4 * Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
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 #define MAX_SEARCHES_AVX16 16
41 #define MAX_SEARCHES_SSE8 8
42 #define MAX_SEARCHES_SSE4 4
43 #define MAX_SEARCHES_SCALAR 2
45 #define GET_NEXT_4BYTES(prm, idx) \
46 (*((const int32_t *)((prm)[(idx)].data + *(prm)[idx].data_index++)))
49 #define RTE_ACL_NODE_INDEX ((uint32_t)~RTE_ACL_NODE_TYPE)
51 #define SCALAR_QRANGE_MULT 0x01010101
52 #define SCALAR_QRANGE_MASK 0x7f7f7f7f
53 #define SCALAR_QRANGE_MIN 0x80808080
56 * Structure to manage N parallel trie traversals.
57 * The runtime trie traversal routines can process 8, 4, or 2 tries
58 * in parallel. Each packet may require multiple trie traversals (up to 4).
59 * This structure is used to fill the slots (0 to n-1) for parallel processing
60 * with the trie traversals needed for each packet.
62 struct acl_flow_data {
64 /* number of packets processed */
66 /* number of trie traversals in progress */
68 /* current trie index (0 to N-1) */
70 uint32_t total_packets;
72 /* number of result categories per packet. */
73 /* maximum number of packets to process */
74 const uint64_t *trans;
77 struct completion *last_cmplt;
78 struct completion *cmplt_array;
82 * Structure to maintain running results for
83 * a single packet (up to 4 tries).
86 uint32_t *results; /* running results. */
87 int32_t priority[RTE_ACL_MAX_CATEGORIES]; /* running priorities. */
88 uint32_t count; /* num of remaining tries */
89 /* true for allocated struct */
90 } __attribute__((aligned(XMM_SIZE)));
93 * One parms structure for each slot in the search engine.
97 /* input data for this packet */
98 const uint32_t *data_index;
99 /* data indirection for this trie */
100 struct completion *cmplt;
101 /* completion data for this packet */
105 * Define an global idle node for unused engine slots
107 static const uint32_t idle[UINT8_MAX + 1];
110 * Allocate a completion structure to manage the tries for a packet.
112 static inline struct completion *
113 alloc_completion(struct completion *p, uint32_t size, uint32_t tries,
118 for (n = 0; n < size; n++) {
120 if (p[n].count == 0) {
122 /* mark as allocated and set number of tries. */
124 p[n].results = results;
129 /* should never get here */
134 * Resolve priority for a single result trie.
137 resolve_single_priority(uint64_t transition, int n,
138 const struct rte_acl_ctx *ctx, struct parms *parms,
139 const struct rte_acl_match_results *p)
141 if (parms[n].cmplt->count == ctx->num_tries ||
142 parms[n].cmplt->priority[0] <=
143 p[transition].priority[0]) {
145 parms[n].cmplt->priority[0] = p[transition].priority[0];
146 parms[n].cmplt->results[0] = p[transition].results[0];
151 * Routine to fill a slot in the parallel trie traversal array (parms) from
152 * the list of packets (flows).
154 static inline uint64_t
155 acl_start_next_trie(struct acl_flow_data *flows, struct parms *parms, int n,
156 const struct rte_acl_ctx *ctx)
160 /* if there are any more packets to process */
161 if (flows->num_packets < flows->total_packets) {
162 parms[n].data = flows->data[flows->num_packets];
163 parms[n].data_index = ctx->trie[flows->trie].data_index;
165 /* if this is the first trie for this packet */
166 if (flows->trie == 0) {
167 flows->last_cmplt = alloc_completion(flows->cmplt_array,
168 flows->cmplt_size, ctx->num_tries,
170 flows->num_packets * flows->categories);
173 /* set completion parameters and starting index for this slot */
174 parms[n].cmplt = flows->last_cmplt;
176 flows->trans[parms[n].data[*parms[n].data_index++] +
177 ctx->trie[flows->trie].root_index];
180 * if this is the last trie for this packet,
181 * then setup next packet.
184 if (flows->trie >= ctx->num_tries) {
186 flows->num_packets++;
189 /* keep track of number of active trie traversals */
192 /* no more tries to process, set slot to an idle position */
194 transition = ctx->idle;
195 parms[n].data = (const uint8_t *)idle;
196 parms[n].data_index = idle;
202 acl_set_flow(struct acl_flow_data *flows, struct completion *cmplt,
203 uint32_t cmplt_size, const uint8_t **data, uint32_t *results,
204 uint32_t data_num, uint32_t categories, const uint64_t *trans)
206 flows->num_packets = 0;
209 flows->last_cmplt = NULL;
210 flows->cmplt_array = cmplt;
211 flows->total_packets = data_num;
212 flows->categories = categories;
213 flows->cmplt_size = cmplt_size;
215 flows->results = results;
216 flows->trans = trans;
219 typedef void (*resolve_priority_t)
220 (uint64_t transition, int n, const struct rte_acl_ctx *ctx,
221 struct parms *parms, const struct rte_acl_match_results *p,
222 uint32_t categories);
225 * Detect matches. If a match node transition is found, then this trie
226 * traversal is complete and fill the slot with the next trie
229 static inline uint64_t
230 acl_match_check(uint64_t transition, int slot,
231 const struct rte_acl_ctx *ctx, struct parms *parms,
232 struct acl_flow_data *flows, resolve_priority_t resolve_priority)
234 const struct rte_acl_match_results *p;
236 p = (const struct rte_acl_match_results *)
237 (flows->trans + ctx->match_index);
239 if (transition & RTE_ACL_NODE_MATCH) {
241 /* Remove flags from index and decrement active traversals */
242 transition &= RTE_ACL_NODE_INDEX;
245 /* Resolve priorities for this trie and running results */
246 if (flows->categories == 1)
247 resolve_single_priority(transition, slot, ctx,
250 resolve_priority(transition, slot, ctx, parms,
251 p, flows->categories);
253 /* Count down completed tries for this search request */
254 parms[slot].cmplt->count--;
256 /* Fill the slot with the next trie or idle trie */
257 transition = acl_start_next_trie(flows, parms, slot, ctx);
263 #endif /* _ACL_RUN_H_ */