1 /* 2 * Copyright (c) 2016 Adrian Chadd <adrian@FreeBSD.org> 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 */ 26 27 #include <sys/param.h> 28 #include <sys/types.h> 29 #include <sys/socket.h> 30 #include <sys/cpuset.h> 31 #include <sys/sysctl.h> 32 33 #include <stdio.h> 34 #include <stdlib.h> 35 #include <unistd.h> 36 #include <strings.h> 37 #include <err.h> 38 #include <fcntl.h> 39 #include <string.h> 40 #include <errno.h> 41 42 #include <netinet/in.h> 43 44 #include "librss.h" 45 46 int 47 rss_sock_set_recvrss(int fd, int af, int val) 48 { 49 int opt, retval; 50 socklen_t optlen; 51 int f1, f2, p; 52 53 switch (af) { 54 case AF_INET: 55 p = IPPROTO_IP; 56 f1 = IP_RECVFLOWID; 57 f2 = IP_RECVRSSBUCKETID; 58 break; 59 case AF_INET6: 60 p = IPPROTO_IPV6; 61 f1 = IPV6_RECVFLOWID; 62 f2 = IPV6_RECVRSSBUCKETID; 63 break; 64 default: 65 return (-1); 66 } 67 68 /* Enable/disable flowid */ 69 opt = val; 70 optlen = sizeof(opt); 71 retval = setsockopt(fd, p, f1, &opt, optlen); 72 if (retval < 0) { 73 warn("%s: setsockopt(IP_RECVFLOWID)", __func__); 74 return (-1); 75 } 76 77 /* Enable/disable RSS bucket reception */ 78 opt = val; 79 optlen = sizeof(opt); 80 retval = setsockopt(fd, p, f2, &opt, optlen); 81 if (retval < 0) { 82 warn("%s: setsockopt(IP_RECVRSSBUCKETID)", __func__); 83 return (-1); 84 } 85 86 return (0); 87 } 88 89 static int 90 rss_getsysctlint(const char *s) 91 { 92 int val, retval; 93 size_t rlen; 94 95 rlen = sizeof(int); 96 retval = sysctlbyname(s, &val, &rlen, NULL, 0); 97 if (retval < 0) { 98 warn("sysctlbyname (%s)", s); 99 return (-1); 100 } 101 102 return (val); 103 } 104 105 static int 106 rss_getbucketmap(int *bucket_map, int nbuckets) 107 { 108 /* XXX I'm lazy; so static string it is */ 109 char bstr[2048]; 110 int retval; 111 size_t rlen; 112 char *s, *ss; 113 int r, b, c; 114 115 /* Paranoia */ 116 memset(bstr, '\0', sizeof(bstr)); 117 118 rlen = sizeof(bstr) - 1; 119 retval = sysctlbyname("net.inet.rss.bucket_mapping", bstr, &rlen, NULL, 0); 120 if (retval < 0) { 121 warn("sysctlbyname (net.inet.rss.bucket_mapping)"); 122 return (-1); 123 } 124 125 ss = bstr; 126 while ((s = strsep(&ss, " ")) != NULL) { 127 r = sscanf(s, "%d:%d", &b, &c); 128 if (r != 2) { 129 fprintf(stderr, "%s: string (%s) not parsable\n", 130 __func__, 131 s); 132 return (-1); 133 } 134 if (b > nbuckets) { 135 fprintf(stderr, "%s: bucket %d > nbuckets %d\n", 136 __func__, 137 b, 138 nbuckets); 139 return (-1); 140 } 141 /* XXX no maxcpu check */ 142 bucket_map[b] = c; 143 } 144 return (0); 145 } 146 147 struct rss_config * 148 rss_config_get(void) 149 { 150 struct rss_config *rc = NULL; 151 152 rc = calloc(1, sizeof(*rc)); 153 if (rc == NULL) { 154 warn("%s: calloc", __func__); 155 goto error; 156 } 157 158 rc->rss_ncpus = rss_getsysctlint("net.inet.rss.ncpus"); 159 if (rc->rss_ncpus < 0) { 160 fprintf(stderr, "%s: couldn't fetch net.inet.rss.ncpus\n", __func__); 161 goto error; 162 } 163 164 rc->rss_nbuckets = rss_getsysctlint("net.inet.rss.buckets"); 165 if (rc->rss_nbuckets < 0) { 166 fprintf(stderr, "%s: couldn't fetch net.inet.rss.nbuckets\n", __func__); 167 goto error; 168 } 169 170 rc->rss_basecpu = rss_getsysctlint("net.inet.rss.basecpu"); 171 if (rc->rss_basecpu< 0) { 172 fprintf(stderr, "%s: couldn't fetch net.inet.rss.basecpu\n", __func__); 173 goto error; 174 } 175 176 rc->rss_bucket_map = calloc(rc->rss_nbuckets, sizeof(int)); 177 if (rc->rss_bucket_map == NULL) { 178 warn("%s: calloc (rss buckets; %d entries)", __func__, rc->rss_nbuckets); 179 goto error; 180 } 181 182 if (rss_getbucketmap(rc->rss_bucket_map, rc->rss_nbuckets) != 0) { 183 fprintf(stderr, "%s: rss_getbucketmap failed\n", __func__); 184 goto error; 185 } 186 187 return (rc); 188 189 error: 190 if (rc != NULL) { 191 free(rc->rss_bucket_map); 192 free(rc); 193 } 194 return (NULL); 195 } 196 197 void 198 rss_config_free(struct rss_config *rc) 199 { 200 201 if ((rc != NULL) && rc->rss_bucket_map) 202 free(rc->rss_bucket_map); 203 if (rc != NULL) 204 free(rc); 205 } 206 207 int 208 rss_config_get_bucket_count(struct rss_config *rc) 209 { 210 211 if (rc == NULL) 212 return (-1); 213 return (rc->rss_nbuckets); 214 } 215 216 int 217 rss_get_bucket_cpuset(struct rss_config *rc, rss_bucket_type_t btype, 218 int bucket, cpuset_t *cs) 219 { 220 221 if (bucket < 0 || bucket >= rc->rss_nbuckets) { 222 errno = EINVAL; 223 return (-1); 224 } 225 226 /* 227 * For now all buckets are the same, but eventually we'll want 228 * to allow administrators to set separate RSS cpusets for 229 * {kernel,user} {tx, rx} combinations. 230 */ 231 if (btype <= RSS_BUCKET_TYPE_NONE || btype > RSS_BUCKET_TYPE_MAX) { 232 errno = ENOTSUP; 233 return (-1); 234 } 235 236 CPU_ZERO(cs); 237 CPU_SET(rc->rss_bucket_map[bucket], cs); 238 239 return (0); 240 } 241 242 int 243 rss_set_bucket_rebalance_cb(rss_bucket_rebalance_cb_t *cb, void *cbdata) 244 { 245 246 (void) cb; 247 (void) cbdata; 248 249 /* 250 * For now there's no rebalance callback, so 251 * just return 0 and ignore it. 252 */ 253 return (0); 254 } 255