AR for Bryce: Get more familiar w/ memory statistics available. AR for Mary: 1. Check if memtoy can be used to verify that newly added memory is being used. 2. Check with Yasunori Goto if cross compiling using PLM of his patches would be useful to him. AR for Martine: 1. At next F2F put out a call to the OSDL members to get their memory specialists to give feedback on Mel's and Yasunori's patches. 2. Oversee and contribute to the effort of creating the list of user level tools that report memory usage that will be considered for test case 6. 3. Look at Mel's methodology (included in his patches) and try to derive a testing case that would be useful to him. Attendees: Joel Schopp, IBM Mary Edie Meredith, OSDL Bryce Harrington, OSDL Natalie Protasevich, Unisys Martine Silbermann, HP * Status on CPU testing Bryce has found that it is not possible to automatically fall back on a known good kernel if the boot fails w/ the current kernel on the ppc OSDL has in their lab. He started doing automated testing for hotplug CPU on his NFS machine that is already configured to run in automated mode. This is a 1 CPU x86 system but has hyperthreading which makes it look like 2 CPUs. The hotplug tests could not be run yet due to a configuration issue. Bryce is working on that. IBM has assigned some resources to do hotplug testing and they're interested in integrating OSDL's CPU tests to their test environment (ABAT: Automated Build And test). * Status on memory testing - memtoy Mary has been working on updating the capabilities document for the next F2F and has not had time to work w/ memtoy. * Report on latest patches for memory hotplug There hasn't been a tremendous amount of activity regarding the memory hotplug patches. Memory add is in mainline and will be picked up by the distros. Yasunori Goto just submitted a series of patches to handle memory hotplug for new nodes. Mel Gorman has submitted patches on reducing fragmentation using zones but these patches have not received much comments from the community. Mel is still maintaining a light version of the original fragmentation avoidance patches. AR for Martine: At next F2F put out a call to the OSDL members to get their memory specialists to give feedback on Mel's and Yasunori's patches. * From memory hotplug test plan to scripts - discussion Test Case 2: Hot memory add 1. Boot the system w/ less memory than is available using boot option mem=xxx, 2. Check memory statistics (/proc/meminfo) 3. Add remaining memory. Memory needs to be added in incremental amounts corresponding to the section size on the specific platform (arch dependent) 4. Check memory statistics to see if added memory is "there" 5. Check if added memory is being used. Can memtoy be used to perform this task? AR for Bryce: Get more familiar w/ memory statistics available. AR for Mary: Check if memtoy can be used to verify that newly added memory is being used. Test case 6: Verify that user tools can handle adding memory (removing will be considered when code is in place) 1. Need a determined list of user level tools that report memory usage. There's an existing wiki, please add your ideas to it. 2. Test if those tools handle the added memory properly. AR for Martine: Oversee and contribute to the effort of creating the list of user level tools that report memory usage and therefore will be considered for test case 6. Test case 1: Fragmentation avoidance What could constitute basic testing to identify the level of fragmentation at any given time. What kind of testing would benefit the developers? AR for Mary: Check with Yasunori Goto if cross compiling using PLM of his patches would be useful to him. (IBM already does cross-compiling in house so this would not be helpful to Mel Gorman). AR for Martine: Look at Mel's methodology (included in his patches) and try to derive a testing case that would be useful to him. The next meeting is scheduled on February 28th at 11:00am -12:00pm PST, 2:00pm - 3:00pm EST. Martine J. Silbermann
