1 .. SPDX-License-Identifier: BSD-3-Clause
2 Copyright(c) 2015-2016 Intel Corporation.
4 Intel(R) QuickAssist (QAT) Crypto Poll Mode Driver
5 ==================================================
7 QAT documentation consists of three parts:
9 * Details of the symmetric crypto service below.
10 * Details of the `compression service <http://dpdk.org/doc/guides/compressdevs/qat_comp.html>`_
11 in the compressdev drivers section.
12 * Details of building the common QAT infrastructure and the PMDs to support the
13 above services. See :ref:`building_qat` below.
16 Symmetric Crypto Service on QAT
17 -------------------------------
19 The QAT crypto PMD provides poll mode crypto driver support for the following
20 hardware accelerator devices:
22 * ``Intel QuickAssist Technology DH895xCC``
23 * ``Intel QuickAssist Technology C62x``
24 * ``Intel QuickAssist Technology C3xxx``
25 * ``Intel QuickAssist Technology D15xx``
31 The QAT PMD has support for:
35 * ``RTE_CRYPTO_CIPHER_3DES_CBC``
36 * ``RTE_CRYPTO_CIPHER_3DES_CTR``
37 * ``RTE_CRYPTO_CIPHER_AES128_CBC``
38 * ``RTE_CRYPTO_CIPHER_AES192_CBC``
39 * ``RTE_CRYPTO_CIPHER_AES256_CBC``
40 * ``RTE_CRYPTO_CIPHER_AES128_CTR``
41 * ``RTE_CRYPTO_CIPHER_AES192_CTR``
42 * ``RTE_CRYPTO_CIPHER_AES256_CTR``
43 * ``RTE_CRYPTO_CIPHER_SNOW3G_UEA2``
44 * ``RTE_CRYPTO_CIPHER_NULL``
45 * ``RTE_CRYPTO_CIPHER_KASUMI_F8``
46 * ``RTE_CRYPTO_CIPHER_DES_CBC``
47 * ``RTE_CRYPTO_CIPHER_AES_DOCSISBPI``
48 * ``RTE_CRYPTO_CIPHER_DES_DOCSISBPI``
49 * ``RTE_CRYPTO_CIPHER_ZUC_EEA3``
53 * ``RTE_CRYPTO_AUTH_SHA1_HMAC``
54 * ``RTE_CRYPTO_AUTH_SHA224_HMAC``
55 * ``RTE_CRYPTO_AUTH_SHA256_HMAC``
56 * ``RTE_CRYPTO_AUTH_SHA384_HMAC``
57 * ``RTE_CRYPTO_AUTH_SHA512_HMAC``
58 * ``RTE_CRYPTO_AUTH_AES_XCBC_MAC``
59 * ``RTE_CRYPTO_AUTH_SNOW3G_UIA2``
60 * ``RTE_CRYPTO_AUTH_MD5_HMAC``
61 * ``RTE_CRYPTO_AUTH_NULL``
62 * ``RTE_CRYPTO_AUTH_KASUMI_F9``
63 * ``RTE_CRYPTO_AUTH_AES_GMAC``
64 * ``RTE_CRYPTO_AUTH_ZUC_EIA3``
66 Supported AEAD algorithms:
68 * ``RTE_CRYPTO_AEAD_AES_GCM``
69 * ``RTE_CRYPTO_AEAD_AES_CCM``
75 * Only supports the session-oriented API implementation (session-less APIs are not supported).
76 * SNOW 3G (UEA2), KASUMI (F8) and ZUC (EEA3) supported only if cipher length and offset fields are byte-multiple.
77 * SNOW 3G (UIA2) and ZUC (EIA3) supported only if hash length and offset fields are byte-multiple.
78 * No BSD support as BSD QAT kernel driver not available.
79 * ZUC EEA3/EIA3 is not supported by dh895xcc devices
80 * Maximum additional authenticated data (AAD) for GCM is 240 bytes long.
81 * Queue pairs are not thread-safe (that is, within a single queue pair, RX and TX from different lcores is not supported).
84 Extra notes on KASUMI F9
85 ~~~~~~~~~~~~~~~~~~~~~~~~
87 When using KASUMI F9 authentication algorithm, the input buffer must be
88 constructed according to the
89 `3GPP KASUMI specification <http://cryptome.org/3gpp/35201-900.pdf>`_
90 (section 4.4, page 13). The input buffer has to have COUNT (4 bytes),
91 FRESH (4 bytes), MESSAGE and DIRECTION (1 bit) concatenated. After the DIRECTION
92 bit, a single '1' bit is appended, followed by between 0 and 7 '0' bits, so that
93 the total length of the buffer is multiple of 8 bits. Note that the actual
94 message can be any length, specified in bits.
96 Once this buffer is passed this way, when creating the crypto operation,
97 length of data to authenticate "op.sym.auth.data.length" must be the length
98 of all the items described above, including the padding at the end.
99 Also, offset of data to authenticate "op.sym.auth.data.offset"
100 must be such that points at the start of the COUNT bytes.
109 A QAT device can host multiple acceleration services:
111 * symmetric cryptography
114 These services are provided to DPDK applications via PMDs which register to
115 implement the corresponding cryptodev and compressdev APIs. The PMDs use
116 common QAT driver code which manages the QAT PCI device. They also depend on a
117 QAT kernel driver being installed on the platform, see :ref:`qat_kernel` below.
120 Configuring and Building the DPDK QAT PMDs
121 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
124 Further information on configuring, building and installing DPDK is described
125 `here <http://dpdk.org/doc/guides/linux_gsg/build_dpdk.html>`_.
128 Quick instructions for QAT cryptodev PMD are as follows:
130 .. code-block:: console
132 cd to the top-level DPDK directory
134 sed -i 's,\(CONFIG_RTE_LIBRTE_PMD_QAT_SYM\)=n,\1=y,' build/.config
137 Quick instructions for QAT compressdev PMD are as follows:
139 .. code-block:: console
141 cd to the top-level DPDK directory
149 These are the build configuration options affecting QAT, and their default values:
151 .. code-block:: console
153 CONFIG_RTE_LIBRTE_PMD_QAT=y
154 CONFIG_RTE_LIBRTE_PMD_QAT_SYM=n
155 CONFIG_RTE_PMD_QAT_MAX_PCI_DEVICES=48
156 CONFIG_RTE_PMD_QAT_COMP_SGL_MAX_SEGMENTS=16
158 CONFIG_RTE_LIBRTE_PMD_QAT must be enabled for any QAT PMD to be built.
160 The QAT cryptodev PMD has an external dependency on libcrypto, so is not
161 built by default. CONFIG_RTE_LIBRTE_PMD_QAT_SYM should be enabled to build it.
163 The QAT compressdev PMD has no external dependencies, so needs no configuration
164 options and is built by default.
166 The number of VFs per PF varies - see table below. If multiple QAT packages are
167 installed on a platform then CONFIG_RTE_PMD_QAT_MAX_PCI_DEVICES should be
168 adjusted to the number of VFs which the QAT common code will need to handle.
169 Note, there is a separate config item for max cryptodevs CONFIG_RTE_CRYPTO_MAX_DEVS,
170 if necessary this should be adjusted to handle the total of QAT and other devices
171 which the process will use.
173 QAT allocates internal structures to handle SGLs. For the compression service
174 CONFIG_RTE_PMD_QAT_COMP_SGL_MAX_SEGMENTS can be changed if more segments are needed.
175 An extra (max_inflight_ops x 16) bytes per queue_pair will be used for every increment.
178 Device and driver naming
179 ~~~~~~~~~~~~~~~~~~~~~~~~
181 * The qat cryptodev driver name is "crypto_qat".
182 The "rte_cryptodev_devices_get()" returns the devices exposed by this driver.
184 * Each qat crypto device has a unique name, in format
185 "<pci bdf>_<service>", e.g. "0000:41:01.0_qat_sym".
186 This name can be passed to "rte_cryptodev_get_dev_id()" to get the device_id.
190 The qat crypto driver name is passed to the dpdk-test-crypto-perf tool in the "-devtype" parameter.
192 The qat crypto device name is in the format of the slave parameter passed to the crypto scheduler.
194 * The qat compressdev driver name is "compress_qat".
195 The rte_compressdev_devices_get() returns the devices exposed by this driver.
197 * Each qat compression device has a unique name, in format
198 <pci bdf>_<service>, e.g. "0000:41:01.0_qat_comp".
199 This name can be passed to rte_compressdev_get_dev_id() to get the device_id.
203 Dependency on the QAT kernel driver
204 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
206 To use QAT an SRIOV-enabled QAT kernel driver is required. The VF
207 devices created and initialised by this driver will be used by the QAT PMDs.
209 Instructions for installation are below, but first an explanation of the
210 relationships between the PF/VF devices and the PMDs visible to
213 Each QuickAssist PF device exposes a number of VF devices. Each VF device can
214 enable one cryptodev PMD and/or one compressdev PMD.
215 These QAT PMDs share the same underlying device and pci-mgmt code, but are
216 enumerated independently on their respective APIs and appear as independent
217 devices to applications.
221 Each VF can only be used by one DPDK process. It is not possible to share
222 the same VF across multiple processes, even if these processes are using
223 different acceleration services.
225 Conversely one DPDK process can use one or more QAT VFs and can expose both
226 cryptodev and compressdev instances on each of those VFs.
229 Available kernel drivers
230 ~~~~~~~~~~~~~~~~~~~~~~~~
232 Kernel drivers for each device are listed in the following table. Scroll right
233 to check that the driver and device supports the service you require.
236 .. _table_qat_pmds_drivers:
238 .. table:: QAT device generations, devices and drivers
240 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
241 | Gen | Device | Driver/ver | Kernel Module | Pci Driver | PF Did | #PFs | VF Did | VFs/PF | cryptodev | compressdev |
242 +=====+==========+===============+===============+============+========+======+========+========+===========+=============+
243 | 1 | DH895xCC | linux/4.4+ | qat_dh895xcc | dh895xcc | 435 | 1 | 443 | 32 | Yes | No |
244 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
245 | " | " | 01.org/4.2.0+ | " | " | " | " | " | " | Yes | No |
246 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
247 | 2 | C62x | linux/4.5+ | qat_c62x | c6xx | 37c8 | 3 | 37c9 | 16 | Yes | No |
248 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
249 | " | " | 01.org/4.2.0+ | " | " | " | " | " | " | Yes | Yes |
250 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
251 | 2 | C3xxx | linux/4.5+ | qat_c3xxx | c3xxx | 19e2 | 1 | 19e3 | 16 | Yes | No |
252 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
253 | " | " | 01.org/4.2.0+ | " | " | " | " | " | " | Yes | Yes |
254 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
255 | 2 | D15xx | p | qat_d15xx | d15xx | 6f54 | 1 | 6f55 | 16 | Yes | No |
256 +-----+----------+---------------+---------------+------------+--------+------+--------+--------+-----------+-------------+
259 The ``Driver`` column indicates either the Linux kernel version in which
260 support for this device was introduced or a driver available on Intel's 01.org
261 website. There are both linux and 01.org kernel drivers available for some
262 devices. p = release pending.
264 If you are running on a kernel which includes a driver for your device, see
265 `Installation using kernel.org driver`_ below. Otherwise see
266 `Installation using 01.org QAT driver`_.
269 Installation using kernel.org driver
270 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
272 The examples below are based on the C62x device, if you have a different device
273 use the corresponding values in the above table.
275 In BIOS ensure that SRIOV is enabled and either:
278 * Enable VT-d and set ``"intel_iommu=on iommu=pt"`` in the grub file.
280 Check that the QAT driver is loaded on your system, by executing::
284 You should see the kernel module for your device listed, e.g.::
287 intel_qat 82336 1 qat_c62x
289 Next, you need to expose the Virtual Functions (VFs) using the sysfs file system.
291 First find the BDFs (Bus-Device-Function) of the physical functions (PFs) of
296 You should see output similar to::
298 1a:00.0 Co-processor: Intel Corporation Device 37c8
299 3d:00.0 Co-processor: Intel Corporation Device 37c8
300 3f:00.0 Co-processor: Intel Corporation Device 37c8
302 Enable the VFs for each PF by echoing the number of VFs per PF to the pci driver::
304 echo 16 > /sys/bus/pci/drivers/c6xx/0000:1a:00.0/sriov_numvfs
305 echo 16 > /sys/bus/pci/drivers/c6xx/0000:3d:00.0/sriov_numvfs
306 echo 16 > /sys/bus/pci/drivers/c6xx/0000:3f:00.0/sriov_numvfs
308 Check that the VFs are available for use. For example ``lspci -d:37c9`` should
309 list 48 VF devices available for a ``C62x`` device.
311 To complete the installation follow the instructions in
312 `Binding the available VFs to the DPDK UIO driver`_.
316 If the QAT kernel modules are not loaded and you see an error like ``Failed
317 to load MMP firmware qat_895xcc_mmp.bin`` in kernel logs, this may be as a
318 result of not using a distribution, but just updating the kernel directly.
320 Download firmware from the `kernel firmware repo
321 <http://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/tree/>`_.
323 Copy qat binaries to ``/lib/firmware``::
325 cp qat_895xcc.bin /lib/firmware
326 cp qat_895xcc_mmp.bin /lib/firmware
328 Change to your linux source root directory and start the qat kernel modules::
330 insmod ./drivers/crypto/qat/qat_common/intel_qat.ko
331 insmod ./drivers/crypto/qat/qat_dh895xcc/qat_dh895xcc.ko
336 If you see the following warning in ``/var/log/messages`` it can be ignored:
337 ``IOMMU should be enabled for SR-IOV to work correctly``.
340 Installation using 01.org QAT driver
341 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
343 Download the latest QuickAssist Technology Driver from `01.org
344 <https://01.org/packet-processing/intel%C2%AE-quickassist-technology-drivers-and-patches>`_.
345 Consult the *Getting Started Guide* at the same URL for further information.
347 The steps below assume you are:
349 * Building on a platform with one ``C62x`` device.
350 * Using package ``qat1.7.l.4.2.0-000xx.tar.gz``.
351 * On Fedora26 kernel ``4.11.11-300.fc26.x86_64``.
353 In the BIOS ensure that SRIOV is enabled and VT-d is disabled.
355 Uninstall any existing QAT driver, for example by running:
357 * ``./installer.sh uninstall`` in the directory where originally installed.
360 Build and install the SRIOV-enabled QAT driver::
365 # Copy the package to this location and unpack
366 tar zxof qat1.7.l.4.2.0-000xx.tar.gz
368 ./configure --enable-icp-sriov=host
371 You can use ``cat /sys/kernel/debug/qat<your device type and bdf>/version/fw`` to confirm the driver is correctly installed and is using firmware version 4.2.0.
372 You can use ``lspci -d:37c9`` to confirm the presence of the 16 VF devices available per ``C62x`` PF.
374 Confirm the driver is correctly installed and is using firmware version 4.2.0::
376 cat /sys/kernel/debug/qat<your device type and bdf>/version/fw
379 Confirm the presence of 48 VF devices - 16 per PF::
384 To complete the installation - follow instructions in `Binding the available VFs to the DPDK UIO driver`_.
388 If using a later kernel and the build fails with an error relating to
389 ``strict_stroul`` not being available apply the following patch:
393 /QAT/QAT1.6/quickassist/utilities/downloader/Target_CoreLibs/uclo/include/linux/uclo_platform.h
394 + #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,18,5)
395 + #define STR_TO_64(str, base, num, endPtr) {endPtr=NULL; if (kstrtoul((str), (base), (num))) printk("Error strtoull convert %s\n", str); }
397 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,38)
398 #define STR_TO_64(str, base, num, endPtr) {endPtr=NULL; if (strict_strtoull((str), (base), (num))) printk("Error strtoull convert %s\n", str); }
400 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25)
401 #define STR_TO_64(str, base, num, endPtr) {endPtr=NULL; strict_strtoll((str), (base), (num));}
403 #define STR_TO_64(str, base, num, endPtr) \
407 *(num) = -(simple_strtoull((str+1), &(endPtr), (base))); \
409 *(num) = simple_strtoull((str), &(endPtr), (base)); \
419 If the build fails due to missing header files you may need to do following::
421 sudo yum install zlib-devel
422 sudo yum install openssl-devel
423 sudo yum install libudev-devel
427 If the build or install fails due to mismatching kernel sources you may need to do the following::
429 sudo yum install kernel-headers-`uname -r`
430 sudo yum install kernel-src-`uname -r`
431 sudo yum install kernel-devel-`uname -r`
434 Binding the available VFs to the DPDK UIO driver
435 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
437 Unbind the VFs from the stock driver so they can be bound to the uio driver.
439 For an Intel(R) QuickAssist Technology DH895xCC device
440 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
442 The unbind command below assumes ``BDFs`` of ``03:01.00-03:04.07``, if your
443 VFs are different adjust the unbind command below::
445 for device in $(seq 1 4); do \
446 for fn in $(seq 0 7); do \
447 echo -n 0000:03:0${device}.${fn} > \
448 /sys/bus/pci/devices/0000\:03\:0${device}.${fn}/driver/unbind; \
452 For an Intel(R) QuickAssist Technology C62x device
453 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
455 The unbind command below assumes ``BDFs`` of ``1a:01.00-1a:02.07``,
456 ``3d:01.00-3d:02.07`` and ``3f:01.00-3f:02.07``, if your VFs are different
457 adjust the unbind command below::
459 for device in $(seq 1 2); do \
460 for fn in $(seq 0 7); do \
461 echo -n 0000:1a:0${device}.${fn} > \
462 /sys/bus/pci/devices/0000\:1a\:0${device}.${fn}/driver/unbind; \
464 echo -n 0000:3d:0${device}.${fn} > \
465 /sys/bus/pci/devices/0000\:3d\:0${device}.${fn}/driver/unbind; \
467 echo -n 0000:3f:0${device}.${fn} > \
468 /sys/bus/pci/devices/0000\:3f\:0${device}.${fn}/driver/unbind; \
472 For Intel(R) QuickAssist Technology C3xxx or D15xx device
473 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
475 The unbind command below assumes ``BDFs`` of ``01:01.00-01:02.07``, if your
476 VFs are different adjust the unbind command below::
478 for device in $(seq 1 2); do \
479 for fn in $(seq 0 7); do \
480 echo -n 0000:01:0${device}.${fn} > \
481 /sys/bus/pci/devices/0000\:01\:0${device}.${fn}/driver/unbind; \
485 Bind to the DPDK uio driver
486 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
488 Install the DPDK igb_uio driver, bind the VF PCI Device id to it and use lspci
489 to confirm the VF devices are now in use by igb_uio kernel driver,
490 e.g. for the C62x device::
492 cd to the top-level DPDK directory
494 insmod ./build/kmod/igb_uio.ko
495 echo "8086 37c9" > /sys/bus/pci/drivers/igb_uio/new_id
499 Another way to bind the VFs to the DPDK UIO driver is by using the
500 ``dpdk-devbind.py`` script::
502 cd to the top-level DPDK directory
503 ./usertools/dpdk-devbind.py -b igb_uio 0000:03:01.1
508 QAT crypto PMD can be tested by running the test application::
513 ./test -l1 -n1 -w <your qat bdf>
514 RTE>>cryptodev_qat_autotest
516 QAT compression PMD can be tested by running the test application::
519 sed -i 's,\(CONFIG_RTE_COMPRESSDEV_TEST\)=n,\1=y,' build/.config
522 ./test -l1 -n1 -w <your qat bdf>
523 RTE>>compressdev_autotest
529 There are 2 sets of trace available via the dynamic logging feature:
531 * pmd.qat_dp exposes trace on the data-path.
532 * pmd.qat_general exposes all other trace.
534 pmd.qat exposes both sets of traces.
535 They can be enabled using the log-level option (where 8=maximum log level) on
536 the process cmdline, e.g. using any of the following::
538 --log-level="pmd.qat_general,8"
539 --log-level="pmd.qat_dp,8"
540 --log-level="pmd.qat,8"
544 The global RTE_LOG_DP_LEVEL overrides data-path trace so must be set to
545 RTE_LOG_DEBUG to see all the trace. This variable is in config/rte_config.h
546 for meson build and config/common_base for gnu make.
547 Also the dynamic global log level overrides both sets of trace, so e.g. no
548 QAT trace would display in this case::
550 --log-level="7" --log-level="pmd.qat_general,8"