xref: /dpdk/doc/guides/prog_guide/env_abstraction_layer.rst (revision 42c3576d44dd48602aba0820b98c07e2c4278c0e)
1..  BSD LICENSE
2    Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
3    All rights reserved.
4
5    Redistribution and use in source and binary forms, with or without
6    modification, are permitted provided that the following conditions
7    are met:
8
9    * Redistributions of source code must retain the above copyright
10    notice, this list of conditions and the following disclaimer.
11    * Redistributions in binary form must reproduce the above copyright
12    notice, this list of conditions and the following disclaimer in
13    the documentation and/or other materials provided with the
14    distribution.
15    * Neither the name of Intel Corporation nor the names of its
16    contributors may be used to endorse or promote products derived
17    from this software without specific prior written permission.
18
19    THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20    "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21    LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22    A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23    OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24    SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25    LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26    DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27    THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28    (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29    OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30
31.. _Environment_Abstraction_Layer:
32
33Environment Abstraction Layer
34=============================
35
36The Environment Abstraction Layer (EAL) is responsible for gaining access to low-level resources such as hardware and memory space.
37It provides a generic interface that hides the environment specifics from the applications and libraries.
38It is the responsibility of the initialization routine to decide how to allocate these resources
39(that is, memory space, PCI devices, timers, consoles, and so on).
40
41Typical services expected from the EAL are:
42
43*   DPDK Loading and Launching:
44    The DPDK and its application are linked as a single application and must be loaded by some means.
45
46*   Core Affinity/Assignment Procedures:
47    The EAL provides mechanisms for assigning execution units to specific cores as well as creating execution instances.
48
49*   System Memory Reservation:
50    The EAL facilitates the reservation of different memory zones, for example, physical memory areas for device interactions.
51
52*   PCI Address Abstraction: The EAL provides an interface to access PCI address space.
53
54*   Trace and Debug Functions: Logs, dump_stack, panic and so on.
55
56*   Utility Functions: Spinlocks and atomic counters that are not provided in libc.
57
58*   CPU Feature Identification: Determine at runtime if a particular feature, for example, Intel® AVX is supported.
59    Determine if the current CPU supports the feature set that the binary was compiled for.
60
61*   Interrupt Handling: Interfaces to register/unregister callbacks to specific interrupt sources.
62
63*   Alarm Functions: Interfaces to set/remove callbacks to be run at a specific time.
64
65EAL in a Linux-userland Execution Environment
66---------------------------------------------
67
68In a Linux user space environment, the DPDK application runs as a user-space application using the pthread library.
69PCI information about devices and address space is discovered through the /sys kernel interface and through kernel modules such as uio_pci_generic, or igb_uio.
70Refer to the UIO: User-space drivers documentation in the Linux kernel. This memory is mmap'd in the application.
71
72The EAL performs physical memory allocation using mmap() in hugetlbfs (using huge page sizes to increase performance).
73This memory is exposed to DPDK service layers such as the :ref:`Mempool Library <Mempool_Library>`.
74
75At this point, the DPDK services layer will be initialized, then through pthread setaffinity calls,
76each execution unit will be assigned to a specific logical core to run as a user-level thread.
77
78The time reference is provided by the CPU Time-Stamp Counter (TSC) or by the HPET kernel API through a mmap() call.
79
80Initialization and Core Launching
81~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
82
83Part of the initialization is done by the start function of glibc.
84A check is also performed at initialization time to ensure that the micro architecture type chosen in the config file is supported by the CPU.
85Then, the main() function is called. The core initialization and launch is done in rte_eal_init() (see the API documentation).
86It consist of calls to the pthread library (more specifically, pthread_self(), pthread_create(), and pthread_setaffinity_np()).
87
88.. _figure_linuxapp_launch:
89
90.. figure:: img/linuxapp_launch.*
91
92   EAL Initialization in a Linux Application Environment
93
94
95.. note::
96
97    Initialization of objects, such as memory zones, rings, memory pools, lpm tables and hash tables,
98    should be done as part of the overall application initialization on the master lcore.
99    The creation and initialization functions for these objects are not multi-thread safe.
100    However, once initialized, the objects themselves can safely be used in multiple threads simultaneously.
101
102Shutdown and Cleanup
103~~~~~~~~~~~~~~~~~~~~
104
105During the initialization of EAL resources such as hugepage backed memory can be
106allocated by core components.  The memory allocated during ``rte_eal_init()``
107can be released by calling the ``rte_eal_cleanup()`` function. Refer to the
108API documentation for details.
109
110Multi-process Support
111~~~~~~~~~~~~~~~~~~~~~
112
113The Linuxapp EAL allows a multi-process as well as a multi-threaded (pthread) deployment model.
114See chapter
115:ref:`Multi-process Support <Multi-process_Support>` for more details.
116
117Memory Mapping Discovery and Memory Reservation
118~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
119
120The allocation of large contiguous physical memory is done using the hugetlbfs kernel filesystem.
121The EAL provides an API to reserve named memory zones in this contiguous memory.
122The physical address of the reserved memory for that memory zone is also returned to the user by the memory zone reservation API.
123
124.. note::
125
126    Memory reservations done using the APIs provided by rte_malloc are also backed by pages from the hugetlbfs filesystem.
127
128PCI Access
129~~~~~~~~~~
130
131The EAL uses the /sys/bus/pci utilities provided by the kernel to scan the content on the PCI bus.
132To access PCI memory, a kernel module called uio_pci_generic provides a /dev/uioX device file
133and resource files in /sys
134that can be mmap'd to obtain access to PCI address space from the application.
135The DPDK-specific igb_uio module can also be used for this. Both drivers use the uio kernel feature (userland driver).
136
137Per-lcore and Shared Variables
138~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
139
140.. note::
141
142    lcore refers to a logical execution unit of the processor, sometimes called a hardware *thread*.
143
144Shared variables are the default behavior.
145Per-lcore variables are implemented using *Thread Local Storage* (TLS) to provide per-thread local storage.
146
147Logs
148~~~~
149
150A logging API is provided by EAL.
151By default, in a Linux application, logs are sent to syslog and also to the console.
152However, the log function can be overridden by the user to use a different logging mechanism.
153
154Trace and Debug Functions
155^^^^^^^^^^^^^^^^^^^^^^^^^
156
157There are some debug functions to dump the stack in glibc.
158The rte_panic() function can voluntarily provoke a SIG_ABORT,
159which can trigger the generation of a core file, readable by gdb.
160
161CPU Feature Identification
162~~~~~~~~~~~~~~~~~~~~~~~~~~
163
164The EAL can query the CPU at runtime (using the rte_cpu_get_features() function) to determine which CPU features are available.
165
166User Space Interrupt Event
167~~~~~~~~~~~~~~~~~~~~~~~~~~
168
169+ User Space Interrupt and Alarm Handling in Host Thread
170
171The EAL creates a host thread to poll the UIO device file descriptors to detect the interrupts.
172Callbacks can be registered or unregistered by the EAL functions for a specific interrupt event
173and are called in the host thread asynchronously.
174The EAL also allows timed callbacks to be used in the same way as for NIC interrupts.
175
176.. note::
177
178    In DPDK PMD, the only interrupts handled by the dedicated host thread are those for link status change
179    (link up and link down notification) and for sudden device removal.
180
181
182+ RX Interrupt Event
183
184The receive and transmit routines provided by each PMD don't limit themselves to execute in polling thread mode.
185To ease the idle polling with tiny throughput, it's useful to pause the polling and wait until the wake-up event happens.
186The RX interrupt is the first choice to be such kind of wake-up event, but probably won't be the only one.
187
188EAL provides the event APIs for this event-driven thread mode.
189Taking linuxapp as an example, the implementation relies on epoll. Each thread can monitor an epoll instance
190in which all the wake-up events' file descriptors are added. The event file descriptors are created and mapped to
191the interrupt vectors according to the UIO/VFIO spec.
192From bsdapp's perspective, kqueue is the alternative way, but not implemented yet.
193
194EAL initializes the mapping between event file descriptors and interrupt vectors, while each device initializes the mapping
195between interrupt vectors and queues. In this way, EAL actually is unaware of the interrupt cause on the specific vector.
196The eth_dev driver takes responsibility to program the latter mapping.
197
198.. note::
199
200    Per queue RX interrupt event is only allowed in VFIO which supports multiple MSI-X vector. In UIO, the RX interrupt
201    together with other interrupt causes shares the same vector. In this case, when RX interrupt and LSC(link status change)
202    interrupt are both enabled(intr_conf.lsc == 1 && intr_conf.rxq == 1), only the former is capable.
203
204The RX interrupt are controlled/enabled/disabled by ethdev APIs - 'rte_eth_dev_rx_intr_*'. They return failure if the PMD
205hasn't support them yet. The intr_conf.rxq flag is used to turn on the capability of RX interrupt per device.
206
207+ Device Removal Event
208
209This event is triggered by a device being removed at a bus level. Its
210underlying resources may have been made unavailable (i.e. PCI mappings
211unmapped). The PMD must make sure that on such occurrence, the application can
212still safely use its callbacks.
213
214This event can be subscribed to in the same way one would subscribe to a link
215status change event. The execution context is thus the same, i.e. it is the
216dedicated interrupt host thread.
217
218Considering this, it is likely that an application would want to close a
219device having emitted a Device Removal Event. In such case, calling
220``rte_eth_dev_close()`` can trigger it to unregister its own Device Removal Event
221callback. Care must be taken not to close the device from the interrupt handler
222context. It is necessary to reschedule such closing operation.
223
224Blacklisting
225~~~~~~~~~~~~
226
227The EAL PCI device blacklist functionality can be used to mark certain NIC ports as blacklisted,
228so they are ignored by the DPDK.
229The ports to be blacklisted are identified using the PCIe* description (Domain:Bus:Device.Function).
230
231Misc Functions
232~~~~~~~~~~~~~~
233
234Locks and atomic operations are per-architecture (i686 and x86_64).
235
236Memory Segments and Memory Zones (memzone)
237------------------------------------------
238
239The mapping of physical memory is provided by this feature in the EAL.
240As physical memory can have gaps, the memory is described in a table of descriptors,
241and each descriptor (called rte_memseg ) describes a contiguous portion of memory.
242
243On top of this, the memzone allocator's role is to reserve contiguous portions of physical memory.
244These zones are identified by a unique name when the memory is reserved.
245
246The rte_memzone descriptors are also located in the configuration structure.
247This structure is accessed using rte_eal_get_configuration().
248The lookup (by name) of a memory zone returns a descriptor containing the physical address of the memory zone.
249
250Memory zones can be reserved with specific start address alignment by supplying the align parameter
251(by default, they are aligned to cache line size).
252The alignment value should be a power of two and not less than the cache line size (64 bytes).
253Memory zones can also be reserved from either 2 MB or 1 GB hugepages, provided that both are available on the system.
254
255
256Multiple pthread
257----------------
258
259DPDK usually pins one pthread per core to avoid the overhead of task switching.
260This allows for significant performance gains, but lacks flexibility and is not always efficient.
261
262Power management helps to improve the CPU efficiency by limiting the CPU runtime frequency.
263However, alternately it is possible to utilize the idle cycles available to take advantage of
264the full capability of the CPU.
265
266By taking advantage of cgroup, the CPU utilization quota can be simply assigned.
267This gives another way to improve the CPU efficiency, however, there is a prerequisite;
268DPDK must handle the context switching between multiple pthreads per core.
269
270For further flexibility, it is useful to set pthread affinity not only to a CPU but to a CPU set.
271
272EAL pthread and lcore Affinity
273~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
274
275The term "lcore" refers to an EAL thread, which is really a Linux/FreeBSD pthread.
276"EAL pthreads"  are created and managed by EAL and execute the tasks issued by *remote_launch*.
277In each EAL pthread, there is a TLS (Thread Local Storage) called *_lcore_id* for unique identification.
278As EAL pthreads usually bind 1:1 to the physical CPU, the *_lcore_id* is typically equal to the CPU ID.
279
280When using multiple pthreads, however, the binding is no longer always 1:1 between an EAL pthread and a specified physical CPU.
281The EAL pthread may have affinity to a CPU set, and as such the *_lcore_id* will not be the same as the CPU ID.
282For this reason, there is an EAL long option '--lcores' defined to assign the CPU affinity of lcores.
283For a specified lcore ID or ID group, the option allows setting the CPU set for that EAL pthread.
284
285The format pattern:
286	--lcores='<lcore_set>[@cpu_set][,<lcore_set>[@cpu_set],...]'
287
288'lcore_set' and 'cpu_set' can be a single number, range or a group.
289
290A number is a "digit([0-9]+)"; a range is "<number>-<number>"; a group is "(<number|range>[,<number|range>,...])".
291
292If a '\@cpu_set' value is not supplied, the value of 'cpu_set' will default to the value of 'lcore_set'.
293
294    ::
295
296    	For example, "--lcores='1,2@(5-7),(3-5)@(0,2),(0,6),7-8'" which means start 9 EAL thread;
297    	    lcore 0 runs on cpuset 0x41 (cpu 0,6);
298    	    lcore 1 runs on cpuset 0x2 (cpu 1);
299    	    lcore 2 runs on cpuset 0xe0 (cpu 5,6,7);
300    	    lcore 3,4,5 runs on cpuset 0x5 (cpu 0,2);
301    	    lcore 6 runs on cpuset 0x41 (cpu 0,6);
302    	    lcore 7 runs on cpuset 0x80 (cpu 7);
303    	    lcore 8 runs on cpuset 0x100 (cpu 8).
304
305Using this option, for each given lcore ID, the associated CPUs can be assigned.
306It's also compatible with the pattern of corelist('-l') option.
307
308non-EAL pthread support
309~~~~~~~~~~~~~~~~~~~~~~~
310
311It is possible to use the DPDK execution context with any user pthread (aka. Non-EAL pthreads).
312In a non-EAL pthread, the *_lcore_id* is always LCORE_ID_ANY which identifies that it is not an EAL thread with a valid, unique, *_lcore_id*.
313Some libraries will use an alternative unique ID (e.g. TID), some will not be impacted at all, and some will work but with limitations (e.g. timer and mempool libraries).
314
315All these impacts are mentioned in :ref:`known_issue_label` section.
316
317Public Thread API
318~~~~~~~~~~~~~~~~~
319
320There are two public APIs ``rte_thread_set_affinity()`` and ``rte_thread_get_affinity()`` introduced for threads.
321When they're used in any pthread context, the Thread Local Storage(TLS) will be set/get.
322
323Those TLS include *_cpuset* and *_socket_id*:
324
325*	*_cpuset* stores the CPUs bitmap to which the pthread is affinitized.
326
327*	*_socket_id* stores the NUMA node of the CPU set. If the CPUs in CPU set belong to different NUMA node, the *_socket_id* will be set to SOCKET_ID_ANY.
328
329
330.. _known_issue_label:
331
332Known Issues
333~~~~~~~~~~~~
334
335+ rte_mempool
336
337  The rte_mempool uses a per-lcore cache inside the mempool.
338  For non-EAL pthreads, ``rte_lcore_id()`` will not return a valid number.
339  So for now, when rte_mempool is used with non-EAL pthreads, the put/get operations will bypass the default mempool cache and there is a performance penalty because of this bypass.
340  Only user-owned external caches can be used in a non-EAL context in conjunction with ``rte_mempool_generic_put()`` and ``rte_mempool_generic_get()`` that accept an explicit cache parameter.
341
342+ rte_ring
343
344  rte_ring supports multi-producer enqueue and multi-consumer dequeue.
345  However, it is non-preemptive, this has a knock on effect of making rte_mempool non-preemptable.
346
347  .. note::
348
349    The "non-preemptive" constraint means:
350
351    - a pthread doing multi-producers enqueues on a given ring must not
352      be preempted by another pthread doing a multi-producer enqueue on
353      the same ring.
354    - a pthread doing multi-consumers dequeues on a given ring must not
355      be preempted by another pthread doing a multi-consumer dequeue on
356      the same ring.
357
358    Bypassing this constraint may cause the 2nd pthread to spin until the 1st one is scheduled again.
359    Moreover, if the 1st pthread is preempted by a context that has an higher priority, it may even cause a dead lock.
360
361  This does not mean it cannot be used, simply, there is a need to narrow down the situation when it is used by multi-pthread on the same core.
362
363  1. It CAN be used for any single-producer or single-consumer situation.
364
365  2. It MAY be used by multi-producer/consumer pthread whose scheduling policy are all SCHED_OTHER(cfs). User SHOULD be aware of the performance penalty before using it.
366
367  3. It MUST not be used by multi-producer/consumer pthreads, whose scheduling policies are SCHED_FIFO or SCHED_RR.
368
369+ rte_timer
370
371  Running  ``rte_timer_manager()`` on a non-EAL pthread is not allowed. However, resetting/stopping the timer from a non-EAL pthread is allowed.
372
373+ rte_log
374
375  In non-EAL pthreads, there is no per thread loglevel and logtype, global loglevels are used.
376
377+ misc
378
379  The debug statistics of rte_ring, rte_mempool and rte_timer are not supported in a non-EAL pthread.
380
381cgroup control
382~~~~~~~~~~~~~~
383
384The following is a simple example of cgroup control usage, there are two pthreads(t0 and t1) doing packet I/O on the same core ($CPU).
385We expect only 50% of CPU spend on packet IO.
386
387  .. code-block:: console
388
389    mkdir /sys/fs/cgroup/cpu/pkt_io
390    mkdir /sys/fs/cgroup/cpuset/pkt_io
391
392    echo $cpu > /sys/fs/cgroup/cpuset/cpuset.cpus
393
394    echo $t0 > /sys/fs/cgroup/cpu/pkt_io/tasks
395    echo $t0 > /sys/fs/cgroup/cpuset/pkt_io/tasks
396
397    echo $t1 > /sys/fs/cgroup/cpu/pkt_io/tasks
398    echo $t1 > /sys/fs/cgroup/cpuset/pkt_io/tasks
399
400    cd /sys/fs/cgroup/cpu/pkt_io
401    echo 100000 > pkt_io/cpu.cfs_period_us
402    echo  50000 > pkt_io/cpu.cfs_quota_us
403
404
405Malloc
406------
407
408The EAL provides a malloc API to allocate any-sized memory.
409
410The objective of this API is to provide malloc-like functions to allow
411allocation from hugepage memory and to facilitate application porting.
412The *DPDK API Reference* manual describes the available functions.
413
414Typically, these kinds of allocations should not be done in data plane
415processing because they are slower than pool-based allocation and make
416use of locks within the allocation and free paths.
417However, they can be used in configuration code.
418
419Refer to the rte_malloc() function description in the *DPDK API Reference*
420manual for more information.
421
422Cookies
423~~~~~~~
424
425When CONFIG_RTE_MALLOC_DEBUG is enabled, the allocated memory contains
426overwrite protection fields to help identify buffer overflows.
427
428Alignment and NUMA Constraints
429~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
430
431The rte_malloc() takes an align argument that can be used to request a memory
432area that is aligned on a multiple of this value (which must be a power of two).
433
434On systems with NUMA support, a call to the rte_malloc() function will return
435memory that has been allocated on the NUMA socket of the core which made the call.
436A set of APIs is also provided, to allow memory to be explicitly allocated on a
437NUMA socket directly, or by allocated on the NUMA socket where another core is
438located, in the case where the memory is to be used by a logical core other than
439on the one doing the memory allocation.
440
441Use Cases
442~~~~~~~~~
443
444This API is meant to be used by an application that requires malloc-like
445functions at initialization time.
446
447For allocating/freeing data at runtime, in the fast-path of an application,
448the memory pool library should be used instead.
449
450Internal Implementation
451~~~~~~~~~~~~~~~~~~~~~~~
452
453Data Structures
454^^^^^^^^^^^^^^^
455
456There are two data structure types used internally in the malloc library:
457
458*   struct malloc_heap - used to track free space on a per-socket basis
459
460*   struct malloc_elem - the basic element of allocation and free-space
461    tracking inside the library.
462
463Structure: malloc_heap
464""""""""""""""""""""""
465
466The malloc_heap structure is used to manage free space on a per-socket basis.
467Internally, there is one heap structure per NUMA node, which allows us to
468allocate memory to a thread based on the NUMA node on which this thread runs.
469While this does not guarantee that the memory will be used on that NUMA node,
470it is no worse than a scheme where the memory is always allocated on a fixed
471or random node.
472
473The key fields of the heap structure and their function are described below
474(see also diagram above):
475
476*   lock - the lock field is needed to synchronize access to the heap.
477    Given that the free space in the heap is tracked using a linked list,
478    we need a lock to prevent two threads manipulating the list at the same time.
479
480*   free_head - this points to the first element in the list of free nodes for
481    this malloc heap.
482
483.. note::
484
485    The malloc_heap structure does not keep track of in-use blocks of memory,
486    since these are never touched except when they are to be freed again -
487    at which point the pointer to the block is an input to the free() function.
488
489.. _figure_malloc_heap:
490
491.. figure:: img/malloc_heap.*
492
493   Example of a malloc heap and malloc elements within the malloc library
494
495
496.. _malloc_elem:
497
498Structure: malloc_elem
499""""""""""""""""""""""
500
501The malloc_elem structure is used as a generic header structure for various
502blocks of memory.
503It is used in three different ways - all shown in the diagram above:
504
505#.  As a header on a block of free or allocated memory - normal case
506
507#.  As a padding header inside a block of memory
508
509#.  As an end-of-memseg marker
510
511The most important fields in the structure and how they are used are described below.
512
513.. note::
514
515    If the usage of a particular field in one of the above three usages is not
516    described, the field can be assumed to have an undefined value in that
517    situation, for example, for padding headers only the "state" and "pad"
518    fields have valid values.
519
520*   heap - this pointer is a reference back to the heap structure from which
521    this block was allocated.
522    It is used for normal memory blocks when they are being freed, to add the
523    newly-freed block to the heap's free-list.
524
525*   prev - this pointer points to the header element/block in the memseg
526    immediately behind the current one. When freeing a block, this pointer is
527    used to reference the previous block to check if that block is also free.
528    If so, then the two free blocks are merged to form a single larger block.
529
530*   next_free - this pointer is used to chain the free-list of unallocated
531    memory blocks together.
532    It is only used in normal memory blocks; on ``malloc()`` to find a suitable
533    free block to allocate and on ``free()`` to add the newly freed element to
534    the free-list.
535
536*   state - This field can have one of three values: ``FREE``, ``BUSY`` or
537    ``PAD``.
538    The former two are to indicate the allocation state of a normal memory block
539    and the latter is to indicate that the element structure is a dummy structure
540    at the end of the start-of-block padding, i.e. where the start of the data
541    within a block is not at the start of the block itself, due to alignment
542    constraints.
543    In that case, the pad header is used to locate the actual malloc element
544    header for the block.
545    For the end-of-memseg structure, this is always a ``BUSY`` value, which
546    ensures that no element, on being freed, searches beyond the end of the
547    memseg for other blocks to merge with into a larger free area.
548
549*   pad - this holds the length of the padding present at the start of the block.
550    In the case of a normal block header, it is added to the address of the end
551    of the header to give the address of the start of the data area, i.e. the
552    value passed back to the application on a malloc.
553    Within a dummy header inside the padding, this same value is stored, and is
554    subtracted from the address of the dummy header to yield the address of the
555    actual block header.
556
557*   size - the size of the data block, including the header itself.
558    For end-of-memseg structures, this size is given as zero, though it is never
559    actually checked.
560    For normal blocks which are being freed, this size value is used in place of
561    a "next" pointer to identify the location of the next block of memory that
562    in the case of being ``FREE``, the two free blocks can be merged into one.
563
564Memory Allocation
565^^^^^^^^^^^^^^^^^
566
567On EAL initialization, all memsegs are setup as part of the malloc heap.
568This setup involves placing a dummy structure at the end with ``BUSY`` state,
569which may contain a sentinel value if ``CONFIG_RTE_MALLOC_DEBUG`` is enabled,
570and a proper :ref:`element header<malloc_elem>` with ``FREE`` at the start
571for each memseg.
572The ``FREE`` element is then added to the ``free_list`` for the malloc heap.
573
574When an application makes a call to a malloc-like function, the malloc function
575will first index the ``lcore_config`` structure for the calling thread, and
576determine the NUMA node of that thread.
577The NUMA node is used to index the array of ``malloc_heap`` structures which is
578passed as a parameter to the ``heap_alloc()`` function, along with the
579requested size, type, alignment and boundary parameters.
580
581The ``heap_alloc()`` function will scan the free_list of the heap, and attempt
582to find a free block suitable for storing data of the requested size, with the
583requested alignment and boundary constraints.
584
585When a suitable free element has been identified, the pointer to be returned
586to the user is calculated.
587The cache-line of memory immediately preceding this pointer is filled with a
588struct malloc_elem header.
589Because of alignment and boundary constraints, there could be free space at
590the start and/or end of the element, resulting in the following behavior:
591
592#. Check for trailing space.
593   If the trailing space is big enough, i.e. > 128 bytes, then the free element
594   is split.
595   If it is not, then we just ignore it (wasted space).
596
597#. Check for space at the start of the element.
598   If the space at the start is small, i.e. <=128 bytes, then a pad header is
599   used, and the remaining space is wasted.
600   If, however, the remaining space is greater, then the free element is split.
601
602The advantage of allocating the memory from the end of the existing element is
603that no adjustment of the free list needs to take place - the existing element
604on the free list just has its size pointer adjusted, and the following element
605has its "prev" pointer redirected to the newly created element.
606
607Freeing Memory
608^^^^^^^^^^^^^^
609
610To free an area of memory, the pointer to the start of the data area is passed
611to the free function.
612The size of the ``malloc_elem`` structure is subtracted from this pointer to get
613the element header for the block.
614If this header is of type ``PAD`` then the pad length is further subtracted from
615the pointer to get the proper element header for the entire block.
616
617From this element header, we get pointers to the heap from which the block was
618allocated and to where it must be freed, as well as the pointer to the previous
619element, and via the size field, we can calculate the pointer to the next element.
620These next and previous elements are then checked to see if they are also
621``FREE``, and if so, they are merged with the current element.
622This means that we can never have two ``FREE`` memory blocks adjacent to one
623another, as they are always merged into a single block.
624