1.. SPDX-License-Identifier: BSD-3-Clause 2 Copyright(c) 2015-2016 Intel Corporation. 3 4Intel(R) QuickAssist (QAT) Crypto Poll Mode Driver 5================================================== 6 7The QAT PMD provides poll mode crypto driver support for the following 8hardware accelerator devices: 9 10* ``Intel QuickAssist Technology DH895xCC`` 11* ``Intel QuickAssist Technology C62x`` 12* ``Intel QuickAssist Technology C3xxx`` 13* ``Intel QuickAssist Technology D15xx`` 14 15 16Features 17-------- 18 19The QAT PMD has support for: 20 21Cipher algorithms: 22 23* ``RTE_CRYPTO_CIPHER_3DES_CBC`` 24* ``RTE_CRYPTO_CIPHER_3DES_CTR`` 25* ``RTE_CRYPTO_CIPHER_AES128_CBC`` 26* ``RTE_CRYPTO_CIPHER_AES192_CBC`` 27* ``RTE_CRYPTO_CIPHER_AES256_CBC`` 28* ``RTE_CRYPTO_CIPHER_AES128_CTR`` 29* ``RTE_CRYPTO_CIPHER_AES192_CTR`` 30* ``RTE_CRYPTO_CIPHER_AES256_CTR`` 31* ``RTE_CRYPTO_CIPHER_SNOW3G_UEA2`` 32* ``RTE_CRYPTO_CIPHER_NULL`` 33* ``RTE_CRYPTO_CIPHER_KASUMI_F8`` 34* ``RTE_CRYPTO_CIPHER_DES_CBC`` 35* ``RTE_CRYPTO_CIPHER_AES_DOCSISBPI`` 36* ``RTE_CRYPTO_CIPHER_DES_DOCSISBPI`` 37* ``RTE_CRYPTO_CIPHER_ZUC_EEA3`` 38 39Hash algorithms: 40 41* ``RTE_CRYPTO_AUTH_SHA1_HMAC`` 42* ``RTE_CRYPTO_AUTH_SHA224_HMAC`` 43* ``RTE_CRYPTO_AUTH_SHA256_HMAC`` 44* ``RTE_CRYPTO_AUTH_SHA384_HMAC`` 45* ``RTE_CRYPTO_AUTH_SHA512_HMAC`` 46* ``RTE_CRYPTO_AUTH_AES_XCBC_MAC`` 47* ``RTE_CRYPTO_AUTH_SNOW3G_UIA2`` 48* ``RTE_CRYPTO_AUTH_MD5_HMAC`` 49* ``RTE_CRYPTO_AUTH_NULL`` 50* ``RTE_CRYPTO_AUTH_KASUMI_F9`` 51* ``RTE_CRYPTO_AUTH_AES_GMAC`` 52* ``RTE_CRYPTO_AUTH_ZUC_EIA3`` 53 54Supported AEAD algorithms: 55 56* ``RTE_CRYPTO_AEAD_AES_GCM`` 57 58 59Limitations 60----------- 61 62* Only supports the session-oriented API implementation (session-less APIs are not supported). 63* SNOW 3G (UEA2), KASUMI (F8) and ZUC (EEA3) supported only if cipher length and offset fields are byte-multiple. 64* SNOW 3G (UIA2) and ZUC (EIA3) supported only if hash length and offset fields are byte-multiple. 65* No BSD support as BSD QAT kernel driver not available. 66* ZUC EEA3/EIA3 is not supported by dh895xcc devices 67* Maximum additional authenticated data (AAD) for GCM is 240 bytes long. 68* Queue pairs are not thread-safe (that is, within a single queue pair, RX and TX from different lcores is not supported). 69 70 71Installation 72------------ 73 74To enable QAT in DPDK, follow the instructions for modifying the compile-time 75configuration file as described `here <http://dpdk.org/doc/guides/linux_gsg/build_dpdk.html>`_. 76 77Quick instructions are as follows: 78 79.. code-block:: console 80 81 cd to the top-level DPDK directory 82 make config T=x86_64-native-linuxapp-gcc 83 sed -i 's,\(CONFIG_RTE_LIBRTE_PMD_QAT\)=n,\1=y,' build/.config 84 make 85 86To use the DPDK QAT PMD an SRIOV-enabled QAT kernel driver is required. The VF 87devices exposed by this driver will be used by the QAT PMD. The devices and 88available kernel drivers and device ids are : 89 90.. _table_qat_pmds_drivers: 91 92.. table:: QAT device generations, devices and drivers 93 94 +-----+----------+--------+---------------+------------+--------+------+--------+--------+ 95 | Gen | Device | Driver | Kernel Module | Pci Driver | PF Did | #PFs | Vf Did | VFs/PF | 96 +=====+==========+========+===============+============+========+======+========+========+ 97 | 1 | DH895xCC | 01.org | icp_qa_al | n/a | 435 | 1 | 443 | 32 | 98 +-----+----------+--------+---------------+------------+--------+------+--------+--------+ 99 | 1 | DH895xCC | 4.4+ | qat_dh895xcc | dh895xcc | 435 | 1 | 443 | 32 | 100 +-----+----------+--------+---------------+------------+--------+------+--------+--------+ 101 | 2 | C62x | 4.5+ | qat_c62x | c6xx | 37c8 | 3 | 37c9 | 16 | 102 +-----+----------+--------+---------------+------------+--------+------+--------+--------+ 103 | 2 | C3xxx | 4.5+ | qat_c3xxx | c3xxx | 19e2 | 1 | 19e3 | 16 | 104 +-----+----------+--------+---------------+------------+--------+------+--------+--------+ 105 | 2 | D15xx | p | qat_d15xx | d15xx | 6f54 | 1 | 6f55 | 16 | 106 +-----+----------+--------+---------------+------------+--------+------+--------+--------+ 107 108 109The ``Driver`` column indicates either the Linux kernel version in which 110support for this device was introduced or a driver available on Intel's 01.org 111website. There are both linux and 01.org kernel drivers available for some 112devices. p = release pending. 113 114If you are running on a kernel which includes a driver for your device, see 115`Installation using kernel.org driver`_ below. Otherwise see 116`Installation using 01.org QAT driver`_. 117 118 119Installation using kernel.org driver 120------------------------------------ 121 122The examples below are based on the C62x device, if you have a different device 123use the corresponding values in the above table. 124 125In BIOS ensure that SRIOV is enabled and either: 126 127* Disable VT-d or 128* Enable VT-d and set ``"intel_iommu=on iommu=pt"`` in the grub file. 129 130Check that the QAT driver is loaded on your system, by executing:: 131 132 lsmod | grep qa 133 134You should see the kernel module for your device listed, e.g.:: 135 136 qat_c62x 5626 0 137 intel_qat 82336 1 qat_c62x 138 139Next, you need to expose the Virtual Functions (VFs) using the sysfs file system. 140 141First find the BDFs (Bus-Device-Function) of the physical functions (PFs) of 142your device, e.g.:: 143 144 lspci -d:37c8 145 146You should see output similar to:: 147 148 1a:00.0 Co-processor: Intel Corporation Device 37c8 149 3d:00.0 Co-processor: Intel Corporation Device 37c8 150 3f:00.0 Co-processor: Intel Corporation Device 37c8 151 152Enable the VFs for each PF by echoing the number of VFs per PF to the pci driver:: 153 154 echo 16 > /sys/bus/pci/drivers/c6xx/0000:1a:00.0/sriov_numvfs 155 echo 16 > /sys/bus/pci/drivers/c6xx/0000:3d:00.0/sriov_numvfs 156 echo 16 > /sys/bus/pci/drivers/c6xx/0000:3f:00.0/sriov_numvfs 157 158Check that the VFs are available for use. For example ``lspci -d:37c9`` should 159list 48 VF devices available for a ``C62x`` device. 160 161To complete the installation follow the instructions in 162`Binding the available VFs to the DPDK UIO driver`_. 163 164.. Note:: 165 166 If the QAT kernel modules are not loaded and you see an error like ``Failed 167 to load MMP firmware qat_895xcc_mmp.bin`` in kernel logs, this may be as a 168 result of not using a distribution, but just updating the kernel directly. 169 170 Download firmware from the `kernel firmware repo 171 <http://git.kernel.org/cgit/linux/kernel/git/firmware/linux-firmware.git/tree/>`_. 172 173 Copy qat binaries to ``/lib/firmware``:: 174 175 cp qat_895xcc.bin /lib/firmware 176 cp qat_895xcc_mmp.bin /lib/firmware 177 178 Change to your linux source root directory and start the qat kernel modules:: 179 180 insmod ./drivers/crypto/qat/qat_common/intel_qat.ko 181 insmod ./drivers/crypto/qat/qat_dh895xcc/qat_dh895xcc.ko 182 183 184.. Note:: 185 186 If you see the following warning in ``/var/log/messages`` it can be ignored: 187 ``IOMMU should be enabled for SR-IOV to work correctly``. 188 189 190Installation using 01.org QAT driver 191------------------------------------ 192 193Download the latest QuickAssist Technology Driver from `01.org 194<https://01.org/packet-processing/intel%C2%AE-quickassist-technology-drivers-and-patches>`_. 195Consult the *Getting Started Guide* at the same URL for further information. 196 197The steps below assume you are: 198 199* Building on a platform with one ``DH895xCC`` device. 200* Using package ``qatmux.l.2.3.0-34.tgz``. 201* On Fedora21 kernel ``3.17.4-301.fc21.x86_64``. 202 203In the BIOS ensure that SRIOV is enabled and VT-d is disabled. 204 205Uninstall any existing QAT driver, for example by running: 206 207* ``./installer.sh uninstall`` in the directory where originally installed. 208 209* or ``rmmod qat_dh895xcc; rmmod intel_qat``. 210 211Build and install the SRIOV-enabled QAT driver:: 212 213 mkdir /QAT 214 cd /QAT 215 216 # Copy qatmux.l.2.3.0-34.tgz to this location 217 tar zxof qatmux.l.2.3.0-34.tgz 218 219 export ICP_WITHOUT_IOMMU=1 220 ./installer.sh install QAT1.6 host 221 222You can use ``cat /proc/icp_dh895xcc_dev0/version`` to confirm the driver is correctly installed. 223You can use ``lspci -d:443`` to confirm the of the 32 VF devices available per ``DH895xCC`` device. 224 225To complete the installation - follow instructions in `Binding the available VFs to the DPDK UIO driver`_. 226 227.. Note:: 228 229 If using a later kernel and the build fails with an error relating to 230 ``strict_stroul`` not being available apply the following patch: 231 232 .. code-block:: diff 233 234 /QAT/QAT1.6/quickassist/utilities/downloader/Target_CoreLibs/uclo/include/linux/uclo_platform.h 235 + #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,18,5) 236 + #define STR_TO_64(str, base, num, endPtr) {endPtr=NULL; if (kstrtoul((str), (base), (num))) printk("Error strtoull convert %s\n", str); } 237 + #else 238 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,38) 239 #define STR_TO_64(str, base, num, endPtr) {endPtr=NULL; if (strict_strtoull((str), (base), (num))) printk("Error strtoull convert %s\n", str); } 240 #else 241 #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25) 242 #define STR_TO_64(str, base, num, endPtr) {endPtr=NULL; strict_strtoll((str), (base), (num));} 243 #else 244 #define STR_TO_64(str, base, num, endPtr) \ 245 do { \ 246 if (str[0] == '-') \ 247 { \ 248 *(num) = -(simple_strtoull((str+1), &(endPtr), (base))); \ 249 }else { \ 250 *(num) = simple_strtoull((str), &(endPtr), (base)); \ 251 } \ 252 } while(0) 253 + #endif 254 #endif 255 #endif 256 257 258.. Note:: 259 260 If the build fails due to missing header files you may need to do following:: 261 262 sudo yum install zlib-devel 263 sudo yum install openssl-devel 264 265.. Note:: 266 267 If the build or install fails due to mismatching kernel sources you may need to do the following:: 268 269 sudo yum install kernel-headers-`uname -r` 270 sudo yum install kernel-src-`uname -r` 271 sudo yum install kernel-devel-`uname -r` 272 273 274Binding the available VFs to the DPDK UIO driver 275------------------------------------------------ 276 277Unbind the VFs from the stock driver so they can be bound to the uio driver. 278 279For an Intel(R) QuickAssist Technology DH895xCC device 280~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 281 282The unbind command below assumes ``BDFs`` of ``03:01.00-03:04.07``, if your 283VFs are different adjust the unbind command below:: 284 285 for device in $(seq 1 4); do \ 286 for fn in $(seq 0 7); do \ 287 echo -n 0000:03:0${device}.${fn} > \ 288 /sys/bus/pci/devices/0000\:03\:0${device}.${fn}/driver/unbind; \ 289 done; \ 290 done 291 292For an Intel(R) QuickAssist Technology C62x device 293~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 294 295The unbind command below assumes ``BDFs`` of ``1a:01.00-1a:02.07``, 296``3d:01.00-3d:02.07`` and ``3f:01.00-3f:02.07``, if your VFs are different 297adjust the unbind command below:: 298 299 for device in $(seq 1 2); do \ 300 for fn in $(seq 0 7); do \ 301 echo -n 0000:1a:0${device}.${fn} > \ 302 /sys/bus/pci/devices/0000\:1a\:0${device}.${fn}/driver/unbind; \ 303 304 echo -n 0000:3d:0${device}.${fn} > \ 305 /sys/bus/pci/devices/0000\:3d\:0${device}.${fn}/driver/unbind; \ 306 307 echo -n 0000:3f:0${device}.${fn} > \ 308 /sys/bus/pci/devices/0000\:3f\:0${device}.${fn}/driver/unbind; \ 309 done; \ 310 done 311 312For Intel(R) QuickAssist Technology C3xxx or D15xx device 313~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 314 315The unbind command below assumes ``BDFs`` of ``01:01.00-01:02.07``, if your 316VFs are different adjust the unbind command below:: 317 318 for device in $(seq 1 2); do \ 319 for fn in $(seq 0 7); do \ 320 echo -n 0000:01:0${device}.${fn} > \ 321 /sys/bus/pci/devices/0000\:01\:0${device}.${fn}/driver/unbind; \ 322 done; \ 323 done 324 325Bind to the DPDK uio driver 326~~~~~~~~~~~~~~~~~~~~~~~~~~~ 327 328Install the DPDK igb_uio driver, bind the VF PCI Device id to it and use lspci 329to confirm the VF devices are now in use by igb_uio kernel driver, 330e.g. for the C62x device:: 331 332 cd to the top-level DPDK directory 333 modprobe uio 334 insmod ./build/kmod/igb_uio.ko 335 echo "8086 37c9" > /sys/bus/pci/drivers/igb_uio/new_id 336 lspci -vvd:37c9 337 338 339Another way to bind the VFs to the DPDK UIO driver is by using the 340``dpdk-devbind.py`` script:: 341 342 cd to the top-level DPDK directory 343 ./usertools/dpdk-devbind.py -b igb_uio 0000:03:01.1 344 345 346Extra notes on KASUMI F9 347------------------------ 348 349When using KASUMI F9 authentication algorithm, the input buffer must be 350constructed according to the 3GPP KASUMI specifications (section 4.4, page 13): 351`<http://cryptome.org/3gpp/35201-900.pdf>`_. 352Input buffer has to have COUNT (4 bytes), FRESH (4 bytes), MESSAGE and DIRECTION (1 bit) 353concatenated. After the DIRECTION bit, a single '1' bit is appended, followed by 354between 0 and 7 '0' bits, so that the total length of the buffer is multiple of 8 bits. 355Note that the actual message can be any length, specified in bits. 356 357Once this buffer is passed this way, when creating the crypto operation, 358length of data to authenticate (op.sym.auth.data.length) must be the length 359of all the items described above, including the padding at the end. 360Also, offset of data to authenticate (op.sym.auth.data.offset) 361must be such that points at the start of the COUNT bytes. 362 363Device and driver naming 364------------------------ 365 366The qat crypto driver name is "crypto_qat". 367This name is passed to the dpdk-test-crypto-perf tool in the -devtype parameter. 368The rte_cryptodev_devices_get() can return the devices exposed by a driver. 369 370Each qat crypto device has a unique name, in format 371<pci bdf>_<service>, e.g. "0000:41:01.0_qat_sym". 372This name can be passed to rte_cryptodev_get_dev_id() to get the device_id. 373This is also the format of the slave parameter passed to the crypto scheduler. 374 375Debugging 376---------------------------------------- 377 378There are 2 sets of trace available via the dynamic logging feature: 379 380* pmd.qat_dp exposes trace on the data-path. 381* pmd.qat_general exposes all other trace. 382 383pmd.qat exposes both sets of traces. 384They can be enabled using the log-level option (where 8=maximum log level) on 385the process cmdline, e.g. using any of the following:: 386 387 --log-level="pmd.qat_general,8" 388 --log-level="pmd.qat_dp,8" 389 --log-level="pmd.qat,8" 390 391.. Note:: 392 393 The global RTE_LOG_DP_LEVEL overrides data-path trace so must be set to 394 RTE_LOG_DEBUG to see all the trace. This variable is in config/rte_config.h 395 for meson build and config/common_base for gnu make. 396 Also the dynamic global log level overrides both sets of trace, so e.g. no 397 QAT trace would display in this case:: 398 399 --log-level="7" --log-level="pmd.qat_general,8" 400