/**
|
* \file
|
* Copyright 2008-2011 Novell Inc
|
* Copyright 2011 Xamarin Inc
|
* Licensed under the MIT license. See LICENSE file in the project root for full license information.
|
*/
|
|
#include "config.h"
|
#include "utils/mono-proclib.h"
|
#include "utils/mono-time.h"
|
|
#include <stdlib.h>
|
#include <stdio.h>
|
#include <string.h>
|
#include <fcntl.h>
|
#ifdef HAVE_UNISTD_H
|
#include <unistd.h>
|
#endif
|
#ifdef HAVE_SCHED_GETAFFINITY
|
#include <sched.h>
|
#endif
|
|
#if defined(_POSIX_VERSION)
|
#ifdef HAVE_SYS_ERRNO_H
|
#include <sys/errno.h>
|
#endif
|
#include <sys/param.h>
|
#include <errno.h>
|
#ifdef HAVE_SYS_TYPES_H
|
#include <sys/types.h>
|
#endif
|
#ifdef HAVE_SYS_SYSCTL_H
|
#include <sys/sysctl.h>
|
#endif
|
#ifdef HAVE_SYS_RESOURCE_H
|
#include <sys/resource.h>
|
#endif
|
#endif
|
#if defined(__HAIKU__)
|
#include <os/kernel/OS.h>
|
#endif
|
#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__)
|
#include <sys/proc.h>
|
#if defined(__APPLE__)
|
#include <mach/mach.h>
|
#endif
|
#ifdef HAVE_SYS_USER_H
|
#include <sys/user.h>
|
#endif
|
#ifdef HAVE_STRUCT_KINFO_PROC_KP_PROC
|
# define kinfo_starttime_member kp_proc.p_starttime
|
# define kinfo_pid_member kp_proc.p_pid
|
# define kinfo_name_member kp_proc.p_comm
|
#elif defined(__NetBSD__)
|
# define kinfo_starttime_member p_ustart_sec
|
# define kinfo_pid_member p_pid
|
# define kinfo_name_member p_comm
|
#elif defined(__OpenBSD__)
|
// Can not figure out how to get the proc's start time on OpenBSD
|
# undef kinfo_starttime_member
|
# define kinfo_pid_member p_pid
|
# define kinfo_name_member p_comm
|
#else
|
#define kinfo_starttime_member ki_start
|
#define kinfo_pid_member ki_pid
|
#define kinfo_name_member ki_comm
|
#endif
|
#define USE_SYSCTL 1
|
#endif
|
|
#ifdef HAVE_SCHED_GETAFFINITY
|
# ifndef GLIBC_HAS_CPU_COUNT
|
static int
|
CPU_COUNT(cpu_set_t *set)
|
{
|
int i, count = 0;
|
|
for (int i = 0; i < CPU_SETSIZE; i++)
|
if (CPU_ISSET(i, set))
|
count++;
|
return count;
|
}
|
# endif
|
#endif
|
|
/**
|
* mono_process_list:
|
* \param size a pointer to a location where the size of the returned array is stored
|
* \returns an array of pid values for the processes currently running on the system.
|
* The size of the array is stored in \p size.
|
*/
|
gpointer*
|
mono_process_list (int *size)
|
{
|
#if USE_SYSCTL
|
int res, i;
|
#ifdef KERN_PROC2
|
int mib [6];
|
size_t data_len = sizeof (struct kinfo_proc2) * 400;
|
struct kinfo_proc2 *processes = g_malloc (data_len);
|
#else
|
int mib [4];
|
size_t data_len = sizeof (struct kinfo_proc) * 16;
|
struct kinfo_proc *processes;
|
int limit = 8;
|
#endif /* KERN_PROC2 */
|
void **buf = NULL;
|
|
if (size)
|
*size = 0;
|
|
#ifdef KERN_PROC2
|
if (!processes)
|
return NULL;
|
|
mib [0] = CTL_KERN;
|
mib [1] = KERN_PROC2;
|
mib [2] = KERN_PROC_ALL;
|
mib [3] = 0;
|
mib [4] = sizeof(struct kinfo_proc2);
|
mib [5] = 400; /* XXX */
|
|
res = sysctl (mib, 6, processes, &data_len, NULL, 0);
|
if (res < 0) {
|
g_free (processes);
|
return NULL;
|
}
|
#else
|
processes = NULL;
|
while (limit) {
|
mib [0] = CTL_KERN;
|
mib [1] = KERN_PROC;
|
mib [2] = KERN_PROC_ALL;
|
mib [3] = 0;
|
|
res = sysctl (mib, 3, NULL, &data_len, NULL, 0);
|
if (res)
|
return NULL;
|
processes = (struct kinfo_proc *) g_malloc (data_len);
|
res = sysctl (mib, 3, processes, &data_len, NULL, 0);
|
if (res < 0) {
|
g_free (processes);
|
if (errno != ENOMEM)
|
return NULL;
|
limit --;
|
} else {
|
break;
|
}
|
}
|
#endif /* KERN_PROC2 */
|
|
#ifdef KERN_PROC2
|
res = data_len/sizeof (struct kinfo_proc2);
|
#else
|
res = data_len/sizeof (struct kinfo_proc);
|
#endif /* KERN_PROC2 */
|
buf = (void **) g_realloc (buf, res * sizeof (void*));
|
for (i = 0; i < res; ++i)
|
buf [i] = GINT_TO_POINTER (processes [i].kinfo_pid_member);
|
g_free (processes);
|
if (size)
|
*size = res;
|
return buf;
|
#elif defined(__HAIKU__)
|
int32 cookie = 0;
|
int32 i = 0;
|
team_info ti;
|
system_info si;
|
|
get_system_info(&si);
|
void **buf = g_calloc(si.used_teams, sizeof(void*));
|
|
while (get_next_team_info(&cookie, &ti) == B_OK && i < si.used_teams) {
|
buf[i++] = GINT_TO_POINTER (ti.team);
|
}
|
*size = i;
|
|
return buf;
|
#else
|
const char *name;
|
void **buf = NULL;
|
int count = 0;
|
int i = 0;
|
GDir *dir = g_dir_open ("/proc/", 0, NULL);
|
if (!dir) {
|
if (size)
|
*size = 0;
|
return NULL;
|
}
|
while ((name = g_dir_read_name (dir))) {
|
int pid;
|
char *nend;
|
pid = strtol (name, &nend, 10);
|
if (pid <= 0 || nend == name || *nend)
|
continue;
|
if (i >= count) {
|
if (!count)
|
count = 16;
|
else
|
count *= 2;
|
buf = (void **)g_realloc (buf, count * sizeof (void*));
|
}
|
buf [i++] = GINT_TO_POINTER (pid);
|
}
|
g_dir_close (dir);
|
if (size)
|
*size = i;
|
return buf;
|
#endif
|
}
|
|
static G_GNUC_UNUSED char*
|
get_pid_status_item_buf (int pid, const char *item, char *rbuf, int blen, MonoProcessError *error)
|
{
|
char buf [256];
|
char *s;
|
FILE *f;
|
size_t len = strlen (item);
|
|
g_snprintf (buf, sizeof (buf), "/proc/%d/status", pid);
|
f = fopen (buf, "r");
|
if (!f) {
|
if (error)
|
*error = MONO_PROCESS_ERROR_NOT_FOUND;
|
return NULL;
|
}
|
while ((s = fgets (buf, sizeof (buf), f))) {
|
if (*item != *buf)
|
continue;
|
if (strncmp (buf, item, len))
|
continue;
|
s = buf + len;
|
while (g_ascii_isspace (*s)) s++;
|
if (*s++ != ':')
|
continue;
|
while (g_ascii_isspace (*s)) s++;
|
fclose (f);
|
len = strlen (s);
|
memcpy (rbuf, s, MIN (len, blen));
|
rbuf [MIN (len, blen) - 1] = 0;
|
if (error)
|
*error = MONO_PROCESS_ERROR_NONE;
|
return rbuf;
|
}
|
fclose (f);
|
if (error)
|
*error = MONO_PROCESS_ERROR_OTHER;
|
return NULL;
|
}
|
|
#if USE_SYSCTL
|
|
#ifdef KERN_PROC2
|
#define KINFO_PROC struct kinfo_proc2
|
#else
|
#define KINFO_PROC struct kinfo_proc
|
#endif
|
|
static gboolean
|
sysctl_kinfo_proc (gpointer pid, KINFO_PROC* processi)
|
{
|
int res;
|
size_t data_len = sizeof (KINFO_PROC);
|
|
#ifdef KERN_PROC2
|
int mib [6];
|
mib [0] = CTL_KERN;
|
mib [1] = KERN_PROC2;
|
mib [2] = KERN_PROC_PID;
|
mib [3] = GPOINTER_TO_UINT (pid);
|
mib [4] = sizeof(KINFO_PROC);
|
mib [5] = 400; /* XXX */
|
|
res = sysctl (mib, 6, processi, &data_len, NULL, 0);
|
#else
|
int mib [4];
|
mib [0] = CTL_KERN;
|
mib [1] = KERN_PROC;
|
mib [2] = KERN_PROC_PID;
|
mib [3] = GPOINTER_TO_UINT (pid);
|
|
res = sysctl (mib, 4, processi, &data_len, NULL, 0);
|
#endif /* KERN_PROC2 */
|
|
if (res < 0 || data_len != sizeof (KINFO_PROC))
|
return FALSE;
|
|
return TRUE;
|
}
|
#endif /* USE_SYSCTL */
|
|
/**
|
* mono_process_get_name:
|
* \param pid pid of the process
|
* \param buf byte buffer where to store the name of the prcoess
|
* \param len size of the buffer \p buf
|
* \returns the name of the process identified by \p pid, storing it
|
* inside \p buf for a maximum of len bytes (including the terminating 0).
|
*/
|
char*
|
mono_process_get_name (gpointer pid, char *buf, int len)
|
{
|
#if USE_SYSCTL
|
KINFO_PROC processi;
|
|
memset (buf, 0, len);
|
|
if (sysctl_kinfo_proc (pid, &processi))
|
memcpy (buf, processi.kinfo_name_member, len - 1);
|
|
return buf;
|
#else
|
char fname [128];
|
FILE *file;
|
char *p;
|
size_t r;
|
sprintf (fname, "/proc/%d/cmdline", GPOINTER_TO_INT (pid));
|
buf [0] = 0;
|
file = fopen (fname, "r");
|
if (!file)
|
return buf;
|
r = fread (buf, 1, len - 1, file);
|
fclose (file);
|
buf [r] = 0;
|
p = strrchr (buf, '/');
|
if (p)
|
return p + 1;
|
if (r == 0) {
|
return get_pid_status_item_buf (GPOINTER_TO_INT (pid), "Name", buf, len, NULL);
|
}
|
return buf;
|
#endif
|
}
|
|
void
|
mono_process_get_times (gpointer pid, gint64 *start_time, gint64 *user_time, gint64 *kernel_time)
|
{
|
if (user_time)
|
*user_time = mono_process_get_data (pid, MONO_PROCESS_USER_TIME);
|
|
if (kernel_time)
|
*kernel_time = mono_process_get_data (pid, MONO_PROCESS_SYSTEM_TIME);
|
|
if (start_time) {
|
*start_time = 0;
|
|
#if USE_SYSCTL && defined(kinfo_starttime_member)
|
{
|
KINFO_PROC processi;
|
|
if (sysctl_kinfo_proc (pid, &processi)) {
|
#if defined(__NetBSD__)
|
struct timeval tv;
|
tv.tv_sec = processi.kinfo_starttime_member;
|
tv.tv_usec = processi.p_ustart_usec;
|
*start_time = mono_100ns_datetime_from_timeval(tv);
|
#else
|
*start_time = mono_100ns_datetime_from_timeval (processi.kinfo_starttime_member);
|
#endif
|
}
|
}
|
#endif
|
|
if (*start_time == 0) {
|
static guint64 boot_time = 0;
|
if (!boot_time)
|
boot_time = mono_100ns_datetime () - mono_msec_boottime () * 10000;
|
|
*start_time = boot_time + mono_process_get_data (pid, MONO_PROCESS_ELAPSED);
|
}
|
}
|
}
|
|
/*
|
* /proc/pid/stat format:
|
* pid (cmdname) S
|
* [0] ppid pgid sid tty_nr tty_pgrp flags min_flt cmin_flt maj_flt cmaj_flt
|
* [10] utime stime cutime cstime prio nice threads 0 start_time vsize
|
* [20] rss rsslim start_code end_code start_stack esp eip pending blocked sigign
|
* [30] sigcatch wchan 0 0 exit_signal cpu rt_prio policy
|
*/
|
|
#define RET_ERROR(err) do { \
|
if (error) *error = (err); \
|
return 0; \
|
} while (0)
|
|
static gint64
|
get_process_stat_item (int pid, int pos, int sum, MonoProcessError *error)
|
{
|
#if defined(__APPLE__)
|
double process_user_time = 0, process_system_time = 0;//, process_percent = 0;
|
task_t task;
|
struct task_basic_info t_info;
|
mach_msg_type_number_t t_info_count = TASK_BASIC_INFO_COUNT, th_count;
|
thread_array_t th_array;
|
size_t i;
|
kern_return_t ret;
|
|
if (pid == getpid ()) {
|
/* task_for_pid () doesn't work on ios, even for the current process */
|
task = mach_task_self ();
|
} else {
|
do {
|
ret = task_for_pid (mach_task_self (), pid, &task);
|
} while (ret == KERN_ABORTED);
|
|
if (ret != KERN_SUCCESS)
|
RET_ERROR (MONO_PROCESS_ERROR_NOT_FOUND);
|
}
|
|
do {
|
ret = task_info (task, TASK_BASIC_INFO, (task_info_t)&t_info, &t_info_count);
|
} while (ret == KERN_ABORTED);
|
|
if (ret != KERN_SUCCESS) {
|
if (pid != getpid ())
|
mach_port_deallocate (mach_task_self (), task);
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
}
|
|
do {
|
ret = task_threads (task, &th_array, &th_count);
|
} while (ret == KERN_ABORTED);
|
|
if (ret != KERN_SUCCESS) {
|
if (pid != getpid ())
|
mach_port_deallocate (mach_task_self (), task);
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
}
|
|
for (i = 0; i < th_count; i++) {
|
double thread_user_time, thread_system_time;//, thread_percent;
|
|
struct thread_basic_info th_info;
|
mach_msg_type_number_t th_info_count = THREAD_BASIC_INFO_COUNT;
|
do {
|
ret = thread_info(th_array[i], THREAD_BASIC_INFO, (thread_info_t)&th_info, &th_info_count);
|
} while (ret == KERN_ABORTED);
|
|
if (ret == KERN_SUCCESS) {
|
thread_user_time = th_info.user_time.seconds + th_info.user_time.microseconds / 1e6;
|
thread_system_time = th_info.system_time.seconds + th_info.system_time.microseconds / 1e6;
|
//thread_percent = (double)th_info.cpu_usage / TH_USAGE_SCALE;
|
|
process_user_time += thread_user_time;
|
process_system_time += thread_system_time;
|
//process_percent += th_percent;
|
}
|
}
|
|
for (i = 0; i < th_count; i++)
|
mach_port_deallocate(task, th_array[i]);
|
|
if (pid != getpid ())
|
mach_port_deallocate (mach_task_self (), task);
|
|
process_user_time += t_info.user_time.seconds + t_info.user_time.microseconds / 1e6;
|
process_system_time += t_info.system_time.seconds + t_info.system_time.microseconds / 1e6;
|
|
if (pos == 10 && sum == TRUE)
|
return (gint64)((process_user_time + process_system_time) * 10000000);
|
else if (pos == 10)
|
return (gint64)(process_user_time * 10000000);
|
else if (pos == 11)
|
return (gint64)(process_system_time * 10000000);
|
|
return 0;
|
#else
|
char buf [512];
|
char *s, *end;
|
FILE *f;
|
size_t len;
|
int i;
|
gint64 value;
|
|
g_snprintf (buf, sizeof (buf), "/proc/%d/stat", pid);
|
f = fopen (buf, "r");
|
if (!f)
|
RET_ERROR (MONO_PROCESS_ERROR_NOT_FOUND);
|
len = fread (buf, 1, sizeof (buf), f);
|
fclose (f);
|
if (len <= 0)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
s = strchr (buf, ')');
|
if (!s)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
s++;
|
while (g_ascii_isspace (*s)) s++;
|
if (!*s)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
/* skip the status char */
|
while (*s && !g_ascii_isspace (*s)) s++;
|
if (!*s)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
for (i = 0; i < pos; ++i) {
|
while (g_ascii_isspace (*s)) s++;
|
if (!*s)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
while (*s && !g_ascii_isspace (*s)) s++;
|
if (!*s)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
}
|
/* we are finally at the needed item */
|
value = strtoul (s, &end, 0);
|
/* add also the following value */
|
if (sum) {
|
while (g_ascii_isspace (*s)) s++;
|
if (!*s)
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
value += strtoul (s, &end, 0);
|
}
|
if (error)
|
*error = MONO_PROCESS_ERROR_NONE;
|
return value;
|
#endif
|
}
|
|
static int
|
get_user_hz (void)
|
{
|
static int user_hz = 0;
|
if (user_hz == 0) {
|
#if defined (_SC_CLK_TCK) && defined (HAVE_SYSCONF)
|
user_hz = sysconf (_SC_CLK_TCK);
|
#endif
|
if (user_hz == 0)
|
user_hz = 100;
|
}
|
return user_hz;
|
}
|
|
static gint64
|
get_process_stat_time (int pid, int pos, int sum, MonoProcessError *error)
|
{
|
gint64 val = get_process_stat_item (pid, pos, sum, error);
|
#if defined(__APPLE__)
|
return val;
|
#else
|
/* return 100ns ticks */
|
return (val * 10000000) / get_user_hz ();
|
#endif
|
}
|
|
static gint64
|
get_pid_status_item (int pid, const char *item, MonoProcessError *error, int multiplier)
|
{
|
#if defined(__APPLE__)
|
// ignore the multiplier
|
|
gint64 ret;
|
task_t task;
|
struct task_basic_info t_info;
|
mach_msg_type_number_t th_count = TASK_BASIC_INFO_COUNT;
|
kern_return_t mach_ret;
|
|
if (pid == getpid ()) {
|
/* task_for_pid () doesn't work on ios, even for the current process */
|
task = mach_task_self ();
|
} else {
|
do {
|
mach_ret = task_for_pid (mach_task_self (), pid, &task);
|
} while (mach_ret == KERN_ABORTED);
|
|
if (mach_ret != KERN_SUCCESS)
|
RET_ERROR (MONO_PROCESS_ERROR_NOT_FOUND);
|
}
|
|
do {
|
mach_ret = task_info (task, TASK_BASIC_INFO, (task_info_t)&t_info, &th_count);
|
} while (mach_ret == KERN_ABORTED);
|
|
if (mach_ret != KERN_SUCCESS) {
|
if (pid != getpid ())
|
mach_port_deallocate (mach_task_self (), task);
|
RET_ERROR (MONO_PROCESS_ERROR_OTHER);
|
}
|
|
if (strcmp (item, "VmRSS") == 0 || strcmp (item, "VmHWM") == 0 || strcmp (item, "VmData") == 0)
|
ret = t_info.resident_size;
|
else if (strcmp (item, "VmSize") == 0 || strcmp (item, "VmPeak") == 0)
|
ret = t_info.virtual_size;
|
else if (strcmp (item, "Threads") == 0)
|
ret = th_count;
|
else
|
ret = 0;
|
|
if (pid != getpid ())
|
mach_port_deallocate (mach_task_self (), task);
|
|
return ret;
|
#else
|
char buf [64];
|
char *s;
|
|
s = get_pid_status_item_buf (pid, item, buf, sizeof (buf), error);
|
if (s)
|
return ((gint64) atol (s)) * multiplier;
|
return 0;
|
#endif
|
}
|
|
/**
|
* mono_process_get_data:
|
* \param pid pid of the process
|
* \param data description of data to return
|
* \returns a data item of a process like user time, memory use etc,
|
* according to the \p data argumet.
|
*/
|
gint64
|
mono_process_get_data_with_error (gpointer pid, MonoProcessData data, MonoProcessError *error)
|
{
|
gint64 val;
|
int rpid = GPOINTER_TO_INT (pid);
|
|
if (error)
|
*error = MONO_PROCESS_ERROR_OTHER;
|
|
switch (data) {
|
case MONO_PROCESS_NUM_THREADS:
|
return get_pid_status_item (rpid, "Threads", error, 1);
|
case MONO_PROCESS_USER_TIME:
|
return get_process_stat_time (rpid, 10, FALSE, error);
|
case MONO_PROCESS_SYSTEM_TIME:
|
return get_process_stat_time (rpid, 11, FALSE, error);
|
case MONO_PROCESS_TOTAL_TIME:
|
return get_process_stat_time (rpid, 10, TRUE, error);
|
case MONO_PROCESS_WORKING_SET:
|
return get_pid_status_item (rpid, "VmRSS", error, 1024);
|
case MONO_PROCESS_WORKING_SET_PEAK:
|
val = get_pid_status_item (rpid, "VmHWM", error, 1024);
|
if (val == 0)
|
val = get_pid_status_item (rpid, "VmRSS", error, 1024);
|
return val;
|
case MONO_PROCESS_PRIVATE_BYTES:
|
return get_pid_status_item (rpid, "VmData", error, 1024);
|
case MONO_PROCESS_VIRTUAL_BYTES:
|
return get_pid_status_item (rpid, "VmSize", error, 1024);
|
case MONO_PROCESS_VIRTUAL_BYTES_PEAK:
|
val = get_pid_status_item (rpid, "VmPeak", error, 1024);
|
if (val == 0)
|
val = get_pid_status_item (rpid, "VmSize", error, 1024);
|
return val;
|
case MONO_PROCESS_FAULTS:
|
return get_process_stat_item (rpid, 6, TRUE, error);
|
case MONO_PROCESS_ELAPSED:
|
return get_process_stat_time (rpid, 18, FALSE, error);
|
case MONO_PROCESS_PPID:
|
return get_process_stat_time (rpid, 0, FALSE, error);
|
case MONO_PROCESS_PAGED_BYTES:
|
return get_pid_status_item (rpid, "VmSwap", error, 1024);
|
|
/* Nothing yet */
|
case MONO_PROCESS_END:
|
return 0;
|
}
|
return 0;
|
}
|
|
gint64
|
mono_process_get_data (gpointer pid, MonoProcessData data)
|
{
|
MonoProcessError error;
|
return mono_process_get_data_with_error (pid, data, &error);
|
}
|
|
#ifndef HOST_WIN32
|
int
|
mono_process_current_pid ()
|
{
|
#if defined(HAVE_UNISTD_H)
|
return (int) getpid ();
|
#else
|
#error getpid
|
#endif
|
}
|
#endif /* !HOST_WIN32 */
|
|
/**
|
* mono_cpu_count:
|
* \returns the number of processors on the system.
|
*/
|
#ifndef HOST_WIN32
|
int
|
mono_cpu_count (void)
|
{
|
#ifdef HOST_ANDROID
|
/* Android tries really hard to save power by powering off CPUs on SMP phones which
|
* means the normal way to query cpu count returns a wrong value with userspace API.
|
* Instead we use /sys entries to query the actual hardware CPU count.
|
*/
|
int count = 0;
|
char buffer[8] = {'\0'};
|
int present = open ("/sys/devices/system/cpu/present", O_RDONLY);
|
/* Format of the /sys entry is a cpulist of indexes which in the case
|
* of present is always of the form "0-(n-1)" when there is more than
|
* 1 core, n being the number of CPU cores in the system. Otherwise
|
* the value is simply 0
|
*/
|
if (present != -1 && read (present, (char*)buffer, sizeof (buffer)) > 3)
|
count = strtol (((char*)buffer) + 2, NULL, 10);
|
if (present != -1)
|
close (present);
|
if (count > 0)
|
return count + 1;
|
#endif
|
|
#if defined(HOST_ARM) || defined (HOST_ARM64)
|
|
/*
|
* Recap from Alexander Köplinger <alex.koeplinger@outlook.com>:
|
*
|
* When we merged the change from PR #2722, we started seeing random failures on ARM in
|
* the MonoTests.System.Threading.ThreadPoolTests.SetAndGetMaxThreads and
|
* MonoTests.System.Threading.ManualResetEventSlimTests.Constructor_Defaults tests. Both
|
* of those tests are dealing with Environment.ProcessorCount to verify some implementation
|
* details.
|
*
|
* It turns out that on the Jetson TK1 board we use on public Jenkins and on ARM kernels
|
* in general, the value returned by sched_getaffinity (or _SC_NPROCESSORS_ONLN) doesn't
|
* contain CPUs/cores that are powered off for power saving reasons. This is contrary to
|
* what happens on x86, where even cores in deep-sleep state are returned [1], [2]. This
|
* means that we would get a processor count of 1 at one point in time and a higher value
|
* when load increases later on as the system wakes CPUs.
|
*
|
* Various runtime pieces like the threadpool and also user code however relies on the
|
* value returned by Environment.ProcessorCount e.g. for deciding how many parallel tasks
|
* to start, thereby limiting the performance when that code thinks we only have one CPU.
|
*
|
* Talking to a few people, this was the reason why we changed to _SC_NPROCESSORS_CONF in
|
* mono#1688 and why we added a special case for Android in mono@de3addc to get the "real"
|
* number of processors in the system.
|
*
|
* Because of those issues Android/Dalvik also switched from _ONLN to _SC_NPROCESSORS_CONF
|
* for the Java API Runtime.availableProcessors() too [3], citing:
|
* > Traditionally this returned the number currently online, but many mobile devices are
|
* able to take unused cores offline to save power, so releases newer than Android 4.2 (Jelly
|
* Bean) return the maximum number of cores that could be made available if there were no
|
* power or heat constraints.
|
*
|
* The problem with sticking to _SC_NPROCESSORS_CONF however is that it breaks down in
|
* constrained environments like Docker or with an explicit CPU affinity set by the Linux
|
* `taskset` command, They'd get a higher CPU count than can be used, start more threads etc.
|
* which results in unnecessary context switches and overloaded systems. That's why we need
|
* to respect sched_getaffinity.
|
*
|
* So while in an ideal world we would be able to rely on sched_getaffinity/_SC_NPROCESSORS_ONLN
|
* to return the number of theoretically available CPUs regardless of power saving measures
|
* everywhere, we can't do this on ARM.
|
*
|
* I think the pragmatic solution is the following:
|
* * use sched_getaffinity (+ fallback to _SC_NPROCESSORS_ONLN in case of error) on x86. This
|
* ensures we're inline with what OpenJDK [4] and CoreCLR [5] do
|
* * use _SC_NPROCESSORS_CONF exclusively on ARM (I think we could eventually even get rid of
|
* the HOST_ANDROID special case)
|
*
|
* Helpful links:
|
*
|
* [1] https://sourceware.org/ml/libc-alpha/2013-07/msg00383.html
|
* [2] https://lists.01.org/pipermail/powertop/2012-September/000433.html
|
* [3] https://android.googlesource.com/platform/libcore/+/750dc634e56c58d1d04f6a138734ac2b772900b5%5E1..750dc634e56c58d1d04f6a138734ac2b772900b5/
|
* [4] https://bugs.openjdk.java.net/browse/JDK-6515172
|
* [5] https://github.com/dotnet/coreclr/blob/7058273693db2555f127ce16e6b0c5b40fb04867/src/pal/src/misc/sysinfo.cpp#L148
|
*/
|
|
#if defined (_SC_NPROCESSORS_CONF) && defined (HAVE_SYSCONF)
|
{
|
int count = sysconf (_SC_NPROCESSORS_CONF);
|
if (count > 0)
|
return count;
|
}
|
#endif
|
|
#else
|
|
#ifdef HAVE_SCHED_GETAFFINITY
|
{
|
cpu_set_t set;
|
if (sched_getaffinity (mono_process_current_pid (), sizeof (set), &set) == 0)
|
return CPU_COUNT (&set);
|
}
|
#endif
|
#if defined (_SC_NPROCESSORS_ONLN) && defined (HAVE_SYSCONF)
|
{
|
int count = sysconf (_SC_NPROCESSORS_ONLN);
|
if (count > 0)
|
return count;
|
}
|
#endif
|
|
#endif /* defined(HOST_ARM) || defined (HOST_ARM64) */
|
|
#ifdef USE_SYSCTL
|
{
|
int count;
|
int mib [2];
|
size_t len = sizeof (int);
|
mib [0] = CTL_HW;
|
mib [1] = HW_NCPU;
|
if (sysctl (mib, 2, &count, &len, NULL, 0) == 0)
|
return count;
|
}
|
#endif
|
/* FIXME: warn */
|
return 1;
|
}
|
#endif /* !HOST_WIN32 */
|
|
static void
|
get_cpu_times (int cpu_id, gint64 *user, gint64 *systemt, gint64 *irq, gint64 *sirq, gint64 *idle)
|
{
|
char buf [256];
|
char *s;
|
int hz = get_user_hz ();
|
guint64 user_ticks = 0, nice_ticks = 0, system_ticks = 0, idle_ticks = 0, irq_ticks = 0, sirq_ticks = 0;
|
FILE *f = fopen ("/proc/stat", "r");
|
if (!f)
|
return;
|
if (cpu_id < 0)
|
hz *= mono_cpu_count ();
|
while ((s = fgets (buf, sizeof (buf), f))) {
|
char *data = NULL;
|
if (cpu_id < 0 && strncmp (s, "cpu", 3) == 0 && g_ascii_isspace (s [3])) {
|
data = s + 4;
|
} else if (cpu_id >= 0 && strncmp (s, "cpu", 3) == 0 && strtol (s + 3, &data, 10) == cpu_id) {
|
if (data == s + 3)
|
continue;
|
data++;
|
} else {
|
continue;
|
}
|
|
user_ticks = strtoull (data, &data, 10);
|
nice_ticks = strtoull (data, &data, 10);
|
system_ticks = strtoull (data, &data, 10);
|
idle_ticks = strtoull (data, &data, 10);
|
/* iowait_ticks = strtoull (data, &data, 10); */
|
irq_ticks = strtoull (data, &data, 10);
|
sirq_ticks = strtoull (data, &data, 10);
|
break;
|
}
|
fclose (f);
|
|
if (user)
|
*user = (user_ticks + nice_ticks) * 10000000 / hz;
|
if (systemt)
|
*systemt = (system_ticks) * 10000000 / hz;
|
if (irq)
|
*irq = (irq_ticks) * 10000000 / hz;
|
if (sirq)
|
*sirq = (sirq_ticks) * 10000000 / hz;
|
if (idle)
|
*idle = (idle_ticks) * 10000000 / hz;
|
}
|
|
/**
|
* mono_cpu_get_data:
|
* \param cpu_id processor number or -1 to get a summary of all the processors
|
* \param data type of data to retrieve
|
* Get data about a processor on the system, like time spent in user space or idle time.
|
*/
|
gint64
|
mono_cpu_get_data (int cpu_id, MonoCpuData data, MonoProcessError *error)
|
{
|
gint64 value = 0;
|
|
if (error)
|
*error = MONO_PROCESS_ERROR_NONE;
|
switch (data) {
|
case MONO_CPU_USER_TIME:
|
get_cpu_times (cpu_id, &value, NULL, NULL, NULL, NULL);
|
break;
|
case MONO_CPU_PRIV_TIME:
|
get_cpu_times (cpu_id, NULL, &value, NULL, NULL, NULL);
|
break;
|
case MONO_CPU_INTR_TIME:
|
get_cpu_times (cpu_id, NULL, NULL, &value, NULL, NULL);
|
break;
|
case MONO_CPU_DCP_TIME:
|
get_cpu_times (cpu_id, NULL, NULL, NULL, &value, NULL);
|
break;
|
case MONO_CPU_IDLE_TIME:
|
get_cpu_times (cpu_id, NULL, NULL, NULL, NULL, &value);
|
break;
|
|
case MONO_CPU_END:
|
/* Nothing yet */
|
return 0;
|
}
|
return value;
|
}
|
|
int
|
mono_atexit (void (*func)(void))
|
{
|
#ifdef HOST_ANDROID
|
/* Some versions of android libc doesn't define atexit () */
|
return 0;
|
#else
|
return atexit (func);
|
#endif
|
}
|
|
/*
|
* This function returns the cpu usage in percentage,
|
* normalized on the number of cores.
|
*
|
* Warning : the percentage returned can be > 100%. This
|
* might happens on systems like Android which, for
|
* battery and performance reasons, shut down cores and
|
* lie about the number of active cores.
|
*/
|
#ifndef HOST_WIN32
|
gint32
|
mono_cpu_usage (MonoCpuUsageState *prev)
|
{
|
gint32 cpu_usage = 0;
|
#ifdef HAVE_GETRUSAGE
|
gint64 cpu_total_time;
|
gint64 cpu_busy_time;
|
struct rusage resource_usage;
|
gint64 current_time;
|
gint64 kernel_time;
|
gint64 user_time;
|
|
if (getrusage (RUSAGE_SELF, &resource_usage) == -1) {
|
g_error ("getrusage() failed, errno is %d (%s)\n", errno, strerror (errno));
|
return -1;
|
}
|
|
current_time = mono_100ns_ticks ();
|
kernel_time = resource_usage.ru_stime.tv_sec * 1000 * 1000 * 10 + resource_usage.ru_stime.tv_usec * 10;
|
user_time = resource_usage.ru_utime.tv_sec * 1000 * 1000 * 10 + resource_usage.ru_utime.tv_usec * 10;
|
|
cpu_busy_time = (user_time - (prev ? prev->user_time : 0)) + (kernel_time - (prev ? prev->kernel_time : 0));
|
cpu_total_time = (current_time - (prev ? prev->current_time : 0)) * mono_cpu_count ();
|
|
if (prev) {
|
prev->kernel_time = kernel_time;
|
prev->user_time = user_time;
|
prev->current_time = current_time;
|
}
|
|
if (cpu_total_time > 0 && cpu_busy_time > 0)
|
cpu_usage = (gint32)(cpu_busy_time * 100 / cpu_total_time);
|
#endif
|
return cpu_usage;
|
}
|
#endif /* !HOST_WIN32 */
|
