sg_get_process_stats, sg_get_process_stats_r, sg_get_process_count, sg_get_process_count_of, sg_get_process_count_r, sg_free_process_count, sg_process_compare_name, sg_process_compare_pid, sg_process_compare_uid, sg_process_compare_gid, sg_process_compare_size, sg_process_compare_res, sg_process_compare_cpu, sg_process_compare_time — get process statistics
#include <statgrab.h>
sg_process_stats *sg_get_process_stats( | entries) ; |
size_t *entries
;sg_process_stats *sg_get_process_stats_r( | entries) ; |
size_t *entries
;sg_error sg_free_process_stats( | data) ; |
sg_process_stats *data
;sg_process_count *sg_get_process_count( | ) ; |
sg_process_count *sg_get_process_count_of( | pcs) ; |
sg_process_count_source pcs
;sg_process_count *sg_get_process_count_r( | whereof) ; |
sg_process_stats const *whereof
;sg_error sg_free_process_count( | data) ; |
sg_process_count *data
;int sg_process_compare_name( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_pid( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_uid( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_gid( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_size( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_res( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_cpu( | va, | |
vb) ; |
const void *va
;const void *vb
;int sg_process_compare_time( | va, | |
vb) ; |
const void *va
;const void *vb
;
The sg_get_process_stats
functions provide
statistics about the currently running processes. Both functions,
sg_get_process_stats
() and
sg_get_process_stats_r
(), take an optional
entries
parameter, which points
(when given) to a size_t to take the number of returned vector
entries.
The functions sg_get_process_count_of
() and
sg_get_process_count_r
() provide an aggregated
view of the process table - they deliver the amount of processes
per process state. The sg_get_process_count
()
is in fact a preprocessor macro for backward compatibility and calls
sg_get_process_count_of
() with the parameter
pcs
of sg_entire_process_count to emulate
the behavior until 0.17.
Table 1. API Shortcut
function | returns | data owner |
---|---|---|
sg_get_process_stats | sg_process_stats * | libstatgrab (thread local) |
sg_get_process_stats_r | sg_process_stats * | caller |
sg_get_process_count_of | sg_process_count * | libstatgrab (thread local) |
sg_get_process_count_r | sg_process_count * | caller |
The sg_process_stats vectors received from
sg_get_process_stats_r
() or the
sg_process_count summaries received from
sg_get_process_count_r
must be
freed using sg_free_process_stats
() or
sg_free_process_count
(), respectively.
The caller is responsible for doing it when the data isn't needed
any more.
sg_process_compare_name |
sg_process_compare_pid |
sg_process_compare_uid |
sg_process_compare_gid |
sg_process_compare_size |
sg_process_compare_res |
sg_process_compare_cpu |
sg_process_compare_time |
These functions compare two sg_process_stats entries, and return an int to represent which one is greater. The main use of these functions is to be passed to qsort to sort the sg_process_stats by the given type.
Example 1. Example
size_t entries; sg_process_stats *proc_stats = NULL; while( NULL != ( proc_stats = sg_get_process_stats_r(&entries) ) ) { /* order entries by comparing the process identifier */ qsort( proc_stats, entries, sizeof(proc_stats[0]), &sg_process_compare_pid ); show_proc_stats( proc_stats ); sg_free_process_stats( proc_stats ); }
The structure returned by sg_get_process_stats is of type sg_process_stats.
typedef struct { char *process_name; char *proctitle; pid_t pid; /* process identifier */ pid_t parent; /* Parent pid */ pid_t pgid; /* process id of process group leader */ pid_t sessid; /* session id of the session the process belongs to */ uid_t uid; uid_t euid; gid_t gid; gid_t egid; unsigned long long context_switches; unsigned long long voluntary_context_switches; unsigned long long involuntary_context_switches; unsigned long long proc_size; /* in bytes */ unsigned long long proc_resident; /* in bytes */ time_t start_time; /* When was the process started */ time_t time_spent; /* time running in seconds */ double cpu_percent; int nice; sg_process_state state; time_t systime; } sg_process_stats;
typedef enum { SG_PROCESS_STATE_RUNNING, SG_PROCESS_STATE_SLEEPING, SG_PROCESS_STATE_STOPPED, SG_PROCESS_STATE_ZOMBIE, SG_PROCESS_STATE_UNKNOWN } sg_process_state;
process_name
The name of the command that was run. The content of this
field heavily depends on the underlying operating system,
some store the basename the executable passes to
the exec
(2) system call, some the entire
path. Most OS restrict the size of this field - some like
the *BSD family to a very low value of 15 bytes.
This field is usually immutable for userland processes.
proctitle
The command line (the "title") of the process. Take note - this can be modified by the process, so isn't guaranteed to be the original command line.
pid
The process ID.
parent
The parent process ID.
pgid
The process ID of the process group leader.
sessid
Session id of the session the process belongs to.
uid
The ID of the user the process is running as.
euid
The ID of the effective user the process is running as.
gid
The ID of the group the process is running as.
egid
The ID of the effective group the process is running as.
context_switches
The number of context switches of this process (voluntary and involuntary).
voluntary_context_switches
The number of voluntary context switches of this process (eg.
by calling sched_yield
() or sleep
()).
involuntary_context_switches
The number of involuntary context switches of this process (eg. time slice exhausted or signal sent).
proc_size
The virtual memory size of the process in bytes.
proc_resident
The size of the process that's resident in memory.
start_time
The time when the process has been started in seconds since epoch.
time_spent
The number of seconds the process has been running (user+system time, without time spent by child processes).
cpu_percent
The current percentage of CPU the process is using.
nice
The nice value of the process.
state
The current state of the process. See sg_process_state for permitted values.
systime
The time in seconds since epoch of the moment where the present statistic has been created. This might be (but doesn't have to be) the same moment for all returned entries, regardless whether they're fetched with one snapshot or puzzled from some kind of procfs.
The structure returned by sg_get_process_count_of and sg_get_process_count_r is of type sg_process_count.
typedef enum sg_process_count_source { sg_entire_process_count, sg_last_process_count } sg_process_count_source;
typedef struct{ unsigned long long total; unsigned long long running; unsigned long long sleeping; unsigned long long stopped; unsigned long long zombie; unsigned long long unknown; time_t systime; }sg_process_count;
total
The total number of processes.
running
The number of running processes.
sleeping
The number of sleeping processes.
stopped
The number of stopped processes.
zombie
The number of zombie processes.
unknown
The number of processes not matching any of above named categories.
systime
The time in seconds since epoch of the moment where the present statistic has been created.
The very first call of
sg_get_process_count_of
(sg_last_process_count)
will return the same as
sg_get_process_count_of
(sg_entire_process_count).
The compare functions exist rather for backward compatibility than for functionality enhancements. Limited flexibility (e.g. reverse order) and lack of optimising opportunities for the compiler leads to the recommendation to implement the required compare routines locally.