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 102Multi-process Support 103~~~~~~~~~~~~~~~~~~~~~ 104 105The Linuxapp EAL allows a multi-process as well as a multi-threaded (pthread) deployment model. 106See chapter 107:ref:`Multi-process Support <Multi-process_Support>` for more details. 108 109Memory Mapping Discovery and Memory Reservation 110~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 111 112The allocation of large contiguous physical memory is done using the hugetlbfs kernel filesystem. 113The EAL provides an API to reserve named memory zones in this contiguous memory. 114The physical address of the reserved memory for that memory zone is also returned to the user by the memory zone reservation API. 115 116.. note:: 117 118 Memory reservations done using the APIs provided by rte_malloc are also backed by pages from the hugetlbfs filesystem. 119 120PCI Access 121~~~~~~~~~~ 122 123The EAL uses the /sys/bus/pci utilities provided by the kernel to scan the content on the PCI bus. 124To access PCI memory, a kernel module called uio_pci_generic provides a /dev/uioX device file 125and resource files in /sys 126that can be mmap'd to obtain access to PCI address space from the application. 127The DPDK-specific igb_uio module can also be used for this. Both drivers use the uio kernel feature (userland driver). 128 129Per-lcore and Shared Variables 130~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 131 132.. note:: 133 134 lcore refers to a logical execution unit of the processor, sometimes called a hardware *thread*. 135 136Shared variables are the default behavior. 137Per-lcore variables are implemented using *Thread Local Storage* (TLS) to provide per-thread local storage. 138 139Logs 140~~~~ 141 142A logging API is provided by EAL. 143By default, in a Linux application, logs are sent to syslog and also to the console. 144However, the log function can be overridden by the user to use a different logging mechanism. 145 146Trace and Debug Functions 147^^^^^^^^^^^^^^^^^^^^^^^^^ 148 149There are some debug functions to dump the stack in glibc. 150The rte_panic() function can voluntarily provoke a SIG_ABORT, 151which can trigger the generation of a core file, readable by gdb. 152 153CPU Feature Identification 154~~~~~~~~~~~~~~~~~~~~~~~~~~ 155 156The EAL can query the CPU at runtime (using the rte_cpu_get_features() function) to determine which CPU features are available. 157 158User Space Interrupt Event 159~~~~~~~~~~~~~~~~~~~~~~~~~~ 160 161+ User Space Interrupt and Alarm Handling in Host Thread 162 163The EAL creates a host thread to poll the UIO device file descriptors to detect the interrupts. 164Callbacks can be registered or unregistered by the EAL functions for a specific interrupt event 165and are called in the host thread asynchronously. 166The EAL also allows timed callbacks to be used in the same way as for NIC interrupts. 167 168.. note:: 169 170 In DPDK PMD, the only interrupts handled by the dedicated host thread are those for link status change 171 (link up and link down notification) and for sudden device removal. 172 173 174+ RX Interrupt Event 175 176The receive and transmit routines provided by each PMD don't limit themselves to execute in polling thread mode. 177To ease the idle polling with tiny throughput, it's useful to pause the polling and wait until the wake-up event happens. 178The RX interrupt is the first choice to be such kind of wake-up event, but probably won't be the only one. 179 180EAL provides the event APIs for this event-driven thread mode. 181Taking linuxapp as an example, the implementation relies on epoll. Each thread can monitor an epoll instance 182in which all the wake-up events' file descriptors are added. The event file descriptors are created and mapped to 183the interrupt vectors according to the UIO/VFIO spec. 184From bsdapp's perspective, kqueue is the alternative way, but not implemented yet. 185 186EAL initializes the mapping between event file descriptors and interrupt vectors, while each device initializes the mapping 187between interrupt vectors and queues. In this way, EAL actually is unaware of the interrupt cause on the specific vector. 188The eth_dev driver takes responsibility to program the latter mapping. 189 190.. note:: 191 192 Per queue RX interrupt event is only allowed in VFIO which supports multiple MSI-X vector. In UIO, the RX interrupt 193 together with other interrupt causes shares the same vector. In this case, when RX interrupt and LSC(link status change) 194 interrupt are both enabled(intr_conf.lsc == 1 && intr_conf.rxq == 1), only the former is capable. 195 196The RX interrupt are controlled/enabled/disabled by ethdev APIs - 'rte_eth_dev_rx_intr_*'. They return failure if the PMD 197hasn't support them yet. The intr_conf.rxq flag is used to turn on the capability of RX interrupt per device. 198 199+ Device Removal Event 200 201This event is triggered by a device being removed at a bus level. Its 202underlying resources may have been made unavailable (i.e. PCI mappings 203unmapped). The PMD must make sure that on such occurrence, the application can 204still safely use its callbacks. 205 206This event can be subscribed to in the same way one would subscribe to a link 207status change event. The execution context is thus the same, i.e. it is the 208dedicated interrupt host thread. 209 210Considering this, it is likely that an application would want to close a 211device having emitted a Device Removal Event. In such case, calling 212``rte_eth_dev_close()`` can trigger it to unregister its own Device Removal Event 213callback. Care must be taken not to close the device from the interrupt handler 214context. It is necessary to reschedule such closing operation. 215 216Blacklisting 217~~~~~~~~~~~~ 218 219The EAL PCI device blacklist functionality can be used to mark certain NIC ports as blacklisted, 220so they are ignored by the DPDK. 221The ports to be blacklisted are identified using the PCIe* description (Domain:Bus:Device.Function). 222 223Misc Functions 224~~~~~~~~~~~~~~ 225 226Locks and atomic operations are per-architecture (i686 and x86_64). 227 228Memory Segments and Memory Zones (memzone) 229------------------------------------------ 230 231The mapping of physical memory is provided by this feature in the EAL. 232As physical memory can have gaps, the memory is described in a table of descriptors, 233and each descriptor (called rte_memseg ) describes a contiguous portion of memory. 234 235On top of this, the memzone allocator's role is to reserve contiguous portions of physical memory. 236These zones are identified by a unique name when the memory is reserved. 237 238The rte_memzone descriptors are also located in the configuration structure. 239This structure is accessed using rte_eal_get_configuration(). 240The lookup (by name) of a memory zone returns a descriptor containing the physical address of the memory zone. 241 242Memory zones can be reserved with specific start address alignment by supplying the align parameter 243(by default, they are aligned to cache line size). 244The alignment value should be a power of two and not less than the cache line size (64 bytes). 245Memory zones can also be reserved from either 2 MB or 1 GB hugepages, provided that both are available on the system. 246 247 248Multiple pthread 249---------------- 250 251DPDK usually pins one pthread per core to avoid the overhead of task switching. 252This allows for significant performance gains, but lacks flexibility and is not always efficient. 253 254Power management helps to improve the CPU efficiency by limiting the CPU runtime frequency. 255However, alternately it is possible to utilize the idle cycles available to take advantage of 256the full capability of the CPU. 257 258By taking advantage of cgroup, the CPU utilization quota can be simply assigned. 259This gives another way to improve the CPU efficiency, however, there is a prerequisite; 260DPDK must handle the context switching between multiple pthreads per core. 261 262For further flexibility, it is useful to set pthread affinity not only to a CPU but to a CPU set. 263 264EAL pthread and lcore Affinity 265~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 266 267The term "lcore" refers to an EAL thread, which is really a Linux/FreeBSD pthread. 268"EAL pthreads" are created and managed by EAL and execute the tasks issued by *remote_launch*. 269In each EAL pthread, there is a TLS (Thread Local Storage) called *_lcore_id* for unique identification. 270As EAL pthreads usually bind 1:1 to the physical CPU, the *_lcore_id* is typically equal to the CPU ID. 271 272When using multiple pthreads, however, the binding is no longer always 1:1 between an EAL pthread and a specified physical CPU. 273The EAL pthread may have affinity to a CPU set, and as such the *_lcore_id* will not be the same as the CPU ID. 274For this reason, there is an EAL long option '--lcores' defined to assign the CPU affinity of lcores. 275For a specified lcore ID or ID group, the option allows setting the CPU set for that EAL pthread. 276 277The format pattern: 278 --lcores='<lcore_set>[@cpu_set][,<lcore_set>[@cpu_set],...]' 279 280'lcore_set' and 'cpu_set' can be a single number, range or a group. 281 282A number is a "digit([0-9]+)"; a range is "<number>-<number>"; a group is "(<number|range>[,<number|range>,...])". 283 284If a '\@cpu_set' value is not supplied, the value of 'cpu_set' will default to the value of 'lcore_set'. 285 286 :: 287 288 For example, "--lcores='1,2@(5-7),(3-5)@(0,2),(0,6),7-8'" which means start 9 EAL thread; 289 lcore 0 runs on cpuset 0x41 (cpu 0,6); 290 lcore 1 runs on cpuset 0x2 (cpu 1); 291 lcore 2 runs on cpuset 0xe0 (cpu 5,6,7); 292 lcore 3,4,5 runs on cpuset 0x5 (cpu 0,2); 293 lcore 6 runs on cpuset 0x41 (cpu 0,6); 294 lcore 7 runs on cpuset 0x80 (cpu 7); 295 lcore 8 runs on cpuset 0x100 (cpu 8). 296 297Using this option, for each given lcore ID, the associated CPUs can be assigned. 298It's also compatible with the pattern of corelist('-l') option. 299 300non-EAL pthread support 301~~~~~~~~~~~~~~~~~~~~~~~ 302 303It is possible to use the DPDK execution context with any user pthread (aka. Non-EAL pthreads). 304In 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*. 305Some 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). 306 307All these impacts are mentioned in :ref:`known_issue_label` section. 308 309Public Thread API 310~~~~~~~~~~~~~~~~~ 311 312There are two public APIs ``rte_thread_set_affinity()`` and ``rte_pthread_get_affinity()`` introduced for threads. 313When they're used in any pthread context, the Thread Local Storage(TLS) will be set/get. 314 315Those TLS include *_cpuset* and *_socket_id*: 316 317* *_cpuset* stores the CPUs bitmap to which the pthread is affinitized. 318 319* *_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. 320 321 322.. _known_issue_label: 323 324Known Issues 325~~~~~~~~~~~~ 326 327+ rte_mempool 328 329 The rte_mempool uses a per-lcore cache inside the mempool. 330 For non-EAL pthreads, ``rte_lcore_id()`` will not return a valid number. 331 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. 332 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. 333 334+ rte_ring 335 336 rte_ring supports multi-producer enqueue and multi-consumer dequeue. 337 However, it is non-preemptive, this has a knock on effect of making rte_mempool non-preemptable. 338 339 .. note:: 340 341 The "non-preemptive" constraint means: 342 343 - a pthread doing multi-producers enqueues on a given ring must not 344 be preempted by another pthread doing a multi-producer enqueue on 345 the same ring. 346 - a pthread doing multi-consumers dequeues on a given ring must not 347 be preempted by another pthread doing a multi-consumer dequeue on 348 the same ring. 349 350 Bypassing this constraint may cause the 2nd pthread to spin until the 1st one is scheduled again. 351 Moreover, if the 1st pthread is preempted by a context that has an higher priority, it may even cause a dead lock. 352 353 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. 354 355 1. It CAN be used for any single-producer or single-consumer situation. 356 357 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. 358 359 3. It MUST not be used by multi-producer/consumer pthreads, whose scheduling policies are SCHED_FIFO or SCHED_RR. 360 361+ rte_timer 362 363 Running ``rte_timer_manager()`` on a non-EAL pthread is not allowed. However, resetting/stopping the timer from a non-EAL pthread is allowed. 364 365+ rte_log 366 367 In non-EAL pthreads, there is no per thread loglevel and logtype, global loglevels are used. 368 369+ misc 370 371 The debug statistics of rte_ring, rte_mempool and rte_timer are not supported in a non-EAL pthread. 372 373cgroup control 374~~~~~~~~~~~~~~ 375 376The 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). 377We expect only 50% of CPU spend on packet IO. 378 379 .. code-block:: console 380 381 mkdir /sys/fs/cgroup/cpu/pkt_io 382 mkdir /sys/fs/cgroup/cpuset/pkt_io 383 384 echo $cpu > /sys/fs/cgroup/cpuset/cpuset.cpus 385 386 echo $t0 > /sys/fs/cgroup/cpu/pkt_io/tasks 387 echo $t0 > /sys/fs/cgroup/cpuset/pkt_io/tasks 388 389 echo $t1 > /sys/fs/cgroup/cpu/pkt_io/tasks 390 echo $t1 > /sys/fs/cgroup/cpuset/pkt_io/tasks 391 392 cd /sys/fs/cgroup/cpu/pkt_io 393 echo 100000 > pkt_io/cpu.cfs_period_us 394 echo 50000 > pkt_io/cpu.cfs_quota_us 395 396 397Malloc 398------ 399 400The EAL provides a malloc API to allocate any-sized memory. 401 402The objective of this API is to provide malloc-like functions to allow 403allocation from hugepage memory and to facilitate application porting. 404The *DPDK API Reference* manual describes the available functions. 405 406Typically, these kinds of allocations should not be done in data plane 407processing because they are slower than pool-based allocation and make 408use of locks within the allocation and free paths. 409However, they can be used in configuration code. 410 411Refer to the rte_malloc() function description in the *DPDK API Reference* 412manual for more information. 413 414Cookies 415~~~~~~~ 416 417When CONFIG_RTE_MALLOC_DEBUG is enabled, the allocated memory contains 418overwrite protection fields to help identify buffer overflows. 419 420Alignment and NUMA Constraints 421~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 422 423The rte_malloc() takes an align argument that can be used to request a memory 424area that is aligned on a multiple of this value (which must be a power of two). 425 426On systems with NUMA support, a call to the rte_malloc() function will return 427memory that has been allocated on the NUMA socket of the core which made the call. 428A set of APIs is also provided, to allow memory to be explicitly allocated on a 429NUMA socket directly, or by allocated on the NUMA socket where another core is 430located, in the case where the memory is to be used by a logical core other than 431on the one doing the memory allocation. 432 433Use Cases 434~~~~~~~~~ 435 436This API is meant to be used by an application that requires malloc-like 437functions at initialization time. 438 439For allocating/freeing data at runtime, in the fast-path of an application, 440the memory pool library should be used instead. 441 442Internal Implementation 443~~~~~~~~~~~~~~~~~~~~~~~ 444 445Data Structures 446^^^^^^^^^^^^^^^ 447 448There are two data structure types used internally in the malloc library: 449 450* struct malloc_heap - used to track free space on a per-socket basis 451 452* struct malloc_elem - the basic element of allocation and free-space 453 tracking inside the library. 454 455Structure: malloc_heap 456"""""""""""""""""""""" 457 458The malloc_heap structure is used to manage free space on a per-socket basis. 459Internally, there is one heap structure per NUMA node, which allows us to 460allocate memory to a thread based on the NUMA node on which this thread runs. 461While this does not guarantee that the memory will be used on that NUMA node, 462it is no worse than a scheme where the memory is always allocated on a fixed 463or random node. 464 465The key fields of the heap structure and their function are described below 466(see also diagram above): 467 468* lock - the lock field is needed to synchronize access to the heap. 469 Given that the free space in the heap is tracked using a linked list, 470 we need a lock to prevent two threads manipulating the list at the same time. 471 472* free_head - this points to the first element in the list of free nodes for 473 this malloc heap. 474 475.. note:: 476 477 The malloc_heap structure does not keep track of in-use blocks of memory, 478 since these are never touched except when they are to be freed again - 479 at which point the pointer to the block is an input to the free() function. 480 481.. _figure_malloc_heap: 482 483.. figure:: img/malloc_heap.* 484 485 Example of a malloc heap and malloc elements within the malloc library 486 487 488.. _malloc_elem: 489 490Structure: malloc_elem 491"""""""""""""""""""""" 492 493The malloc_elem structure is used as a generic header structure for various 494blocks of memory. 495It is used in three different ways - all shown in the diagram above: 496 497#. As a header on a block of free or allocated memory - normal case 498 499#. As a padding header inside a block of memory 500 501#. As an end-of-memseg marker 502 503The most important fields in the structure and how they are used are described below. 504 505.. note:: 506 507 If the usage of a particular field in one of the above three usages is not 508 described, the field can be assumed to have an undefined value in that 509 situation, for example, for padding headers only the "state" and "pad" 510 fields have valid values. 511 512* heap - this pointer is a reference back to the heap structure from which 513 this block was allocated. 514 It is used for normal memory blocks when they are being freed, to add the 515 newly-freed block to the heap's free-list. 516 517* prev - this pointer points to the header element/block in the memseg 518 immediately behind the current one. When freeing a block, this pointer is 519 used to reference the previous block to check if that block is also free. 520 If so, then the two free blocks are merged to form a single larger block. 521 522* next_free - this pointer is used to chain the free-list of unallocated 523 memory blocks together. 524 It is only used in normal memory blocks; on ``malloc()`` to find a suitable 525 free block to allocate and on ``free()`` to add the newly freed element to 526 the free-list. 527 528* state - This field can have one of three values: ``FREE``, ``BUSY`` or 529 ``PAD``. 530 The former two are to indicate the allocation state of a normal memory block 531 and the latter is to indicate that the element structure is a dummy structure 532 at the end of the start-of-block padding, i.e. where the start of the data 533 within a block is not at the start of the block itself, due to alignment 534 constraints. 535 In that case, the pad header is used to locate the actual malloc element 536 header for the block. 537 For the end-of-memseg structure, this is always a ``BUSY`` value, which 538 ensures that no element, on being freed, searches beyond the end of the 539 memseg for other blocks to merge with into a larger free area. 540 541* pad - this holds the length of the padding present at the start of the block. 542 In the case of a normal block header, it is added to the address of the end 543 of the header to give the address of the start of the data area, i.e. the 544 value passed back to the application on a malloc. 545 Within a dummy header inside the padding, this same value is stored, and is 546 subtracted from the address of the dummy header to yield the address of the 547 actual block header. 548 549* size - the size of the data block, including the header itself. 550 For end-of-memseg structures, this size is given as zero, though it is never 551 actually checked. 552 For normal blocks which are being freed, this size value is used in place of 553 a "next" pointer to identify the location of the next block of memory that 554 in the case of being ``FREE``, the two free blocks can be merged into one. 555 556Memory Allocation 557^^^^^^^^^^^^^^^^^ 558 559On EAL initialization, all memsegs are setup as part of the malloc heap. 560This setup involves placing a dummy structure at the end with ``BUSY`` state, 561which may contain a sentinel value if ``CONFIG_RTE_MALLOC_DEBUG`` is enabled, 562and a proper :ref:`element header<malloc_elem>` with ``FREE`` at the start 563for each memseg. 564The ``FREE`` element is then added to the ``free_list`` for the malloc heap. 565 566When an application makes a call to a malloc-like function, the malloc function 567will first index the ``lcore_config`` structure for the calling thread, and 568determine the NUMA node of that thread. 569The NUMA node is used to index the array of ``malloc_heap`` structures which is 570passed as a parameter to the ``heap_alloc()`` function, along with the 571requested size, type, alignment and boundary parameters. 572 573The ``heap_alloc()`` function will scan the free_list of the heap, and attempt 574to find a free block suitable for storing data of the requested size, with the 575requested alignment and boundary constraints. 576 577When a suitable free element has been identified, the pointer to be returned 578to the user is calculated. 579The cache-line of memory immediately preceding this pointer is filled with a 580struct malloc_elem header. 581Because of alignment and boundary constraints, there could be free space at 582the start and/or end of the element, resulting in the following behavior: 583 584#. Check for trailing space. 585 If the trailing space is big enough, i.e. > 128 bytes, then the free element 586 is split. 587 If it is not, then we just ignore it (wasted space). 588 589#. Check for space at the start of the element. 590 If the space at the start is small, i.e. <=128 bytes, then a pad header is 591 used, and the remaining space is wasted. 592 If, however, the remaining space is greater, then the free element is split. 593 594The advantage of allocating the memory from the end of the existing element is 595that no adjustment of the free list needs to take place - the existing element 596on the free list just has its size pointer adjusted, and the following element 597has its "prev" pointer redirected to the newly created element. 598 599Freeing Memory 600^^^^^^^^^^^^^^ 601 602To free an area of memory, the pointer to the start of the data area is passed 603to the free function. 604The size of the ``malloc_elem`` structure is subtracted from this pointer to get 605the element header for the block. 606If this header is of type ``PAD`` then the pad length is further subtracted from 607the pointer to get the proper element header for the entire block. 608 609From this element header, we get pointers to the heap from which the block was 610allocated and to where it must be freed, as well as the pointer to the previous 611element, and via the size field, we can calculate the pointer to the next element. 612These next and previous elements are then checked to see if they are also 613``FREE``, and if so, they are merged with the current element. 614This means that we can never have two ``FREE`` memory blocks adjacent to one 615another, as they are always merged into a single block. 616