On Sat, Sep 04, 2010 at 08:47:39PM -0500, Christoph Lameter wrote: > On Sat, 4 Sep 2010, Christian Riesch wrote: > > > Richard's idea is to support clock hardware for IEEE 1588 (PTP, Precision Time > > Protocol). Have a look at the earlier discussions: > > > > http://lkml.org/lkml/2010/8/16/90 > > A superficial scan does not give me any definite agreement on an approach. Welcome to the club. I am also not too sure how to bring everything that was said together. I have attempted to synthesize the various ideas from the previous threads into a concrete form (a new patch set) in order to see: 1. If I understood what the commentators were asking for. 2. Whether people now agree on the approach. > Ok this goes more into something that may prove useful but it would be > very helpful still to have a clear discussion as to why this is all > needed. From what I can tell the functionality is already there and > another clock device (CLOCK_SGI_CYCLE) is already providing a model for > how a PTP clock could be controlled and used. John Stultz warned against that approach in his mail from August 18. http://lkml.org/lkml/2010/8/18/264 At this point, I think it is clear that, in order to support IEEE 1588, some kind of changes are needed. I have argued and discussed the issues at some length. I am willing to continue the discussion, but I would also appreciate more than a "superficial scan." PTP is quite a large topic that probably cannot be adequately treated in a single email. Having said that, if you are convinced that the existing interfaces are good enough to support PTP (and possible future methods, as discussed in the threads), then I am all ears to find out how to do make it work. Thanks, Richard -- To unsubscribe from this list: send the line "unsubscribe linux-api" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
