Difference between revisions of "GTIMER"

Latest revision as of 15:51, 4 February 2016

µGFX comes with its own timer abstraction module. The GTIMER module provides high level, simple and hardware independent timers. Due to their nature of being virtual (software) timers they are not very accurate. However, the accuracy is more than enough for any GUI requirements.

The reason why µGFX has it's own timer abstraction is because virtual timers provided by most RTOSs are interrupt context only. While great for what they are designed for, they make coding of the input drivers much more complex. For non-performance critical drivers like these input drivers, it would also hog an in-ordinate amount of critical (interrupt locked) system time. This contrary to the goals of a real-time operating system. So a user-land (thread based) timer mechanism is also required.

Even if GTIMER is originally meant for internal use only, you can still use it yourself. This might become handy in some applications.

API reference

The API reference of the GTIMER module can be found here.

Usage

A new GTimer job (from here on refered to as a timer) can be created using gtimerInit(). Creating a new timer doesn't start it yet, it just initializes the GTimer object. A timer is started by using gtimerStart(). The function parameters specify the the timer interval, whether it's a single-shot or a periodic timer and the callback function. When the timer fires, the callback function will be executed.

Timers can be stopped at any time using gtimerStop(). Calling gtimerJab() will cause the timer period to expire immediately.

The callback function needs to have the following signature:

void myCallback(void* param);

Theory of operation

When the GTIMER module has been enabled in the configuration file, a new thread will be created upon calling gfxInit(). The GTIMER thread maintains a list of all created timers. It iterates through that list periodically. When a timer period has expired the callback function that has been registered with that timer will be executed.

Precision

GTimers are software timers and therefore their precision is totally unpredictable. The precision depends on how many threads and with which priority are currently running on the system. You can modify the GTimer thread priority and stack size in your configuration file:

#define GTIMER_THREAD_PRIORITY			HIGH_PRIORITY

See threading for more information about thread priorities.

Stack size

All GTimer jobs are executed in the same thread. This means that the GTimer thread needs to have a sufficient stack size to execute the callback function with the largest stack requirements. The default stack size for the GTimer thread is 2048 bytes to provide enough stack for the GTimer jobs used internally by some of the uGFX modules.

#define GTIMER_THREAD_WORKAREA_SIZE		2048

Example

The following is a simple, self explanatory example of how a GTimer can be used. For more features, please have a look at the corresponding api reference.

#include "gfx.h"
 
GTimer GT1;
GTimer GT2;
 
void callback1(void* arg)
{
    (void)arg;
 
    palTogglePad(GPIOD, GPIOD_LED3);
}
 
void callback2(void* arg)
{
    (void)arg;
 
    palSetPad(GPIOD, GPIOD_LED4);
}
 
int main(void)
{ 
    /* initialize µGFX and the underlying system */
    gfxInit();
 
    /* initialize the timers */
    gtimerInit(&GT1);
    gtimerInit(&GT2);
 
    /* continious mode - callback1() called without any argument every 250ms */
    gtimerStart(&GT1, callback1, NULL, TRUE, 250);
 
    /* single shot mode - callback2() called without any argument once after 1s */
    gtimerStart(&GT2, callback2, NULL, FALSE, 1000);
 
    while (TRUE) {
        gfxSleepMilliseconds(500);
    }   
}