CODEWITHSUNDEEP
clinuxlinux-kernellinux-device-driver
Anandhakrishnan Ramasamy
Anandhakrishnan Ramasamy

Reputation: 1

How to implement a clock in c similar to linux system clock

Normally when the linux system boots up it actually takes the reference time from RTC and runs a software timer on its own [i.e, generally known as system clock/wall clock]. When the system is about to shutdown it sync its wall clock time with RTC. I am looking for a method to implement a wall clock in c as similar to this. Can any body suggest some idea for me?

Thanks in advance,

Anandhakrishnan Ramasamy.

Upvotes: 0

Views: 1151

Answers (3)

Rafael Baptista
Rafael Baptista

Reputation: 11499

Its hard to do this 100% correctly. You will have to detect times when the CPU goes to sleep, if the system is suspended, and also any time someone changes the timezone, or when daylight savings time starts or ends. You would have to do all these things yourself.

All CPUs today have a high resolution timer. Its just a register that increments every CPU clock cycle. If you know the frequency of the CPU, and you read that register on a regular basis ( e.g. often enough that it doesn't overflow ) you can measure time.

On linux there is a family of functions that reads this register for you, and figures out the CPU frequency, and returns the time in that register in nano-seconds:

timespec ts;
clock_gettime( CLOCK_MONOTONIC_RAW, &ts );
u64 timeInNanoSeconds = ts.tv_nsec + ( ts.tv_sec * 1000000000LL );

That time will wrap around every 5 minutes or so. So you have to read it pretty often, so you can detect the wrap around. So any time you read it, if ts.tv.nsec is smaller than the last time you called it, they you had an overflow, and you have to account for it.

Once you can accurately measure the passage of a second, then you can build your wall clock from there.

Upvotes: 0

Claudio
Claudio

Reputation: 10947

You can't do it in plain C without relying on functionalities provided by the OS. The reason is that the OS schedules several applications through multiprogramming, and your C application can't have knowledge about when it has been suspended by the scheduler.

Therefore, you have to use Posix functions like gettimeofday(), time() and so on.

Upvotes: 1

danish
danish

Reputation: 359

What OS usually do is they fetch the system startup time from RTC or HPET or any other timer device. And after they load PIC or APIC with a value to receive periodic interrupts from them (e.g after every 100ms). Based on these interrupts value of system clock or wall clock gets updated.

Upvotes: 1

Related Questions