| 1 |
|
CPU Statistics |
| 2 |
+ |
============== |
| 3 |
|
|
| 4 |
< |
cpu_states_t *get_cpu_totals(); |
| 4 |
< |
cpu_states_t *get_cpu_diff(); |
| 5 |
< |
cpu_percent_t *cpu_percent_usage(); |
| 4 |
> |
$Id$ |
| 5 |
|
|
| 6 |
+ |
Data Structures |
| 7 |
+ |
--------------- |
| 8 |
+ |
|
| 9 |
+ |
There are two structures returned by the CPU statistics functions. |
| 10 |
+ |
They look like this: |
| 11 |
+ |
|
| 12 |
|
typedef struct{ |
| 13 |
|
long long user; |
| 14 |
|
long long kernel; |
| 30 |
|
time_t time_taken; |
| 31 |
|
}cpu_percent_t; |
| 32 |
|
|
| 33 |
+ |
user, kernel, idle, iowait, swap, and nice are different CPU states. |
| 34 |
+ |
systime and time_taken are the time in seconds since the last call |
| 35 |
+ |
of the function. |
| 36 |
+ |
|
| 37 |
+ |
Functions |
| 38 |
+ |
--------- |
| 39 |
+ |
|
| 40 |
+ |
cpu_states_t *get_cpu_totals(); |
| 41 |
+ |
cpu_states_t *get_cpu_diff(); |
| 42 |
+ |
cpu_percent_t *cpu_percent_usage(); |
| 43 |
+ |
|
| 44 |
|
get_cpu_totals() and get_cpu_diff() both return static pointers of |
| 45 |
< |
type cpu_states_t. get_cpu_totals() returns the total amount of "ticks" |
| 46 |
< |
the OS has spent in each of the different states. get_cpu_diff() returns |
| 47 |
< |
the difference in "ticks" for each of the states since last time get_cpu_diff() |
| 48 |
< |
or get_cpu_totals() was called. If it has never been called, it will return |
| 49 |
< |
the result of get_cpu_totals() |
| 45 |
> |
type cpu_states_t. get_cpu_totals() returns the total amount of |
| 46 |
> |
"ticks" the operating system has spent in each of the different |
| 47 |
> |
states. get_cpu_diff() returns the difference in "ticks" for each |
| 48 |
> |
of the states since last time get_cpu_diff() or get_cpu_totals() |
| 49 |
> |
was called. If it has never been called, it will return the result |
| 50 |
> |
of get_cpu_totals() |
| 51 |
|
|
| 52 |
< |
The vaule stored (the "ticks") will vary from OS to OS. E.g. solaris has a |
| 53 |
< |
total of 100 per second. Linux has substatially more than that. Also, different |
| 54 |
< |
OS's store different information. E.g. solaris doesn't do nice. |
| 52 |
> |
The value stored (the "ticks") will vary between operating systems. |
| 53 |
> |
For example Solaris has a total of 100 per second, while Linux has |
| 54 |
> |
substantially more. Also, different operating systems store different |
| 55 |
> |
information - you won't find nice cpu on Solaris for example. |
| 56 |
|
|
| 57 |
< |
Because of this, ideally you will always want to work on a scale against the |
| 58 |
< |
total, or in percentages. |
| 57 |
> |
Because of this, you will ideally always want to work on a scale |
| 58 |
> |
against the total, or in percentages. |
| 59 |
|
|
| 60 |
< |
cpu_percent_usage() returns a pointer to a static cpu_percent_t. The function calls |
| 61 |
< |
get_cpu_diff() and changes the values into percentages. If its never been called |
| 62 |
< |
before (and nor has get_cpu_totals() and get_cpu_diff() ), the returned percentages |
| 63 |
< |
will be the systems total ever since its uptime. (Unless the counters of cycled) |
| 60 |
> |
cpu_percent_usage() returns a pointer to a static cpu_percent_t. |
| 61 |
> |
The function calls get_cpu_diff() and changes the values into |
| 62 |
> |
percentages. If it has never been called before (and nor has |
| 63 |
> |
get_cpu_totals() or get_cpu_diff() ), the returned percentages will |
| 64 |
> |
be the systems total ever since its uptime. (Unless the counters |
| 65 |
> |
have cycled) |
| 66 |
> |
|
| 67 |
> |
Example |
| 68 |
> |
------- |
| 69 |
> |
|
| 70 |
> |
A basic example can be found in examples/cpu_usage.c |
