Timer resolution too low

[<kim@fa...>]

Hi

Some time ago I told that xmode2 showed that the pulse train coming from my
remote wasn't the same each time a button was pressed, even though I have
confirmed with an oscilloscope how the pulses should be fore each button at
my remote.

Today I think I've found the answer. My VIA EPIA EN12000 apparantly has a
terrible low timer resolution. clock_getres() returns 0sec, 999848ns, i.e.
almost 1ms resolotion.
Related to the fact that my remote pulses are either 1,6ms or 540ms long
separated by 600us silence it is suddenly obvious why lirc_serial doesn't
work.

Any of you who have experienced something similar?
Do you know of a workaround?

/KimBP

Timer resolution too low

[<kim@fa...>]

Hi

Some time ago I told that xmode2 showed that the pulse train coming from my
remote wasn't the same each time a button was pressed, even though I have
confirmed with an oscilloscope how the pulses should be fore each button at
my remote.

Today I think I've found the answer. My VIA EPIA EN12000 apparantly has a
terrible low timer resolution. clock_getres() returns 0sec, 999848ns, i.e.
almost 1ms resolotion.
Related to the fact that my remote pulses are either 1,6ms or 540ms long
separated by 600us silence it is suddenly obvious why lirc_serial doesn't
work.

Any of you who have experienced something similar?
Do you know of a workaround?

/KimBP

Re: Timer resolution too low

[<lirc@ba...>]

Hi!

Kim Bondergaard Poulsen "kim@..." wrote:
[...]
> Today I think I've found the answer. My VIA EPIA EN12000 apparantly has a
> terrible low timer resolution. clock_getres() returns 0sec, 999848ns, i.e.
> almost 1ms resolotion.

I don't think the problem is directly related to what clock_getres  
returns. I think it will only reflect the HZ setting of your kernel.
I get 0s 4000250ns on my system and I don't have the time resolution  
problem.

But looking at the do_gettimeofday function might give some hints.

Christoph