Viewing file:  ebpf_swap.c (21.8 KB) -rw-r--r--
Select action/file-type:
 (+) |  (+) |  (+) | Code (+) | Session (+) |  (+) | SDB (+) |  (+) |  (+) |  (+) |  (+) |  (+) |
// SPDX-License-Identifier: GPL-3.0-or-later
#include "ebpf.h"
#include "ebpf_swap.h"
static char *swap_dimension_name[NETDATA_SWAP_END] = { "read", "write" };
static netdata_syscall_stat_t swap_aggregated_data[NETDATA_SWAP_END];
static netdata_publish_syscall_t swap_publish_aggregated[NETDATA_SWAP_END];
static int read_thread_closed = 1;
netdata_publish_swap_t *swap_vector = NULL;
static netdata_idx_t swap_hash_values[NETDATA_SWAP_END];
static netdata_idx_t *swap_values = NULL;
netdata_publish_swap_t **swap_pid = NULL;
struct config swap_config = { .first_section = NULL,
.last_section = NULL,
.mutex = NETDATA_MUTEX_INITIALIZER,
.index = { .avl_tree = { .root = NULL, .compar = appconfig_section_compare },
.rwlock = AVL_LOCK_INITIALIZER } };
static ebpf_local_maps_t swap_maps[] = {{.name = "tbl_pid_swap", .internal_input = ND_EBPF_DEFAULT_PID_SIZE,
.user_input = 0,
.type = NETDATA_EBPF_MAP_RESIZABLE | NETDATA_EBPF_MAP_PID,
.map_fd = ND_EBPF_MAP_FD_NOT_INITIALIZED},
{.name = "swap_ctrl", .internal_input = NETDATA_CONTROLLER_END,
.user_input = 0,
.type = NETDATA_EBPF_MAP_CONTROLLER,
.map_fd = ND_EBPF_MAP_FD_NOT_INITIALIZED},
{.name = "tbl_swap", .internal_input = NETDATA_SWAP_END,
.user_input = 0,
.type = NETDATA_EBPF_MAP_STATIC,
.map_fd = ND_EBPF_MAP_FD_NOT_INITIALIZED},
{.name = NULL, .internal_input = 0, .user_input = 0}};
static struct bpf_link **probe_links = NULL;
static struct bpf_object *objects = NULL;
struct netdata_static_thread swap_threads = {"SWAP KERNEL", NULL, NULL, 1,
NULL, NULL, NULL};
/*****************************************************************
*
* FUNCTIONS TO CLOSE THE THREAD
*
*****************************************************************/
/**
* Clean swap structure
*/
void clean_swap_pid_structures() {
struct pid_stat *pids = root_of_pids;
while (pids) {
freez(swap_pid[pids->pid]);
pids = pids->next;
}
}
/**
* Clean up the main thread.
*
* @param ptr thread data.
*/
static void ebpf_swap_cleanup(void *ptr)
{
ebpf_module_t *em = (ebpf_module_t *)ptr;
if (!em->enabled)
return;
heartbeat_t hb;
heartbeat_init(&hb);
uint32_t tick = 2 * USEC_PER_MS;
while (!read_thread_closed) {
usec_t dt = heartbeat_next(&hb, tick);
UNUSED(dt);
}
ebpf_cleanup_publish_syscall(swap_publish_aggregated);
freez(swap_vector);
freez(swap_values);
if (probe_links) {
struct bpf_program *prog;
size_t i = 0 ;
bpf_object__for_each_program(prog, objects) {
bpf_link__destroy(probe_links[i]);
i++;
}
bpf_object__close(objects);
}
}
/*****************************************************************
*
* COLLECTOR THREAD
*
*****************************************************************/
/**
* Apps Accumulator
*
* Sum all values read from kernel and store in the first address.
*
* @param out the vector with read values.
*/
static void swap_apps_accumulator(netdata_publish_swap_t *out)
{
int i, end = (running_on_kernel >= NETDATA_KERNEL_V4_15) ? ebpf_nprocs : 1;
netdata_publish_swap_t *total = &out[0];
for (i = 1; i < end; i++) {
netdata_publish_swap_t *w = &out[i];
total->write += w->write;
total->read += w->read;
}
}
/**
* Fill PID
*
* Fill PID structures
*
* @param current_pid pid that we are collecting data
* @param out values read from hash tables;
*/
static void swap_fill_pid(uint32_t current_pid, netdata_publish_swap_t *publish)
{
netdata_publish_swap_t *curr = swap_pid[current_pid];
if (!curr) {
curr = callocz(1, sizeof(netdata_publish_swap_t));
swap_pid[current_pid] = curr;
}
memcpy(curr, publish, sizeof(netdata_publish_swap_t));
}
/**
* Update cgroup
*
* Update cgroup data based in
*/
static void ebpf_update_swap_cgroup()
{
ebpf_cgroup_target_t *ect ;
netdata_publish_swap_t *cv = swap_vector;
int fd = swap_maps[NETDATA_PID_SWAP_TABLE].map_fd;
size_t length = sizeof(netdata_publish_swap_t)*ebpf_nprocs;
pthread_mutex_lock(&mutex_cgroup_shm);
for (ect = ebpf_cgroup_pids; ect; ect = ect->next) {
struct pid_on_target2 *pids;
for (pids = ect->pids; pids; pids = pids->next) {
int pid = pids->pid;
netdata_publish_swap_t *out = &pids->swap;
if (likely(swap_pid) && swap_pid[pid]) {
netdata_publish_swap_t *in = swap_pid[pid];
memcpy(out, in, sizeof(netdata_publish_swap_t));
} else {
memset(cv, 0, length);
if (!bpf_map_lookup_elem(fd, &pid, cv)) {
swap_apps_accumulator(cv);
memcpy(out, cv, sizeof(netdata_publish_swap_t));
}
}
}
}
pthread_mutex_unlock(&mutex_cgroup_shm);
}
/**
* Read APPS table
*
* Read the apps table and store data inside the structure.
*/
static void read_apps_table()
{
netdata_publish_swap_t *cv = swap_vector;
uint32_t key;
struct pid_stat *pids = root_of_pids;
int fd = swap_maps[NETDATA_PID_SWAP_TABLE].map_fd;
size_t length = sizeof(netdata_publish_swap_t)*ebpf_nprocs;
while (pids) {
key = pids->pid;
if (bpf_map_lookup_elem(fd, &key, cv)) {
pids = pids->next;
continue;
}
swap_apps_accumulator(cv);
swap_fill_pid(key, cv);
// We are cleaning to avoid passing data read from one process to other.
memset(cv, 0, length);
pids = pids->next;
}
}
/**
* Send global
*
* Send global charts to Netdata
*/
static void swap_send_global()
{
write_io_chart(NETDATA_MEM_SWAP_CHART, NETDATA_EBPF_SYSTEM_GROUP,
swap_publish_aggregated[NETDATA_KEY_SWAP_WRITEPAGE_CALL].dimension,
(long long) swap_hash_values[NETDATA_KEY_SWAP_WRITEPAGE_CALL],
swap_publish_aggregated[NETDATA_KEY_SWAP_READPAGE_CALL].dimension,
(long long) swap_hash_values[NETDATA_KEY_SWAP_READPAGE_CALL]);
}
/**
* Read global counter
*
* Read the table with number of calls to all functions
*/
static void read_global_table()
{
netdata_idx_t *stored = swap_values;
netdata_idx_t *val = swap_hash_values;
int fd = swap_maps[NETDATA_SWAP_GLOBAL_TABLE].map_fd;
uint32_t i, end = NETDATA_SWAP_END;
for (i = NETDATA_KEY_SWAP_READPAGE_CALL; i < end; i++) {
if (!bpf_map_lookup_elem(fd, &i, stored)) {
int j;
int last = ebpf_nprocs;
netdata_idx_t total = 0;
for (j = 0; j < last; j++)
total += stored[j];
val[i] = total;
}
}
}
/**
* Swap read hash
*
* This is the thread callback.
*
* @param ptr It is a NULL value for this thread.
*
* @return It always returns NULL.
*/
void *ebpf_swap_read_hash(void *ptr)
{
read_thread_closed = 0;
heartbeat_t hb;
heartbeat_init(&hb);
ebpf_module_t *em = (ebpf_module_t *)ptr;
usec_t step = NETDATA_SWAP_SLEEP_MS * em->update_every;
while (!close_ebpf_plugin) {
usec_t dt = heartbeat_next(&hb, step);
(void)dt;
read_global_table();
}
read_thread_closed = 1;
return NULL;
}
/**
* Sum PIDs
*
* Sum values for all targets.
*
* @param swap
* @param root
*/
static void ebpf_swap_sum_pids(netdata_publish_swap_t *swap, struct pid_on_target *root)
{
uint64_t local_read = 0;
uint64_t local_write = 0;
while (root) {
int32_t pid = root->pid;
netdata_publish_swap_t *w = swap_pid[pid];
if (w) {
local_write += w->write;
local_read += w->read;
}
root = root->next;
}
// These conditions were added, because we are using incremental algorithm
swap->write = (local_write >= swap->write) ? local_write : swap->write;
swap->read = (local_read >= swap->read) ? local_read : swap->read;
}
/**
* Send data to Netdata calling auxiliar functions.
*
* @param root the target list.
*/
void ebpf_swap_send_apps_data(struct target *root)
{
struct target *w;
for (w = root; w; w = w->next) {
if (unlikely(w->exposed && w->processes)) {
ebpf_swap_sum_pids(&w->swap, w->root_pid);
}
}
write_begin_chart(NETDATA_APPS_FAMILY, NETDATA_MEM_SWAP_READ_CHART);
for (w = root; w; w = w->next) {
if (unlikely(w->exposed && w->processes)) {
write_chart_dimension(w->name, (long long) w->swap.read);
}
}
write_end_chart();
write_begin_chart(NETDATA_APPS_FAMILY, NETDATA_MEM_SWAP_WRITE_CHART);
for (w = root; w; w = w->next) {
if (unlikely(w->exposed && w->processes)) {
write_chart_dimension(w->name, (long long) w->swap.write);
}
}
write_end_chart();
}
/**
* Sum PIDs
*
* Sum values for all targets.
*
* @param swap
* @param root
*/
static void ebpf_swap_sum_cgroup_pids(netdata_publish_swap_t *swap, struct pid_on_target2 *pids)
{
uint64_t local_read = 0;
uint64_t local_write = 0;
while (pids) {
netdata_publish_swap_t *w = &pids->swap;
local_write += w->write;
local_read += w->read;
pids = pids->next;
}
// These conditions were added, because we are using incremental algorithm
swap->write = (local_write >= swap->write) ? local_write : swap->write;
swap->read = (local_read >= swap->read) ? local_read : swap->read;
}
/**
* Send Systemd charts
*
* Send collected data to Netdata.
*
* @return It returns the status for chart creation, if it is necessary to remove a specific dimension, zero is returned
* otherwise function returns 1 to avoid chart recreation
*/
static int ebpf_send_systemd_swap_charts()
{
int ret = 1;
ebpf_cgroup_target_t *ect;
write_begin_chart(NETDATA_SERVICE_FAMILY, NETDATA_MEM_SWAP_READ_CHART);
for (ect = ebpf_cgroup_pids; ect ; ect = ect->next) {
if (unlikely(ect->systemd) && unlikely(ect->updated)) {
write_chart_dimension(ect->name, (long long) ect->publish_systemd_swap.read);
} else
ret = 0;
}
write_end_chart();
write_begin_chart(NETDATA_SERVICE_FAMILY, NETDATA_MEM_SWAP_WRITE_CHART);
for (ect = ebpf_cgroup_pids; ect ; ect = ect->next) {
if (unlikely(ect->systemd) && unlikely(ect->updated)) {
write_chart_dimension(ect->name, (long long) ect->publish_systemd_swap.write);
}
}
write_end_chart();
return ret;
}
/**
* Create specific swap charts
*
* Create charts for cgroup/application.
*
* @param type the chart type.
* @param update_every value to overwrite the update frequency set by the server.
*/
static void ebpf_create_specific_swap_charts(char *type, int update_every)
{
ebpf_create_chart(type, NETDATA_MEM_SWAP_READ_CHART,
"Calls to function <code>swap_readpage</code>.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_CGROUP_SWAP_SUBMENU,
NETDATA_CGROUP_SWAP_READ_CONTEXT, NETDATA_EBPF_CHART_TYPE_LINE,
NETDATA_CHART_PRIO_CGROUPS_CONTAINERS + 5100,
ebpf_create_global_dimension,
swap_publish_aggregated, 1, update_every, NETDATA_EBPF_MODULE_NAME_SWAP);
ebpf_create_chart(type, NETDATA_MEM_SWAP_WRITE_CHART,
"Calls to function <code>swap_writepage</code>.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_CGROUP_SWAP_SUBMENU,
NETDATA_CGROUP_SWAP_WRITE_CONTEXT, NETDATA_EBPF_CHART_TYPE_LINE,
NETDATA_CHART_PRIO_CGROUPS_CONTAINERS + 5101,
ebpf_create_global_dimension,
&swap_publish_aggregated[NETDATA_KEY_SWAP_WRITEPAGE_CALL], 1,
update_every, NETDATA_EBPF_MODULE_NAME_SWAP);
}
/**
* Create specific swap charts
*
* Create charts for cgroup/application.
*
* @param type the chart type.
* @param update_every value to overwrite the update frequency set by the server.
*/
static void ebpf_obsolete_specific_swap_charts(char *type, int update_every)
{
ebpf_write_chart_obsolete(type, NETDATA_MEM_SWAP_READ_CHART,"Calls to function <code>swap_readpage</code>.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_CGROUP_SWAP_SUBMENU,
NETDATA_EBPF_CHART_TYPE_LINE, NETDATA_CGROUP_SWAP_READ_CONTEXT,
NETDATA_CHART_PRIO_CGROUPS_CONTAINERS + 5100, update_every);
ebpf_write_chart_obsolete(type, NETDATA_MEM_SWAP_WRITE_CHART, "Calls to function <code>swap_writepage</code>.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_CGROUP_SWAP_SUBMENU,
NETDATA_EBPF_CHART_TYPE_LINE, NETDATA_CGROUP_SWAP_WRITE_CONTEXT,
NETDATA_CHART_PRIO_CGROUPS_CONTAINERS + 5101, update_every);
}
/*
* Send Specific Swap data
*
* Send data for specific cgroup/apps.
*
* @param type chart type
* @param values structure with values that will be sent to netdata
*/
static void ebpf_send_specific_swap_data(char *type, netdata_publish_swap_t *values)
{
write_begin_chart(type, NETDATA_MEM_SWAP_READ_CHART);
write_chart_dimension(swap_publish_aggregated[NETDATA_KEY_SWAP_READPAGE_CALL].name, (long long) values->read);
write_end_chart();
write_begin_chart(type, NETDATA_MEM_SWAP_WRITE_CHART);
write_chart_dimension(swap_publish_aggregated[NETDATA_KEY_SWAP_WRITEPAGE_CALL].name, (long long) values->write);
write_end_chart();
}
/**
* Create Systemd Swap Charts
*
* Create charts when systemd is enabled
*
* @param update_every value to overwrite the update frequency set by the server.
**/
static void ebpf_create_systemd_swap_charts(int update_every)
{
ebpf_create_charts_on_systemd(NETDATA_MEM_SWAP_READ_CHART,
"Calls to <code>swap_readpage</code>.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_CGROUP_SWAP_SUBMENU,
NETDATA_EBPF_CHART_TYPE_STACKED, 20191,
ebpf_algorithms[NETDATA_EBPF_INCREMENTAL_IDX], NETDATA_SYSTEMD_SWAP_READ_CONTEXT,
NETDATA_EBPF_MODULE_NAME_SWAP, update_every);
ebpf_create_charts_on_systemd(NETDATA_MEM_SWAP_WRITE_CHART,
"Calls to function <code>swap_writepage</code>.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_CGROUP_SWAP_SUBMENU,
NETDATA_EBPF_CHART_TYPE_STACKED, 20192,
ebpf_algorithms[NETDATA_EBPF_INCREMENTAL_IDX], NETDATA_SYSTEMD_SWAP_WRITE_CONTEXT,
NETDATA_EBPF_MODULE_NAME_SWAP, update_every);
}
/**
* Send data to Netdata calling auxiliar functions.
*
* @param update_every value to overwrite the update frequency set by the server.
*/
void ebpf_swap_send_cgroup_data(int update_every)
{
if (!ebpf_cgroup_pids)
return;
pthread_mutex_lock(&mutex_cgroup_shm);
ebpf_cgroup_target_t *ect;
for (ect = ebpf_cgroup_pids; ect ; ect = ect->next) {
ebpf_swap_sum_cgroup_pids(&ect->publish_systemd_swap, ect->pids);
}
int has_systemd = shm_ebpf_cgroup.header->systemd_enabled;
if (has_systemd) {
static int systemd_charts = 0;
if (!systemd_charts) {
ebpf_create_systemd_swap_charts(update_every);
systemd_charts = 1;
fflush(stdout);
}
systemd_charts = ebpf_send_systemd_swap_charts();
}
for (ect = ebpf_cgroup_pids; ect ; ect = ect->next) {
if (ect->systemd)
continue;
if (!(ect->flags & NETDATA_EBPF_CGROUP_HAS_SWAP_CHART) && ect->updated) {
ebpf_create_specific_swap_charts(ect->name, update_every);
ect->flags |= NETDATA_EBPF_CGROUP_HAS_SWAP_CHART;
}
if (ect->flags & NETDATA_EBPF_CGROUP_HAS_SWAP_CHART) {
if (ect->updated) {
ebpf_send_specific_swap_data(ect->name, &ect->publish_systemd_swap);
} else {
ebpf_obsolete_specific_swap_charts(ect->name, update_every);
ect->flags &= ~NETDATA_EBPF_CGROUP_HAS_SWAP_CHART;
}
}
}
pthread_mutex_unlock(&mutex_cgroup_shm);
}
/**
* Main loop for this collector.
*/
static void swap_collector(ebpf_module_t *em)
{
swap_threads.thread = mallocz(sizeof(netdata_thread_t));
swap_threads.start_routine = ebpf_swap_read_hash;
netdata_thread_create(swap_threads.thread, swap_threads.name, NETDATA_THREAD_OPTION_JOINABLE,
ebpf_swap_read_hash, em);
int apps = em->apps_charts;
int cgroup = em->cgroup_charts;
int update_every = em->update_every;
int counter = update_every - 1;
while (!close_ebpf_plugin) {
pthread_mutex_lock(&collect_data_mutex);
pthread_cond_wait(&collect_data_cond_var, &collect_data_mutex);
if (++counter == update_every) {
counter = 0;
if (apps)
read_apps_table();
if (cgroup)
ebpf_update_swap_cgroup();
pthread_mutex_lock(&lock);
swap_send_global();
if (apps)
ebpf_swap_send_apps_data(apps_groups_root_target);
if (cgroup)
ebpf_swap_send_cgroup_data(update_every);
pthread_mutex_unlock(&lock);
}
pthread_mutex_unlock(&collect_data_mutex);
}
}
/*****************************************************************
*
* INITIALIZE THREAD
*
*****************************************************************/
/**
* Create apps charts
*
* Call ebpf_create_chart to create the charts on apps submenu.
*
* @param em a pointer to the structure with the default values.
*/
void ebpf_swap_create_apps_charts(struct ebpf_module *em, void *ptr)
{
struct target *root = ptr;
ebpf_create_charts_on_apps(NETDATA_MEM_SWAP_READ_CHART,
"Calls to function <code>swap_readpage</code>.",
EBPF_COMMON_DIMENSION_CALL,
NETDATA_SWAP_SUBMENU,
NETDATA_EBPF_CHART_TYPE_STACKED,
20191,
ebpf_algorithms[NETDATA_EBPF_INCREMENTAL_IDX],
root, em->update_every, NETDATA_EBPF_MODULE_NAME_SWAP);
ebpf_create_charts_on_apps(NETDATA_MEM_SWAP_WRITE_CHART,
"Calls to function <code>swap_writepage</code>.",
EBPF_COMMON_DIMENSION_CALL,
NETDATA_SWAP_SUBMENU,
NETDATA_EBPF_CHART_TYPE_STACKED,
20192,
ebpf_algorithms[NETDATA_EBPF_INCREMENTAL_IDX],
root, em->update_every, NETDATA_EBPF_MODULE_NAME_SWAP);
}
/**
* Allocate vectors used with this thread.
*
* We are not testing the return, because callocz does this and shutdown the software
* case it was not possible to allocate.
*
* @param apps is apps enabled?
*/
static void ebpf_swap_allocate_global_vectors(int apps)
{
if (apps)
swap_pid = callocz((size_t)pid_max, sizeof(netdata_publish_swap_t *));
swap_vector = callocz((size_t)ebpf_nprocs, sizeof(netdata_publish_swap_t));
swap_values = callocz((size_t)ebpf_nprocs, sizeof(netdata_idx_t));
memset(swap_hash_values, 0, sizeof(swap_hash_values));
}
/*****************************************************************
*
* MAIN THREAD
*
*****************************************************************/
/**
* Create global charts
*
* Call ebpf_create_chart to create the charts for the collector.
*
* @param update_every value to overwrite the update frequency set by the server.
*/
static void ebpf_create_swap_charts(int update_every)
{
ebpf_create_chart(NETDATA_EBPF_SYSTEM_GROUP, NETDATA_MEM_SWAP_CHART,
"Calls to internal functions used to access swap.",
EBPF_COMMON_DIMENSION_CALL, NETDATA_SYSTEM_SWAP_SUBMENU,
NULL,
NETDATA_EBPF_CHART_TYPE_LINE,
202,
ebpf_create_global_dimension,
swap_publish_aggregated, NETDATA_SWAP_END,
update_every, NETDATA_EBPF_MODULE_NAME_SWAP);
}
/**
* SWAP thread
*
* Thread used to make swap thread
*
* @param ptr a pointer to `struct ebpf_module`
*
* @return It always return NULL
*/
void *ebpf_swap_thread(void *ptr)
{
netdata_thread_cleanup_push(ebpf_swap_cleanup, ptr);
ebpf_module_t *em = (ebpf_module_t *)ptr;
em->maps = swap_maps;
ebpf_update_pid_table(&swap_maps[NETDATA_PID_SWAP_TABLE], em);
if (!em->enabled)
goto endswap;
probe_links = ebpf_load_program(ebpf_plugin_dir, em, kernel_string, &objects);
if (!probe_links) {
goto endswap;
}
ebpf_swap_allocate_global_vectors(em->apps_charts);
int algorithms[NETDATA_SWAP_END] = { NETDATA_EBPF_INCREMENTAL_IDX, NETDATA_EBPF_INCREMENTAL_IDX };
ebpf_global_labels(swap_aggregated_data, swap_publish_aggregated, swap_dimension_name, swap_dimension_name,
algorithms, NETDATA_SWAP_END);
pthread_mutex_lock(&lock);
ebpf_create_swap_charts(em->update_every);
pthread_mutex_unlock(&lock);
swap_collector(em);
endswap:
netdata_thread_cleanup_pop(1);
return NULL;
}
|