1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
|
/*
* Copyright (C) 2013-2017 Canonical, Ltd.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* This code is a complete clean re-write of the stress tool by
* Colin Ian King <colin.king@canonical.com> and attempts to be
* backwardly compatible with the stress tool by Amos Waterland
* <apw@rossby.metr.ou.edu> but has more stress tests and more
* functionality.
*
*/
#include "stress-ng.h"
#if defined(__linux__) && \
defined(__NR_get_mempolicy) && \
defined(__NR_mbind) && \
defined(__NR_migrate_pages) && \
defined(__NR_move_pages) && \
defined(__NR_set_mempolicy)
#define BITS_PER_BYTE (8)
#define NUMA_LONG_BITS (sizeof(unsigned long) * BITS_PER_BYTE)
#define MPOL_DEFAULT (0)
#define MPOL_PREFERRED (1)
#define MPOL_BIND (2)
#define MPOL_INTERLEAVE (3)
#define MPOL_LOCAL (4)
#define MPOL_F_NODE (1 << 0)
#define MPOL_F_ADDR (1 << 1)
#define MPOL_F_MEMS_ALLOWED (1 << 2)
#define MPOL_MF_STRICT (1 << 0)
#define MPOL_MF_MOVE (1 << 1)
#define MPOL_MF_MOVE_ALL (1 << 2)
#define MMAP_SZ (4 * MB)
typedef struct node {
uint32_t node_id;
struct node *next;
} node_t;
/*
* stress_numa_get_max_nodes()
* probe for maximum number of nodes
*/
static unsigned long stress_numa_get_max_nodes(void)
{
unsigned long sz = BITS_PER_BYTE, *mask = NULL;
do {
int mode = 0;
unsigned long *newmask = realloc(mask, sz / BITS_PER_BYTE);
if (!newmask)
break;
mask = newmask;
if (shim_get_mempolicy(&mode, mask, sz, 0, 0) == 0)
goto done;
sz <<= 1;
} while (sz < 0x100000 && errno == EINVAL);
/* Failed */
sz = 0;
done:
free(mask);
return sz;
}
/*
* stress_numa_free_nodes()
* free circular list of node info
*/
static void stress_numa_free_nodes(node_t *nodes)
{
node_t *n = nodes;
while (n) {
node_t *next = n->next;
free(n);
n = next;
if (n == nodes)
break;
}
}
/*
* hex_to_int()
* convert ASCII hex digit to integer
*/
static inline int hex_to_int(const char ch)
{
if (ch >= '0' && ch <= '9')
return ch - '0';
if (ch >= 'a' && ch <= 'f')
return ch - 'a' + 10;
if (ch >= 'A' && ch <= 'F')
return ch - 'F' + 10;
return -1;
}
/*
* stress_numa_get_mem_nodes(void)
* collect number of NUMA memory nodes, add them to a
* circular linked list
*/
static int stress_numa_get_mem_nodes(node_t **node_ptr)
{
FILE *fp;
unsigned long n = 0, node_id = 0;
node_t *tail = NULL;
*node_ptr = NULL;
char buffer[8192], *str = NULL, *ptr;
fp = fopen("/proc/self/status", "r");
if (!fp)
return -1;
while (fgets(buffer, sizeof(buffer), fp)) {
if (!strncmp(buffer, "Mems_allowed:", 13)) {
str = buffer + 13;
break;
}
}
(void)fclose(fp);
if (!str)
return -1;
ptr = buffer + strlen(buffer) - 2;
/*
* Parse hex digits into NUMA node ids, these
* are listed with least significant node last
* so we need to scan backwards from the end of
* the string back to the start.
*/
while (*ptr != ' ' && (ptr > str)) {
int val, i;
/* Skip commas */
if (*ptr == ',') {
ptr--;
continue;
}
val = hex_to_int(*ptr);
if (val < 0)
return -1;
/* Each hex digit represent 4 memory nodes */
for (i = 0; i < 4; i++) {
if (val & (1 << i)) {
node_t *node = calloc(1, sizeof(*node));
if (!node)
return -1;
node->node_id = node_id;
node->next = *node_ptr;
*node_ptr = node;
if (!tail)
tail = node;
tail->next = node;
n++;
}
node_id++;
}
ptr--;
}
return n;
}
/*
* stress_numa()
* stress the Linux NUMA interfaces
*/
int stress_numa(
uint64_t *const counter,
const uint32_t instance,
const uint64_t max_ops,
const char *name)
{
long numa_nodes;
unsigned long max_nodes;
const unsigned long lbits = NUMA_LONG_BITS;
const unsigned long page_sz = stress_get_pagesize();
const unsigned long num_pages = MMAP_SZ / page_sz;
uint8_t *buf;
node_t *n;
const pid_t mypid = getpid();
int rc = EXIT_FAILURE;
numa_nodes = stress_numa_get_mem_nodes(&n);
if (numa_nodes < 1) {
pr_inf(stdout, "%s: no NUMA nodes not found, "
"aborting test\n", name);
rc = EXIT_NO_RESOURCE;
goto numa_free;
}
max_nodes = stress_numa_get_max_nodes();
if (max_nodes == 0) {
pr_inf(stderr, "%s: cannot determine maximum number "
"of NUMA nodes, aborting test\n", name);
rc = EXIT_NO_RESOURCE;
goto numa_free;
}
if (!instance) {
pr_inf(stdout, "%s: system has %lu of a maximum %lu memory NUMA nodes\n",
name, numa_nodes, max_nodes);
}
/*
* We need a buffer to migrate around NUMA nodes
*/
buf = mmap(NULL, MMAP_SZ, PROT_READ | PROT_WRITE,
MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
if (buf == MAP_FAILED) {
rc = exit_status(errno);
pr_fail(stderr, "%s: mmap'd region of %zu bytes failed",
name, (size_t)MMAP_SZ);
goto numa_free;
}
do {
int j, mode, ret, status[num_pages], dest_nodes[num_pages];
unsigned long i, node_mask[lbits], old_node_mask[lbits];
void *pages[num_pages];
uint8_t *ptr;
node_t *n_tmp;
unsigned cpu, curr_node;
/*
* Fetch memory policy
*/
ret = shim_get_mempolicy(&mode, node_mask, max_nodes,
(unsigned long)buf, MPOL_F_ADDR);
if (ret < 0) {
pr_fail_err(name, "get_mempolicy");
goto err;
}
if (!opt_do_run)
break;
ret = shim_set_mempolicy(MPOL_PREFERRED, NULL, max_nodes);
if (ret < 0) {
pr_fail_err(name, "set_mempolicy");
goto err;
}
memset(buf, 0xff, MMAP_SZ);
if (!opt_do_run)
break;
/*
* Fetch CPU and node, we just waste some cycled
* doing this for stress reasons only
*/
(void)shim_getcpu(&cpu, &curr_node, NULL);
/*
* mbind the buffer, first try MPOL_STRICT which
* may fail with EIO
*/
memset(node_mask, 0, sizeof(node_mask));
STRESS_SETBIT(node_mask, n->node_id);
ret = shim_mbind(buf, MMAP_SZ, MPOL_BIND, node_mask,
max_nodes, MPOL_MF_STRICT);
if (ret < 0) {
if (errno != EIO) {
pr_fail_err(name, "mbind");
goto err;
}
} else {
memset(buf, 0xaa, MMAP_SZ);
}
if (!opt_do_run)
break;
/*
* mbind the buffer, now try MPOL_DEFAULT
*/
memset(node_mask, 0, sizeof(node_mask));
STRESS_SETBIT(node_mask, n->node_id);
ret = shim_mbind(buf, MMAP_SZ, MPOL_BIND, node_mask,
max_nodes, MPOL_DEFAULT);
if (ret < 0) {
if (errno != EIO) {
pr_fail_err(name, "mbind");
goto err;
}
} else {
memset(buf, 0x5c, MMAP_SZ);
}
if (!opt_do_run)
break;
/* Move to next node */
n = n->next;
/*
* Migrate all this processes pages to the current new node
*/
memset(old_node_mask, 0xff, sizeof(old_node_mask));
memset(node_mask, 0, sizeof(node_mask));
STRESS_SETBIT(node_mask, n->node_id);
ret = shim_migrate_pages(mypid, max_nodes,
old_node_mask, node_mask);
if (ret < 0) {
pr_fail_err(name, "migrate_pages");
goto err;
}
if (!opt_do_run)
break;
n_tmp = n;
for (j = 0; j < 16; j++) {
/*
* Now move pages to lots of different numa nodes
*/
for (ptr = buf, i = 0; i < num_pages; i++, ptr += page_sz, n_tmp = n_tmp->next) {
pages[i] = ptr;
dest_nodes[i] = n_tmp->node_id;
}
memset(status, 0, sizeof(status));
ret = shim_move_pages(mypid, num_pages, pages,
dest_nodes, status, MPOL_MF_MOVE);
if (ret < 0) {
pr_fail_err(name, "move_pages");
goto err;
}
memset(buf, j, MMAP_SZ);
if (!opt_do_run)
break;
}
(*counter)++;
} while (opt_do_run && (!max_ops || *counter < max_ops));
rc = EXIT_SUCCESS;
err:
munmap(buf, MMAP_SZ);
numa_free:
stress_numa_free_nodes(n);
return rc;
}
#else
int stress_numa(
uint64_t *const counter,
const uint32_t instance,
const uint64_t max_ops,
const char *name)
{
return stress_not_implemented(counter, instance, max_ops, name);
}
#endif
|