Messages in this thread |  | | From | "Ulrich Windl" <> | Date | Wed, 23 Oct 1996 08:28:03 +0200 | Subject | patch for NTP PPS support updated |
| |
Hello,
Before I for get to mention the FTP address again, I'm telling you that an updated version of my PPS patch is located on pcphy4.physik.uni-regensburg.de in /pub/wiu09524/PPS/design.txt.
The previous path still works, but is a symbolic link now. The new version is about 30kB, but is also available gzipped. In case you have problems with the new patch, the old patch is still available with the date suffix.
OK, now, what's new? Well almost nothing; the patch is just almost complete now. It includes my EMail address as a patch to the CREDITS file and the new proposed Documentation/Kernel-time.txt. Also the cycle counting is default off now so that no additional messages are produced in the syslog (they weren't many anyway). So far nobody ailing me complained about severe problems, except about electrical issues how to connect an arbitrary source to the carrier detect pin.
So far the god news. Now the bad thing: Currently the PPS interrupt is edge-triggered, but unfortunately on both edges! Thanks for Ted for remembering me of the fact that an interrupt is generated on any level change. Obviously the existing code can already cope with that kind of noise, but definitely I'll have to work on a mechanism to allow only _one_ edge to call hardware_pps(). Up to now I have not decided whether to make it configurable at runtime, or at compilation time. Of course, I'd prefer runtime, but we'll have to pass the parameter in some way. Suggestions?
Ulrich (I'll be away from work tomorrow, thus no replies)
|  |