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.
35 * Derived from FreeBSD's bufring.c
37 **************************************************************************
39 * Copyright (c) 2007,2008 Kip Macy kmacy@freebsd.org
40 * All rights reserved.
42 * Redistribution and use in source and binary forms, with or without
43 * modification, are permitted provided that the following conditions are met:
45 * 1. Redistributions of source code must retain the above copyright notice,
46 * this list of conditions and the following disclaimer.
48 * 2. The name of Kip Macy nor the names of other
49 * contributors may be used to endorse or promote products derived from
50 * this software without specific prior written permission.
52 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
53 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
56 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
57 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
58 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
59 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
60 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
61 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
62 * POSSIBILITY OF SUCH DAMAGE.
64 ***************************************************************************/
72 #include <sys/queue.h>
74 #include <rte_common.h>
76 #include <rte_memory.h>
77 #include <rte_memzone.h>
78 #include <rte_launch.h>
79 #include <rte_tailq.h>
81 #include <rte_eal_memconfig.h>
82 #include <rte_atomic.h>
83 #include <rte_per_lcore.h>
84 #include <rte_lcore.h>
85 #include <rte_branch_prediction.h>
86 #include <rte_errno.h>
87 #include <rte_string_fns.h>
88 #include <rte_spinlock.h>
92 TAILQ_HEAD(rte_ring_list, rte_ring);
94 /* true if x is a power of 2 */
95 #define POWEROF2(x) ((((x)-1) & (x)) == 0)
97 /* return the size of memory occupied by a ring */
99 rte_ring_get_memsize(unsigned count)
103 /* count must be a power of 2 */
104 if ((!POWEROF2(count)) || (count > RTE_RING_SZ_MASK )) {
106 "Requested size is invalid, must be power of 2, and "
107 "do not exceed the size limit %u\n", RTE_RING_SZ_MASK);
111 sz = sizeof(struct rte_ring) + count * sizeof(void *);
112 sz = RTE_ALIGN(sz, CACHE_LINE_SIZE);
117 rte_ring_init(struct rte_ring *r, const char *name, unsigned count,
120 /* compilation-time checks */
121 RTE_BUILD_BUG_ON((sizeof(struct rte_ring) &
122 CACHE_LINE_MASK) != 0);
123 #ifdef RTE_RING_SPLIT_PROD_CONS
124 RTE_BUILD_BUG_ON((offsetof(struct rte_ring, cons) &
125 CACHE_LINE_MASK) != 0);
127 RTE_BUILD_BUG_ON((offsetof(struct rte_ring, prod) &
128 CACHE_LINE_MASK) != 0);
129 #ifdef RTE_LIBRTE_RING_DEBUG
130 RTE_BUILD_BUG_ON((sizeof(struct rte_ring_debug_stats) &
131 CACHE_LINE_MASK) != 0);
132 RTE_BUILD_BUG_ON((offsetof(struct rte_ring, stats) &
133 CACHE_LINE_MASK) != 0);
136 /* init the ring structure */
137 memset(r, 0, sizeof(*r));
138 rte_snprintf(r->name, sizeof(r->name), "%s", name);
140 r->prod.watermark = count;
141 r->prod.sp_enqueue = !!(flags & RING_F_SP_ENQ);
142 r->cons.sc_dequeue = !!(flags & RING_F_SC_DEQ);
143 r->prod.size = r->cons.size = count;
144 r->prod.mask = r->cons.mask = count-1;
145 r->prod.head = r->cons.head = 0;
146 r->prod.tail = r->cons.tail = 0;
151 /* create the ring */
153 rte_ring_create(const char *name, unsigned count, int socket_id,
156 char mz_name[RTE_MEMZONE_NAMESIZE];
158 const struct rte_memzone *mz;
161 struct rte_ring_list* ring_list = NULL;
163 /* check that we have an initialised tail queue */
165 RTE_TAILQ_LOOKUP_BY_IDX(RTE_TAILQ_RING, rte_ring_list)) == NULL) {
166 rte_errno = E_RTE_NO_TAILQ;
170 ring_size = rte_ring_get_memsize(count);
172 rte_errno = ring_size;
176 rte_snprintf(mz_name, sizeof(mz_name), "%s%s", RTE_RING_MZ_PREFIX, name);
178 rte_rwlock_write_lock(RTE_EAL_TAILQ_RWLOCK);
180 /* reserve a memory zone for this ring. If we can't get rte_config or
181 * we are secondary process, the memzone_reserve function will set
182 * rte_errno for us appropriately - hence no check in this this function */
183 mz = rte_memzone_reserve(mz_name, ring_size, socket_id, mz_flags);
186 /* no need to check return value here, we already checked the
188 rte_ring_init(r, name, count, flags);
189 TAILQ_INSERT_TAIL(ring_list, r, next);
192 RTE_LOG(ERR, RING, "Cannot reserve memory\n");
194 rte_rwlock_write_unlock(RTE_EAL_TAILQ_RWLOCK);
200 * change the high water mark. If *count* is 0, water marking is
204 rte_ring_set_water_mark(struct rte_ring *r, unsigned count)
206 if (count >= r->prod.size)
209 /* if count is 0, disable the watermarking */
211 count = r->prod.size;
213 r->prod.watermark = count;
217 /* dump the status of the ring on the console */
219 rte_ring_dump(FILE *f, const struct rte_ring *r)
221 #ifdef RTE_LIBRTE_RING_DEBUG
222 struct rte_ring_debug_stats sum;
226 fprintf(f, "ring <%s>@%p\n", r->name, r);
227 fprintf(f, " flags=%x\n", r->flags);
228 fprintf(f, " size=%"PRIu32"\n", r->prod.size);
229 fprintf(f, " ct=%"PRIu32"\n", r->cons.tail);
230 fprintf(f, " ch=%"PRIu32"\n", r->cons.head);
231 fprintf(f, " pt=%"PRIu32"\n", r->prod.tail);
232 fprintf(f, " ph=%"PRIu32"\n", r->prod.head);
233 fprintf(f, " used=%u\n", rte_ring_count(r));
234 fprintf(f, " avail=%u\n", rte_ring_free_count(r));
235 if (r->prod.watermark == r->prod.size)
236 fprintf(f, " watermark=0\n");
238 fprintf(f, " watermark=%"PRIu32"\n", r->prod.watermark);
240 /* sum and dump statistics */
241 #ifdef RTE_LIBRTE_RING_DEBUG
242 memset(&sum, 0, sizeof(sum));
243 for (lcore_id = 0; lcore_id < RTE_MAX_LCORE; lcore_id++) {
244 sum.enq_success_bulk += r->stats[lcore_id].enq_success_bulk;
245 sum.enq_success_objs += r->stats[lcore_id].enq_success_objs;
246 sum.enq_quota_bulk += r->stats[lcore_id].enq_quota_bulk;
247 sum.enq_quota_objs += r->stats[lcore_id].enq_quota_objs;
248 sum.enq_fail_bulk += r->stats[lcore_id].enq_fail_bulk;
249 sum.enq_fail_objs += r->stats[lcore_id].enq_fail_objs;
250 sum.deq_success_bulk += r->stats[lcore_id].deq_success_bulk;
251 sum.deq_success_objs += r->stats[lcore_id].deq_success_objs;
252 sum.deq_fail_bulk += r->stats[lcore_id].deq_fail_bulk;
253 sum.deq_fail_objs += r->stats[lcore_id].deq_fail_objs;
255 fprintf(f, " size=%"PRIu32"\n", r->prod.size);
256 fprintf(f, " enq_success_bulk=%"PRIu64"\n", sum.enq_success_bulk);
257 fprintf(f, " enq_success_objs=%"PRIu64"\n", sum.enq_success_objs);
258 fprintf(f, " enq_quota_bulk=%"PRIu64"\n", sum.enq_quota_bulk);
259 fprintf(f, " enq_quota_objs=%"PRIu64"\n", sum.enq_quota_objs);
260 fprintf(f, " enq_fail_bulk=%"PRIu64"\n", sum.enq_fail_bulk);
261 fprintf(f, " enq_fail_objs=%"PRIu64"\n", sum.enq_fail_objs);
262 fprintf(f, " deq_success_bulk=%"PRIu64"\n", sum.deq_success_bulk);
263 fprintf(f, " deq_success_objs=%"PRIu64"\n", sum.deq_success_objs);
264 fprintf(f, " deq_fail_bulk=%"PRIu64"\n", sum.deq_fail_bulk);
265 fprintf(f, " deq_fail_objs=%"PRIu64"\n", sum.deq_fail_objs);
267 fprintf(f, " no statistics available\n");
271 /* dump the status of all rings on the console */
273 rte_ring_list_dump(FILE *f)
275 const struct rte_ring *mp;
276 struct rte_ring_list *ring_list;
278 /* check that we have an initialised tail queue */
280 RTE_TAILQ_LOOKUP_BY_IDX(RTE_TAILQ_RING, rte_ring_list)) == NULL) {
281 rte_errno = E_RTE_NO_TAILQ;
285 rte_rwlock_read_lock(RTE_EAL_TAILQ_RWLOCK);
287 TAILQ_FOREACH(mp, ring_list, next) {
288 rte_ring_dump(f, mp);
291 rte_rwlock_read_unlock(RTE_EAL_TAILQ_RWLOCK);
294 /* search a ring from its name */
296 rte_ring_lookup(const char *name)
299 struct rte_ring_list *ring_list;
301 /* check that we have an initialized tail queue */
303 RTE_TAILQ_LOOKUP_BY_IDX(RTE_TAILQ_RING, rte_ring_list)) == NULL) {
304 rte_errno = E_RTE_NO_TAILQ;
308 rte_rwlock_read_lock(RTE_EAL_TAILQ_RWLOCK);
310 TAILQ_FOREACH(r, ring_list, next) {
311 if (strncmp(name, r->name, RTE_RING_NAMESIZE) == 0)
315 rte_rwlock_read_unlock(RTE_EAL_TAILQ_RWLOCK);