--- /dev/null
+/**
+ * collectd - src/intel_rdt.c
+ *
+ * Copyright(c) 2016 Intel Corporation. All rights reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ * Authors:
+ * Serhiy Pshyk <serhiyx.pshyk@intel.com>
+ **/
+
+#include "collectd.h"
+#include "common.h"
+
+#include <pqos.h>
+
+#define RDT_PLUGIN "intel_rdt"
+
+#define RDT_MAX_SOCKETS 8
+#define RDT_MAX_SOCKET_CORES 64
+#define RDT_MAX_CORES (RDT_MAX_SOCKET_CORES * RDT_MAX_SOCKETS)
+
+struct rdt_core_group_s {
+ char *desc;
+ size_t num_cores;
+ unsigned *cores;
+ enum pqos_mon_event events;
+};
+typedef struct rdt_core_group_s rdt_core_group_t;
+
+struct rdt_ctx_s {
+ rdt_core_group_t cgroups[RDT_MAX_CORES];
+ struct pqos_mon_data *pgroups[RDT_MAX_CORES];
+ size_t num_groups;
+ const struct pqos_cpuinfo *pqos_cpu;
+ const struct pqos_cap *pqos_cap;
+ const struct pqos_capability *cap_mon;
+};
+typedef struct rdt_ctx_s rdt_ctx_t;
+
+static rdt_ctx_t *g_rdt = NULL;
+
+static int isdup(const uint64_t *nums, size_t size, uint64_t val) {
+ for (size_t i = 0; i < size; i++)
+ if (nums[i] == val)
+ return 1;
+ return 0;
+}
+
+static int strtouint64(const char *s, uint64_t *n) {
+ char *endptr = NULL;
+
+ assert(s != NULL);
+ assert(n != NULL);
+
+ *n = strtoull(s, &endptr, 0);
+
+ if (!(*s != '\0' && *endptr == '\0')) {
+ DEBUG(RDT_PLUGIN ": Error converting '%s' to unsigned number.", s);
+ return (-EINVAL);
+ }
+
+ return (0);
+}
+
+/*
+ * NAME
+ * strlisttonums
+ *
+ * DESCRIPTION
+ * Converts string of characters representing list of numbers into array of
+ * numbers. Allowed formats are:
+ * 0,1,2,3
+ * 0-10,20-18
+ * 1,3,5-8,10,0x10-12
+ *
+ * Numbers can be in decimal or hexadecimal format.
+ *
+ * PARAMETERS
+ * `s' String representing list of unsigned numbers.
+ * `nums' Array to put converted numeric values into.
+ * `max' Maximum number of elements that nums can accommodate.
+ *
+ * RETURN VALUE
+ * Number of elements placed into nums.
+ */
+static size_t strlisttonums(char *s, uint64_t *nums, size_t max) {
+ int ret;
+ size_t index = 0;
+ char *saveptr = NULL;
+
+ if (s == NULL || nums == NULL || max == 0)
+ return index;
+
+ for (;;) {
+ char *p = NULL;
+ char *token = NULL;
+
+ token = strtok_r(s, ",", &saveptr);
+ if (token == NULL)
+ break;
+
+ s = NULL;
+
+ while (isspace(*token))
+ token++;
+ if (*token == '\0')
+ continue;
+
+ p = strchr(token, '-');
+ if (p != NULL) {
+ uint64_t n, start, end;
+ *p = '\0';
+ ret = strtouint64(token, &start);
+ if (ret < 0)
+ return (0);
+ ret = strtouint64(p + 1, &end);
+ if (ret < 0)
+ return (0);
+ if (start > end) {
+ return (0);
+ }
+ for (n = start; n <= end; n++) {
+ if (!(isdup(nums, index, n))) {
+ nums[index] = n;
+ index++;
+ }
+ if (index >= max)
+ return index;
+ }
+ } else {
+ uint64_t val;
+
+ ret = strtouint64(token, &val);
+ if (ret < 0)
+ return (0);
+
+ if (!(isdup(nums, index, val))) {
+ nums[index] = val;
+ index++;
+ }
+ if (index >= max)
+ return index;
+ }
+ }
+
+ return index;
+}
+
+/*
+ * NAME
+ * cgroup_cmp
+ *
+ * DESCRIPTION
+ * Function to compare cores in 2 core groups.
+ *
+ * PARAMETERS
+ * `cg_a' Pointer to core group a.
+ * `cg_b' Pointer to core group b.
+ *
+ * RETURN VALUE
+ * 1 if both groups contain the same cores
+ * 0 if none of their cores match
+ * -1 if some but not all cores match
+ */
+static int cgroup_cmp(const rdt_core_group_t *cg_a,
+ const rdt_core_group_t *cg_b) {
+ int found = 0;
+
+ assert(cg_a != NULL);
+ assert(cg_b != NULL);
+
+ const int sz_a = cg_a->num_cores;
+ const int sz_b = cg_b->num_cores;
+ const unsigned *tab_a = cg_a->cores;
+ const unsigned *tab_b = cg_b->cores;
+
+ for (int i = 0; i < sz_a; i++) {
+ for (int j = 0; j < sz_b; j++)
+ if (tab_a[i] == tab_b[j])
+ found++;
+ }
+ /* if no cores are the same */
+ if (!found)
+ return 0;
+ /* if group contains same cores */
+ if (sz_a == sz_b && sz_b == found)
+ return 1;
+ /* if not all cores are the same */
+ return -1;
+}
+
+static int cgroup_set(rdt_core_group_t *cg, char *desc, uint64_t *cores,
+ size_t num_cores) {
+ assert(cg != NULL);
+ assert(desc != NULL);
+ assert(cores != NULL);
+ assert(num_cores > 0);
+
+ cg->cores = calloc(num_cores, sizeof(unsigned));
+ if (cg->cores == NULL) {
+ ERROR(RDT_PLUGIN ": Error allocating core group table");
+ return (-ENOMEM);
+ }
+ cg->num_cores = num_cores;
+ cg->desc = strdup(desc);
+ if (cg->desc == NULL) {
+ ERROR(RDT_PLUGIN ": Error allocating core group description");
+ sfree(cg->cores);
+ return (-ENOMEM);
+ }
+
+ for (size_t i = 0; i < num_cores; i++)
+ cg->cores[i] = (unsigned)cores[i];
+
+ return 0;
+}
+
+/*
+ * NAME
+ * oconfig_to_cgroups
+ *
+ * DESCRIPTION
+ * Function to set the descriptions and cores for each core group.
+ * Takes a config option containing list of strings that are used to set
+ * core group values.
+ *
+ * PARAMETERS
+ * `item' Config option containing core groups.
+ * `groups' Table of core groups to set values in.
+ * `max_groups' Maximum number of core groups allowed.
+ * `max_core' Maximum allowed core value.
+ *
+ * RETURN VALUE
+ * On success, the number of core groups set up. On error, appropriate
+ * negative error value.
+ */
+static int oconfig_to_cgroups(oconfig_item_t *item, rdt_core_group_t *groups,
+ size_t max_groups, uint64_t max_core) {
+ int index = 0;
+
+ assert(groups != NULL);
+ assert(max_groups > 0);
+ assert(item != NULL);
+
+ for (int j = 0; j < item->values_num; j++) {
+ int ret;
+ size_t n;
+ uint64_t cores[RDT_MAX_CORES] = {0};
+ char value[DATA_MAX_NAME_LEN];
+
+ if ((item->values[j].value.string == NULL) ||
+ (strlen(item->values[j].value.string) == 0))
+ continue;
+
+ sstrncpy(value, item->values[j].value.string, sizeof(value));
+
+ n = strlisttonums(value, cores, STATIC_ARRAY_SIZE(cores));
+ if (n == 0) {
+ ERROR(RDT_PLUGIN ": Error parsing core group (%s)",
+ item->values[j].value.string);
+ return (-EINVAL);
+ }
+
+ for (int i = 0; i < n; i++) {
+ if (cores[i] > max_core) {
+ ERROR(RDT_PLUGIN ": Core group (%s) contains invalid core id (%d)",
+ item->values[j].value.string, (int)cores[i]);
+ return (-EINVAL);
+ }
+ }
+
+ /* set core group info */
+ ret = cgroup_set(&groups[index], item->values[j].value.string, cores, n);
+ if (ret < 0)
+ return ret;
+
+ index++;
+
+ if (index >= max_groups) {
+ WARNING(RDT_PLUGIN ": Too many core groups configured");
+ return index;
+ }
+ }
+
+ return index;
+}
+
+#if COLLECT_DEBUG
+static void rdt_dump_cgroups(void) {
+ char cores[RDT_MAX_CORES * 4];
+
+ if (g_rdt == NULL)
+ return;
+
+ DEBUG(RDT_PLUGIN ": Core Groups Dump");
+ DEBUG(RDT_PLUGIN ": groups count: %zu", g_rdt->num_groups);
+
+ for (int i = 0; i < g_rdt->num_groups; i++) {
+
+ memset(cores, 0, sizeof(cores));
+ for (int j = 0; j < g_rdt->cgroups[i].num_cores; j++) {
+ snprintf(cores + strlen(cores), sizeof(cores) - strlen(cores) - 1, " %d",
+ g_rdt->cgroups[i].cores[j]);
+ }
+
+ DEBUG(RDT_PLUGIN ": group[%d]:", i);
+ DEBUG(RDT_PLUGIN ": description: %s", g_rdt->cgroups[i].desc);
+ DEBUG(RDT_PLUGIN ": cores: %s", cores);
+ DEBUG(RDT_PLUGIN ": events: 0x%X", g_rdt->cgroups[i].events);
+ }
+
+ return;
+}
+
+static inline double bytes_to_kb(const double bytes) { return bytes / 1024.0; }
+
+static inline double bytes_to_mb(const double bytes) {
+ return bytes / (1024.0 * 1024.0);
+}
+
+static void rdt_dump_data(void) {
+ /*
+ * CORE - monitored group of cores
+ * RMID - Resource Monitoring ID associated with the monitored group
+ * LLC - last level cache occupancy
+ * MBL - local memory bandwidth
+ * MBR - remote memory bandwidth
+ */
+ DEBUG(" CORE RMID LLC[KB] MBL[MB] MBR[MB]");
+ for (int i = 0; i < g_rdt->num_groups; i++) {
+
+ const struct pqos_event_values *pv = &g_rdt->pgroups[i]->values;
+
+ double llc = bytes_to_kb(pv->llc);
+ double mbr = bytes_to_mb(pv->mbm_remote_delta);
+ double mbl = bytes_to_mb(pv->mbm_local_delta);
+
+ DEBUG(" [%s] %8u %10.1f %10.1f %10.1f", g_rdt->cgroups[i].desc,
+ g_rdt->pgroups[i]->poll_ctx[0].rmid, llc, mbl, mbr);
+ }
+}
+#endif /* COLLECT_DEBUG */
+
+static void rdt_free_cgroups(void) {
+ for (int i = 0; i < RDT_MAX_CORES; i++) {
+ sfree(g_rdt->cgroups[i].desc);
+
+ sfree(g_rdt->cgroups[i].cores);
+ g_rdt->cgroups[i].num_cores = 0;
+
+ sfree(g_rdt->pgroups[i]);
+ }
+}
+
+static int rdt_default_cgroups(void) {
+ int ret;
+
+ /* configure each core in separate group */
+ for (unsigned i = 0; i < g_rdt->pqos_cpu->num_cores; i++) {
+ char desc[DATA_MAX_NAME_LEN];
+ uint64_t core = i;
+
+ ssnprintf(desc, sizeof(desc), "%d", g_rdt->pqos_cpu->cores[i].lcore);
+
+ /* set core group info */
+ ret = cgroup_set(&g_rdt->cgroups[i], desc, &core, 1);
+ if (ret < 0)
+ return ret;
+ }
+
+ return g_rdt->pqos_cpu->num_cores;
+}
+
+static int rdt_config_cgroups(oconfig_item_t *item) {
+ int n = 0;
+ enum pqos_mon_event events = 0;
+
+ if (item == NULL) {
+ DEBUG(RDT_PLUGIN ": cgroups_config: Invalid argument.");
+ return (-EINVAL);
+ }
+
+ DEBUG(RDT_PLUGIN ": Core groups [%d]:", item->values_num);
+ for (int j = 0; j < item->values_num; j++) {
+ if (item->values[j].type != OCONFIG_TYPE_STRING) {
+ ERROR(RDT_PLUGIN ": given core group value is not a string [idx=%d]", j);
+ return (-EINVAL);
+ }
+ DEBUG(RDT_PLUGIN ": [%d]: %s", j, item->values[j].value.string);
+ }
+
+ n = oconfig_to_cgroups(item, g_rdt->cgroups, RDT_MAX_CORES,
+ g_rdt->pqos_cpu->num_cores - 1);
+ if (n < 0) {
+ rdt_free_cgroups();
+ ERROR(RDT_PLUGIN ": Error parsing core groups configuration.");
+ return (-EINVAL);
+ }
+
+ if (n == 0) {
+ /* create default core groups if "Cores" config option is empty */
+ n = rdt_default_cgroups();
+ if (n < 0) {
+ rdt_free_cgroups();
+ ERROR(RDT_PLUGIN ": Error creating default core groups configuration.");
+ return n;
+ }
+ INFO(RDT_PLUGIN
+ ": No core groups configured. Default core groups created.");
+ }
+
+ /* Get all available events on this platform */
+ for (int i = 0; i < g_rdt->cap_mon->u.mon->num_events; i++)
+ events |= g_rdt->cap_mon->u.mon->events[i].type;
+
+ events &= ~(PQOS_PERF_EVENT_LLC_MISS);
+
+ DEBUG(RDT_PLUGIN ": Number of cores in the system: %u",
+ g_rdt->pqos_cpu->num_cores);
+ DEBUG(RDT_PLUGIN ": Available events to monitor: %#x", events);
+
+ g_rdt->num_groups = n;
+ for (int i = 0; i < n; i++) {
+ for (int j = 0; j < i; j++) {
+ int found = 0;
+ found = cgroup_cmp(&g_rdt->cgroups[j], &g_rdt->cgroups[i]);
+ if (found != 0) {
+ rdt_free_cgroups();
+ ERROR(RDT_PLUGIN ": Cannot monitor same cores in different groups.");
+ return (-EINVAL);
+ }
+ }
+
+ g_rdt->cgroups[i].events = events;
+ g_rdt->pgroups[i] = calloc(1, sizeof(*g_rdt->pgroups[i]));
+ if (g_rdt->pgroups[i] == NULL) {
+ rdt_free_cgroups();
+ ERROR(RDT_PLUGIN ": Failed to allocate memory for monitoring data.");
+ return (-ENOMEM);
+ }
+ }
+
+ return (0);
+}
+
+static int rdt_preinit(void) {
+ int ret;
+
+ if (g_rdt != NULL) {
+ /* already initialized if config callback was called before init callback */
+ return (0);
+ }
+
+ g_rdt = calloc(1, sizeof(*g_rdt));
+ if (g_rdt == NULL) {
+ ERROR(RDT_PLUGIN ": Failed to allocate memory for rdt context.");
+ return (-ENOMEM);
+ }
+
+ /* In case previous instance of the application was not closed properly
+ * call fini and ignore return code. */
+ pqos_fini();
+
+ /* TODO:
+ * stdout should not be used here. Will be reworked when support of log
+ * callback is added to PQoS library.
+ */
+ ret = pqos_init(&(struct pqos_config){.fd_log = STDOUT_FILENO});
+ if (ret != PQOS_RETVAL_OK) {
+ ERROR(RDT_PLUGIN ": Error initializing PQoS library!");
+ goto rdt_preinit_error1;
+ }
+
+ ret = pqos_cap_get(&g_rdt->pqos_cap, &g_rdt->pqos_cpu);
+ if (ret != PQOS_RETVAL_OK) {
+ ERROR(RDT_PLUGIN ": Error retrieving PQoS capabilities.");
+ goto rdt_preinit_error2;
+ }
+
+ ret = pqos_cap_get_type(g_rdt->pqos_cap, PQOS_CAP_TYPE_MON, &g_rdt->cap_mon);
+ if (ret == PQOS_RETVAL_PARAM) {
+ ERROR(RDT_PLUGIN ": Error retrieving monitoring capabilities.");
+ goto rdt_preinit_error2;
+ }
+
+ if (g_rdt->cap_mon == NULL) {
+ ERROR(
+ RDT_PLUGIN
+ ": Monitoring capability not detected. Nothing to do for the plugin.");
+ goto rdt_preinit_error2;
+ }
+
+ return (0);
+
+rdt_preinit_error2:
+ pqos_fini();
+
+rdt_preinit_error1:
+
+ sfree(g_rdt);
+
+ return (-1);
+}
+
+static int rdt_config(oconfig_item_t *ci) {
+ int ret = 0;
+
+ ret = rdt_preinit();
+ if (ret != 0)
+ return ret;
+
+ for (int i = 0; i < ci->children_num; i++) {
+ oconfig_item_t *child = ci->children + i;
+
+ if (strcasecmp("Cores", child->key) == 0) {
+
+ ret = rdt_config_cgroups(child);
+ if (ret != 0)
+ return ret;
+
+#if COLLECT_DEBUG
+ rdt_dump_cgroups();
+#endif /* COLLECT_DEBUG */
+
+ } else {
+ ERROR(RDT_PLUGIN ": Unknown configuration parameter \"%s\".", child->key);
+ }
+ }
+
+ return (0);
+}
+
+static void rdt_submit_derive(char *cgroup, char *type, char *type_instance,
+ derive_t value) {
+ value_list_t vl = VALUE_LIST_INIT;
+
+ vl.values = &(value_t){.derive = value};
+ vl.values_len = 1;
+
+ sstrncpy(vl.plugin, RDT_PLUGIN, sizeof(vl.plugin));
+ snprintf(vl.plugin_instance, sizeof(vl.plugin_instance), "%s", cgroup);
+ sstrncpy(vl.type, type, sizeof(vl.type));
+ if (type_instance)
+ sstrncpy(vl.type_instance, type_instance, sizeof(vl.type_instance));
+
+ plugin_dispatch_values(&vl);
+}
+
+static void rdt_submit_gauge(char *cgroup, char *type, char *type_instance,
+ gauge_t value) {
+ value_list_t vl = VALUE_LIST_INIT;
+
+ vl.values = &(value_t){.gauge = value};
+ vl.values_len = 1;
+
+ sstrncpy(vl.plugin, RDT_PLUGIN, sizeof(vl.plugin));
+ snprintf(vl.plugin_instance, sizeof(vl.plugin_instance), "%s", cgroup);
+ sstrncpy(vl.type, type, sizeof(vl.type));
+ if (type_instance)
+ sstrncpy(vl.type_instance, type_instance, sizeof(vl.type_instance));
+
+ plugin_dispatch_values(&vl);
+}
+
+static int rdt_read(__attribute__((unused)) user_data_t *ud) {
+ int ret;
+
+ if (g_rdt == NULL) {
+ ERROR(RDT_PLUGIN ": rdt_read: plugin not initialized.");
+ return (-EINVAL);
+ }
+
+ ret = pqos_mon_poll(&g_rdt->pgroups[0], (unsigned)g_rdt->num_groups);
+ if (ret != PQOS_RETVAL_OK) {
+ ERROR(RDT_PLUGIN ": Failed to poll monitoring data.");
+ return (-1);
+ }
+
+#if COLLECT_DEBUG
+ rdt_dump_data();
+#endif /* COLLECT_DEBUG */
+
+ for (int i = 0; i < g_rdt->num_groups; i++) {
+ enum pqos_mon_event mbm_events =
+ (PQOS_MON_EVENT_LMEM_BW | PQOS_MON_EVENT_TMEM_BW |
+ PQOS_MON_EVENT_RMEM_BW);
+
+ const struct pqos_event_values *pv = &g_rdt->pgroups[i]->values;
+
+ /* Submit only monitored events data */
+
+ if (g_rdt->cgroups[i].events & PQOS_MON_EVENT_L3_OCCUP)
+ rdt_submit_gauge(g_rdt->cgroups[i].desc, "bytes", "llc", pv->llc);
+
+ if (g_rdt->cgroups[i].events & PQOS_PERF_EVENT_IPC)
+ rdt_submit_gauge(g_rdt->cgroups[i].desc, "ipc", NULL, pv->ipc);
+
+ if (g_rdt->cgroups[i].events & mbm_events) {
+ rdt_submit_derive(g_rdt->cgroups[i].desc, "memory_bandwidth", "local",
+ pv->mbm_local_delta);
+ rdt_submit_derive(g_rdt->cgroups[i].desc, "memory_bandwidth", "remote",
+ pv->mbm_remote_delta);
+ }
+ }
+
+ return (0);
+}
+
+static int rdt_init(void) {
+ int ret;
+
+ ret = rdt_preinit();
+ if (ret != 0)
+ return ret;
+
+ /* Start monitoring */
+ for (int i = 0; i < g_rdt->num_groups; i++) {
+ rdt_core_group_t *cg = &g_rdt->cgroups[i];
+
+ ret = pqos_mon_start(cg->num_cores, cg->cores, cg->events, (void *)cg->desc,
+ g_rdt->pgroups[i]);
+
+ if (ret != PQOS_RETVAL_OK)
+ ERROR(RDT_PLUGIN ": Error starting monitoring group %s (pqos status=%d)",
+ cg->desc, ret);
+ }
+
+ return (0);
+}
+
+static int rdt_shutdown(void) {
+ int ret;
+
+ DEBUG(RDT_PLUGIN ": rdt_shutdown.");
+
+ if (g_rdt == NULL)
+ return (0);
+
+ /* Stop monitoring */
+ for (int i = 0; i < g_rdt->num_groups; i++) {
+ pqos_mon_stop(g_rdt->pgroups[i]);
+ }
+
+ ret = pqos_fini();
+ if (ret != PQOS_RETVAL_OK)
+ ERROR(RDT_PLUGIN ": Error shutting down PQoS library.");
+
+ rdt_free_cgroups();
+ sfree(g_rdt);
+
+ return (0);
+}
+
+void module_register(void) {
+ plugin_register_init(RDT_PLUGIN, rdt_init);
+ plugin_register_complex_config(RDT_PLUGIN, rdt_config);
+ plugin_register_complex_read(NULL, RDT_PLUGIN, rdt_read, 0, NULL);
+ plugin_register_shutdown(RDT_PLUGIN, rdt_shutdown);
+}
projects / collectd.git / commitdiff