Michael Hunter wrote: > On Fri, 28 Aug 2009 17:18:42 -0700 > Michael Hunter <Michael.Hunter at Sun.COM> wrote: > > >> On Fri, 28 Aug 2009 15:55:27 -0400 >> "Anurag S. Maskey" <Anurag.Maskey at Sun.COM> wrote: >> >> >>> The README definitely needs to be expanded/added with the QUEUE_QUIET >>> stuff and the event handling logic. I think I know how it works, but >>> not entirely sure. >>> >> DEFER (I'll probably get it written before I push but I don't want it >> to hold up the functionality). >> > > Written. Please review. > spectacular. Thanks! > [...ncu_phys.c...] > >>> why is 1774-1776 moved out from 1798? We don't have NCUs with lower >>> priority (ie bigger numbers) going to up state. >>> >>> also, why 1830-1832 outside of priority group checking? >>> >> I need to investigate and write more on these (and almost certainly >> comment). FWIR the issue was that we would lose state for something >> that we later needed state for when we were transitioning to it. But I >> have a dinner date so I'll have to write that later :) >> > [...] > > Conceptually we need to keep track of the state of the link NCUs > independent of which ones we decide to build interfaces on. > > Up until a few minutes ago I could get multiple interfaces to show up > when we didn't do that. Now I can't :( I'm putting the code back to > the way it was and we'll have to figure out what the event sequence is > if it can be reproduced. > > New webrev. > ncu_phys.c:1802 - moving to offline does check the priority group, but the state is changing to offline*, up which ultimately ends up in online, up. There's no checking in online,up.
Anurag
