On Wed, 2005-03-09 at 12:10 -0500, Brown, Len wrote: > >> What do people think? > > I agree that we can improve on last year. > > There was no shortage of interest at the KS -- it seemed > that everybody except Linus was running Linux on a laptop > and had a personal interest in saving power, particularly > through suspend/resume. > > But when we started to dig into a discussion and Alan was > talking about probing devices on resume like we do at boot, > Linus cut him off and basically said "stop talking about > implementation -- go cut some code, and if the code works > people will use it and it will improve over time." Len, I think that Linus simply doesn't have a good idea of what is involved in PM matters here. We should just gather all interested parties and try to make progress. >From my experience (and pmac had _working_ PM suspend/resume for ages in Linux, hint hint hint :), we are very close from some stable framworks for system wide PM, +/- the usual issues with video chips, ACPI, and other platform specific details (don't laugh please), but very far from a model that would properly deal with dynamic power management (runtime PM of devices/buses). > I found this extremely frustrating, because I don't think > there is a consensus on what the requirements should be, > and so it is difficult for people to effectively work > together and cut code if they don't share a common > vision of where we're going. There is none, and there is variable amount of experiences with the issues involves, which means discussions are difficult at best. Add that the context of KS with limited time frames and everybody wanting to get off and have a beer... > At OLS, there was also no shortage of interest. I was > amazed that a whole room full of people showed up for > my talk on ACPI and also for the Linux-laptop BOF. > A flock of people followed me around after that to > make sure I was aware of exactly how their laptops > were not working:-) > > But these forums were not particularly productive > WRT tangible planning or problem solving. They were > probably most useful just in getting various players > on the same page WRT where we are. Agreed. > So I think at both of these forums what would go over > really well would be a crisp articulation of what we think > the requirements should be -- where we think we need > to go -- and what the issues are getting those > features in place. The outcome will be that at least > everybody present will have a better understanding > of the problems, and thus be better able to contribute > with their ideas for solutions. > > For a PM-summit to be most effective, it would have > to occur _before_ the KS, like the networking guys did > last year. (either that, or as I think Patrick suggested, > there should be a focused PM BOF during the KS itself). > A BOF later at OLS would be good in any case b/c it > can have a broader participation than KS. But it would > be more effective to have a pre-meeting preceeding the KS > so that crew could at least weigh-in on what we think > the requirements and challenges are. I totally agree. I think we badly need a PM workshop of some sort, either before or after KS/OLS, if possible at the same place, with a least a couple of days so we have time to 1) expose all the issues, 2) get over the usual misunderstanding, 3) agree on a set of solutions. Ben.
