- From: Askar Ibragimov <askar.ibragimov_at_gmail.com>
- Date: Mon, 27 Aug 2007 15:01:39 +0300
Hello Frank, thank you for the reply. 2007/8/27, Frank Mori Hess <fmhess_at_speakeasy.net>: > On Monday 27 August 2007 02:09, Askar Ibragimov wrote: > > Thank you Frank, > > > > though, I need asyncronous read of the counters, as I need to know, > > how many counts happened per second. > > Take the difference between successive reads. If it will be done from userspace program, than these successive reads are not really acceptable, as some process like HDD reading can substantially delay the system. > > > I guess I figured out a solution: I will write an interrupt handler > > and make it called from system timer (init_timer, etc.) similar to the > > driver of comedi_test. This routine will read the counters and pass > > their data to a buffer. I do not need hard realtime here so the > > precision of the computer clock is enough. What do you think about > > this schema? > > If the hardware doesn't support timed acquisition, don't implement some > lame software timed pseudo-command. If you want that, use > comedi_rt_timer. But I don't see any reason you need either. Since I need quite stable readings, say, each 1 +/- 0.1 second. I hope kernel timer can keep it up like that? And, to properly use rt_timer - do I need RTAI ? Cheers, Askar
Received on 2007-08-27Z11:01:39