This month's meetings

<<MonthCalendar: execution failed [Argument "anniversary" must be a boolean value, not "workinggroups/kernel/meetingtemplate"] (see also the log)>>

Morning Attendees

  • Loïc Minier:
  • Paul McKenney: X

  • Arnd Bergmann: X
  • Anand Gadyar: X
  • Shaju Abraham:
  • Thomas Abraham:
  • Per Forlin: X
  • Mian Yousaf Kaukab: X
  • John Rigby:
  • Steve Sakoman: X
  • Jeremy Kerr:
  • Mounir Bsaibes: X

Evening Attendees

  • John Stultz: X
  • Eric Miao: X
  • Lennart Buytenhek:
  • Jason Hui:
  • Jason Liu:
  • Grant Likely: X
  • Aneesh V: X
  • Paul McKenney: X

Agenda

Action Items

  • What's on what's next on the main page info is outdated: https://wiki.linaro.org/WorkingGroups/Kernel. [Paul]

  • Arnd, Per, Mian, Steve Sakoman: Get together on Flash work. [Done.]
  • Per: Push 5500 Flash-related configuration code (this week or next week). [Done.]
  • Mian: Push PIO driver to mainline staging tree. [Done. Next step to publish externally after acks. Done.]
  • Steve Sakomen: Submit the five uboot patches along with any required fixups. [Done.] Check on uboot -next. [Done.]
  • All: Verify contact information in https://wiki.linaro.org/Internal/EngineeringTeamContacts and https://wiki.linaro.org/EngineeringTeam.

  • Nico: Complete wiki page on Linaro tree policy [Pending], push patches [Done].
  • Carry Over
    • Set a plan handle continuity of Steve's work - Paul/Steve/Loic [Pending]
    • Loic to dig through the blueprints to see which one has u-boot work in them - Loic
    • Loic to talk to other groups to handle kernel testing - Loic
    • Complete blueprints specification - All [Good progress, some left.]
    • Weekly updates sent to Loic and Paul - All [Much improved.]
    • John Stultz will be contacting John Rigby on info/guidance on building Android - John [Done]
    • John Stultz, Jason, Paul: Finish K6 blueprints/specs. [Done, but what work left to do? Dallas sprint?]
    • Anand: Finish K8 kexec blueprints/specs. [Progress, first draft.]

Minutes

  • Paul McKenney: Additional blueprint work. All kernel-configuration use cases (K6) appear to be covered via Kconfig, so not clear that a tool is needed: will propose this topic for the Dallas Sprint. Progress on RCU priority boosting, additional energy-management tasks and scalability tasks look to be easily handled.

  • Arnd Bergmann: Munich with Uli Weigand presenting to Siemens developers. Embedded Linux work across half a dozen architectures. Much interest, but not yet interested in actively contributing on a large scale (though a few groups within Siemens are interested). One group contributing to KVM and a few other areas, mostly in architecture-independent code. Most people interested only in getting their stuff running. Using OpenEmbedded, working on consumer devices such as washing machines. Working on getting Legal procedures in place to ensure compliance with licenses. Reviewed Freescale BSP: graphics driver of concern, as it includes binary-only user-space components. Many drivers, a number of which are close to being upstreamable. Graphics driver is one of the largest drivers. Split out five or six simple branches in preparation for upstreaming. One issue is that commits are too large, often a single commit includes multiple unrelated changes. Could meet with Freescale people in Austin at the end of the week prior to the Dallas Sprint. Looked at gitdm tool, considering sending in a patch to list Linaro people as belonging to Linaro instead of to whatever gitdm currently accounts them to.

  • Anand Gadyar: Distracted by TI work. Finished kexec() blueprint, will get it reviewed. Mini-sprint in Bangalore for PM working group. Many people attended across all of Linaro: Amit K plus Samsung and TI folks. [Ken Werner of toolchain group also playing with kexec.]
  • Per Forlin: Started running IOZone tests, checking performance after adding double-buffering. Initial numbers shows 5% performance gains for large writes and 10% performance gains for large reads. Arnd suggests submitting his benchmarking tool, Per was enthusiastic about this. Work applies to common code between SD and eMMC, probably less percentage improvement on SD due the fact that SD is slower.
  • Mian Yousaf Kaoukab: X Short week (holiday on Friday). USB patches for 8500 have been accepted into Greg KH's tree, and Felipe Balbi's tree -- so on its way to mainline. Battery manager and charging folks within STEricsson can use OTG driver, though changes may be needed. Also, other drivers export similar information, which should be consolidated into a common framework. Should be able to start pushing OTG framework upstream this week or next week. Travel approved for Dallas sprint.
  • John Rigby:
  • Steve Sakoman: Jury duty. u-boot is in the final phases of the 2010-12 release, but some instability remains for relocation. Tested some proposed fixes, some of which seem to be toolchain-version specific. Patch from last week (environment code would incorrectly report correctable errors) accepted in mainline, some comments on 3-patch series, working fixes. Proposed patches within TI for ARMv7 cache management, which are to be upstreamed. Main focus on OMAP u-boot stability. Additional battery-monitoring patch also pending. The three-patch series and the additional patch probably will be taken care of in next release.
  • Jeremy Kerr:
  • Mounir Bsaibes: Analyzed some existing open-source project-management tools that seem to cover what is missing in Launchpad. Scheduling, sizing and load-balancing functionality lacking from Launchpad, so looking at open-source alternatives. Working a getting-started page for developers, with common page for things common across all engineering working groups in conjunction with working-group-specific information. Used BIRT to generate status report from the blueprint helper function output http://ex.seabright.co.nz/helpers/api/blueprints.

  • John Stultz: RTC issues: timer list and timer queue which are precursors to getting the Android alarms upstreamed. Patches in -tip slated for 2.6.38. LWN article for timers and alarms. Getting trivial Android patches organized for submission. One issue: CLOCK_MONOTONIC stops ticking during suspend, which means that you cannot use CLOCK_MONOTONIC to wake up from suspend. Might need another clock type that is like CLOCK_MONOTONIC, but that continues ticking during suspend.
  • Eric Miao: Moving to Freescale landing team, iMX51/53 platforms. Possible Android work: getting Android tree set up. Some things in the process of being upstreamed, but many more candidates for that work. Should talk to Russell King about kernel configuration management.
  • Grant Likely: Lennert working IRQ code in core code in support of virtual IRQ, working no Beagle board (Serial port and eMMC), will move to Panda (or other OMAP board). Grant working on QEMU model of Versatile Express and the Nvidia Tegra (Serial port and eMMC). ITC SPC GPIO device support planned. Dusting off u-boot work that John Rigby has done so that device tree is passed correctly to kernel. Kernel configuration management solution in the works -- "Kconfig fragment". Configuration process loads an alternative per-board Kconfig file that sets defaults for Kconfig, and then sets up with the full Kconfig. Sort of an abbreviated defconfig file, so that you can have definitions set up for a family of SoCs or boards. Sort of like the plat- and mach-, but only in use at defconfig time. Grant's patchset Stephen Rothwell's patchset

  • Aneesh V: TI four years, LSI before that. Most recent area of focus is bootloader.

WorkingGroups/KernelArchived/Meetings/2010-12-13 (last modified 2013-01-14 19:38:11)