numam-dpdk/examples/vm_power_manager/power_manager.c
Rory Sexton 95f648ff9e examples/vm_power: make branch ratio threshold per core
This modification allows for the branch ratio threshold to be set
per core rather than system wide. This gives greater flexibility to
the branch ratio monitoring allowing it to manage different
workloads with different characteristics on the same system.

Signed-off-by: Rory Sexton <rory.sexton@intel.com>
Reviewed-by: David Hunt <david.hunt@intel.com>
Acked-by: Reshma Pattan <reshma.pattan@intel.com>
2020-07-17 14:40:56 +02:00

257 lines
5.3 KiB
C

/* SPDX-License-Identifier: BSD-3-Clause
* Copyright(c) 2010-2014 Intel Corporation
*/
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <inttypes.h>
#include <fcntl.h>
#include <unistd.h>
#include <dirent.h>
#include <errno.h>
#include <sys/sysinfo.h>
#include <sys/types.h>
#include <rte_log.h>
#include <rte_power.h>
#include <rte_spinlock.h>
#include "channel_manager.h"
#include "power_manager.h"
#include "oob_monitor.h"
#define POWER_SCALE_CORE(DIRECTION, core_num , ret) do { \
if (core_num >= ci.core_count) \
return -1; \
if (!(ci.cd[core_num].global_enabled_cpus)) \
return -1; \
rte_spinlock_lock(&global_core_freq_info[core_num].power_sl); \
ret = rte_power_freq_##DIRECTION(core_num); \
rte_spinlock_unlock(&global_core_freq_info[core_num].power_sl); \
} while (0)
struct freq_info {
rte_spinlock_t power_sl;
uint32_t freqs[RTE_MAX_LCORE_FREQS];
unsigned num_freqs;
} __rte_cache_aligned;
static struct freq_info global_core_freq_info[RTE_MAX_LCORE];
struct core_info ci;
#define SYSFS_CPU_PATH "/sys/devices/system/cpu/cpu%u/topology/core_id"
struct core_info *
get_core_info(void)
{
return &ci;
}
int
core_info_init(void)
{
struct core_info *ci;
int i;
ci = get_core_info();
ci->core_count = get_nprocs_conf();
ci->cd = malloc(ci->core_count * sizeof(struct core_details));
memset(ci->cd, 0, ci->core_count * sizeof(struct core_details));
if (!ci->cd) {
RTE_LOG(ERR, POWER_MANAGER, "Failed to allocate memory for core info.");
return -1;
}
for (i = 0; i < ci->core_count; i++) {
ci->cd[i].global_enabled_cpus = 1;
ci->cd[i].branch_ratio_threshold = BRANCH_RATIO_THRESHOLD;
}
printf("%d cores in system\n", ci->core_count);
return 0;
}
int
power_manager_init(void)
{
unsigned int i, num_cpus = 0, num_freqs = 0;
int ret = 0;
struct core_info *ci;
unsigned int max_core_num;
rte_power_set_env(PM_ENV_NOT_SET);
ci = get_core_info();
if (!ci) {
RTE_LOG(ERR, POWER_MANAGER,
"Failed to get core info!\n");
return -1;
}
if (ci->core_count > RTE_MAX_LCORE)
max_core_num = RTE_MAX_LCORE;
else
max_core_num = ci->core_count;
for (i = 0; i < max_core_num; i++) {
if (ci->cd[i].global_enabled_cpus) {
if (rte_power_init(i) < 0)
RTE_LOG(ERR, POWER_MANAGER,
"Unable to initialize power manager "
"for core %u\n", i);
num_cpus++;
num_freqs = rte_power_freqs(i,
global_core_freq_info[i].freqs,
RTE_MAX_LCORE_FREQS);
if (num_freqs == 0) {
RTE_LOG(ERR, POWER_MANAGER,
"Unable to get frequency list for core %u\n",
i);
ci->cd[i].oob_enabled = 0;
ret = -1;
}
global_core_freq_info[i].num_freqs = num_freqs;
rte_spinlock_init(&global_core_freq_info[i].power_sl);
}
if (ci->cd[i].oob_enabled)
add_core_to_monitor(i);
}
RTE_LOG(INFO, POWER_MANAGER, "Managing %u cores out of %u available host cores\n",
num_cpus, ci->core_count);
return ret;
}
uint32_t
power_manager_get_current_frequency(unsigned core_num)
{
uint32_t freq, index;
if (core_num >= RTE_MAX_LCORE) {
RTE_LOG(ERR, POWER_MANAGER, "Core(%u) is out of range 0...%d\n",
core_num, RTE_MAX_LCORE-1);
return -1;
}
if (!(ci.cd[core_num].global_enabled_cpus))
return 0;
rte_spinlock_lock(&global_core_freq_info[core_num].power_sl);
index = rte_power_get_freq(core_num);
rte_spinlock_unlock(&global_core_freq_info[core_num].power_sl);
if (index >= RTE_MAX_LCORE_FREQS)
freq = 0;
else
freq = global_core_freq_info[core_num].freqs[index];
return freq;
}
int
power_manager_exit(void)
{
unsigned int i;
int ret = 0;
struct core_info *ci;
unsigned int max_core_num;
ci = get_core_info();
if (!ci) {
RTE_LOG(ERR, POWER_MANAGER,
"Failed to get core info!\n");
return -1;
}
if (ci->core_count > RTE_MAX_LCORE)
max_core_num = RTE_MAX_LCORE;
else
max_core_num = ci->core_count;
for (i = 0; i < max_core_num; i++) {
if (ci->cd[i].global_enabled_cpus) {
if (rte_power_exit(i) < 0) {
RTE_LOG(ERR, POWER_MANAGER, "Unable to shutdown power manager "
"for core %u\n", i);
ret = -1;
}
ci->cd[i].global_enabled_cpus = 0;
}
remove_core_from_monitor(i);
}
return ret;
}
int
power_manager_scale_core_up(unsigned core_num)
{
int ret = 0;
POWER_SCALE_CORE(up, core_num, ret);
return ret;
}
int
power_manager_scale_core_down(unsigned core_num)
{
int ret = 0;
POWER_SCALE_CORE(down, core_num, ret);
return ret;
}
int
power_manager_scale_core_min(unsigned core_num)
{
int ret = 0;
POWER_SCALE_CORE(min, core_num, ret);
return ret;
}
int
power_manager_scale_core_max(unsigned core_num)
{
int ret = 0;
POWER_SCALE_CORE(max, core_num, ret);
return ret;
}
int
power_manager_enable_turbo_core(unsigned int core_num)
{
int ret = 0;
POWER_SCALE_CORE(enable_turbo, core_num, ret);
return ret;
}
int
power_manager_disable_turbo_core(unsigned int core_num)
{
int ret = 0;
POWER_SCALE_CORE(disable_turbo, core_num, ret);
return ret;
}
int
power_manager_scale_core_med(unsigned int core_num)
{
int ret = 0;
struct core_info *ci;
ci = get_core_info();
if (core_num >= RTE_MAX_LCORE)
return -1;
if (!(ci->cd[core_num].global_enabled_cpus))
return -1;
rte_spinlock_lock(&global_core_freq_info[core_num].power_sl);
ret = rte_power_set_freq(core_num,
global_core_freq_info[core_num].num_freqs / 2);
rte_spinlock_unlock(&global_core_freq_info[core_num].power_sl);
return ret;
}