1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2010-2014 Intel Corporation.
4 .. _Environment_Abstraction_Layer:
6 Environment Abstraction Layer
7 =============================
9 The Environment Abstraction Layer (EAL) is responsible for gaining access to low-level resources such as hardware and memory space.
10 It provides a generic interface that hides the environment specifics from the applications and libraries.
11 It is the responsibility of the initialization routine to decide how to allocate these resources
12 (that is, memory space, devices, timers, consoles, and so on).
14 Typical services expected from the EAL are:
16 * DPDK Loading and Launching:
17 The DPDK and its application are linked as a single application and must be loaded by some means.
19 * Core Affinity/Assignment Procedures:
20 The EAL provides mechanisms for assigning execution units to specific cores as well as creating execution instances.
22 * System Memory Reservation:
23 The EAL facilitates the reservation of different memory zones, for example, physical memory areas for device interactions.
25 * Trace and Debug Functions: Logs, dump_stack, panic and so on.
27 * Utility Functions: Spinlocks and atomic counters that are not provided in libc.
29 * CPU Feature Identification: Determine at runtime if a particular feature, for example, IntelĀ® AVX is supported.
30 Determine if the current CPU supports the feature set that the binary was compiled for.
32 * Interrupt Handling: Interfaces to register/unregister callbacks to specific interrupt sources.
34 * Alarm Functions: Interfaces to set/remove callbacks to be run at a specific time.
36 EAL in a Linux-userland Execution Environment
37 ---------------------------------------------
39 In a Linux user space environment, the DPDK application runs as a user-space application using the pthread library.
41 The EAL performs physical memory allocation using mmap() in hugetlbfs (using huge page sizes to increase performance).
42 This memory is exposed to DPDK service layers such as the :ref:`Mempool Library <Mempool_Library>`.
44 At this point, the DPDK services layer will be initialized, then through pthread setaffinity calls,
45 each execution unit will be assigned to a specific logical core to run as a user-level thread.
47 The time reference is provided by the CPU Time-Stamp Counter (TSC) or by the HPET kernel API through a mmap() call.
49 Initialization and Core Launching
50 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
52 Part of the initialization is done by the start function of glibc.
53 A check is also performed at initialization time to ensure that the micro architecture type chosen in the config file is supported by the CPU.
54 Then, the main() function is called. The core initialization and launch is done in rte_eal_init() (see the API documentation).
55 It consist of calls to the pthread library (more specifically, pthread_self(), pthread_create(), and pthread_setaffinity_np()).
57 .. _figure_linux_launch:
59 .. figure:: img/linuxapp_launch.*
61 EAL Initialization in a Linux Application Environment
66 Initialization of objects, such as memory zones, rings, memory pools, lpm tables and hash tables,
67 should be done as part of the overall application initialization on the main lcore.
68 The creation and initialization functions for these objects are not multi-thread safe.
69 However, once initialized, the objects themselves can safely be used in multiple threads simultaneously.
74 During the initialization of EAL resources such as hugepage backed memory can be
75 allocated by core components. The memory allocated during ``rte_eal_init()``
76 can be released by calling the ``rte_eal_cleanup()`` function. Refer to the
77 API documentation for details.
82 The Linux EAL allows a multi-process as well as a multi-threaded (pthread) deployment model.
84 :ref:`Multi-process Support <Multi-process_Support>` for more details.
86 Memory Mapping Discovery and Memory Reservation
87 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
89 The allocation of large contiguous physical memory is done using hugepages.
90 The EAL provides an API to reserve named memory zones in this contiguous memory.
91 The physical address of the reserved memory for that memory zone is also returned to the user by the memory zone reservation API.
93 There are two modes in which DPDK memory subsystem can operate: dynamic mode,
94 and legacy mode. Both modes are explained below.
98 Memory reservations done using the APIs provided by rte_malloc
99 are also backed by hugepages unless ``--no-huge`` option is given.
104 Currently, this mode is only supported on Linux and Windows.
106 In this mode, usage of hugepages by DPDK application will grow and shrink based
107 on application's requests. Any memory allocation through ``rte_malloc()``,
108 ``rte_memzone_reserve()`` or other methods, can potentially result in more
109 hugepages being reserved from the system. Similarly, any memory deallocation can
110 potentially result in hugepages being released back to the system.
112 Memory allocated in this mode is not guaranteed to be IOVA-contiguous. If large
113 chunks of IOVA-contiguous are required (with "large" defined as "more than one
114 page"), it is recommended to either use VFIO driver for all physical devices (so
115 that IOVA and VA addresses can be the same, thereby bypassing physical addresses
116 entirely), or use legacy memory mode.
118 For chunks of memory which must be IOVA-contiguous, it is recommended to use
119 ``rte_memzone_reserve()`` function with ``RTE_MEMZONE_IOVA_CONTIG`` flag
120 specified. This way, memory allocator will ensure that, whatever memory mode is
121 in use, either reserved memory will satisfy the requirements, or the allocation
124 There is no need to preallocate any memory at startup using ``-m`` or
125 ``--socket-mem`` command-line parameters, however it is still possible to do so,
126 in which case preallocate memory will be "pinned" (i.e. will never be released
127 by the application back to the system). It will be possible to allocate more
128 hugepages, and deallocate those, but any preallocated pages will not be freed.
129 If neither ``-m`` nor ``--socket-mem`` were specified, no memory will be
130 preallocated, and all memory will be allocated at runtime, as needed.
132 Another available option to use in dynamic memory mode is
133 ``--single-file-segments`` command-line option. This option will put pages in
134 single files (per memseg list), as opposed to creating a file per page. This is
135 normally not needed, but can be useful for use cases like userspace vhost, where
136 there is limited number of page file descriptors that can be passed to VirtIO.
138 If the application (or DPDK-internal code, such as device drivers) wishes to
139 receive notifications about newly allocated memory, it is possible to register
140 for memory event callbacks via ``rte_mem_event_callback_register()`` function.
141 This will call a callback function any time DPDK's memory map has changed.
143 If the application (or DPDK-internal code, such as device drivers) wishes to be
144 notified about memory allocations above specified threshold (and have a chance
145 to deny them), allocation validator callbacks are also available via
146 ``rte_mem_alloc_validator_callback_register()`` function.
148 A default validator callback is provided by EAL, which can be enabled with a
149 ``--socket-limit`` command-line option, for a simple way to limit maximum amount
150 of memory that can be used by DPDK application.
153 Memory subsystem uses DPDK IPC internally, so memory allocations/callbacks
154 and IPC must not be mixed: it is not safe to allocate/free memory inside
155 memory-related or IPC callbacks, and it is not safe to use IPC inside
156 memory-related callbacks. See chapter
157 :ref:`Multi-process Support <Multi-process_Support>` for more details about
163 This mode is enabled by specifying ``--legacy-mem`` command-line switch to the
164 EAL. This switch will have no effect on FreeBSD as FreeBSD only supports
167 This mode mimics historical behavior of EAL. That is, EAL will reserve all
168 memory at startup, sort all memory into large IOVA-contiguous chunks, and will
169 not allow acquiring or releasing hugepages from the system at runtime.
171 If neither ``-m`` nor ``--socket-mem`` were specified, the entire available
172 hugepage memory will be preallocated.
174 Hugepage Allocation Matching
175 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
177 This behavior is enabled by specifying the ``--match-allocations`` command-line
178 switch to the EAL. This switch is Linux-only and not supported with
179 ``--legacy-mem`` nor ``--no-huge``.
181 Some applications using memory event callbacks may require that hugepages be
182 freed exactly as they were allocated. These applications may also require
183 that any allocation from the malloc heap not span across allocations
184 associated with two different memory event callbacks. Hugepage allocation
185 matching can be used by these types of applications to satisfy both of these
186 requirements. This can result in some increased memory usage which is
187 very dependent on the memory allocation patterns of the application.
192 Additional restrictions are present when running in 32-bit mode. In dynamic
193 memory mode, by default maximum of 2 gigabytes of VA space will be preallocated,
194 and all of it will be on main lcore NUMA node unless ``--socket-mem`` flag is
197 In legacy mode, VA space will only be preallocated for segments that were
198 requested (plus padding, to keep IOVA-contiguousness).
200 Maximum Amount of Memory
201 ^^^^^^^^^^^^^^^^^^^^^^^^
203 All possible virtual memory space that can ever be used for hugepage mapping in
204 a DPDK process is preallocated at startup, thereby placing an upper limit on how
205 much memory a DPDK application can have. DPDK memory is stored in segment lists,
206 each segment is strictly one physical page. It is possible to change the amount
207 of virtual memory being preallocated at startup by editing the following config
210 * ``RTE_MAX_MEMSEG_LISTS`` controls how many segment lists can DPDK have
211 * ``RTE_MAX_MEM_MB_PER_LIST`` controls how much megabytes of memory each
212 segment list can address
213 * ``RTE_MAX_MEMSEG_PER_LIST`` controls how many segments each segment list
215 * ``RTE_MAX_MEMSEG_PER_TYPE`` controls how many segments each memory type
216 can have (where "type" is defined as "page size + NUMA node" combination)
217 * ``RTE_MAX_MEM_MB_PER_TYPE`` controls how much megabytes of memory each
218 memory type can address
219 * ``RTE_MAX_MEM_MB`` places a global maximum on the amount of memory
222 Normally, these options do not need to be changed.
226 Preallocated virtual memory is not to be confused with preallocated hugepage
227 memory! All DPDK processes preallocate virtual memory at startup. Hugepages
228 can later be mapped into that preallocated VA space (if dynamic memory mode
229 is enabled), and can optionally be mapped into it at startup.
234 Below is an overview of methods used for each OS to obtain hugepages,
235 explaining why certain limitations and options exist in EAL.
236 See the user guide for a specific OS for configuration details.
238 FreeBSD uses ``contigmem`` kernel module
239 to reserve a fixed number of hugepages at system start,
240 which are mapped by EAL at initialization using a specific ``sysctl()``.
242 Windows EAL allocates hugepages from the OS as needed using Win32 API,
243 so available amount depends on the system load.
244 It uses ``virt2phys`` kernel module to obtain physical addresses,
245 unless running in IOVA-as-VA mode (e.g. forced with ``--iova-mode=va``).
247 Linux allows to select any combination of the following:
249 * use files in hugetlbfs (the default)
250 or anonymous mappings (``--in-memory``);
251 * map each hugepage from its own file (the default)
252 or map multiple hugepages from one big file (``--single-file-segments``).
254 Mapping hugepages from files in hugetlbfs is essential for multi-process,
255 because secondary processes need to map the same hugepages.
256 EAL creates files like ``rtemap_0``
257 in directories specified with ``--huge-dir`` option
258 (or in the mount point for a specific hugepage size).
259 The ``rte`` prefix can be changed using ``--file-prefix``.
260 This may be needed for running multiple primary processes
261 that share a hugetlbfs mount point.
262 Each backing file by default corresponds to one hugepage,
263 it is opened and locked for the entire time the hugepage is used.
264 This may exhaust the number of open files limit (``NOFILE``).
265 See :ref:`segment-file-descriptors` section
266 on how the number of open backing file descriptors can be reduced.
268 In dynamic memory mode, EAL removes a backing hugepage file
269 when all pages mapped from it are freed back to the system.
270 However, backing files may persist after the application terminates
271 in case of a crash or a leak of DPDK memory (e.g. ``rte_free()`` is missing).
272 This reduces the number of hugepages available to other processes
273 as reported by ``/sys/kernel/mm/hugepages/hugepages-*/free_hugepages``.
274 EAL can remove the backing files after opening them for mapping
275 if ``--huge-unlink`` is given to avoid polluting hugetlbfs.
276 However, since it disables multi-process anyway,
277 using anonymous mapping (``--in-memory``) is recommended instead.
279 :ref:`EAL memory allocator <malloc>` relies on hugepages being zero-filled.
280 Hugepages are cleared by the kernel when a file in hugetlbfs or its part
281 is mapped for the first time system-wide
282 to prevent data leaks from previous users of the same hugepage.
283 EAL ensures this behavior by removing existing backing files at startup
284 and by recreating them before opening for mapping (as a precaution).
286 Anonymous mapping does not allow multi-process architecture.
287 This mode does not use hugetlbfs
288 and thus does not require root permissions for memory management
289 (the limit of locked memory amount, ``MEMLOCK``, still applies).
290 It is free of filename conflict and leftover file issues.
291 If ``memfd_create(2)`` is supported both at build and run time,
292 DPDK memory manager can provide file descriptors for memory segments,
293 which are required for VirtIO with vhost-user backend.
294 This can exhaust the number of open files limit (``NOFILE``)
295 despite not creating any visible files.
296 See :ref:`segment-file-descriptors` section
297 on how the number of open file descriptors used by EAL can be reduced.
299 .. _segment-file-descriptors:
301 Segment File Descriptors
302 ^^^^^^^^^^^^^^^^^^^^^^^^
304 On Linux, in most cases, EAL will store segment file descriptors in EAL. This
305 can become a problem when using smaller page sizes due to underlying limitations
306 of ``glibc`` library. For example, Linux API calls such as ``select()`` may not
307 work correctly because ``glibc`` does not support more than certain number of
310 There are two possible solutions for this problem. The recommended solution is
311 to use ``--single-file-segments`` mode, as that mode will not use a file
312 descriptor per each page, and it will keep compatibility with Virtio with
313 vhost-user backend. This option is not available when using ``--legacy-mem``
316 Another option is to use bigger page sizes. Since fewer pages are required to
317 cover the same memory area, fewer file descriptors will be stored internally
320 Support for Externally Allocated Memory
321 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
323 It is possible to use externally allocated memory in DPDK. There are two ways in
324 which using externally allocated memory can work: the malloc heap API's, and
325 manual memory management.
327 + Using heap API's for externally allocated memory
329 Using a set of malloc heap API's is the recommended way to use externally
330 allocated memory in DPDK. In this way, support for externally allocated memory
331 is implemented through overloading the socket ID - externally allocated heaps
332 will have socket ID's that would be considered invalid under normal
333 circumstances. Requesting an allocation to take place from a specified
334 externally allocated memory is a matter of supplying the correct socket ID to
335 DPDK allocator, either directly (e.g. through a call to ``rte_malloc``) or
336 indirectly (through data structure-specific allocation API's such as
337 ``rte_ring_create``). Using these API's also ensures that mapping of externally
338 allocated memory for DMA is also performed on any memory segment that is added
339 to a DPDK malloc heap.
341 Since there is no way DPDK can verify whether memory is available or valid, this
342 responsibility falls on the shoulders of the user. All multiprocess
343 synchronization is also user's responsibility, as well as ensuring that all
344 calls to add/attach/detach/remove memory are done in the correct order. It is
345 not required to attach to a memory area in all processes - only attach to memory
348 The expected workflow is as follows:
350 * Get a pointer to memory area
351 * Create a named heap
352 * Add memory area(s) to the heap
353 - If IOVA table is not specified, IOVA addresses will be assumed to be
354 unavailable, and DMA mappings will not be performed
355 - Other processes must attach to the memory area before they can use it
356 * Get socket ID used for the heap
357 * Use normal DPDK allocation procedures, using supplied socket ID
358 * If memory area is no longer needed, it can be removed from the heap
359 - Other processes must detach from this memory area before it can be removed
360 * If heap is no longer needed, remove it
361 - Socket ID will become invalid and will not be reused
363 For more information, please refer to ``rte_malloc`` API documentation,
364 specifically the ``rte_malloc_heap_*`` family of function calls.
366 + Using externally allocated memory without DPDK API's
368 While using heap API's is the recommended method of using externally allocated
369 memory in DPDK, there are certain use cases where the overhead of DPDK heap API
370 is undesirable - for example, when manual memory management is performed on an
371 externally allocated area. To support use cases where externally allocated
372 memory will not be used as part of normal DPDK workflow, there is also another
373 set of API's under the ``rte_extmem_*`` namespace.
375 These API's are (as their name implies) intended to allow registering or
376 unregistering externally allocated memory to/from DPDK's internal page table, to
377 allow API's like ``rte_mem_virt2memseg`` etc. to work with externally allocated
378 memory. Memory added this way will not be available for any regular DPDK
379 allocators; DPDK will leave this memory for the user application to manage.
381 The expected workflow is as follows:
383 * Get a pointer to memory area
384 * Register memory within DPDK
385 - If IOVA table is not specified, IOVA addresses will be assumed to be
387 - Other processes must attach to the memory area before they can use it
388 * Perform DMA mapping with ``rte_dev_dma_map`` if needed
389 * Use the memory area in your application
390 * If memory area is no longer needed, it can be unregistered
391 - If the area was mapped for DMA, unmapping must be performed before
393 - Other processes must detach from the memory area before it can be
396 Since these externally allocated memory areas will not be managed by DPDK, it is
397 therefore up to the user application to decide how to use them and what to do
398 with them once they're registered.
400 Per-lcore and Shared Variables
401 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
405 lcore refers to a logical execution unit of the processor, sometimes called a hardware *thread*.
407 Shared variables are the default behavior.
408 Per-lcore variables are implemented using *Thread Local Storage* (TLS) to provide per-thread local storage.
413 A logging API is provided by EAL.
414 By default, in a Linux application, logs are sent to syslog and also to the console.
415 However, the log function can be overridden by the user to use a different logging mechanism.
417 Trace and Debug Functions
418 ^^^^^^^^^^^^^^^^^^^^^^^^^
420 There are some debug functions to dump the stack in glibc.
421 The rte_panic() function can voluntarily provoke a SIG_ABORT,
422 which can trigger the generation of a core file, readable by gdb.
424 CPU Feature Identification
425 ~~~~~~~~~~~~~~~~~~~~~~~~~~
427 The EAL can query the CPU at runtime (using the rte_cpu_get_features() function) to determine which CPU features are available.
429 User Space Interrupt Event
430 ~~~~~~~~~~~~~~~~~~~~~~~~~~
432 + User Space Interrupt and Alarm Handling in Host Thread
434 The EAL creates a host thread to poll the UIO device file descriptors to detect the interrupts.
435 Callbacks can be registered or unregistered by the EAL functions for a specific interrupt event
436 and are called in the host thread asynchronously.
437 The EAL also allows timed callbacks to be used in the same way as for NIC interrupts.
441 In DPDK PMD, the only interrupts handled by the dedicated host thread are those for link status change
442 (link up and link down notification) and for sudden device removal.
447 The receive and transmit routines provided by each PMD don't limit themselves to execute in polling thread mode.
448 To ease the idle polling with tiny throughput, it's useful to pause the polling and wait until the wake-up event happens.
449 The RX interrupt is the first choice to be such kind of wake-up event, but probably won't be the only one.
451 EAL provides the event APIs for this event-driven thread mode.
452 Taking Linux as an example, the implementation relies on epoll. Each thread can monitor an epoll instance
453 in which all the wake-up events' file descriptors are added. The event file descriptors are created and mapped to
454 the interrupt vectors according to the UIO/VFIO spec.
455 From FreeBSD's perspective, kqueue is the alternative way, but not implemented yet.
457 EAL initializes the mapping between event file descriptors and interrupt vectors, while each device initializes the mapping
458 between interrupt vectors and queues. In this way, EAL actually is unaware of the interrupt cause on the specific vector.
459 The eth_dev driver takes responsibility to program the latter mapping.
463 Per queue RX interrupt event is only allowed in VFIO which supports multiple MSI-X vector. In UIO, the RX interrupt
464 together with other interrupt causes shares the same vector. In this case, when RX interrupt and LSC(link status change)
465 interrupt are both enabled(intr_conf.lsc == 1 && intr_conf.rxq == 1), only the former is capable.
467 The RX interrupt are controlled/enabled/disabled by ethdev APIs - 'rte_eth_dev_rx_intr_*'. They return failure if the PMD
468 hasn't support them yet. The intr_conf.rxq flag is used to turn on the capability of RX interrupt per device.
470 + Device Removal Event
472 This event is triggered by a device being removed at a bus level. Its
473 underlying resources may have been made unavailable (i.e. PCI mappings
474 unmapped). The PMD must make sure that on such occurrence, the application can
475 still safely use its callbacks.
477 This event can be subscribed to in the same way one would subscribe to a link
478 status change event. The execution context is thus the same, i.e. it is the
479 dedicated interrupt host thread.
481 Considering this, it is likely that an application would want to close a
482 device having emitted a Device Removal Event. In such case, calling
483 ``rte_eth_dev_close()`` can trigger it to unregister its own Device Removal Event
484 callback. Care must be taken not to close the device from the interrupt handler
485 context. It is necessary to reschedule such closing operation.
490 The EAL PCI device block list functionality can be used to mark certain NIC ports as unavailable,
491 so they are ignored by the DPDK.
492 The ports to be blocked are identified using the PCIe* description (Domain:Bus:Device.Function).
497 Locks and atomic operations are per-architecture (i686 and x86_64).
502 IOVA Mode is selected by considering what the current usable Devices on the
503 system require and/or support.
505 On FreeBSD, RTE_IOVA_PA is always the default. On Linux, the IOVA mode is
506 detected based on a 2-step heuristic detailed below.
508 For the first step, EAL asks each bus its requirement in terms of IOVA mode
509 and decides on a preferred IOVA mode.
511 - if all buses report RTE_IOVA_PA, then the preferred IOVA mode is RTE_IOVA_PA,
512 - if all buses report RTE_IOVA_VA, then the preferred IOVA mode is RTE_IOVA_VA,
513 - if all buses report RTE_IOVA_DC, no bus expressed a preference, then the
514 preferred mode is RTE_IOVA_DC,
515 - if the buses disagree (at least one wants RTE_IOVA_PA and at least one wants
516 RTE_IOVA_VA), then the preferred IOVA mode is RTE_IOVA_DC (see below with the
517 check on Physical Addresses availability),
519 If the buses have expressed no preference on which IOVA mode to pick, then a
520 default is selected using the following logic:
522 - if physical addresses are not available, RTE_IOVA_VA mode is used
523 - if /sys/kernel/iommu_groups is not empty, RTE_IOVA_VA mode is used
524 - otherwise, RTE_IOVA_PA mode is used
526 In the case when the buses had disagreed on their preferred IOVA mode, part of
527 the buses won't work because of this decision.
529 The second step checks if the preferred mode complies with the Physical
530 Addresses availability since those are only available to root user in recent
531 kernels. Namely, if the preferred mode is RTE_IOVA_PA but there is no access to
532 Physical Addresses, then EAL init fails early, since later probing of the
533 devices would fail anyway.
537 The RTE_IOVA_VA mode is preferred as the default in most cases for the
540 - All drivers are expected to work in RTE_IOVA_VA mode, irrespective of
541 physical address availability.
542 - By default, the mempool, first asks for IOVA-contiguous memory using
543 ``RTE_MEMZONE_IOVA_CONTIG``. This is slow in RTE_IOVA_PA mode and it may
544 affect the application boot time.
545 - It is easy to enable large amount of IOVA-contiguous memory use cases
546 with IOVA in VA mode.
548 It is expected that all PCI drivers work in both RTE_IOVA_PA and
551 If a PCI driver does not support RTE_IOVA_PA mode, the
552 ``RTE_PCI_DRV_NEED_IOVA_AS_VA`` flag is used to dictate that this PCI
553 driver can only work in RTE_IOVA_VA mode.
555 When the KNI kernel module is detected, RTE_IOVA_PA mode is preferred as a
556 performance penalty is expected in RTE_IOVA_VA mode.
558 IOVA Mode Configuration
559 ~~~~~~~~~~~~~~~~~~~~~~~
561 Auto detection of the IOVA mode, based on probing the bus and IOMMU configuration, may not report
562 the desired addressing mode when virtual devices that are not directly attached to the bus are present.
563 To facilitate forcing the IOVA mode to a specific value the EAL command line option ``--iova-mode`` can
564 be used to select either physical addressing('pa') or virtual addressing('va').
566 .. _max_simd_bitwidth:
572 The EAL provides a single setting to limit the max SIMD bitwidth used by DPDK,
573 which is used in determining the vector path, if any, chosen by a component.
574 The value can be set at runtime by an application using the
575 'rte_vect_set_max_simd_bitwidth(uint16_t bitwidth)' function,
576 which should only be called once at initialization, before EAL init.
577 The value can be overridden by the user using the EAL command-line option '--force-max-simd-bitwidth'.
579 When choosing a vector path, along with checking the CPU feature support,
580 the value of the max SIMD bitwidth must also be checked, and can be retrieved using the
581 'rte_vect_get_max_simd_bitwidth()' function.
582 The value should be compared against the enum values for accepted max SIMD bitwidths:
586 enum rte_vect_max_simd {
587 RTE_VECT_SIMD_DISABLED = 64,
588 RTE_VECT_SIMD_128 = 128,
589 RTE_VECT_SIMD_256 = 256,
590 RTE_VECT_SIMD_512 = 512,
591 RTE_VECT_SIMD_MAX = INT16_MAX + 1,
594 if (rte_vect_get_max_simd_bitwidth() >= RTE_VECT_SIMD_512)
595 /* Take AVX-512 vector path */
596 else if (rte_vect_get_max_simd_bitwidth() >= RTE_VECT_SIMD_256)
597 /* Take AVX2 vector path */
600 Memory Segments and Memory Zones (memzone)
601 ------------------------------------------
603 The mapping of physical memory is provided by this feature in the EAL.
604 As physical memory can have gaps, the memory is described in a table of descriptors,
605 and each descriptor (called rte_memseg ) describes a physical page.
607 On top of this, the memzone allocator's role is to reserve contiguous portions of physical memory.
608 These zones are identified by a unique name when the memory is reserved.
610 The rte_memzone descriptors are also located in the configuration structure.
611 This structure is accessed using rte_eal_get_configuration().
612 The lookup (by name) of a memory zone returns a descriptor containing the physical address of the memory zone.
614 Memory zones can be reserved with specific start address alignment by supplying the align parameter
615 (by default, they are aligned to cache line size).
616 The alignment value should be a power of two and not less than the cache line size (64 bytes).
617 Memory zones can also be reserved from either 2 MB or 1 GB hugepages, provided that both are available on the system.
619 Both memsegs and memzones are stored using ``rte_fbarray`` structures. Please
620 refer to *DPDK API Reference* for more information.
626 DPDK usually pins one pthread per core to avoid the overhead of task switching.
627 This allows for significant performance gains, but lacks flexibility and is not always efficient.
629 Power management helps to improve the CPU efficiency by limiting the CPU runtime frequency.
630 However, alternately it is possible to utilize the idle cycles available to take advantage of
631 the full capability of the CPU.
633 By taking advantage of cgroup, the CPU utilization quota can be simply assigned.
634 This gives another way to improve the CPU efficiency, however, there is a prerequisite;
635 DPDK must handle the context switching between multiple pthreads per core.
637 For further flexibility, it is useful to set pthread affinity not only to a CPU but to a CPU set.
639 EAL pthread and lcore Affinity
640 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
642 The term "lcore" refers to an EAL thread, which is really a Linux/FreeBSD pthread.
643 "EAL pthreads" are created and managed by EAL and execute the tasks issued by *remote_launch*.
644 In each EAL pthread, there is a TLS (Thread Local Storage) called *_lcore_id* for unique identification.
645 As EAL pthreads usually bind 1:1 to the physical CPU, the *_lcore_id* is typically equal to the CPU ID.
647 When using multiple pthreads, however, the binding is no longer always 1:1 between an EAL pthread and a specified physical CPU.
648 The EAL pthread may have affinity to a CPU set, and as such the *_lcore_id* will not be the same as the CPU ID.
649 For this reason, there is an EAL long option '--lcores' defined to assign the CPU affinity of lcores.
650 For a specified lcore ID or ID group, the option allows setting the CPU set for that EAL pthread.
653 --lcores='<lcore_set>[@cpu_set][,<lcore_set>[@cpu_set],...]'
655 'lcore_set' and 'cpu_set' can be a single number, range or a group.
657 A number is a "digit([0-9]+)"; a range is "<number>-<number>"; a group is "(<number|range>[,<number|range>,...])".
659 If a '\@cpu_set' value is not supplied, the value of 'cpu_set' will default to the value of 'lcore_set'.
663 For example, "--lcores='1,2@(5-7),(3-5)@(0,2),(0,6),7-8'" which means start 9 EAL thread;
664 lcore 0 runs on cpuset 0x41 (cpu 0,6);
665 lcore 1 runs on cpuset 0x2 (cpu 1);
666 lcore 2 runs on cpuset 0xe0 (cpu 5,6,7);
667 lcore 3,4,5 runs on cpuset 0x5 (cpu 0,2);
668 lcore 6 runs on cpuset 0x41 (cpu 0,6);
669 lcore 7 runs on cpuset 0x80 (cpu 7);
670 lcore 8 runs on cpuset 0x100 (cpu 8).
672 Using this option, for each given lcore ID, the associated CPUs can be assigned.
673 It's also compatible with the pattern of corelist('-l') option.
675 non-EAL pthread support
676 ~~~~~~~~~~~~~~~~~~~~~~~
678 It is possible to use the DPDK execution context with any user pthread (aka. non-EAL pthreads).
679 There are two kinds of non-EAL pthreads:
681 - a registered non-EAL pthread with a valid *_lcore_id* that was successfully assigned by calling ``rte_thread_register()``,
682 - a non registered non-EAL pthread with a LCORE_ID_ANY,
684 For non registered non-EAL pthread (with a LCORE_ID_ANY *_lcore_id*), some 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).
686 All these impacts are mentioned in :ref:`known_issue_label` section.
691 There are two public APIs ``rte_thread_set_affinity()`` and ``rte_thread_get_affinity()`` introduced for threads.
692 When they're used in any pthread context, the Thread Local Storage(TLS) will be set/get.
694 Those TLS include *_cpuset* and *_socket_id*:
696 * *_cpuset* stores the CPUs bitmap to which the pthread is affinitized.
698 * *_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.
704 It is possible to create Control Threads using the public API
705 ``rte_ctrl_thread_create()``.
706 Those threads can be used for management/infrastructure tasks and are used
707 internally by DPDK for multi process support and interrupt handling.
709 Those threads will be scheduled on CPUs part of the original process CPU
710 affinity from which the dataplane and service lcores are excluded.
712 For example, on a 8 CPUs system, starting a dpdk application with -l 2,3
713 (dataplane cores), then depending on the affinity configuration which can be
714 controlled with tools like taskset (Linux) or cpuset (FreeBSD),
716 - with no affinity configuration, the Control Threads will end up on
718 - with affinity restricted to 2-4, the Control Threads will end up on
720 - with affinity restricted to 2-3, the Control Threads will end up on
721 CPU 2 (main lcore, which is the default when no CPU is available).
723 .. _known_issue_label:
730 The rte_mempool uses a per-lcore cache inside the mempool.
731 For unregistered non-EAL pthreads, ``rte_lcore_id()`` will not return a valid number.
732 So for now, when rte_mempool is used with unregistered non-EAL pthreads, the put/get operations will bypass the default mempool cache and there is a performance penalty because of this bypass.
733 Only user-owned external caches can be used in an unregistered non-EAL context in conjunction with ``rte_mempool_generic_put()`` and ``rte_mempool_generic_get()`` that accept an explicit cache parameter.
737 rte_ring supports multi-producer enqueue and multi-consumer dequeue.
738 However, it is non-preemptive, this has a knock on effect of making rte_mempool non-preemptible.
742 The "non-preemptive" constraint means:
744 - a pthread doing multi-producers enqueues on a given ring must not
745 be preempted by another pthread doing a multi-producer enqueue on
747 - a pthread doing multi-consumers dequeues on a given ring must not
748 be preempted by another pthread doing a multi-consumer dequeue on
751 Bypassing this constraint may cause the 2nd pthread to spin until the 1st one is scheduled again.
752 Moreover, if the 1st pthread is preempted by a context that has an higher priority, it may even cause a dead lock.
754 This means, use cases involving preemptible pthreads should consider using rte_ring carefully.
756 1. It CAN be used for preemptible single-producer and single-consumer use case.
758 2. It CAN be used for non-preemptible multi-producer and preemptible single-consumer use case.
760 3. It CAN be used for preemptible single-producer and non-preemptible multi-consumer use case.
762 4. It MAY be used by preemptible multi-producer and/or preemptible multi-consumer pthreads whose scheduling policy are all SCHED_OTHER(cfs), SCHED_IDLE or SCHED_BATCH. User SHOULD be aware of the performance penalty before using it.
764 5. It MUST not be used by multi-producer/consumer pthreads, whose scheduling policies are SCHED_FIFO or SCHED_RR.
766 Alternatively, applications can use the lock-free stack mempool handler. When
767 considering this handler, note that:
769 - It is currently limited to the aarch64 and x86_64 platforms, because it uses
770 an instruction (16-byte compare-and-swap) that is not yet available on other
772 - It has worse average-case performance than the non-preemptive rte_ring, but
773 software caching (e.g. the mempool cache) can mitigate this by reducing the
774 number of stack accesses.
778 Running ``rte_timer_manage()`` on an unregistered non-EAL pthread is not allowed. However, resetting/stopping the timer from a non-EAL pthread is allowed.
782 In unregistered non-EAL pthreads, there is no per thread loglevel and logtype, global loglevels are used.
786 The debug statistics of rte_ring, rte_mempool and rte_timer are not supported in an unregistered non-EAL pthread.
791 The 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).
792 We expect only 50% of CPU spend on packet IO.
794 .. code-block:: console
796 mkdir /sys/fs/cgroup/cpu/pkt_io
797 mkdir /sys/fs/cgroup/cpuset/pkt_io
799 echo $cpu > /sys/fs/cgroup/cpuset/cpuset.cpus
801 echo $t0 > /sys/fs/cgroup/cpu/pkt_io/tasks
802 echo $t0 > /sys/fs/cgroup/cpuset/pkt_io/tasks
804 echo $t1 > /sys/fs/cgroup/cpu/pkt_io/tasks
805 echo $t1 > /sys/fs/cgroup/cpuset/pkt_io/tasks
807 cd /sys/fs/cgroup/cpu/pkt_io
808 echo 100000 > pkt_io/cpu.cfs_period_us
809 echo 50000 > pkt_io/cpu.cfs_quota_us
816 The EAL provides a malloc API to allocate any-sized memory.
818 The objective of this API is to provide malloc-like functions to allow
819 allocation from hugepage memory and to facilitate application porting.
820 The *DPDK API Reference* manual describes the available functions.
822 Typically, these kinds of allocations should not be done in data plane
823 processing because they are slower than pool-based allocation and make
824 use of locks within the allocation and free paths.
825 However, they can be used in configuration code.
827 Refer to the rte_malloc() function description in the *DPDK API Reference*
828 manual for more information.
831 Alignment and NUMA Constraints
832 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
834 The rte_malloc() takes an align argument that can be used to request a memory
835 area that is aligned on a multiple of this value (which must be a power of two).
837 On systems with NUMA support, a call to the rte_malloc() function will return
838 memory that has been allocated on the NUMA socket of the core which made the call.
839 A set of APIs is also provided, to allow memory to be explicitly allocated on a
840 NUMA socket directly, or by allocated on the NUMA socket where another core is
841 located, in the case where the memory is to be used by a logical core other than
842 on the one doing the memory allocation.
847 This API is meant to be used by an application that requires malloc-like
848 functions at initialization time.
850 For allocating/freeing data at runtime, in the fast-path of an application,
851 the memory pool library should be used instead.
853 Internal Implementation
854 ~~~~~~~~~~~~~~~~~~~~~~~
859 There are two data structure types used internally in the malloc library:
861 * struct malloc_heap - used to track free space on a per-socket basis
863 * struct malloc_elem - the basic element of allocation and free-space
864 tracking inside the library.
866 Structure: malloc_heap
867 """"""""""""""""""""""
869 The malloc_heap structure is used to manage free space on a per-socket basis.
870 Internally, there is one heap structure per NUMA node, which allows us to
871 allocate memory to a thread based on the NUMA node on which this thread runs.
872 While this does not guarantee that the memory will be used on that NUMA node,
873 it is no worse than a scheme where the memory is always allocated on a fixed
876 The key fields of the heap structure and their function are described below
877 (see also diagram above):
879 * lock - the lock field is needed to synchronize access to the heap.
880 Given that the free space in the heap is tracked using a linked list,
881 we need a lock to prevent two threads manipulating the list at the same time.
883 * free_head - this points to the first element in the list of free nodes for
886 * first - this points to the first element in the heap.
888 * last - this points to the last element in the heap.
890 .. _figure_malloc_heap:
892 .. figure:: img/malloc_heap.*
894 Example of a malloc heap and malloc elements within the malloc library
899 Structure: malloc_elem
900 """"""""""""""""""""""
902 The malloc_elem structure is used as a generic header structure for various
904 It is used in two different ways - all shown in the diagram above:
906 #. As a header on a block of free or allocated memory - normal case
908 #. As a padding header inside a block of memory
910 The most important fields in the structure and how they are used are described below.
912 Malloc heap is a doubly-linked list, where each element keeps track of its
913 previous and next elements. Due to the fact that hugepage memory can come and
914 go, neighboring malloc elements may not necessarily be adjacent in memory.
915 Also, since a malloc element may span multiple pages, its contents may not
916 necessarily be IOVA-contiguous either - each malloc element is only guaranteed
917 to be virtually contiguous.
921 If the usage of a particular field in one of the above three usages is not
922 described, the field can be assumed to have an undefined value in that
923 situation, for example, for padding headers only the "state" and "pad"
924 fields have valid values.
926 * heap - this pointer is a reference back to the heap structure from which
927 this block was allocated.
928 It is used for normal memory blocks when they are being freed, to add the
929 newly-freed block to the heap's free-list.
931 * prev - this pointer points to previous header element/block in memory. When
932 freeing a block, this pointer is used to reference the previous block to
933 check if that block is also free. If so, and the two blocks are immediately
934 adjacent to each other, then the two free blocks are merged to form a single
937 * next - this pointer points to next header element/block in memory. When
938 freeing a block, this pointer is used to reference the next block to check
939 if that block is also free. If so, and the two blocks are immediately
940 adjacent to each other, then the two free blocks are merged to form a single
943 * free_list - this is a structure pointing to previous and next elements in
944 this heap's free list.
945 It is only used in normal memory blocks; on ``malloc()`` to find a suitable
946 free block to allocate and on ``free()`` to add the newly freed element to
949 * state - This field can have one of three values: ``FREE``, ``BUSY`` or
951 The former two are to indicate the allocation state of a normal memory block
952 and the latter is to indicate that the element structure is a dummy structure
953 at the end of the start-of-block padding, i.e. where the start of the data
954 within a block is not at the start of the block itself, due to alignment
956 In that case, the pad header is used to locate the actual malloc element
957 header for the block.
959 * dirty - this flag is only meaningful when ``state`` is ``FREE``.
960 It indicates that the content of the element is not fully zero-filled.
961 Memory from such blocks must be cleared when requested via ``rte_zmalloc*()``.
963 * pad - this holds the length of the padding present at the start of the block.
964 In the case of a normal block header, it is added to the address of the end
965 of the header to give the address of the start of the data area, i.e. the
966 value passed back to the application on a malloc.
967 Within a dummy header inside the padding, this same value is stored, and is
968 subtracted from the address of the dummy header to yield the address of the
971 * size - the size of the data block, including the header itself.
976 On EAL initialization, all preallocated memory segments are setup as part of the
977 malloc heap. This setup involves placing an :ref:`element header<malloc_elem>`
978 with ``FREE`` at the start of each virtually contiguous segment of memory.
979 The ``FREE`` element is then added to the ``free_list`` for the malloc heap.
981 This setup also happens whenever memory is allocated at runtime (if supported),
982 in which case newly allocated pages are also added to the heap, merging with any
983 adjacent free segments if there are any.
985 When an application makes a call to a malloc-like function, the malloc function
986 will first index the ``lcore_config`` structure for the calling thread, and
987 determine the NUMA node of that thread.
988 The NUMA node is used to index the array of ``malloc_heap`` structures which is
989 passed as a parameter to the ``heap_alloc()`` function, along with the
990 requested size, type, alignment and boundary parameters.
992 The ``heap_alloc()`` function will scan the free_list of the heap, and attempt
993 to find a free block suitable for storing data of the requested size, with the
994 requested alignment and boundary constraints.
996 When a suitable free element has been identified, the pointer to be returned
997 to the user is calculated.
998 The cache-line of memory immediately preceding this pointer is filled with a
999 struct malloc_elem header.
1000 Because of alignment and boundary constraints, there could be free space at
1001 the start and/or end of the element, resulting in the following behavior:
1003 #. Check for trailing space.
1004 If the trailing space is big enough, i.e. > 128 bytes, then the free element
1006 If it is not, then we just ignore it (wasted space).
1008 #. Check for space at the start of the element.
1009 If the space at the start is small, i.e. <=128 bytes, then a pad header is
1010 used, and the remaining space is wasted.
1011 If, however, the remaining space is greater, then the free element is split.
1013 The advantage of allocating the memory from the end of the existing element is
1014 that no adjustment of the free list needs to take place - the existing element
1015 on the free list just has its size value adjusted, and the next/previous elements
1016 have their "prev"/"next" pointers redirected to the newly created element.
1018 In case when there is not enough memory in the heap to satisfy allocation
1019 request, EAL will attempt to allocate more memory from the system (if supported)
1020 and, following successful allocation, will retry reserving the memory again. In
1021 a multiprocessing scenario, all primary and secondary processes will synchronize
1022 their memory maps to ensure that any valid pointer to DPDK memory is guaranteed
1023 to be valid at all times in all currently running processes.
1025 Failure to synchronize memory maps in one of the processes will cause allocation
1026 to fail, even though some of the processes may have allocated the memory
1027 successfully. The memory is not added to the malloc heap unless primary process
1028 has ensured that all other processes have mapped this memory successfully.
1030 Any successful allocation event will trigger a callback, for which user
1031 applications and other DPDK subsystems can register. Additionally, validation
1032 callbacks will be triggered before allocation if the newly allocated memory will
1033 exceed threshold set by the user, giving a chance to allow or deny allocation.
1037 Any allocation of new pages has to go through primary process. If the
1038 primary process is not active, no memory will be allocated even if it was
1039 theoretically possible to do so. This is because primary's process map acts
1040 as an authority on what should or should not be mapped, while each secondary
1041 process has its own, local memory map. Secondary processes do not update the
1042 shared memory map, they only copy its contents to their local memory map.
1047 To free an area of memory, the pointer to the start of the data area is passed
1048 to the free function.
1049 The size of the ``malloc_elem`` structure is subtracted from this pointer to get
1050 the element header for the block.
1051 If this header is of type ``PAD`` then the pad length is further subtracted from
1052 the pointer to get the proper element header for the entire block.
1054 From this element header, we get pointers to the heap from which the block was
1055 allocated and to where it must be freed, as well as the pointer to the previous
1056 and next elements. These next and previous elements are then checked to see if
1057 they are also ``FREE`` and are immediately adjacent to the current one, and if
1058 so, they are merged with the current element. This means that we can never have
1059 two ``FREE`` memory blocks adjacent to one another, as they are always merged
1060 into a single block.
1062 If deallocating pages at runtime is supported, and the free element encloses
1063 one or more pages, those pages can be deallocated and be removed from the heap.
1064 If DPDK was started with command-line parameters for preallocating memory
1065 (``-m`` or ``--socket-mem``), then those pages that were allocated at startup
1066 will not be deallocated.
1068 Any successful deallocation event will trigger a callback, for which user
1069 applications and other DPDK subsystems can register.