glib/docs/reference/glib/tmpl/memory_slices.sgml

<!-- ##### SECTION Title ##### -->
Memory Slices

<!-- ##### SECTION Short_Description ##### -->
efficient way to allocate groups of equal-sized chunks of memory

<!-- ##### SECTION Long_Description ##### -->
<para>
Memory slices provide a space-efficient and multi-processing scalable
way to allocate equal-sized pieces of memory, just like the original
#GMemChunks (from GLib &lt;= 2.8), while avoiding their excessive
memory-waste, scalability and performance problems.
</para>

<para>
To achieve these goals, the slice allocator uses a sophisticated, 
layered design that has been inspired by Bonwick's slab allocator
<footnote><para>
<ulink url="http://citeseer.ist.psu.edu/bonwick94slab.html">[Bonwick94]</ulink> Jeff Bonwick, The slab allocator: An object-caching kernel
memory allocator. USENIX 1994, and  
<ulink url="http://citeseer.ist.psu.edu/bonwick01magazines.html">[Bonwick01]</ulink> Bonwick and Jonathan Adams, Magazines and vmem: Extending the
slab allocator to many cpu's and arbitrary resources. USENIX 2001
</para></footnote>.
It uses posix_memalign() to optimize allocations of many equally-sized 
chunks, and has per-thread free lists (the so-called magazine layer) 
to quickly satisfy allocation requests of already known structure sizes. 
This is accompanied by extra caching logic to keep freed memory around 
for some time before returning it to the system. Memory that is unused 
due to alignment constraints is used for cache colorization (random 
distribution of chunk addresses) to improve CPU cache utilization. The 
caching layer of the slice allocator adapts itself to high lock contention 
to improve scalability.
</para>

<para>
The slice allocator can allocate blocks as small as two pointers, and
unlike malloc(), it does not reserve extra space per block. For large block 
sizes, g_slice_new() and g_slice_alloc() will automatically delegate to the
system malloc() implementation. For newly written code it is recommended
to use the new <literal>g_slice</literal> API instead of g_malloc() and 
friends, as long as objects are not resized during their lifetime and the 
object size used at allocation time is still available when freeing.
</para>

<example>
<title>Using the slice allocator</title>
<programlisting>
  gchar *mem[10000];
  gint i;

  /* Allocate 10000 blocks. */
  for (i = 0; i &lt; 10000; i++)
    {
      mem[i] = g_slice_alloc (50);

      /* Fill in the memory with some junk. */
      for (j = 0; j &lt; 50; j++)
	mem[i][j] = i * j;
    }

  /* Now free all of the blocks. */
  for (i = 0; i &lt; 10000; i++)
    {
      g_slice_free1 (50, mem[i]);
    }
</programlisting></example>

<example>
<title>Using the slice allocator with data structures</title>
<programlisting>
  GRealArray *array;

  /* Allocate one block, using the g_slice_new(<!-- -->) macro. */
  array = g_slice_new (GRealArray);

  /* We can now use array just like a normal pointer to a structure. */
  array->data            = NULL;
  array->len             = 0;
  array->alloc           = 0;
  array->zero_terminated = (zero_terminated ? 1 : 0);
  array->clear           = (clear ? 1 : 0);
  array->elt_size        = elt_size;

  /* We can free the block, so it can be reused. */
  g_slice_free (GRealArray, array);
</programlisting></example>

<!-- ##### SECTION See_Also ##### -->
<para>
</para>

<!-- ##### SECTION Stability_Level ##### -->


<!-- ##### SECTION Image ##### -->


<!-- ##### FUNCTION g_slice_alloc ##### -->
<para>
Allocates a block of memory from the slice allocator.
The block adress handed out can be expected to be aligned
to at least <literal>1 * sizeof (void*)</literal>,
though in general slices are 2 * sizeof (void*) bytes aligned,
if a malloc() fallback implementation is used instead,
the alignment may be reduced in a libc dependent fashion.
Note that the underlying slice allocation mechanism can
be changed with the <link linkend="G_SLICE">G_SLICE=always-malloc</link>
environment variable.
</para>

@block_size: the number of bytes to allocate
@Returns: a pointer to the allocated memory block
@Since: 2.10


<!-- ##### FUNCTION g_slice_alloc0 ##### -->
<para>
Allocates a block of memory via g_slice_alloc()
and initialize the returned memory to 0.
Note that the underlying slice allocation mechanism can
be changed with the <link linkend="G_SLICE">G_SLICE=always-malloc</link>
environment variable.
</para>

@block_size: the number of bytes to allocate
@Returns: a pointer to the allocated block
@Since: 2.10


<!-- ##### FUNCTION g_slice_copy ##### -->
<para>
Allocates a block of memory from the slice allocator and copies
@block_size bytes into it from @mem_block.
</para>

@block_size: the number of bytes to allocate
@mem_block: the memory to copy
@Returns: a pointer to the allocated memory block
@Since: 2.14


<!-- ##### FUNCTION g_slice_free1 ##### -->
<para>
Frees a block of memory. The memory must have been allocated via
g_slice_alloc() or g_slice_alloc0()
and the @block_size has to match the size specified upon allocation.
Note that the exact release behaviour can be changed with the
<link linkend="G_DEBUG">G_DEBUG=gc-friendly</link> environment variable,
also see <link linkend="G_SLICE">G_SLICE</link> for related debugging options.
</para>

@block_size: the size of the block
@mem_block: a pointer to the block to free
@Since: 2.10


<!-- ##### FUNCTION g_slice_free_chain_with_offset ##### -->
<para>
Frees a linked list of memory blocks of structure type @type.
The memory blocks must be equal-sized, allocated via
g_slice_alloc() or g_slice_alloc0()
and linked together by a @next pointer (similar to #GSList). The offset 
of the @next field in each block is passed as third argument.
Note that the exact release behaviour can be changed with the
<link linkend="G_DEBUG">G_DEBUG=gc-friendly</link> environment variable,
also see <link linkend="G_SLICE">G_SLICE</link> for related debugging options.
</para>

@block_size: the size of the blocks
@mem_chain:  a pointer to the first block of the chain
@next_offset: the offset of the @next field in the blocks
@Since: 2.10


<!-- ##### MACRO g_slice_new ##### -->
<para>
A convenience macro to allocate a block of memory from the slice allocator.
It calls g_slice_alloc() with <literal>sizeof (@type)</literal> and casts 
the returned pointer to a pointer of the given type, avoiding a type cast 
in the source code.
Note that the underlying slice allocation mechanism can
be changed with the <link linkend="G_SLICE">G_SLICE=always-malloc</link>
environment variable.
</para>

@type: the type to allocate, typically a structure name
@Returns: a pointer to the allocated block, cast to a pointer to @type.
@Since: 2.10


<!-- ##### MACRO g_slice_new0 ##### -->
<para>
A convenience macro to allocate a block of memory from the slice allocator
and set the memory to 0. It calls g_slice_alloc0() with 
<literal>sizeof (@type)</literal> and casts the returned pointer to a pointer 
of the given type, avoiding a type cast in the source code.
Note that the underlying slice allocation mechanism can
be changed with the <link linkend="G_SLICE">G_SLICE=always-malloc</link>
environment variable.
</para>

@type: the type to allocate, typically a structure name
@Returns: a pointer to the allocated block, cast to a pointer to @type.
@Since: 2.10


<!-- ##### MACRO g_slice_dup ##### -->
<para>
A convenience macro to duplicate a block of memory using the slice allocator.
It calls g_slice_copy() with <literal>sizeof (@type)</literal> and casts 
the returned pointer to a pointer of the given type, avoiding a type cast 
in the source code.
Note that the underlying slice allocation mechanism can
be changed with the <link linkend="G_SLICE">G_SLICE=always-malloc</link>
environment variable.
</para>

@type: the type to duplicate, typically a structure name
@mem: the memory to copy into the allocated block
@Returns: a pointer to the allocated block, cast to a pointer to @type.
@Since: 2.14


<!-- ##### MACRO g_slice_free ##### -->
<para>
A convenience macro to free a block of memory that has been allocated
from the slice allocator. It calls g_slice_free1() using 
<literal>sizeof (type)</literal> as the block size.
Note that the exact release behaviour can be changed with the
<link linkend="G_DEBUG">G_DEBUG=gc-friendly</link> environment variable,
also see <link linkend="G_SLICE">G_SLICE</link> for related debugging options.
</para>

@type: the type of the block to free, typically a structure name
@mem: a pointer to the block to free
@Since: 2.10


<!-- ##### MACRO g_slice_free_chain ##### -->
<para>
Frees a linked list of memory blocks of structure type @type.
The memory blocks must be equal-sized, allocated via
g_slice_alloc() or g_slice_alloc0() and linked together by a 
@next pointer (similar to #GSList). The name of the
@next field in @type is passed as third argument.
Note that the exact release behaviour can be changed with the
<link linkend="G_DEBUG">G_DEBUG=gc-friendly</link> environment variable,
also see <link linkend="G_SLICE">G_SLICE</link> for related debugging options.
</para>

@type:        the type of the @mem_chain blocks
@mem_chain:   a pointer to the first block of the chain
@next:        the field name of the next pointer in @type
@Since: 2.10