-local garbage in order to deallocate
        // it as soon as possible.
        if mem::size_of::<T>() * new_cap >= FLUSH_THRESHOLD_BYTES {
            guard.flush();
        }
    }

    /// Reserves enough capacity so that `reserve_cap` elements can be pushed without growing the
    /// buffer.
    fn reserve(&self, reserve_cap: usize) {
        if reserve_cap > 0 {
            // Compute the current length.
            let b = self.inner.back.load(Ordering::Relaxed);
            let f = self.inner.front.load(Ordering::SeqCst);
            let len = b.wrapping_sub(f) as usize;

            // The current capacity.
            let cap = self.cached_buffer.get().cap;

            // Is there enough capacity to push `reserve_cap` elements?
            if cap - len < reserve_cap {
                // Keep doubling the capacity as much as is needed.
                let mut new_cap = cap * 2;
                while new_cap - len < reserve_cap {
                    new_cap *= 2;
                }

                // Resize the buffer.
                unsafe {
                    self.resize(new_cap);
                }
            }
        }
    }

    /// Returns `true` if the deque is empty.
    ///
    /// ```
    /// use crossbeam_deque as deque;
    ///
    /// let (w, _) = deque::lifo();
    /// assert!(w.is_empty());
    /// w.push(1);
    /// assert!(!w.is_empty());
    /// ```
    pub fn is_empty(&self) -> bool {
        let b = self.inner.back.load(Ordering::Relaxed);
        let f = self.inner.front.load(Ordering::SeqCst);
        b.wrapping_sub(f) <= 0
    }

    /// Pushes an element into the back of the deque.
    ///
    /// # Examples
    ///
    /// ```
    /// use crossbeam_deque as deque;
    ///
    /// let (w, _) = deque::lifo();
    /// w.push(1);
    /// w.push(2);
    /// ```
    pub fn push(&self, value: T) {
        // Load the back index, front index, and buffer.
        let b = self.inner.back.load(Ordering::Relaxed);
        let f = self.inner.front.load(Ordering::Acquire);
        let mut buffer = self.cached_buffer.get();

        // Calculate the length of the deque.
        let len = b.wrapping_sub(f);

        // Is the deque full?
        if len >= buffer.cap as isize {
            // Yes. Grow the underlying buffer.
            unsafe {
                self.resize(2 * buffer.cap);
            }
            buffer = self.cached_buffer.get();
        }

        // Write `value` into the slot.
        unsafe {
            buffer.write(b, value);
        }

        atomic::fence(Ordering::Release);

        // Increment the back index.
        //
        // This ordering could be `Relaxed`, but then thread sanitizer would falsely report data
        // races because it doesn't understand fences.
        self.inner.back.store(b.wrapping_add(1), Ordering::Release);
    }

    /// Pops an element from the deque.
    ///
    /// Which end of the deque is used depends on the strategy:
    ///
    /// * If this is a FIFO deque, an element is popped from the front.
    /// * If this is a LIFO deque, an element is popped from the back.
    ///
    /// # Examples
    ///
    /// ```
    /// use crossbeam_deque::{self as deque, Pop};
    ///
    /// let (w, _) = deque::fifo();
    /// w.push(1);
    /// w.push(2);
    ///
    /// assert_eq!(w.pop(), Pop::Data(1));
    /// assert_eq!(w.pop(), Pop::Data(2));
    /// assert_eq!(w.pop(), Pop::Empty);
    /// ```
    pub fn pop(&self) -> Pop<T> {
        // Load the back and front index.
        let b = self.inner.back.load(Ordering::Relaxed);
        let f = self.inner.front.load(Ordering::Relaxed);

        // Calculate the length of the deque.
        let len = b.wrapping_sub(f);

        // Is the deque empty?
        if len <= 0 {
            return Pop::Empty;
        }

        match self.flavor {
            // Pop from the front of the deque.
            Flavor::Fifo => {
                // Try incrementing the front index to pop the value.
                if self
                    .inner
                    .front
                    .compare_exchange(f, f.wrapping_add(1), Ordering::SeqCst, Ordering::Relaxed)
                    .is_ok()
                {
                    unsafe {
                        // Read the popped value.
                        let buffer = self.cached_buffer.get();
                        let data = buffer.read(f);

                        // Shrink the buffer if `len - 1` is less than one fourth of the capacity.
                        if buffer.cap > MIN_CAP && len <= buffer.cap as isize / 4 {
                            self.resize(buffer.cap / 2);
                        }

                        return Pop::Data(data);
                    }
                }

                Pop::Retry
            }

            // Pop from the back of the deque.
            Flavor::Lifo => {
                // Decrement the back index.
                let b = b.wrapping_sub(1);
                self.inner.back.store(b, Ordering::Relaxed);

                atomic::fence(Ordering::SeqCst);

                // Load the front index.
                let f = self.inner.front.load(Ordering::Relaxed);

                // Compute the length after the back index was decremented.
                let len = b.wrapping_sub(f);

                if len < 0 {
                    // The deque is empty. Restore the back index to the original value.
                    self.inner.back.store(b.wrapping_add(1), Ordering::Relaxed);
                    Pop::Empty
                } else {
                    // Read the value to be popped.
                    let buffer = self.cached_buffer.get();
                    let mut value = unsafe { Some(buffer.read(b)) };

                    // Are we popping the last element from the deque?
                    if len == 0 {
                        // Try incrementing the front index.
                        if self
                            .inner
                            .front
                            .compare_exchange(
                                f,
                                f.wrapping_add(1),
                                Ordering::SeqCst,
                                Ordering::Relaxed,
                            ).is_err()
                        {
                            // Failed. We didn't pop anything.
                            mem::forget(value.take());
                        }

                        // Restore the back index to the original value.
                        self.inner.back.store(b.wrapping_add(1), Ordering::Relaxed);
                    } else {
                        // Shrink the buffer if `len` is less than one fourth of the capacity.
                        if buffer.cap > MIN_CAP && len < buffer.cap as isize / 4 {
                            unsafe {
                                self.resize(buffer.cap / 2);
                            }
                        }
                    }

                    match value {
                        None => Pop::Empty,
                        Some(data) => Pop::Data(data),
                    }
                }
            }
        }
    }
}

impl<T> fmt::Debug for Worker<T> {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.pad("Worker { .. }")
    }
}

/// The stealer side of a deque.
///
/// Stealers can only steal elements from the front of the deque.
///
/// Stealers are cloneable so that they can be easily shared among multiple threads.
pub struct Stealer<T> {
    /// A reference to the inner representation of the deque.
    inner: Arc<CachePadded<Inner<T>>>,

    /// The flavor of the deque.
    flavor: Flavor,
}

unsafe impl<T: Send> Send for Stealer<T> {}
unsafe impl<T: Send> Sync for Stealer<T> {}

impl<T> Stealer<T> {
    /// Returns `true` if the deque is empty.
    ///
    /// ```
    /// use crossbeam_deque as deque;
    ///
    /// let (w, s) = deque::lifo();
    /// assert!(s.is_empty());
    /// w.push(1);
    /// assert!(!s.is_empty());
    /// ```
    pub fn is_empty(&self) -> bool {
        let f = self.inner.front.load(Ordering::Acquire);
        atomic::fence(Ordering::SeqCst);
        let b = self.inner.back.load(Ordering::Acquire);
        b.wrapping_sub(f) <= 0
    }

    /// Steals an element from the front of the deque.
    ///
    /// # Examples
    ///
    /// ```
    /// use crossbeam_deque::{self as deque, Steal};
    ///
    /// let (w, s) = deque::lifo();
    /// w.push(1);
    /// w.push(2);
    ///
    /// assert_eq!(s.steal(), Steal::Data(1));
    /// assert_eq!(s.steal(), Steal::Data(2));
    /// assert_eq!(s.steal(), Steal::Empty);
    /// ```
    pub fn steal(&self) -> Steal<T> {
        // Load the front index.
        let f = self.inner.front.load(Ordering::Acquire);

        // A SeqCst fence is needed here.
        //
        // If the current thread is already pinned (reentrantly), we must manually issue the
        // fence. Otherwise, the following pinning will issue the fence anyway, so we don't
        // have to.
        if epoch::is_pinned() {
            atomic::fence(Ordering::SeqCst);
        }

        let guard = &epoch::pin();

        // Load the back index.
        let b = self.inner.back.load(Ordering::Acquire);

        // Is the deque empty?
        if b.wrapping_sub(f) <= 0 {
            return Steal::Empty;
        }

        // Load the buffer and read the value at the front.
        let buffer = self.inner.buffer.load(Ordering::Acquire, guard);
        let value = unsafe { buffer.deref().read(f) };

        // Try incrementing the front index to steal the value.
        if self
            .inner
            .front
            .compare_exchange(f, f.wrapping_add(1), Ordering::SeqCst, Ordering::Relaxed)
            .is_err()
        {
            // We didn't steal this value, forget it.
            mem::forget(value);
            return Steal::Retry;
        }

        // Return the stolen value.
        Steal::Data(value)
    }

    /// Steals elements from the front of the deque.
    ///
    /// If at least one element can be stolen, it will be returned. Additionally, some of the
    /// remaining elements will be stolen and pushed into the back of worker `dest` in order to
    /// balance the work among deques. There is no hard guarantee on exactly how many elements will
    /// be stolen, but it should be around half of the deque.
    ///
    /// # Examples
    ///
    /// ```
    /// use crossbeam_deque::{self as deque, Steal};
    ///
    /// let (w1, s1) = deque::fifo();
    /// let (w2, s2) = deque::fifo();
    ///
    /// w1.push(1);
    /// w1.push(2);
    /// w1.push(3);
    /// w1.push(4);
    ///
    /// assert_eq!(s1.steal_many(&w2), Steal::Data(1));
    /// assert_eq!(s2.steal(), Steal::Data(2));
    /// ```
    pub fn steal_many(&self, dest: &Worker<T>) -> Steal<T> {
        // Load the front index.
        let mut f = self.inner.front.load(Ordering::Acquire);

        // A SeqCst fence is needed here.
        //
        // If the current thread is already pinned (reentrantly), we must manually issue the
        // fence. Otherwise, the following pinning will issue the fence anyway, so we don't
        // have to.
        if epoch::is_pinned() {
            atomic::fence(Ordering::SeqCst);
        }

        let guard = &epoch::pin();

        // Load the back index.
        let b = self.inner.back.load(Ordering::Acquire);

        // Is the deque empty?
        let len = b.wrapping_sub(f);
        if len <= 0 {
            return Steal::Empty;
        }

        // Reserve capacity for the stolen additional elements.
        let additional = cmp::min((len as usize - 1) / 2, MAX_BATCH);
        dest.reserve(additional);
        let additional = additional as isize;

        // Get the destination buffer and back index.
        let dest_buffer = dest.cached_buffer.get();
        let mut dest_b = dest.inner.back.load(Ordering::Relaxed);

        // Load the buffer and read the value at the front.
        let buffer = self.inner.buffer.load(Ordering::Acquire, guard);
        let value = unsafe { buffer.deref().read(f) };

        match self.flavor {
            // Steal a batch of elements from the front at once.
            Flavor::Fifo => {
                // Copy the additional elements from the source to the destination buffer.
                for i in 0..additional {
                    unsafe {
                        let value = buffer.deref().read(f.wrapping_add(i + 1));
                        dest_buffer.write(dest_b.wrapping_add(i), value);
                    }
                }

                // Try incrementing the front index to steal the batch.
                if self
                    .inner
                    .front
                    .compare_exchange(
                        f,
                        f.wrapping_add(additional + 1),
                        Ordering::SeqCst,
                        Ordering::Relaxed,
                    ).is_err()
                {
                    // We didn't steal this value, forget it.
                    mem::forget(value);
                    return Steal::Retry;
                }

                atomic::fence(Ordering::Release);

                // Success! Update the back index in the destination deque.
                //
                // This ordering could be `Relaxed`, but then thread sanitizer would falsely report
                // data races because it doesn't understand fences.
                dest.inner
                    .back
                    .store(dest_b.wrapping_add(additional), Ordering::Release);

                // Return the first stolen value.
                Steal::Data(value)
            }

            // Steal a batch of elements from the front one by one.
            Flavor::Lifo => {
                // Try incrementing the front index to steal the value.
                if self
                    .inner
                    .front
                    .compare_exchange(f, f.wrapping_add(1), Ordering::SeqCst, Ordering::Relaxed)
                    .is_err()
                {
                    // We didn't steal this value, forget it.
                    mem::forget(value);
                    return Steal::Retry;
                }

                // Move the front index one step forward.
                f = f.wrapping_add(1);

                // Repeat the same procedure for the additional steals.
                for _ in 0..additional {
                    // We've already got the current front index. Now execute the fence to
                    // synchronize with other threads.
                    atomic::fence(Ordering::SeqCst);

                    // Load the back index.
                    let b = self.inner.back.load(Ordering::Acquire);

                    // Is the deque empty?
                    if b.wrapping_sub(f) <= 0 {
                        break;
                    }

                    // Read the value at the front.
                    let value = unsafe { buffer.deref().read(f) };

                    // Try incrementing the front index to steal the value.
                    if self
                        .inner
                        .front
                        .compare_exchange(f, f.wrapping_add(1), Ordering::SeqCst, Ordering::Relaxed)
                        .is_err()
                    {
                        // We didn't steal this value, forget it and break from the loop.
                        mem::forget(value);
                        break;
                    }

                    // Write the stolen value into the destination buffer.
                    unsafe {
                        dest_buffer.write(dest_b, value);
                    }

                    // Move the source front index and the destination back index one step forward.
                    f = f.wrapping_add(1);
                    dest_b = dest_b.wrapping_add(1);

                    atomic::fence(Ordering::Release);

                    // Update the destination back index.
                    //
                    // This ordering could be `Relaxed`, but then thread sanitizer would falsely
                    // report data races because it doesn't understand fences.
                    dest.inner.back.store(dest_b, Ordering::Release);
                }

                // Return the first stolen value.
                Steal::Data(value)
            }
        }
    }
}

impl<T> Clone for Stealer<T> {
    fn clone(&self) -> Stealer<T> {
        Stealer {
            inner: self.inner.clone(),
            flavor: self.flavor,
        }
    }
}

impl<T> fmt::Debug for Stealer<T> {
    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
        f.pad("Stealer { .. }")
    }
}