This file is indexed.

/usr/include/trilinos/Kokkos_CudaSpace.hpp is in libtrilinos-kokkos-dev 12.12.1-5.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

  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
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
/*
//@HEADER
// ************************************************************************
//
//                        Kokkos v. 2.0
//              Copyright (2014) Sandia Corporation
//
// Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
// the U.S. Government retains certain rights in this software.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact  H. Carter Edwards (hcedwar@sandia.gov)
//
// ************************************************************************
//@HEADER
*/

#ifndef KOKKOS_CUDASPACE_HPP
#define KOKKOS_CUDASPACE_HPP

#include <Kokkos_Macros.hpp>
#if defined( KOKKOS_ENABLE_CUDA )

#include <Kokkos_Core_fwd.hpp>

#include <iosfwd>
#include <typeinfo>
#include <string>

#include <Kokkos_HostSpace.hpp>

#include <Cuda/Kokkos_Cuda_abort.hpp>

/*--------------------------------------------------------------------------*/

namespace Kokkos {

/** \brief  Cuda on-device memory management */

class CudaSpace {
public:

  //! Tag this class as a kokkos memory space
  typedef CudaSpace             memory_space ;
  typedef Kokkos::Cuda          execution_space ;
  typedef Kokkos::Device<execution_space,memory_space> device_type;

  typedef unsigned int          size_type ;

  /*--------------------------------*/

  CudaSpace();
  CudaSpace( CudaSpace && rhs ) = default ;
  CudaSpace( const CudaSpace & rhs ) = default ;
  CudaSpace & operator = ( CudaSpace && rhs ) = default ;
  CudaSpace & operator = ( const CudaSpace & rhs ) = default ;
  ~CudaSpace() = default ;

  /**\brief  Allocate untracked memory in the cuda space */
  void * allocate( const size_t arg_alloc_size ) const ;

  /**\brief  Deallocate untracked memory in the cuda space */
  void deallocate( void * const arg_alloc_ptr
                 , const size_t arg_alloc_size ) const ;

  /**\brief Return Name of the MemorySpace */
  static constexpr const char* name();

  /*--------------------------------*/
  /** \brief  Error reporting for HostSpace attempt to access CudaSpace */
  static void access_error();
  static void access_error( const void * const );

private:

  int  m_device ; ///< Which Cuda device

  static constexpr const char* m_name = "Cuda";
  friend class Kokkos::Impl::SharedAllocationRecord< Kokkos::CudaSpace , void > ;
};

namespace Impl {
/// \brief Initialize lock array for arbitrary size atomics.
///
/// Arbitrary atomics are implemented using a hash table of locks
/// where the hash value is derived from the address of the
/// object for which an atomic operation is performed.
/// This function initializes the locks to zero (unset).
void init_lock_arrays_cuda_space();

/// \brief Retrieve the pointer to the lock array for arbitrary size atomics.
///
/// Arbitrary atomics are implemented using a hash table of locks
/// where the hash value is derived from the address of the
/// object for which an atomic operation is performed.
/// This function retrieves the lock array pointer.
/// If the array is not yet allocated it will do so.
int* atomic_lock_array_cuda_space_ptr(bool deallocate = false);

/// \brief Retrieve the pointer to the scratch array for team and thread private global memory.
///
/// Team and Thread private scratch allocations in
/// global memory are aquired via locks.
/// This function retrieves the lock array pointer.
/// If the array is not yet allocated it will do so.
int* scratch_lock_array_cuda_space_ptr(bool deallocate = false);

/// \brief Retrieve the pointer to the scratch array for unique identifiers.
///
/// Unique identifiers in the range 0-Cuda::concurrency
/// are provided via locks.
/// This function retrieves the lock array pointer.
/// If the array is not yet allocated it will do so.
int* threadid_lock_array_cuda_space_ptr(bool deallocate = false);
}
} // namespace Kokkos

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {

/** \brief  Cuda memory that is accessible to Host execution space
 *          through Cuda's unified virtual memory (UVM) runtime.
 */
class CudaUVMSpace {
public:

  //! Tag this class as a kokkos memory space
  typedef CudaUVMSpace          memory_space ;
  typedef Cuda                  execution_space ;
  typedef Kokkos::Device<execution_space,memory_space> device_type;
  typedef unsigned int          size_type ;

  /** \brief  If UVM capability is available */
  static bool available();


  /*--------------------------------*/
  /** \brief  CudaUVMSpace specific routine */
  static int number_of_allocations();

  /*--------------------------------*/


  /*--------------------------------*/

  CudaUVMSpace();
  CudaUVMSpace( CudaUVMSpace && rhs ) = default ;
  CudaUVMSpace( const CudaUVMSpace & rhs ) = default ;
  CudaUVMSpace & operator = ( CudaUVMSpace && rhs ) = default ;
  CudaUVMSpace & operator = ( const CudaUVMSpace & rhs ) = default ;
  ~CudaUVMSpace() = default ;

  /**\brief  Allocate untracked memory in the cuda space */
  void * allocate( const size_t arg_alloc_size ) const ;

  /**\brief  Deallocate untracked memory in the cuda space */
  void deallocate( void * const arg_alloc_ptr
                 , const size_t arg_alloc_size ) const ;

  /**\brief Return Name of the MemorySpace */
  static constexpr const char* name();

  /*--------------------------------*/

private:
  int  m_device ; ///< Which Cuda device

  static constexpr const char* m_name = "CudaUVM";

};

} // namespace Kokkos

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {

/** \brief  Host memory that is accessible to Cuda execution space
 *          through Cuda's host-pinned memory allocation.
 */
class CudaHostPinnedSpace {
public:

  //! Tag this class as a kokkos memory space
  /** \brief  Memory is in HostSpace so use the HostSpace::execution_space */
  typedef HostSpace::execution_space  execution_space ;
  typedef CudaHostPinnedSpace         memory_space ;
  typedef Kokkos::Device<execution_space,memory_space> device_type;
  typedef unsigned int                size_type ;

  /*--------------------------------*/

  CudaHostPinnedSpace();
  CudaHostPinnedSpace( CudaHostPinnedSpace && rhs ) = default ;
  CudaHostPinnedSpace( const CudaHostPinnedSpace & rhs ) = default ;
  CudaHostPinnedSpace & operator = ( CudaHostPinnedSpace && rhs ) = default ;
  CudaHostPinnedSpace & operator = ( const CudaHostPinnedSpace & rhs ) = default ;
  ~CudaHostPinnedSpace() = default ;

  /**\brief  Allocate untracked memory in the space */
  void * allocate( const size_t arg_alloc_size ) const ;

  /**\brief  Deallocate untracked memory in the space */
  void deallocate( void * const arg_alloc_ptr
                 , const size_t arg_alloc_size ) const ;

  /**\brief Return Name of the MemorySpace */
  static constexpr const char* name();

private:

  static constexpr const char* m_name = "CudaHostPinned";

  /*--------------------------------*/
};

} // namespace Kokkos

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {
namespace Impl {

static_assert( Kokkos::Impl::MemorySpaceAccess< Kokkos::CudaSpace , Kokkos::CudaSpace >::assignable , "" );
static_assert( Kokkos::Impl::MemorySpaceAccess< Kokkos::CudaUVMSpace , Kokkos::CudaUVMSpace >::assignable , "" );
static_assert( Kokkos::Impl::MemorySpaceAccess< Kokkos::CudaHostPinnedSpace , Kokkos::CudaHostPinnedSpace >::assignable , "" );

//----------------------------------------

template<>
struct MemorySpaceAccess< Kokkos::HostSpace , Kokkos::CudaSpace > {
  enum { assignable = false };
  enum { accessible = false };
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::HostSpace , Kokkos::CudaUVMSpace > {
  // HostSpace::execution_space != CudaUVMSpace::execution_space
  enum { assignable = false };
  enum { accessible = true };
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::HostSpace , Kokkos::CudaHostPinnedSpace > {
  // HostSpace::execution_space == CudaHostPinnedSpace::execution_space
  enum { assignable = true };
  enum { accessible = true };
  enum { deepcopy   = true };
};

//----------------------------------------

template<>
struct MemorySpaceAccess< Kokkos::CudaSpace , Kokkos::HostSpace > {
  enum { assignable = false };
  enum { accessible = false };
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::CudaSpace , Kokkos::CudaUVMSpace > {
  // CudaSpace::execution_space == CudaUVMSpace::execution_space
  enum { assignable = true };
  enum { accessible = true };
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::CudaSpace , Kokkos::CudaHostPinnedSpace > {
  // CudaSpace::execution_space != CudaHostPinnedSpace::execution_space
  enum { assignable = false };
  enum { accessible = true }; // CudaSpace::execution_space
  enum { deepcopy   = true };
};

//----------------------------------------
// CudaUVMSpace::execution_space == Cuda
// CudaUVMSpace accessible to both Cuda and Host

template<>
struct MemorySpaceAccess< Kokkos::CudaUVMSpace , Kokkos::HostSpace > {
  enum { assignable = false };
  enum { accessible = false }; // Cuda cannot access HostSpace
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::CudaUVMSpace , Kokkos::CudaSpace > {
  // CudaUVMSpace::execution_space == CudaSpace::execution_space
  // Can access CudaUVMSpace from Host but cannot access CudaSpace from Host
  enum { assignable = false };

  // CudaUVMSpace::execution_space can access CudaSpace
  enum { accessible = true };
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::CudaUVMSpace , Kokkos::CudaHostPinnedSpace > {
  // CudaUVMSpace::execution_space != CudaHostPinnedSpace::execution_space
  enum { assignable = false };
  enum { accessible = true }; // CudaUVMSpace::execution_space
  enum { deepcopy   = true };
};


//----------------------------------------
// CudaHostPinnedSpace::execution_space == HostSpace::execution_space
// CudaHostPinnedSpace accessible to both Cuda and Host

template<>
struct MemorySpaceAccess< Kokkos::CudaHostPinnedSpace , Kokkos::HostSpace > {
  enum { assignable = false }; // Cannot access from Cuda
  enum { accessible = true };  // CudaHostPinnedSpace::execution_space
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::CudaHostPinnedSpace , Kokkos::CudaSpace > {
  enum { assignable = false }; // Cannot access from Host
  enum { accessible = false };
  enum { deepcopy   = true };
};

template<>
struct MemorySpaceAccess< Kokkos::CudaHostPinnedSpace , Kokkos::CudaUVMSpace > {
  enum { assignable = false }; // different execution_space
  enum { accessible = true };  // same accessibility
  enum { deepcopy   = true };
};

//----------------------------------------

}} // namespace Kokkos::Impl

/*--------------------------------------------------------------------------*/
/*--------------------------------------------------------------------------*/

namespace Kokkos {
namespace Impl {

void DeepCopyAsyncCuda( void * dst , const void * src , size_t n);

template<> struct DeepCopy< CudaSpace , CudaSpace , Cuda>
{
  DeepCopy( void * dst , const void * src , size_t );
  DeepCopy( const Cuda & , void * dst , const void * src , size_t );
};

template<> struct DeepCopy< CudaSpace , HostSpace , Cuda >
{
  DeepCopy( void * dst , const void * src , size_t );
  DeepCopy( const Cuda & , void * dst , const void * src , size_t );
};

template<> struct DeepCopy< HostSpace , CudaSpace , Cuda >
{
  DeepCopy( void * dst , const void * src , size_t );
  DeepCopy( const Cuda & , void * dst , const void * src , size_t );
};

template<class ExecutionSpace> struct DeepCopy< CudaSpace , CudaSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaSpace , HostSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda>( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< HostSpace , CudaSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaSpace , CudaUVMSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaSpace , CudaHostPinnedSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};


template<class ExecutionSpace>
struct DeepCopy< CudaUVMSpace , CudaSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaUVMSpace , CudaUVMSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace>
struct DeepCopy< CudaUVMSpace , CudaHostPinnedSpace , ExecutionSpace>
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaUVMSpace , HostSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< CudaSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};


template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , CudaSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , CudaUVMSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , CudaHostPinnedSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< CudaHostPinnedSpace , HostSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};


template<class ExecutionSpace> struct DeepCopy< HostSpace , CudaUVMSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , CudaSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

template<class ExecutionSpace> struct DeepCopy< HostSpace , CudaHostPinnedSpace , ExecutionSpace >
{
  inline
  DeepCopy( void * dst , const void * src , size_t n )
  { (void) DeepCopy< HostSpace , HostSpace , Cuda >( dst , src , n ); }

  inline
  DeepCopy( const ExecutionSpace& exec, void * dst , const void * src , size_t n )
  {
    exec.fence();
    DeepCopyAsyncCuda (dst,src,n);
  }
};

} // namespace Impl
} // namespace Kokkos

//----------------------------------------------------------------------------
//----------------------------------------------------------------------------

namespace Kokkos {
namespace Impl {

/** Running in CudaSpace attempting to access HostSpace: error */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::CudaSpace , Kokkos::HostSpace >
{
  enum { value = false };
  KOKKOS_INLINE_FUNCTION static void verify( void )
    { Kokkos::abort("Cuda code attempted to access HostSpace memory"); }

  KOKKOS_INLINE_FUNCTION static void verify( const void * )
    { Kokkos::abort("Cuda code attempted to access HostSpace memory"); }
};

/** Running in CudaSpace accessing CudaUVMSpace: ok */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::CudaSpace , Kokkos::CudaUVMSpace >
{
  enum { value = true };
  KOKKOS_INLINE_FUNCTION static void verify( void ) { }
  KOKKOS_INLINE_FUNCTION static void verify( const void * ) { }
};

/** Running in CudaSpace accessing CudaHostPinnedSpace: ok */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::CudaSpace , Kokkos::CudaHostPinnedSpace >
{
  enum { value = true };
  KOKKOS_INLINE_FUNCTION static void verify( void ) { }
  KOKKOS_INLINE_FUNCTION static void verify( const void * ) { }
};

/** Running in CudaSpace attempting to access an unknown space: error */
template< class OtherSpace >
struct VerifyExecutionCanAccessMemorySpace<
  typename enable_if< ! is_same<Kokkos::CudaSpace,OtherSpace>::value , Kokkos::CudaSpace >::type ,
  OtherSpace >
{
  enum { value = false };
  KOKKOS_INLINE_FUNCTION static void verify( void )
    { Kokkos::abort("Cuda code attempted to access unknown Space memory"); }

  KOKKOS_INLINE_FUNCTION static void verify( const void * )
    { Kokkos::abort("Cuda code attempted to access unknown Space memory"); }
};

//----------------------------------------------------------------------------
/** Running in HostSpace attempting to access CudaSpace */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::HostSpace , Kokkos::CudaSpace >
{
  enum { value = false };
  inline static void verify( void ) { CudaSpace::access_error(); }
  inline static void verify( const void * p ) { CudaSpace::access_error(p); }
};

/** Running in HostSpace accessing CudaUVMSpace is OK */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::HostSpace , Kokkos::CudaUVMSpace >
{
  enum { value = true };
  inline static void verify( void ) { }
  inline static void verify( const void * ) { }
};

/** Running in HostSpace accessing CudaHostPinnedSpace is OK */
template<>
struct VerifyExecutionCanAccessMemorySpace< Kokkos::HostSpace , Kokkos::CudaHostPinnedSpace >
{
  enum { value = true };
  KOKKOS_INLINE_FUNCTION static void verify( void ) {}
  KOKKOS_INLINE_FUNCTION static void verify( const void * ) {}
};

} // namespace Impl
} // namespace Kokkos

//----------------------------------------------------------------------------
//----------------------------------------------------------------------------

namespace Kokkos {
namespace Impl {

template<>
class SharedAllocationRecord< Kokkos::CudaSpace , void >
  : public SharedAllocationRecord< void , void >
{
private:

  friend class SharedAllocationRecord< Kokkos::CudaUVMSpace , void > ;

  typedef SharedAllocationRecord< void , void >  RecordBase ;

  SharedAllocationRecord( const SharedAllocationRecord & ) = delete ;
  SharedAllocationRecord & operator = ( const SharedAllocationRecord & ) = delete ;

  static void deallocate( RecordBase * );

  static ::cudaTextureObject_t
  attach_texture_object( const unsigned sizeof_alias
                       , void * const   alloc_ptr
                       , const size_t   alloc_size );

  static RecordBase s_root_record ;

  ::cudaTextureObject_t   m_tex_obj ;
  const Kokkos::CudaSpace m_space ;

protected:

  ~SharedAllocationRecord();
  SharedAllocationRecord() : RecordBase(), m_tex_obj(0), m_space() {}

  SharedAllocationRecord( const Kokkos::CudaSpace        & arg_space
                        , const std::string              & arg_label
                        , const size_t                     arg_alloc_size
                        , const RecordBase::function_type  arg_dealloc = & deallocate
                        );

public:

  std::string get_label() const ;

  static SharedAllocationRecord * allocate( const Kokkos::CudaSpace &  arg_space
                                          , const std::string       &  arg_label
                                          , const size_t               arg_alloc_size );

  /**\brief  Allocate tracked memory in the space */
  static
  void * allocate_tracked( const Kokkos::CudaSpace & arg_space
                         , const std::string & arg_label
                         , const size_t arg_alloc_size );

  /**\brief  Reallocate tracked memory in the space */
  static
  void * reallocate_tracked( void * const arg_alloc_ptr
                           , const size_t arg_alloc_size );

  /**\brief  Deallocate tracked memory in the space */
  static
  void deallocate_tracked( void * const arg_alloc_ptr );

  static SharedAllocationRecord * get_record( void * arg_alloc_ptr );

  template< typename AliasType >
  inline
  ::cudaTextureObject_t attach_texture_object()
    {
      static_assert( ( std::is_same< AliasType , int >::value ||
                       std::is_same< AliasType , ::int2 >::value ||
                       std::is_same< AliasType , ::int4 >::value )
                   , "Cuda texture fetch only supported for alias types of int, ::int2, or ::int4" );

      if ( m_tex_obj == 0 ) {
        m_tex_obj = attach_texture_object( sizeof(AliasType)
                                         , (void*) RecordBase::m_alloc_ptr
                                         , RecordBase::m_alloc_size );
      }

      return m_tex_obj ;
    }

  template< typename AliasType >
  inline
  int attach_texture_object_offset( const AliasType * const ptr )
    {
      // Texture object is attached to the entire allocation range
      return ptr - reinterpret_cast<AliasType*>( RecordBase::m_alloc_ptr );
    }

  static void print_records( std::ostream & , const Kokkos::CudaSpace & , bool detail = false );
};


template<>
class SharedAllocationRecord< Kokkos::CudaUVMSpace , void >
  : public SharedAllocationRecord< void , void >
{
private:

  typedef SharedAllocationRecord< void , void >  RecordBase ;

  SharedAllocationRecord( const SharedAllocationRecord & ) = delete ;
  SharedAllocationRecord & operator = ( const SharedAllocationRecord & ) = delete ;

  static void deallocate( RecordBase * );

  static RecordBase s_root_record ;

  ::cudaTextureObject_t      m_tex_obj ;
  const Kokkos::CudaUVMSpace m_space ;

protected:

  ~SharedAllocationRecord();
  SharedAllocationRecord() : RecordBase(), m_tex_obj(0), m_space() {}

  SharedAllocationRecord( const Kokkos::CudaUVMSpace     & arg_space
                        , const std::string              & arg_label
                        , const size_t                     arg_alloc_size
                        , const RecordBase::function_type  arg_dealloc = & deallocate
                        );

public:

  std::string get_label() const ;

  static SharedAllocationRecord * allocate( const Kokkos::CudaUVMSpace &  arg_space
                                          , const std::string          &  arg_label
                                          , const size_t                  arg_alloc_size
                                          );

  /**\brief  Allocate tracked memory in the space */
  static
  void * allocate_tracked( const Kokkos::CudaUVMSpace & arg_space
                         , const std::string & arg_label
                         , const size_t arg_alloc_size );

  /**\brief  Reallocate tracked memory in the space */
  static
  void * reallocate_tracked( void * const arg_alloc_ptr
                           , const size_t arg_alloc_size );

  /**\brief  Deallocate tracked memory in the space */
  static
  void deallocate_tracked( void * const arg_alloc_ptr );

  static SharedAllocationRecord * get_record( void * arg_alloc_ptr );


  template< typename AliasType >
  inline
  ::cudaTextureObject_t attach_texture_object()
    {
      static_assert( ( std::is_same< AliasType , int >::value ||
                       std::is_same< AliasType , ::int2 >::value ||
                       std::is_same< AliasType , ::int4 >::value )
                   , "Cuda texture fetch only supported for alias types of int, ::int2, or ::int4" );

      if ( m_tex_obj == 0 ) {
        m_tex_obj = SharedAllocationRecord< Kokkos::CudaSpace , void >::
          attach_texture_object( sizeof(AliasType)
                               , (void*) RecordBase::m_alloc_ptr
                               , RecordBase::m_alloc_size );
      }

      return m_tex_obj ;
    }

  template< typename AliasType >
  inline
  int attach_texture_object_offset( const AliasType * const ptr )
    {
      // Texture object is attached to the entire allocation range
      return ptr - reinterpret_cast<AliasType*>( RecordBase::m_alloc_ptr );
    }

  static void print_records( std::ostream & , const Kokkos::CudaUVMSpace & , bool detail = false );
};

template<>
class SharedAllocationRecord< Kokkos::CudaHostPinnedSpace , void >
  : public SharedAllocationRecord< void , void >
{
private:

  typedef SharedAllocationRecord< void , void >  RecordBase ;

  SharedAllocationRecord( const SharedAllocationRecord & ) = delete ;
  SharedAllocationRecord & operator = ( const SharedAllocationRecord & ) = delete ;

  static void deallocate( RecordBase * );

  static RecordBase s_root_record ;

  const Kokkos::CudaHostPinnedSpace m_space ;

protected:

  ~SharedAllocationRecord();
  SharedAllocationRecord() : RecordBase(), m_space() {}

  SharedAllocationRecord( const Kokkos::CudaHostPinnedSpace     & arg_space
                        , const std::string              & arg_label
                        , const size_t                     arg_alloc_size
                        , const RecordBase::function_type  arg_dealloc = & deallocate
                        );

public:

  std::string get_label() const ;

  static SharedAllocationRecord * allocate( const Kokkos::CudaHostPinnedSpace &  arg_space
                                          , const std::string          &  arg_label
                                          , const size_t                  arg_alloc_size
                                          );
  /**\brief  Allocate tracked memory in the space */
  static
  void * allocate_tracked( const Kokkos::CudaHostPinnedSpace & arg_space
                         , const std::string & arg_label
                         , const size_t arg_alloc_size );

  /**\brief  Reallocate tracked memory in the space */
  static
  void * reallocate_tracked( void * const arg_alloc_ptr
                           , const size_t arg_alloc_size );

  /**\brief  Deallocate tracked memory in the space */
  static
  void deallocate_tracked( void * const arg_alloc_ptr );


  static SharedAllocationRecord * get_record( void * arg_alloc_ptr );

  static void print_records( std::ostream & , const Kokkos::CudaHostPinnedSpace & , bool detail = false );
};

} // namespace Impl
} // namespace Kokkos

//----------------------------------------------------------------------------
//----------------------------------------------------------------------------

#endif /* #if defined( KOKKOS_ENABLE_CUDA ) */
#endif /* #define KOKKOS_CUDASPACE_HPP */