ballocator_doc.txt

来自「xml大全 可读写调用率很高 xml大全 可读写调用率很高」· 文本 代码 · 共 375 行 · 第 1/2 页

			BITMAPPED ALLOCATOR
			===================

2004-03-11  Dhruv Matani  <[e-mail]>

---------------------------------------------------------------------

As this name suggests, this allocator uses a bit-map to keep track of
the used and unused memory locations for it's book-keeping purposes.

This allocator will make use of 1 single bit to keep track of whether
it has been allocated or not. A bit 1 indicates free, while 0
indicates allocated. This has been done so that you can easily check a
collection of bits for a free block. This kind of Bitmapped strategy
works best for single object allocations, and with the STL type
parameterized allocators, we do not need to choose any size for the
block which will be represented by a single bit. This will be the size
of the parameter around which the allocator has been
parameterized. Thus, close to optimal performance will result. Hence,
this should be used for node based containers which call the allocate
function with an argument of 1.

The bitmapped allocator's internal pool is exponentially
growing. Meaning that internally, the blocks acquired from the Free
List Store will double every time the bitmapped allocator runs out of
memory.

--------------------------------------------------------------------

The macro __GTHREADS decides whether to use Mutex Protection around
every allocation/deallocation.  The state of the macro is picked up
automatically from the gthr abstration layer.

----------------------------------------------------------------------

What is the Free List Store?
----------------------------

The Free List Store (referred to as FLS for the remaining part of this
document) is the Global memory pool that is shared by all instances of
the bitmapped allocator instantiated for any type. This maintains a
sorted order of all free memory blocks given back to it by the
bitmapped allocator, and is also responsible for giving memory to the
bitmapped allocator when it asks for more.

Internally, there is a Free List threshold which indicates the Maximum
number of free lists that the FLS can hold internally
(cache). Currently, this value is set at 64. So, if there are more
than 64 free lists coming in, then some of them will be given back to
the OS using operator delete so that at any given time the Free List's
size does not exceed 64 entries. This is done because a Binary Search
is used to locate an entry in a free list when a request for memory
comes along. Thus, the run-time complexity of the search would go up
given an increasing size, for 64 entries however, lg(64) == 6
comparisons are enough to locate the correct free list if it exists.

Suppose the free list size has reached it's threshold, then the
largest block from among those in the list and the new block will be
selected and given back to the OS. This is done because it reduces
external fragmentation, and allows the OS to use the larger blocks
later in an orderly fashion, possibly merging them later. Also, on
some systems, large blocks are obtained via calls to mmap, so giving
them back to free system resources becomes most important.

The function _S_should_i_give decides the policy that determines
whether the current block of memory should be given to the allocator
for the request that it has made. That's because we may not always
have exact fits for the memory size that the allocator requests. We do
this mainly to prevent external fragmentation at the cost of a little
internal fragmentation. Now, the value of this internal fragmentation
has to be decided by this function. I can see 3 possibilities right
now. Please add more as and when you find better strategies.

1. Equal size check. Return true only when the 2 blocks are of equal
   size.

2. Difference Threshold: Return true only when the _block_size is
   greater than or equal to the _required_size, and if the _BS is >
   _RS by a difference of less than some THRESHOLD value, then return
   true, else return false.  

3. Percentage Threshold. Return true only when the _block_size is
   greater than or equal to the _required_size, and if the _BS is >
   _RS by a percentage of less than some THRESHOLD value, then return
   true, else return false.

Currently, (3) is being used with a value of 36% Maximum wastage per
Super Block.

--------------------------------------------------------------------

1) What is a super block? Why is it needed?

   A super block is the block of memory acquired from the FLS from
   which the bitmap allocator carves out memory for single objects and
   satisfies the user's requests. These super blocks come in sizes that
   are powers of 2 and multiples of 32 (_Bits_Per_Block). Yes both at
   the same time! That's because the next super block acquired will be
   2 times the previous one, and also all super blocks have to be
   multiples of the _Bits_Per_Block value.

2) How does it interact with the free list store?

   The super block is contained in the FLS, and the FLS is responsible
   for getting / returning Super Bocks to and from the OS using
   operator new as defined by the C++ standard.

---------------------------------------------------------------------

How does the allocate function Work?
------------------------------------

The allocate function is specialized for single object allocation
ONLY. Thus, ONLY if n == 1, will the bitmap_allocator's specialized
algorithm be used. Otherwise, the request is satisfied directly by
calling operator new.

Suppose n == 1, then the allocator does the following:

1. Checks to see whether the a free block exists somewhere in a region
   of memory close to the last satisfied request. If so, then that
   block is marked as allocated in the bit map and given to the
   user. If not, then (2) is executed.

2. Is there a free block anywhere after the current block right upto
   the end of the memory that we have? If so, that block is found, and
   the same procedure is applied as above, and returned to the
   user. If not, then (3) is executed.

3. Is there any block in whatever region of memory that we own free?
   This is done by checking (a) The use count for each super block,
   and if that fails then (b) The individual bit-maps for each super
   block. Note: Here we are never touching any of the memory that the
   user will be given, and we are confining all memory accesses to a
   small region of memory! This helps reduce cache misses. If this
   succeeds then we apply the same procedure on that bit-map as (1),
   and return that block of memory to the user. However, if this
   process fails, then we resort to (4).

4. This process involves Refilling the internal exponentially growing
   memory pool. The said effect is achieved by calling _S_refill_pool
   which does the following:
	 (a). Gets more memory from the Global Free List of the
	      Required size.
	 (b). Adjusts the size for the next call to itself.
	 (c). Writes the appropriate headers in the bit-maps.
	 (d). Sets the use count for that super-block just allocated
	      to 0 (zero).
	 (e). All of the above accounts to maintaining the basic
	      invariant for the allocator. If the invariant is
	      maintained, we are sure that all is well.
   Now, the same process is applied on the newly acquired free blocks,
   which are dispatched accordingly.

Thus, you can clearly see that the allocate function is nothing but a
combination of the next-fit and first-fit algorithm optimized ONLY for
single object allocations.


-------------------------------------------------------------------------

How does the deallocate function work?
--------------------------------------

The deallocate function again is specialized for single objects ONLY.
For all n belonging to > 1, the operator delete is called without
further ado, and the deallocate function returns.

However for n == 1, a series of steps are performed:

1. We first need to locate that super-block which holds the memory
   location given to us by the user. For that purpose, we maintain a
   static variable _S_last_dealloc_index, which holds the index into
   the vector of block pairs which indicates the index of the last
   super-block from which memory was freed. We use this strategy in
   the hope that the user will deallocate memory in a region close to
   what he/she deallocated the last time around. If the check for
   belongs_to succeeds, then we determine the bit-map for the given
   pointer, and locate the index into that bit-map, and mark that bit
   as free by setting it.

2. If the _S_last_dealloc_index does not point to the memory block
   that we're looking for, then we do a linear search on the block
   stored in the vector of Block Pairs. This vector in code is called
   _S_mem_blocks. When the corresponding super-block is found, we
   apply the same procedure as we did for (1) to mark the block as
   free in the bit-map.