Get a clock ID for a given process and thread
#include <sys/neutrino.h> #include <inttypes.h> extern int ClockId( pid_t pid, int tid ); extern int ClockId_r( pid_t pid, int tid );
libc
Use the -l c option to qcc to link against this library. This library is usually included automatically.
The ClockId() and ClockId_r() kernel calls return an integer that you can pass as a clockid_t to ClockTime(); one of:
While the processor isn't in a power-saving mode, CLOCK_SOFTTIME behaves the same as CLOCK_REALTIME.
When you pass this clock ID to ClockTime(), the function returns (in the location pointed to by old) the number of nanoseconds that the specified thread of the specified process has executed.
The ClockId() and ClockId_r() functions are identical except in the way they indicate errors. See the Returns section for details.
Instead of using these kernel calls directly, consider calling clock_getcpuclockid() or pthread_getcpuclockid(). |
If the tid is zero, the number of nanoseconds that the process as a whole has executed is returned. On an SMP box, this number may exceed the realtime number of nanoseconds that have elapsed because multiple threads in the process can run on several CPUs at the same time.
This call doesn't block.
Here's how you can determine how busy a system is:
id = ClockId(1, 1); for( ;; ) { ClockTime(id, NULL, &start); sleep(1); ClockTime(id, NULL, &stop); printf("load = %f%%\n", (1000000000.0 - (stop-start)) / 10000000.0); }
Safety: | |
---|---|
Cancellation point | No |
Interrupt handler | No |
Signal handler | Yes |
Thread | Yes |
ClockTime(), clock_getcpuclockid(), pthread_getcpuclockid()
Clocks, Timers, and Getting a Kick Every So Often chapter of Getting Started with QNX Neutrino