OSDev.org

Posted: **Sun Nov 24, 2013 12:47 am**

Now I'm trying to use HPET instead of PIT.
But I have a problem:I'm using QEMU.I try to configure HPET and let it produce 1000 interrupt per second.

First,I get the period of HPET.
Next,I clear the counter of HPET and enable it.
Then I wait a minute and check if the counter is counting.The result is true.
Finally,I start the interrupt of HPET and set the configure of the timer0.

Code: Select all

/*Index = 0.*/
u32 conf = hpetIn(HPET_TIMER_CONF_REG(index));
conf |= HPET_TIMER_CONF_PERIODIC;
conf |= HPET_TIMER_CONF_ENABLE;
hpetOut(HPET_TIMER_CONF_REG(index),conf);

And set the compare register of timer0:

Code: Select all

hpetOut(HPET_TIMER_CMP_REG(index),time);

I want to change "time" to change how many interrupts HPET produce per second.
If I set "time" to a big number,it will be successful!
But when I set "time" to a small number even 1,I failed!It can only produce about 100 interrupts per second.

I want HPET to produce about 1000 interrupts per second,how should I do?
(I also try to use one shot mode and change the compare register when interrupts come,but I also failed.)
Thank you very mush!

Posted: **Sun Nov 24, 2013 1:40 am**

Hi,

windows8 wrote:Now I'm trying to use HPET instead of PIT.
But I have a problem:I'm using QEMU.

I'd begin by testing on other emulators and real computers - I wouldn't be surprised if the problem was Qemu.

For example, internally Qemu might be using something like "HPET_main_counter += (clock() - previous_clock) * HPET_TICKS_PER_SECOND / CLOCKS_PER_SEC" on a host OS where "CLOCKS_PER_SEC == 100".

Note: If the problem is Qemu, then I also wouldn't be too surprised if there are sane reasons for it (e.g. guest performance or portability).

Cheers,

Brendan

Posted: **Sun Nov 24, 2013 3:26 am**

Brendan wrote:Hi,

windows8 wrote:Now I'm trying to use HPET instead of PIT.
But I have a problem:I'm using QEMU.
I'd begin by testing on other emulators and real computers - I wouldn't be surprised if the problem was Qemu.

For example, internally Qemu might be using something like "HPET_main_counter += (clock() - previous_clock) * HPET_TICKS_PER_SECOND / CLOCKS_PER_SEC" on a host OS where "CLOCKS_PER_SEC == 100".

Note: If the problem is Qemu, then I also wouldn't be too surprised if there are sane reasons for it (e.g. guest performance or portability).

Cheers,

Brendan

I think you are right,I am going to try it not on QEMU.
Thanks a lot.

Posted: **Sun Nov 24, 2013 6:51 am**

Oh,I tried to run it in VirtualBox just now,it runs very successfully.
It can produce 1000 interrupts per second!
The problem is QEMU,thank you again.

OSDev.org

HPET can only produce 100 interrupts per second?

HPET can only produce 100 interrupts per second?

Re: HPET can only produce 100 interrupts per second?

Re: HPET can only produce 100 interrupts per second?

Re: HPET can only produce 100 interrupts per second?