Port-sparc archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Hardware interrupt list?



On 9/8/2011 17:37, Ken Hornstein wrote:
Actually, there is a change, it's worse.  Without the modification, ntpd
opens the port using PPS and I'm able to start gpsd right behind it once
PPS is active.  With the modification, I can't use the port at all after
PPS takes over, it just keeps gpsd locked out.

Sorry man, I am out of ideas.  Although when you say it "keeps gpsd locked
out", what exactly does that mean?  An open() of that device returns
EBUSY?  Or you just don't get any input?

The fact that this change made it WORSE makes me think that this is close
to the core problem.  You might need to sit down with the Z8530 datasheet
and add some judicious printf()s to figure out what's going wrong, though.
I suspect you're in the distant minority when it comes to ntpd hardware
these days.


Yeah, it reports device busy at all times. Once in a while it is somehow able to open the port but then gpsd spins out of control trying to read data (upwards of 50% cpu time when it normally runs 0.1%).

I know it's not a critical bit of hardware for developers. The backup plan is to just use two serial ports and sort of work through the real serial problems.

Another person on the ntp list actually reported that the same problem occurs in a Sun emulator though perhaps for somewhat different reasons.

Home | Main Index | Thread Index | Old Index