Theses are the minutes from the Hotplug SIG con call held 12/14. Attendees: Renier Morales, IBM Bruce Vessey, Unysis George Mann, BULL Randall Loomis, Wind River Chris Johnson, SUN Russ Anderson, SGI Mark Wong, OSDL Mary Edie Meredith, OSDL Martine Silbermann, HP * Status of memory hotplug regression testing Mark Wong from OSDL has joined the Hotplug SIG and will be doing most of the regression testing work. The hardware donated by IBM is currently being set up in OSDL's lab and should be available shortly. In the meantime Mark has started running memory hotplug regression tests on his ia32 system. While running tests-041126 against the 2.6.10-rc1-mm5-mhp1 he got an oops related to aio. Dave Hansen submitted a fix that works for x86 and ppc64 but might need to be revisited for other architectures. Mark also submitted patches from Dave Hansen's tree to PLM. For example 2.6.10-rc2-mm4-mhp3 for which compilation failed on most architectures except for ppc64, x86_64 and ia32. We should address that issue so that compilation passes on all architectures of interest to the DCL group. Mark's projects for the near future are: - to set up the systems in the lab, - to gather the latest memory hotplug patches and include them in PLM - to get a better understanding of the existing tests - to generate a test plan. With the help of the community we need to identify intermediate steps that constitute milestones in the memory hotplug code and that can be tested. * Status of CPU hotplug regression testing, Joel send a set of scripts to do the regression testing. Mark will look at them and determine what exactly is being tested and if that fits all the needs or if further tests need to be developed. There are definitely gaps in the tool chain to take advantage or even support some of the existing kernel code. We need to determine what is there and what needs to be developed for proper support of CPU hotplug. * Hotplug at the HPI/IPMI level, Randall reported on his side discussions with Tariq. Those were mostly focused on bringing him up to speed on the current development of OpenHPI. Tariq plans to visit Wind River after the release of OpenHPI 2.0 which should happen at the end of the month. When asked how the SIG could help the OpenHPI development in terms of hotplug Randall suggested that our group can set up management and test cases and that OSDL's lab can be used as a test site. Renier commented on the fact that OpenHPI has a plugin which supports sysfs and that that could be a natural way for hotplug to talk to the OpenHPI. Another way the SIG could contribute to the development of OpenHPI is to extend the range of hardware it supports. In terms of current adoption of OpenHPI by the distros, it seems that SLES9 has already included it and so did Force Computers. Future release from Wind River for CGL (and maybe from MontaVista) will support OpenHPI as well. * Status of documentation of CPU hotplug, Martine submitted a first draft for review to this group. Joel provided some feedback which included a suggestion to extend the documentation to cover every aspect of hotplug from the hardware all the way to the user. Everyone on the call agreed with that suggestion. * Other topics: - LWE in Boston, we should start promoting the hotplug BOF. AR: Martine will contact Joel on what the current status on that BOF is. - DCL folks have a DCL Tech Board meeting in January at Beaverton, we might have a special hotplug meeting w/ phone access for the people that aren't present in the room if folks feel that that would be useful. Next meeting: January 4th, 2005 at 11:00am Pacific, 2:00pm Eastern Happy Holidays to all of you and Best Wishes for a great and successful Year 2005. Martine J. Silbermann
