On Fri, 2005-08-12 at 10:37 -0400, Martine Silbermann wrote: > On Thu, 2005-08-11 at 13:43 -0700, Bryce Harrington wrote: > > On Tue, May 24, 2005 at 11:06:57AM -0400, Martine Silbermann wrote: > > > Here are the slides that I presented at last week's face to face. I > > > wanted you to have a copy before this afternoon's meeting since we'll be > > > discussing them. > > > > > > Martine > > > > Hi Martine, > > > > I was just reviewing this set of slides and have some more questions: > > > > * CPU hotplug tests - four of the six tests are written; I think MarkD > > is working on another; we're waiting on additional info from Mary and > > MarkW on the other. > > > > For regression testing, it was requested that we run this against all > > the -mm and -git patchsets. Over the last couple weeks I've gotten > > most of the infrastructure for this put into place. I now have all of > > these linux patches being downloaded, applied to their base trees, and > > compiled. I've also tested running the test suite on it. Next I need > > to automate booting the machine to the new kernels, and after that > > will need to develop a reporting tool, but those should be pretty > > straightforward; if I don't get it done tomorrow I hope to tackle it > > after I get back from vacation. > > > Good job puting all that infrastructure together. You might want to > check with the hotplug community on the format of the report, there may > be ways to present the results that make it easier on them to look them > over. You might want to add some sort of notifier in case of errors. > > > Once we have fully automated these six CPU hotplug regression tests on > > all kernel releases, is there any further testing needed for lchp? > > > I believe that running the scripts on the -mm and -git patchsets and > also on the rc trees should cover it. I assume it'll be done on at least > ppc64 and ia64 architectures. > > > * Memory hotplug - I've also set up the CPU patches to compile using the > > same framework as for CPU hotplug, so we'll have this regression > > testing covered as well, and we will also have the ability to run > > additional tests. I asked Dave if he wanted to see any particular > > tests run on this, and he didn't have any in mind. Do you know of any > > existing tests that would be worth running on these kernels? > > > It may be useful to check if there's any existing written tests for > memory migration and defragmentation which would help too in the > progress of getting memory removal to work. I think the user level tools that track memory utilization should be checked as well, for both memory add and memory remove when it is available. It also may be time to start a list of common regressions so we can eventually create a test plan like we did for hotplug cpu. Perhaps this could be a topic on the next SIG call and we can use the wiki to track them? > > > * ia64 system - I have access to this machine, and am working on getting > > it configured to be able to add it to the above framework, so we can > > get compile results. However, I need some direction about > > specifically what testing we want to do on it, beyond just boot > > testing? > > > I thing for CPU hotplug you should run the scripts on it and for memory > you should do the boot testing....until we get other tests to run. > > > Also, do our testing objectives include testing on hardware that has > > memory hotplug capabilities? If so, is there a machine I should be > > looking at for doing this on? > > > I don't think that at this time this is required but we'll keep it in > mind and revisit this in a few months. > > Thanks - Martine > > Thanks, > > Bryce > > _______________________________________________ > Hotplug_sig mailing list > Hotplug_sig@xxxxxxxxxxxxxx > https://lists.osdl.org/mailman/listinfo/hotplug_sig
