summaryrefslogtreecommitdiff
path: root/boost/interprocess/mem_algo/detail/mem_algo_common.hpp
blob: 87293d18faf1927301996c09b4fcededc22d2222 (plain)
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
//////////////////////////////////////////////////////////////////////////////
//
// (C) Copyright Ion Gaztanaga 2005-2012. Distributed under the Boost
// Software License, Version 1.0. (See accompanying file
// LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
//
// See http://www.boost.org/libs/interprocess for documentation.
//
//////////////////////////////////////////////////////////////////////////////

#ifndef BOOST_INTERPROCESS_DETAIL_MEM_ALGO_COMMON_HPP
#define BOOST_INTERPROCESS_DETAIL_MEM_ALGO_COMMON_HPP

#ifndef BOOST_CONFIG_HPP
#  include <boost/config.hpp>
#endif
#
#if defined(BOOST_HAS_PRAGMA_ONCE)
#  pragma once
#endif

#include <boost/interprocess/detail/config_begin.hpp>
#include <boost/interprocess/detail/workaround.hpp>

// interprocess
#include <boost/interprocess/interprocess_fwd.hpp>
#include <boost/interprocess/containers/allocation_type.hpp>
// interprocess/detail
#include <boost/interprocess/detail/math_functions.hpp>
#include <boost/interprocess/detail/min_max.hpp>
#include <boost/interprocess/detail/type_traits.hpp>
#include <boost/interprocess/detail/utilities.hpp>
// container/detail
#include <boost/container/detail/multiallocation_chain.hpp>
#include <boost/container/detail/placement_new.hpp>
// move
#include <boost/move/utility_core.hpp>
// other boost
#include <boost/static_assert.hpp>
#include <boost/assert.hpp>

//!\file
//!Implements common operations for memory algorithms.

namespace boost {
namespace interprocess {
namespace ipcdetail {

template<class VoidPointer>
class basic_multiallocation_chain
   : public boost::container::container_detail::
      basic_multiallocation_chain<VoidPointer>
{
   BOOST_MOVABLE_BUT_NOT_COPYABLE(basic_multiallocation_chain)
   typedef boost::container::container_detail::
      basic_multiallocation_chain<VoidPointer> base_t;
   public:

   basic_multiallocation_chain()
      :  base_t()
   {}

   basic_multiallocation_chain(BOOST_RV_REF(basic_multiallocation_chain) other)
      :  base_t(::boost::move(static_cast<base_t&>(other)))
   {}

   basic_multiallocation_chain& operator=(BOOST_RV_REF(basic_multiallocation_chain) other)
   {
      this->base_t::operator=(::boost::move(static_cast<base_t&>(other)));
      return *this;
   }

   void *pop_front()
   {
      return boost::interprocess::ipcdetail::to_raw_pointer(this->base_t::pop_front());
   }
};


//!This class implements several allocation functions shared by different algorithms
//!(aligned allocation, multiple allocation...).
template<class MemoryAlgorithm>
class memory_algorithm_common
{
   public:
   typedef typename MemoryAlgorithm::void_pointer              void_pointer;
   typedef typename MemoryAlgorithm::block_ctrl                block_ctrl;
   typedef typename MemoryAlgorithm::multiallocation_chain     multiallocation_chain;
   typedef memory_algorithm_common<MemoryAlgorithm>            this_type;
   typedef typename MemoryAlgorithm::size_type                 size_type;

   static const size_type Alignment              = MemoryAlgorithm::Alignment;
   static const size_type MinBlockUnits          = MemoryAlgorithm::MinBlockUnits;
   static const size_type AllocatedCtrlBytes     = MemoryAlgorithm::AllocatedCtrlBytes;
   static const size_type AllocatedCtrlUnits     = MemoryAlgorithm::AllocatedCtrlUnits;
   static const size_type BlockCtrlBytes         = MemoryAlgorithm::BlockCtrlBytes;
   static const size_type BlockCtrlUnits         = MemoryAlgorithm::BlockCtrlUnits;
   static const size_type UsableByPreviousChunk  = MemoryAlgorithm::UsableByPreviousChunk;

   static void assert_alignment(const void *ptr)
   {  assert_alignment((std::size_t)ptr); }

   static void assert_alignment(size_type uint_ptr)
   {
      (void)uint_ptr;
      BOOST_ASSERT(uint_ptr % Alignment == 0);
   }

   static bool check_alignment(const void *ptr)
   {  return (((std::size_t)ptr) % Alignment == 0);   }

   static size_type ceil_units(size_type size)
   {  return get_rounded_size(size, Alignment)/Alignment; }

   static size_type floor_units(size_type size)
   {  return size/Alignment;  }

   static size_type multiple_of_units(size_type size)
   {  return get_rounded_size(size, Alignment);  }

   static void allocate_many
      (MemoryAlgorithm *memory_algo, size_type elem_bytes, size_type n_elements, multiallocation_chain &chain)
   {
      return this_type::priv_allocate_many(memory_algo, &elem_bytes, n_elements, 0, chain);
   }

   static void deallocate_many(MemoryAlgorithm *memory_algo, multiallocation_chain &chain)
   {
      return this_type::priv_deallocate_many(memory_algo, chain);
   }

   static bool calculate_lcm_and_needs_backwards_lcmed
      (size_type backwards_multiple, size_type received_size, size_type size_to_achieve,
      size_type &lcm_out, size_type &needs_backwards_lcmed_out)
   {
      // Now calculate lcm_val
      size_type max = backwards_multiple;
      size_type min = Alignment;
      size_type needs_backwards;
      size_type needs_backwards_lcmed;
      size_type lcm_val;
      size_type current_forward;
      //Swap if necessary
      if(max < min){
         size_type tmp = min;
         min = max;
         max = tmp;
      }
      //Check if it's power of two
      if((backwards_multiple & (backwards_multiple-1)) == 0){
         if(0 != (size_to_achieve & ((backwards_multiple-1)))){
            return false;
         }

         lcm_val = max;
         //If we want to use minbytes data to get a buffer between maxbytes
         //and minbytes if maxbytes can't be achieved, calculate the
         //biggest of all possibilities
         current_forward = get_truncated_size_po2(received_size, backwards_multiple);
         needs_backwards = size_to_achieve - current_forward;
         BOOST_ASSERT((needs_backwards % backwards_multiple) == 0);
         needs_backwards_lcmed = get_rounded_size_po2(needs_backwards, lcm_val);
         lcm_out = lcm_val;
         needs_backwards_lcmed_out = needs_backwards_lcmed;
         return true;
      }
      //Check if it's multiple of alignment
      else if((backwards_multiple & (Alignment - 1u)) == 0){
         lcm_val = backwards_multiple;
         current_forward = get_truncated_size(received_size, backwards_multiple);
         //No need to round needs_backwards because backwards_multiple == lcm_val
         needs_backwards_lcmed = needs_backwards = size_to_achieve - current_forward;
         BOOST_ASSERT((needs_backwards_lcmed & (Alignment - 1u)) == 0);
         lcm_out = lcm_val;
         needs_backwards_lcmed_out = needs_backwards_lcmed;
         return true;
      }
      //Check if it's multiple of the half of the alignmment
      else if((backwards_multiple & ((Alignment/2u) - 1u)) == 0){
         lcm_val = backwards_multiple*2u;
         current_forward = get_truncated_size(received_size, backwards_multiple);
         needs_backwards_lcmed = needs_backwards = size_to_achieve - current_forward;
         if(0 != (needs_backwards_lcmed & (Alignment-1)))
         //while(0 != (needs_backwards_lcmed & (Alignment-1)))
            needs_backwards_lcmed += backwards_multiple;
         BOOST_ASSERT((needs_backwards_lcmed % lcm_val) == 0);
         lcm_out = lcm_val;
         needs_backwards_lcmed_out = needs_backwards_lcmed;
         return true;
      }
      //Check if it's multiple of the quarter of the alignmment
      else if((backwards_multiple & ((Alignment/4u) - 1u)) == 0){
         size_type remainder;
         lcm_val = backwards_multiple*4u;
         current_forward = get_truncated_size(received_size, backwards_multiple);
         needs_backwards_lcmed = needs_backwards = size_to_achieve - current_forward;
         //while(0 != (needs_backwards_lcmed & (Alignment-1)))
            //needs_backwards_lcmed += backwards_multiple;
         if(0 != (remainder = ((needs_backwards_lcmed & (Alignment-1))>>(Alignment/8u)))){
            if(backwards_multiple & Alignment/2u){
               needs_backwards_lcmed += (remainder)*backwards_multiple;
            }
            else{
               needs_backwards_lcmed += (4-remainder)*backwards_multiple;
            }
         }
         BOOST_ASSERT((needs_backwards_lcmed % lcm_val) == 0);
         lcm_out = lcm_val;
         needs_backwards_lcmed_out = needs_backwards_lcmed;
         return true;
      }
      else{
         lcm_val = lcm(max, min);
      }
      //If we want to use minbytes data to get a buffer between maxbytes
      //and minbytes if maxbytes can't be achieved, calculate the
      //biggest of all possibilities
      current_forward = get_truncated_size(received_size, backwards_multiple);
      needs_backwards = size_to_achieve - current_forward;
      BOOST_ASSERT((needs_backwards % backwards_multiple) == 0);
      needs_backwards_lcmed = get_rounded_size(needs_backwards, lcm_val);
      lcm_out = lcm_val;
      needs_backwards_lcmed_out = needs_backwards_lcmed;
      return true;
   }

   static void allocate_many
      ( MemoryAlgorithm *memory_algo
      , const size_type *elem_sizes
      , size_type n_elements
      , size_type sizeof_element
      , multiallocation_chain &chain)
   {
      this_type::priv_allocate_many(memory_algo, elem_sizes, n_elements, sizeof_element, chain);
   }

   static void* allocate_aligned
      (MemoryAlgorithm *memory_algo, size_type nbytes, size_type alignment)
   {

      //Ensure power of 2
      if ((alignment & (alignment - size_type(1u))) != 0){
         //Alignment is not power of two
         BOOST_ASSERT((alignment & (alignment - size_type(1u))) == 0);
         return 0;
      }

      size_type real_size = nbytes;
      if(alignment <= Alignment){
         void *ignore_reuse = 0;
         return memory_algo->priv_allocate
            (boost::interprocess::allocate_new, nbytes, real_size, ignore_reuse);
      }

      if(nbytes > UsableByPreviousChunk)
         nbytes -= UsableByPreviousChunk;

      //We can find a aligned portion if we allocate a block that has alignment
      //nbytes + alignment bytes or more.
      size_type minimum_allocation = max_value
         (nbytes + alignment, size_type(MinBlockUnits*Alignment));
      //Since we will split that block, we must request a bit more memory
      //if the alignment is near the beginning of the buffer, because otherwise,
      //there is no space for a new block before the alignment.
      //
      //            ____ Aligned here
      //           |
      //  -----------------------------------------------------
      // | MBU |
      //  -----------------------------------------------------
      size_type request =
         minimum_allocation + (2*MinBlockUnits*Alignment - AllocatedCtrlBytes
         //prevsize - UsableByPreviousChunk
         );

      //Now allocate the buffer
      real_size = request;
      void *ignore_reuse = 0;
      void *buffer = memory_algo->priv_allocate(boost::interprocess::allocate_new, request, real_size, ignore_reuse);
      if(!buffer){
         return 0;
      }
      else if ((((std::size_t)(buffer)) % alignment) == 0){
         //If we are lucky and the buffer is aligned, just split it and
         //return the high part
         block_ctrl *first  = memory_algo->priv_get_block(buffer);
         size_type old_size = first->m_size;
         const size_type first_min_units =
            max_value(ceil_units(nbytes) + AllocatedCtrlUnits, size_type(MinBlockUnits));
         //We can create a new block in the end of the segment
         if(old_size >= (first_min_units + MinBlockUnits)){
            block_ctrl *second =  reinterpret_cast<block_ctrl *>
               (reinterpret_cast<char*>(first) + Alignment*first_min_units);
            first->m_size  = first_min_units;
            second->m_size = old_size - first->m_size;
            BOOST_ASSERT(second->m_size >= MinBlockUnits);
            memory_algo->priv_mark_new_allocated_block(first);
            memory_algo->priv_mark_new_allocated_block(second);
            memory_algo->priv_deallocate(memory_algo->priv_get_user_buffer(second));
         }
         return buffer;
      }

      //Buffer not aligned, find the aligned part.
      //
      //                    ____ Aligned here
      //                   |
      //  -----------------------------------------------------
      // | MBU +more | ACB |
      //  -----------------------------------------------------
      char *pos = reinterpret_cast<char*>
         (reinterpret_cast<std::size_t>(static_cast<char*>(buffer) +
            //This is the minimum size of (2)
            (MinBlockUnits*Alignment - AllocatedCtrlBytes) +
            //This is the next MBU for the aligned memory
            AllocatedCtrlBytes +
            //This is the alignment trick
            alignment - 1) & -alignment);

      //Now obtain the address of the blocks
      block_ctrl *first  = memory_algo->priv_get_block(buffer);
      block_ctrl *second = memory_algo->priv_get_block(pos);
      BOOST_ASSERT(pos <= (reinterpret_cast<char*>(first) + first->m_size*Alignment));
      BOOST_ASSERT(first->m_size >= 2*MinBlockUnits);
      BOOST_ASSERT((pos + MinBlockUnits*Alignment - AllocatedCtrlBytes + nbytes*Alignment/Alignment) <=
             (reinterpret_cast<char*>(first) + first->m_size*Alignment));
      //Set the new size of the first block
      size_type old_size = first->m_size;
      first->m_size  = (size_type)(reinterpret_cast<char*>(second) - reinterpret_cast<char*>(first))/Alignment;
      memory_algo->priv_mark_new_allocated_block(first);

      //Now check if we can create a new buffer in the end
      //
      //              __"second" block
      //             |      __Aligned here
      //             |     |      __"third" block
      //  -----------|-----|-----|------------------------------
      // | MBU +more | ACB | (3) | BCU |
      //  -----------------------------------------------------
      //This size will be the minimum size to be able to create a
      //new block in the end.
      const size_type second_min_units = max_value(size_type(MinBlockUnits),
                        ceil_units(nbytes) + AllocatedCtrlUnits );

      //Check if we can create a new block (of size MinBlockUnits) in the end of the segment
      if((old_size - first->m_size) >= (second_min_units + MinBlockUnits)){
         //Now obtain the address of the end block
         block_ctrl *third = new (reinterpret_cast<char*>(second) + Alignment*second_min_units)block_ctrl;
         second->m_size = second_min_units;
         third->m_size  = old_size - first->m_size - second->m_size;
         BOOST_ASSERT(third->m_size >= MinBlockUnits);
         memory_algo->priv_mark_new_allocated_block(second);
         memory_algo->priv_mark_new_allocated_block(third);
         memory_algo->priv_deallocate(memory_algo->priv_get_user_buffer(third));
      }
      else{
         second->m_size = old_size - first->m_size;
         BOOST_ASSERT(second->m_size >= MinBlockUnits);
         memory_algo->priv_mark_new_allocated_block(second);
      }

      memory_algo->priv_deallocate(memory_algo->priv_get_user_buffer(first));
      return memory_algo->priv_get_user_buffer(second);
   }

   static bool try_shrink
      (MemoryAlgorithm *memory_algo, void *ptr
      ,const size_type max_size, size_type &received_size)
   {
      size_type const preferred_size = received_size;
      (void)memory_algo;
      //Obtain the real block
      block_ctrl *block = memory_algo->priv_get_block(ptr);
      size_type old_block_units = (size_type)block->m_size;

      //The block must be marked as allocated
      BOOST_ASSERT(memory_algo->priv_is_allocated_block(block));

      //Check if alignment and block size are right
      assert_alignment(ptr);

      //Put this to a safe value
      received_size = (old_block_units - AllocatedCtrlUnits)*Alignment + UsableByPreviousChunk;

      //Now translate it to Alignment units
      const size_type max_user_units       = floor_units(max_size - UsableByPreviousChunk);
      const size_type preferred_user_units = ceil_units(preferred_size - UsableByPreviousChunk);

      //Check if rounded max and preferred are possible correct
      if(max_user_units < preferred_user_units)
         return false;

      //Check if the block is smaller than the requested minimum
      size_type old_user_units = old_block_units - AllocatedCtrlUnits;

      if(old_user_units < preferred_user_units)
         return false;

      //If the block is smaller than the requested minimum
      if(old_user_units == preferred_user_units)
         return true;

      size_type shrunk_user_units =
         ((BlockCtrlUnits - AllocatedCtrlUnits) >= preferred_user_units)
         ? (BlockCtrlUnits - AllocatedCtrlUnits)
         : preferred_user_units;

      //Some parameter checks
      if(max_user_units < shrunk_user_units)
         return false;

      //We must be able to create at least a new empty block
      if((old_user_units - shrunk_user_units) < BlockCtrlUnits ){
         return false;
      }

      //Update new size
      received_size = shrunk_user_units*Alignment + UsableByPreviousChunk;
      return true;
   }

   static bool shrink
      (MemoryAlgorithm *memory_algo, void *ptr
      ,const size_type max_size, size_type &received_size)
   {
      size_type const preferred_size = received_size;
      //Obtain the real block
      block_ctrl *block = memory_algo->priv_get_block(ptr);
      size_type old_block_units = (size_type)block->m_size;

      if(!try_shrink(memory_algo, ptr, max_size, received_size)){
         return false;
      }

      //Check if the old size was just the shrunk size (no splitting)
      if((old_block_units - AllocatedCtrlUnits) == ceil_units(preferred_size - UsableByPreviousChunk))
         return true;

      //Now we can just rewrite the size of the old buffer
      block->m_size = (received_size-UsableByPreviousChunk)/Alignment + AllocatedCtrlUnits;
      BOOST_ASSERT(block->m_size >= BlockCtrlUnits);

      //We create the new block
      block_ctrl *new_block = reinterpret_cast<block_ctrl*>
                  (reinterpret_cast<char*>(block) + block->m_size*Alignment);
      //Write control data to simulate this new block was previously allocated
      //and deallocate it
      new_block->m_size = old_block_units - block->m_size;
      BOOST_ASSERT(new_block->m_size >= BlockCtrlUnits);
      memory_algo->priv_mark_new_allocated_block(block);
      memory_algo->priv_mark_new_allocated_block(new_block);
      memory_algo->priv_deallocate(memory_algo->priv_get_user_buffer(new_block));
      return true;
   }

   private:
   static void priv_allocate_many
      ( MemoryAlgorithm *memory_algo
      , const size_type *elem_sizes
      , size_type n_elements
      , size_type sizeof_element
      , multiallocation_chain &chain)
   {
      //Note: sizeof_element == 0 indicates that we want to
      //allocate n_elements of the same size "*elem_sizes"

      //Calculate the total size of all requests
      size_type total_request_units = 0;
      size_type elem_units = 0;
      const size_type ptr_size_units = memory_algo->priv_get_total_units(sizeof(void_pointer));
      if(!sizeof_element){
         elem_units = memory_algo->priv_get_total_units(*elem_sizes);
         elem_units = ptr_size_units > elem_units ? ptr_size_units : elem_units;
         total_request_units = n_elements*elem_units;
      }
      else{
         for(size_type i = 0; i < n_elements; ++i){
            if(multiplication_overflows(elem_sizes[i], sizeof_element)){
               total_request_units = 0;
               break;
            }
            elem_units = memory_algo->priv_get_total_units(elem_sizes[i]*sizeof_element);
            elem_units = ptr_size_units > elem_units ? ptr_size_units : elem_units;
            if(sum_overflows(total_request_units, elem_units)){
               total_request_units = 0;
               break;
            }
            total_request_units += elem_units;
         }
      }

      if(total_request_units && !multiplication_overflows(total_request_units, Alignment)){
         size_type low_idx = 0;
         while(low_idx < n_elements){
            size_type total_bytes = total_request_units*Alignment - AllocatedCtrlBytes + UsableByPreviousChunk;
            size_type min_allocation = (!sizeof_element)
               ?  elem_units
               :  memory_algo->priv_get_total_units(elem_sizes[low_idx]*sizeof_element);
            min_allocation = min_allocation*Alignment - AllocatedCtrlBytes + UsableByPreviousChunk;

            size_type received_size = total_bytes;
            void *ignore_reuse = 0;
            void *ret = memory_algo->priv_allocate
               (boost::interprocess::allocate_new, min_allocation, received_size, ignore_reuse);
            if(!ret){
               break;
            }

            block_ctrl *block = memory_algo->priv_get_block(ret);
            size_type received_units = (size_type)block->m_size;
            char *block_address = reinterpret_cast<char*>(block);

            size_type total_used_units = 0;
            while(total_used_units < received_units){
               if(sizeof_element){
                  elem_units = memory_algo->priv_get_total_units(elem_sizes[low_idx]*sizeof_element);
                  elem_units = ptr_size_units > elem_units ? ptr_size_units : elem_units;
               }
               if(total_used_units + elem_units > received_units)
                  break;
               total_request_units -= elem_units;
               //This is the position where the new block must be created
               block_ctrl *new_block = reinterpret_cast<block_ctrl *>(block_address);
               assert_alignment(new_block);

               //The last block should take all the remaining space
               if((low_idx + 1) == n_elements ||
                  (total_used_units + elem_units +
                  ((!sizeof_element)
                     ? elem_units
               : max_value(memory_algo->priv_get_total_units(elem_sizes[low_idx+1]*sizeof_element), ptr_size_units))
                   > received_units)){
                  //By default, the new block will use the rest of the buffer
                  new_block->m_size = received_units - total_used_units;
                  memory_algo->priv_mark_new_allocated_block(new_block);

                  //If the remaining units are bigger than needed and we can
                  //split it obtaining a new free memory block do it.
                  if((received_units - total_used_units) >= (elem_units + MemoryAlgorithm::BlockCtrlUnits)){
                     size_type shrunk_request = elem_units*Alignment - AllocatedCtrlBytes + UsableByPreviousChunk;
                     size_type shrunk_received = shrunk_request;
                     bool shrink_ok = shrink
                           (memory_algo
                           ,memory_algo->priv_get_user_buffer(new_block)
                           ,shrunk_request
                           ,shrunk_received);
                     (void)shrink_ok;
                     //Shrink must always succeed with passed parameters
                     BOOST_ASSERT(shrink_ok);
                     //Some sanity checks
                     BOOST_ASSERT(shrunk_request == shrunk_received);
                     BOOST_ASSERT(elem_units == ((shrunk_request-UsableByPreviousChunk)/Alignment + AllocatedCtrlUnits));
                     //"new_block->m_size" must have been reduced to elem_units by "shrink"
                     BOOST_ASSERT(new_block->m_size == elem_units);
                     //Now update the total received units with the reduction
                     received_units = elem_units + total_used_units;
                  }
               }
               else{
                  new_block->m_size = elem_units;
                  memory_algo->priv_mark_new_allocated_block(new_block);
               }

               block_address += new_block->m_size*Alignment;
               total_used_units += (size_type)new_block->m_size;
               //Check we have enough room to overwrite the intrusive pointer
               BOOST_ASSERT((new_block->m_size*Alignment - AllocatedCtrlUnits) >= sizeof(void_pointer));
               void_pointer p = ::new(memory_algo->priv_get_user_buffer(new_block), boost_container_new_t())void_pointer(0);
               chain.push_back(p);
               ++low_idx;
            }
            //Sanity check
            BOOST_ASSERT(total_used_units == received_units);
         }

         if(low_idx != n_elements){
            priv_deallocate_many(memory_algo, chain);
         }
      }
   }

   static void priv_deallocate_many(MemoryAlgorithm *memory_algo, multiallocation_chain &chain)
   {
      while(!chain.empty()){
         memory_algo->priv_deallocate(to_raw_pointer(chain.pop_front()));
      }
   }
};

}  //namespace ipcdetail {
}  //namespace interprocess {
}  //namespace boost {

#include <boost/interprocess/detail/config_end.hpp>

#endif   //#ifndef BOOST_INTERPROCESS_DETAIL_MEM_ALGO_COMMON_HPP