4 * Copyright(c) 2015 Vladimir Medvedkin <medvedkinv@gmail.com>
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in
15 * the documentation and/or other materials provided with the
17 * * Neither the name of Intel Corporation nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
24 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
25 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
26 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
27 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
28 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
29 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
30 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
31 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
40 * toeplitz hash functions.
48 * Software implementation of the Toeplitz hash function used by RSS.
49 * Can be used either for packet distribution on single queue NIC
50 * or for simulating of RSS computation on specific NIC (for example
51 * after GRE header decapsulating)
55 #include <rte_byteorder.h>
63 /* Byte swap mask used for converting IPv6 address
64 * 4-byte chunks to CPU byte order
66 static const __m128i rte_thash_ipv6_bswap_mask = {
67 0x0405060700010203ULL, 0x0C0D0E0F08090A0BULL};
71 * length in dwords of input tuple to
72 * calculate hash of ipv4 header only
74 #define RTE_THASH_V4_L3_LEN ((sizeof(struct rte_ipv4_tuple) - \
75 sizeof(((struct rte_ipv4_tuple *)0)->sctp_tag)) / 4)
78 * length in dwords of input tuple to
79 * calculate hash of ipv4 header +
82 #define RTE_THASH_V4_L4_LEN ((sizeof(struct rte_ipv4_tuple)) / 4)
85 * length in dwords of input tuple to
86 * calculate hash of ipv6 header only
88 #define RTE_THASH_V6_L3_LEN ((sizeof(struct rte_ipv6_tuple) - \
89 sizeof(((struct rte_ipv6_tuple *)0)->sctp_tag)) / 4)
92 * length in dwords of input tuple to
93 * calculate hash of ipv6 header +
96 #define RTE_THASH_V6_L4_LEN ((sizeof(struct rte_ipv6_tuple)) / 4)
100 * addresses and ports/sctp_tag have to be CPU byte order
102 struct rte_ipv4_tuple {
116 * Addresses have to be filled by rte_thash_load_v6_addr()
117 * ports/sctp_tag have to be CPU byte order
119 struct rte_ipv6_tuple {
120 uint8_t src_addr[16];
121 uint8_t dst_addr[16];
131 union rte_thash_tuple {
132 struct rte_ipv4_tuple v4;
133 struct rte_ipv6_tuple v6;
135 } __attribute__((aligned(XMM_SIZE)));
141 * Prepare special converted key to use with rte_softrss_be()
143 * pointer to original RSS key
145 * pointer to target RSS key
150 rte_convert_rss_key(const uint32_t *orig, uint32_t *targ, int len)
154 for (i = 0; i < (len >> 2); i++)
155 targ[i] = rte_be_to_cpu_32(orig[i]);
159 * Prepare and load IPv6 addresses (src and dst)
162 * Pointer to ipv6 header of the original packet
164 * Pointer to rte_ipv6_tuple structure
167 rte_thash_load_v6_addrs(const struct ipv6_hdr *orig, union rte_thash_tuple *targ)
170 __m128i ipv6 = _mm_loadu_si128((const __m128i *)orig->src_addr);
171 *(__m128i *)targ->v6.src_addr =
172 _mm_shuffle_epi8(ipv6, rte_thash_ipv6_bswap_mask);
173 ipv6 = _mm_loadu_si128((const __m128i *)orig->dst_addr);
174 *(__m128i *)targ->v6.dst_addr =
175 _mm_shuffle_epi8(ipv6, rte_thash_ipv6_bswap_mask);
178 for (i = 0; i < 4; i++) {
179 *((uint32_t *)targ->v6.src_addr + i) =
180 rte_be_to_cpu_32(*((const uint32_t *)orig->src_addr + i));
181 *((uint32_t *)targ->v6.dst_addr + i) =
182 rte_be_to_cpu_32(*((const uint32_t *)orig->dst_addr + i));
188 * Generic implementation. Can be used with original rss_key
190 * Pointer to input tuple
192 * Length of input_tuple in 4-bytes chunks
194 * Pointer to RSS hash key.
196 * Calculated hash value.
198 static inline uint32_t
199 rte_softrss(uint32_t *input_tuple, uint32_t input_len,
200 const uint8_t *rss_key)
202 uint32_t i, j, ret = 0;
204 for (j = 0; j < input_len; j++) {
205 for (i = 0; i < 32; i++) {
206 if (input_tuple[j] & (1 << (31 - i))) {
207 ret ^= rte_cpu_to_be_32(((const uint32_t *)rss_key)[j]) << i |
208 (uint32_t)((uint64_t)(rte_cpu_to_be_32(((const uint32_t *)rss_key)[j + 1])) >>
217 * Optimized implementation.
218 * If you want the calculated hash value matches NIC RSS value
219 * you have to use special converted key with rte_convert_rss_key() fn.
221 * Pointer to input tuple
223 * Length of input_tuple in 4-bytes chunks
225 * Pointer to RSS hash key.
227 * Calculated hash value.
229 static inline uint32_t
230 rte_softrss_be(uint32_t *input_tuple, uint32_t input_len,
231 const uint8_t *rss_key)
233 uint32_t i, j, ret = 0;
235 for (j = 0; j < input_len; j++) {
236 for (i = 0; i < 32; i++) {
237 if (input_tuple[j] & (1 << (31 - i))) {
238 ret ^= ((const uint32_t *)rss_key)[j] << i |
239 (uint32_t)((uint64_t)(((const uint32_t *)rss_key)[j + 1]) >> (32 - i));
250 #endif /* _RTE_THASH_H */