* Discussion to generalize the maintenance process of kernel branches
for the periodic Linaro releases. This still needs documenting in the wiki.
- Merged a set of patches put together by Anand Gadiyar to add display
support to the PandaBoard.
Merged the following additional patches:
- d983450 cpufreq: Add documentation for sampling_down_factor
- 054fcd5 ARM: S5P: Fix end address in memory resource information for UART devices
- bf47520 ARM: make SWP emulation explicit on !CPU_USE_DOMAINS
- 459967a ARM: fixup SMP alternatives in modules
- 7144fbc ARM: 6654/1: perf/oprofile: fix off-by-one in stack check
- 067dfdf ARM: 6659/1: Thumb-2: Make CONFIG_OABI_COMPAT depend on !CONFIG_THUMB2_KERNEL
- cc0f308 ARM: 6656/1: hw_breakpoint: avoid UNPREDICTABLE behaviour when reading DBGDSCR
- 40ef21c ARM: 6657/1: hw_breakpoint: fix ptrace breakpoint advertising on unsupported arch
- 9e97118 ARM: ptrace: remove single-step emulation code
== Upstream oriented activities ==
* Review of Arnd Bergmann's flash card article for LWN.
* Incorporation of feedback to the patch adding Thumb2 support to the P2V branch.
* Another look at the Thumb-2 compatibility fixes for OMAP from Dave Martin.
* Review of a patch series adding support for SDHCI v3.00.
* Posted patches:
* Rework of the kernel decompressor code to improve efficiency * Removal of the 4x expansion presumption while decompressing the kernel * kprobes insn decoding fix * Ignore mdesc->boot_params if out of range * Fold lookup_machine_type() into setup_machine()
== Linaro kernel activities ==
* Looked at some bugs:
* Bug 660811 * Bug 720055
* Merged 2.6.37.1 into linaro-2.6.37
* Merged Dave Martin's Thumb2 compatibility patches for OMAP.
* Merged core ARM ffixes from RMK.
* Merged OMAP fixes from Tony Lindgren.
* Opened up the linaro-2.6.38 branch.
Nicolas
== Upstream oriented activities ==
* WIP patch to clean up register allocation in kernel/head.S.
* Couple random patch reviews from the lak mailing list plus related discussions.
== Linaro kernel activities ==
* Populating the linaro-2.6.38 branch with a couple topics: * Rebased patches from linaro-2.6.37 that didn't make it to 2.6.38-rc5. * Merged OMAP4/Panda display support compiled by Anand. * Many patches from RMK's repository as found on Feb 21: * the "fixes" branch * the "devel-stable" branch * the "p2v-stable" branch * the "v6v7" branch * the "misc" branch * Merged Thumb2 addition and some misc to the p2v branch from myself. * Merged Dave Martin's Thumb2 support patches (mainly for OMAP). * Merged device tree support patches from Grant Likely. * Merged small suspend/resume fix for Samsung.
* Announcement of the linaro-2.6.38 availability, asking people to consider it for stuff they want to show up in the 11.05 Linaro release.
* Pondering on the LTTng merge for linaro-2.6.38.
Nicolas
== Upstream oriented activities ==
* Review and comments on John Bonesio's patch to allow the concatenation of a dtb with the kernel zImage.
* Fix for the patch removing the fixed 4x zImage inflation.
* Couple additional patch reviews from the lak mailing list.
== Linaro kernel activities ==
* Merged the following into the linaro-2.6.38 tree:
* OMAP Cpuidle optimization patches from Vishwanath Sripathy * OMAP4 MPU DVFS patches from Vishwanath Sripathy * Workaround for bad module relocation with Thumb2 from Dave Martin * The 'fixes' branch from RMK * OMAP fixes from Tony Lindgren * The mainline tree up to 2.6.38-rc7. * Addition of errata 753970 and USB to UX500's defconfig from Linus Walleij * Some OMAP I2C fixes from Andy Green (might have to be revisited after some comments from Benoit Cousson)
* Some investigations for bug 720055 (nothing concrete so far, although dmart seems to have zoomed even closer to the problematic code).
* Initial look at the Mali kernel tree which pretty much stopped at the diffstat (more than 30k lines in a single commit).
== Absence ==
* I will be away next week, back on March 14th.
Nicolas
== Upstream oriented activities ==
* Another review on John Bonesio's patch to allow the concatenation of a dtb with the kernel zImage.
* Discussion with Dave Martin and " Tixy" about kprobes for Thumb2.
* tracking of merge progress in mainline during the merge window.
* Patch to kernel/sleep.S: fix Thumb2 compilation issues
== Linaro kernel activities ==
* Merged the following into the linaro-2.6.38 tree:
* The mainline v2.6.38 release * The 'rmk/fixes' branch * The 'rmk/devel-stable' branch * Fix assembler breakage in sleep34xx.S from John Rigby * The '(rmk/misc' branch * OMAP4: PandaBoard: remove unused power regulators
== Plans ==
* Merge LTTng core support (the location for the extra LTTng modules needs to be sorted out).
* Merge more of the ARM support from the mainline merge window into linaro-2.6.38, as well as the last bits from Linaro folks.
* Let people know that the real kernel freeze is coming.
Nicolas
== Upstream oriented activities ==
* Review/comments on Jon Medhurst's (aka Tixy) initial kprobes patches for Thumb2.
* Discussion on LKML about board specific quirks applied to generic USB drivers, that originated from a set of RFC patches from Andy Green. A solution involving two fronts was found which should improve ethernet configuration on the Panda and similar boards.
* Final review of John Bonesio's patch to append a DTB to kernel zImage.
* More tracking of merge progress in mainline during the merge window.
== Linaro kernel activities ==
* Briefly discussed LTTng issues. Modules will be separately built. Still waitting a bit more before merging LTTng for OMAP4 fixes.
* Rebuilt the linaro-2.6.38 branch from scratch, and propose it to replace the current kernel tree. This includes: * Linux v2.6.38 * RMK's branches for mainline (devel, devel-stable, for-linus, etc.) * Latest USB, MMC, I2C, MFD, and async_tx support from mainline * Latest merges for OMAP, MSM, Tegra, (available outside of RMK's branches) * The v2.6.38.1 stable branch * OMAP fixes for I2C and SMC95xx MAC/dev_name * More minor fixes.
Nicolas
On Mon, 28 Mar 2011, Nicolas Pitre wrote:
== Upstream oriented activities ==
* More kprobes discussions about current defficiencies in the ARM mode version, and what to do with undefined instructions.
* Major discussion about the current state of the ARM support in mainline. (http://news.gmane.org/group/gmane.linux.kernel/thread=1114495)
== Linaro kernel activities ==
* Discussed during the kwg meeting if the rebuilt Linaro kernel I produced should replace the previous one or not. General agreement for going forward with the new branch. However, since the Linaro Android tree is a consumer of the Linaro kernel tree, I merged the old and the new branch together instead, after patching the old branch to make it identical to the new branch, so not to force any rebase operations on consumers.
* Merged v2.6.38.2.
* Re-merged the latest DT support after the rebuild of the kernel tree.
* Merged various small fixes.
* Re-merged OMAP DVFS after the rebuild of the kernel tree.
* Merged LTTng support.
== Upstream oriented activities ==
* Review and queuing of several ARM kprobes patches from Tixy.
* Review dmart's FIQ rework patch.
* Fix a couple issues with personality handling in the kernel on ARM which affected the gdb test suite.
== Linaro kernel activities ==
* Merged dmart's patch moving ptrace to regsets framework allowing VFP state in coredumps.
* Merged more DT support collected by Grant.
* Merged more generic fixes from RMK.
* Discussion with Andy Green about the possible merge of a bunch of OMAP4/Panda patches he collected. This bunch is quite large making me somewhat nervous... Needs further considerations.
Nicolas
== Upstream oriented activities ==
* More reviewing and queuing of several ARM kprobes patches from Tixy.
* Pushed a few patch upstream to fix issues with the ADDR_NO_RANDOMIZE personality flag (allows for the gdb test suite to pass).
* Investigation about the kernel decompressor acting funny with the LZMA decompressor, or even corrupting DTB when the DT append patch is used. No resolution yet.
* Initial review of tglx's IRQ consolidation patchset.
== Linaro kernel activities ==
* Merged v2.6.38.3 into linaro-2.6.38
* Merged the following fixes: * support pmu/perf on OMAP4 * vexpress: Update basic DT to use skeleton.dtsi file * Fix broken dtbs rule. * MUSB: shutdown: Make sure block is awake before doing shutdown * OMAP: Fixed gpio polarity of gpio USB-phy reset. * basic DT support on vexpress * Preserve the VFP state during fork * perf: ensure pass through zero is counted on overflow * EXYNOS4: Fix card insert/removal event detection on smdkv310 board * EXYNOS4: CPUIDLE Support
* More pondering about merging Andy Green's OMAP4 branch (still needs confirmation it doesn't break OMAP3).
* Initial investigation about IRQ afinity not working properly (affects mainline too).
* Much communication (emails, etc.) about getting the work from landing teams merged in the Linaro kernel tree.
== Upstream oriented activities ==
* On-going reviewing and queuing of ARM kprobes patches from Tixy.
* More investigation about the kernel decompressor acting funny with the LZMA decompressor, or even corrupting DTB when the DT append patch is used. Two separate issues were identified so far with cumulative effects: global variables marked static causing GOTOFF relocations instead of the expected GOT reference, and a misaligned stack in some circumstances reaching havoc with LDRD/STRD instructions. Those fixes plus a couple related cleanups were posted to LAK. One report indicates problems on a specific target may still exist.
* Initial review of tglx's sched_clock consolidation patchset.
* Review of Will Deacon's patch for SMP futex support.
== Linaro kernel activities ==
* Call for testing: OMAP4 extra support from Andy Green.
* Review of the proposed ST-Ericsson kernel repository. Sent comments to Lee Jones.
* Merged the following in linaro-2.6.38: * Fix infinite loop in ARM user perf_event backtrace code * branch 'omap-edid' of git://git.linaro.org/people/doanac/linux-linaro-2.6.38 * a couple zImage fixes and cleanups from me described above.
Nicolas
== Upstream oriented activities ==
* Git pull request for Tixy's ARM kprobes patches to RMK.
* More work on the kernel decompressor acting funny. A third issue was discovered by Tony Lindgren. Reworked some of my previous patches and prepared a Git branch for RMK to pull.
* Looked at some issue with __aeabi_uldivmod unexpectedly being referenced in the kernel. Turned out to be a toolchain problem.
== Linaro kernel activities ==
* Initial review of the proposed Freescale kernel patches. Sent comments/questions to Eric Miao.
* Merged the following in linaro-2.6.38: * Merged v2.6.38.4. * Merged Tixy's kprobes work. * Replaced SMP futex support with Will Daecon's version, plus assorted generic improvements from 2.6.39-rc1. * Merged my latest zImage fixes, replacing older patch versions when applicable.
* Started a discussion with Angus Ainslie about merging Samsung support in the Linaro kernel tree.
== Upstream oriented activities ==
* Git pull request to RMK: switch Dove to ARMv7 mode only
* Review Per Forlin's SDIO single function optimizations patch.
* Public pondering about possible removal of kernel XIP support. Is it still relevant and used these days? Apparently it is.
* Initial investigation for removing a lot of machine specific memory.h files by always relying on the p2v patching.
* Git pull request to RMK: zImage fixes
== Linaro kernel activities ==
* More direct landing team prodding to get their work merged before LDS. This was generally a failure as about nothing has actually been merged.
* Merged the following in linaro-2.6.38: * omap3: alsa-soc: Remove restrictive checks for cpu type * mmc: mmci: Add ARM variant with extended FIFO * ARM: Thumb-2: Relax relocation requirements for non-function symbols * Merge commit 'v2.6.38.5' into linaro-2.6.38 * ARM: zImage: direct comparison with pc is not allowed in Thumb2 mode
* Lots of miscelaneous email discussions with a couple people (plus time to perform some research in some cases).
== Upstream oriented activities ==
* Reviewpatch series consolidating Marvell Orion/Kirkwood/Dove from Andrew Lunn
* Git pull request to RMK for the above and the generic IRQ chip changes.
i Review patch to fix highmem detection when using SPARSEMEM.
* Review patch to indicate thumbness of the kernel build and investigate alternatives.
* Couple email responses related to ARM subarchitecture maintainership.
* Initial work for removing a lot of machine specific memory.h files by always relying on the p2v patching. So far the stats are: 55 files changed, 113 insertions(+), 1330 deletions(-).
* Review of Paul McKenney's LWN article on Linaro/ARM kernel work.
== Linaro kernel activities ==
* Merged the following:
* v2.6.38.6 * smsc95xx: fix reset check * ptrace: Prepare to fix racy accesses on task breakpoints * arm, hw_breakpoints: Fix racy access to ptrace breakpoints * mmc: omap: Fix possible NULL pointer deref * ARM: 6890/1: memmap: only free allocated memmap entries when using SPARSEMEM * ARM: 6870/1: The mandatory barrier rmb() must be a dsb() in for device accesses * branch 'rsalveti-andy-for-nicolas' of git://kernel.ubuntu.com/rsalveti/linux-linaro-2.6.38 * fix highmem detection when using SPARSEMEM * Add unwinding support for division functions * v2.6.38.7
== Upstream oriented activities ==
* More work on the elimination of arch/arm/*/include/{mach,plat}/memory.h which is one more step towards a single kernel binary for multiple SOCs.
* Investigation of the kernel side support for GCC's 64bit __sync_* ops on ARM which resulted in a preliminary patch.
* Adaptation of John Bonesio's DT append patch to latest changes in the zImage decompressor code.
* Mailing list monitoring and interactions related to the upstream merge window and other issues.
== Linaro kernel activities ==
* Merged Beagle fixes from Ricardo Salveti.
* Tagged linaro-11.05-2.6.38 for the 11.05 release.
* Preparation for the linaro-2.6.39 tree.
Nicolas
== Upstream oriented activities ==
* REview and discussion for a fix to the SDIO single IRq optimization.
* Discussion of the kernel side support for GCC's 64bit __sync_* ops on ARM and the best way for user space to test for its availability.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Production of the initial linaro-2.6.39 tree. This also contains the following from mainline which can be found in v3.0-rc1:
* RMK's 'ep93xx', 'fixes', 'misc', 'mmci', 'remove', 'spear' and 'devel-stable' branches; * Rafael Wysocki's PM work; * Dave Jones' cpufreq 'next' branch; * Tony Lindgren's omap-for-linus branch; * Anton Vorontsov's battery subsystem branch; * dwmw2's MTD branch; * GCL's GPIO consolidation branch; * Kukjin Kim's 'for-rmk' branch for Samsung support; * Sekhar Nori's 'davinci-next' branch; * GCL's 'devicetree/arm-next' branch; * MMC 'for-linus' branch from cjb; * OMAP DSS 'for-paul' branch from Tomi Valkeinen; * RMK's timers-clocksource-for-linus branch; * ALSA's 'for-linus' branch from Takashi Iwai; * Greg KH's usb-next branch; * plus a few random fixes.
* Initial look at merging the LPAE branch from Catalin.
== Upstream oriented activities ==
* Review of a patch for mmp clockevent handling race.
* Revisited the issue surounding dynamic patching of the kernel binary happening in discarded sections.
* Debugging hunt to solve non booting kernels affecting both mainline and linaro-2.6.39. One prevented successful boot with Thumb2, the other one required CONFIG_OF and no DTB provided. Two patches resulted which were pushed upstream.
* Rebase and cleanup of some patches allowing the appending of a DTB to the kernel zImage. Resulting patches posted for review.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Merge of the following in linaro-2.6.39:
* A bunch of OMAP fixes gathered by Tony Lindgren * The v2.6.39.1 stable branch * Fix compilation error for exynos4_defconfig * Revert a few problematic TTBR0/ASID related patches * The two patches preventing boot mentioned above * Two patches from agreen to fix up random or missing MAC addresses for eth0 and wlan0 * A bunch of ARM fixes gathered by RMK: 6947/2, 6948/1, 6951/1, 6952/1, 6894/1.
* Planning the coordination for the monthly Linaro kernelsnapshots.
Nicolas
== Upstream oriented activities ==
* Creation of a compatibility layer to zImage, based on an earlier patch from John Bonesio, allowing runtime translation of ATAGs from legacy bootloaders to FDT nodes. Posted patches for review. Still requires some work.
* review of RMK's cpu_suspend patch series.
* Review of some kprobes infrastructure patches from Tixy.
* Review of patches adding generic proc/arch struct definition macros from Dave Martin.
* Review of the patch series to use nonblock mmc requests to minimize latency from Per Forlin.
* Discussions about: * user space alignment faults * usage of the packed and aligned attributes together * implementation of the IO accessors on ARM
* Cleanup of the kuser helpers documentation, and addition of a cmpxchg64 version. Posted patches for review.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Merge of the following in linaro-2.6.39:
* some ARM fixes from RMK * MMC patches from Shawn Guo fixing bug #754254 * OMAP I2C cleanups/fixes from Andy Green * PMU support for OMAP4 from Ming Lei
== WARNING ==
* The linaro-2.6.39 tree will be frozen from this Thursday for the 11.06 release which should happen Thursday of next week. Only bug fixes will be accepted during that period.
* OMAP4 HDMI support is still reported as broken.
Nicolas
== Upstream oriented activities ==
* Work on the zImage ATAG-DTB compatibility layer to create needed FDT nodes when missing.
* REview of Linus Walleij's LED trigger patches for CPU activity.
* Review of Dave Martin's patch series to clean up the CPU specific low-level assembly support code.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Merge of the following in linaro-2.6.39:
* The OMAP I2C cleanup series from agreen. * A bunch of 3.0-rc1 MMC patches that I intended to merge early on but somehow missed. * The board specific basic DT enablement branch from gcl. * PMU support for OMAP4/Panda. * Asynchronous request support for MMC from Per Forlin. * Fixes for basic UX500 support from linusw. * Fixes for SMP issues from RMK. * HDMI support for OMAP4. * The DTB appended to zImage patch, plus ATAG conversion.
* Tagged linux-linaro-2.6.39-2011.06-0.
== Upstream oriented activities ==
* Work on patch series removing hardcoded absolute physical address in mdesc->boot_params.
* Work on patch series removing compile time usage of ARM_DMA_ZONE_SIZE.
* Continue work on patch series removing many instances of arch/arm/*/include/mach/memory.h.
* REview of idle notifiers from Todd Poynor.
* Posted RFC patches removing incomplete/unused ARM machine code.
* Review patch series making mach/hardware.h optional from Rob Herring.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Merged only one patch to linaro-2.6.39 fixing a bug in the MMC test module. If the lack of more fixes is indicative of anything, linux-linaro-2.6.39 for 11.06 should be close to perfection.
== Upstream oriented activities ==
* Posted the following patch series to contribute to the single kernel image goal: * removal of hardcoded absolute physical address in mdesc->boot_params * removal of compile time usage of ARM_DMA_ZONE_SIZE * removal of a few StrongARM cache-related build time constants * removal of many arch/arm/*/include/mach/memory.h instances * removal of more unused/incomplete ARM machine code
* Processed feedback on the above from the mailing list (fixes, acked-by tags, etc.)
* Review of the following patches and series: * poisoning initmem when it is freed * introduction of a led trigger for CPU activity * linker layouts reorganization * support for XZ compressed kernels
* Discussion of a couple design issues with the Thumb2 kprobe implementation.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Merged the following to linaro-2.6.39: * OMAP: Overo: remove duplicate call to overo_ads7846_init() * mmc: documentation of mmc non-blocking request usage and design. * omap2 overo fix video issue
* Reverted "arm: omap4: support pmu" from linaro-2.6.39 as this is apparently causing boot hangs for some people, and no progress has been made towards a fix after a week.
== Upstream oriented activities ==
* Review of the kprobes patches from Tixy (all 75 of them). This is now merged in RMK's tree.
* Additional cleanups to the patch series removing per machine hardcoded DMA region size, plus various ACKs. Need to rebase on top of RMK's "replace ISA_DMA_THRESHOLD with a variable" commit.
* Quick look at possible solutions to remove per machine hardcoded VMALLOC_END.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Merged the following into linaro-2.6.39: * mmc: block: fixed NULL pointer dereference * Add ARM cpu topology definition
* Initial preparation of a linaro-3.0 tree.
Nicolas
== Upstream oriented activities ==
* Rebased patch series removing build time DMA region size on top of RMK's latest changes in that area, pushed upstream.
* Review proposed solution to remove the 2MB RAM offset in the MSM subarch making uniformization difficult.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Creation of linaro-3.0, primed with the following: * The comprehensive ARM kprobes work from Jon (Tixy) Medhurst. * The new processor struct macros from Dave Martin. * The ARM cpu topology definition from Vincent Guittot. * Basic Cortex A15 support from Will Deacon & Pawel Moll. * DMA infrastructure cleanups from Russell King * Removal of the build time DMA region size from myself (mentioned above). * A kernel helper to perform 64-bit atomic operations from myself. * Enhanced DT support for boards from Grant Likely, Shawn Guo, Manjunath Kondaiah, and Thomas Abraham.
* The above was frozen for 11.07 and tagged as linux-linaro-3.0-2011.07-1.
Nicolas
== Upstream oriented activities ==
* Review few patch from Dave Martin for some unaligned access issues.
* Production of a task list document for the single zImage consolidation effort, posted to linaro-dev and linaro-kernel: http://article.gmane.org/gmane.linux.linaro.devel/6008
* Produced a patch to remove the unused ARM_PATCH_PHYS_VIRT_16BIT once msm moves to the real RAM boundary.
* Another look at bug 709245 "panda: USB disk IO slow" which is converging towards a generic SMP scheduler bug. More investigation needed.
* Usual mailing list monitoring.
== Linaro kernel activities ==
* Not much. The Linaro 11.07 kernel is released as linux-linaro-3.0-2011.07-1.
== Linaro Connect ==
* Work on single zImage consolidation, on the arch_idle() cleanup in particular.
* Meeting with RMK and other people to discuss about CMA implementation details.
* Several presentation attendances.
== vacation ==
I will be on vacation the week of August 8th.
Nicolas
== PATCH review ==
* Fix Thumb-2 undef handling for mixed-arch kernels (from dmart)
* Enabling runtime P2V by default (from RMK)
* allow sched_clock to be selected at runtime (from Marc Zyngier)
* Add safe diagnostic to indicate when __cpu_architecture isn't set up (from dmart)
* Setup consistent dma size at boot time (from Tixy)
* use kconfig choice for selecting DEBUG_LL UART (from Will Deacon, which later turned into a lively discussion with RMK)
== Upstream contributions ==
* Started work on a patch series for the removal of all occurrences of <mach/vmalloc.h>. Some additions to the core vm code is required.
* Rebased my patch series for the boot_params to atag_offset conversion, submitted pull request upstream.
== Linaro kernel maintenance ==
* Merged branch 'for-linus' of master.kernel.org:/home/rmk/linux-2.6-arm as found in mainline commit b6844e8f64.
* Merged fixes collected by RMK for the above.
* Merged v3.0.1, v3.0.2, and v3.0.3.
* Tagged linux-linaro-3.0-2011.08-0.
* Investigation of a build failure reported by John Rigby, which turned out to be toolchain related.
== PATCH review ==
* NR_IRQS consolidation/removal patch series (from svenkatr)
* MMU disabling code and kexec fixes (from Will Daecon)
* change from __backtrace to dump_stack_regs in show_regs (from Laura Abbott)
== Upstream contributions ==
* Rebased and resubmitted my patch to get rid of a few StrongARM cache-related build time constants (prerequisite to another patch series removing most instances of <mach/memory.h>)
* Continued work on removal of <mach/vmalloc.h>, explored ways to ddynamically set VMALLOC_START at run time. Nothing satisfactory so far.
== Linaro kernel maintenance ==
* Merged those fixes:
* Fix Thumb-2 undef handling for multi-CPU kernels * EXYNOS4: Add restart hook for proper reboot
== Week of August 29th ==
* Reviewed enable_fiq/disable_fiq changes from svenkatr
* Initial review of Tixy's kprobes test code
* Posted "musb_debugfs.c: remove unneeded includes on ARM" for comments
* Scratched the surface of the EHCI performance problem on Cortex-A9 with SMP enabled
* Posted a set of patches revamping the low level debug code for OMAP
* Continued work on the big mach/memory.h removal patch series
* Administrivia for the ARM kernel workshop and ELC in Prague, immediately followed by Linaro Connect in Orlando (flight booking, etc)
== Week of September 5th ==
* Off on Sept 5th (Labor Day).
* Sent RMK a pull request for the big series removing most mach/memory.h instances, including all the needed prerequisites
* Reviewed Thumb2 compatibility patches for Marvell PJ4 from dmart
* Experimented with the new Kbuild vdir facility from Arnaud Lacombe, hoping to clean up the zImage DTB append patch series with it. Unfortunately it doesn't work for this particular case without more changes which are becoming a bit too numerous for my taste, so dropped this idea for now.
* Review of the per SoC descriptor patch series from Marc Zyngier
* More review of the ARM kprobes test code from Tixy
* Started work on the vmalloc.h removal
== Week of September 12th ==
* Ffixed second SPI initialization call on Dove
* Posted latest incarnation of the zImage DTB appending patch series
* Continued work on the vmalloc.h removal
* Sent RMK a pull request for the zImage DTB appending series
* Posted initial patch series for the vmalloc.h removal
Nicolas
== Week of September 19th ==
* Review 64-bit division unwinding patch
* Pull request reminder for the mach/memory.h removal series
* Merged v3.0.4 into the Linaro kernel tree
* Patch to remove the unmaintained mach-nuc93x
* Released linux-linaro-3.0-2011.09-0
* Inspection of the shmobile code trying to find out why it uses such a humongous amount of consistemt DMA memory to no avail. Pinged the maintainers.
* Review of the patch series removing mdesc argument to fixup function from RMK.
* Review of the patch series shrinking sched_clock some more from RMK.
* Review of the futex fix from Will Daecon.
* Pull request for the mach/vmalloc.h removal series.
== Week of September 26st ==
* Followed up on the feedback from RMK for the mach/memory.h removal series. Removed the part that he's unable to test and resubmitted.
* Feedback on the mach/vmalloc.h removal series. Investigation, suggestions for a fix, and merging of a couple additional patches from Rob Herring.
* Review of a patch to binfmt_elf Fixing PIE execution with randomization disabled that happened to be wrong.
* More work on the mach/vmalloc.h series, after testing on OMAP revealed multiple bugs coming from pre-existing abuses of the current interface in the OMAP code.
* Started reviewing Matt Waddel's Git training course.
Nicolas
== Patch Review ==
* LED trigger series from Bryan Wu
* ARM kprobes test framework from Tixy
* Thumb2 build fixes for PXA/PJ4 iWMMXt code from Dave Martin
* Kbuild addition allowing code re-use across different directories from Arnaud Lacombe. While this has value, it doesn't do what I need for the DT append to zImage patch series.
* Per SoC descriptor patch series from Marc Zyngier (couple itterations)
* SMP preemption/task migration cache coherency fixups from George Davis
== Upstream Contributions ==
* Posted patch series to rework the low level UART debugging code on OMAP1, OMAP2 and Davinci. Acked by Tony Lindgren and Kevin Hilman. This is a prerequisite to the mach/memory.h removal patch series. Now in RMK's tree.
* Produced and posted a mach/memory.h removal patch series to clean up now redundant and unneeded code. The summary stat is a net reduction of almost 1000 lines, and another step towards the ability to combine multiple SOCs in the same kernel binary. Now in RMK's tree.
* Reworked and posted the patch series for the appending of a device tree binary to the kernel zImage. Non trivial cleanups to the kernel decompressor were required. Now in RMK's tree.
* Reworked and posted patch series removing all instances of mach/vmalloc.h. This also allowed for a common ioremap optimization where static mappings are reused whenever possible, allowing all machines to benefit instead of having special __arch_ioremap() implementations duplicated everywhere. Testing on Panda revealed problems due to reliance on implementation details with the OMAP version of __arch_ioremap() that this series replaces. A plea for help to the OMAP guys produced all the necessary cleanups in record breaking time, however the dependencies on stuff queued for the v3.2 merge window are many. I therefore decided to postpone this series for the v3.3 merge window.
* Initial work on the zreladdr cleanup/removal.
== Linaro Kernel Maintenance ==
* Merged the following in linux-linaro-3.1:
- Thumb-2 undef handling fix for multi-CPU kernels from Dave Martin - ARM cpu topology definition from Vincent Guittot - RMK's devel-stable branch from Oct 17 that contains (among others): - runtime consistent dma size init from Tixy (single zImage topic) - boot_params to atag_offset transition from myself (single zImage topic) - ARM pmu/perf updates from Will Deacon - the ability to append a DT to zImage from myself and others - the ARM kprobes test infrastructure from Tixy - suspend/resume consolidation and cleanups from RMK - cpu pm notifiers on ARM from Colin Cross - MX51 cpuidle support from Robert Lee
* Merged the following in linux-linaro-3.0:
- stable v3.0.5 - SPI initialization fix for Dove - stable v3.0.7
* Tagged linux-linaro-3.1-2011.10-0, quickly followed by linux-linaro-3.1-2011.10-1. Problems are being reported now (missing DT support patches) so another tag is likely in the near future.
== Miscellaneous ==
* Planning committee for the upcoming ARM subarchitecture workshop at Kernel Summit in Prague
* Trip arrangements and registrations for ARM workshop (Prague), ELC-E (Prague), and Linaro Connect (Orlando).
== Patch Review ==
* The LPAE patch series from Catalin Marinas
* The arch_reset patch series from RMK
* Multi IRQ handler patch series from Marc Zyngier
* Initial series to move towards DT support for Orion/Kirkwood/Dove
== Upstream Contributions ==
* Rebased my patch series reworking static IO mappings and removing mach/vmalloc.h to v3.2-rc1. Still need to retest and repost.
== Linaro Kernel Maintenance ==
* Merged the following in linux-linaro-3.1:
- DT support for a couple boards from Grant Likely - LPAE support from Catalin Marinas - Samsung cpuidle work gathered together by Amit Kachhap - The v3.1.1 stable branch from Greg KH
* Announced that the Linaro 11.11 kernel freeze will happen on Nov 14 for a release on Nov 16.
== Miscellaneous ==
* Discussion with Wolfgang Denk about a set of patches from Stephen Warren fixing some shortcoming in u-Boot for booting Linux in the context of a single zImage for multiple SOCs.
== Patch Review ==
* Top-down mmap layout to fix mmap greater than 2GB from Rob Herring
* Orion mbus setup changes from Andrew Lunn
== Upstream Contributions ==
* Posted my patch series reworking static IO mappings and removing mach/vmalloc.h. I should send a pull request soon.
* Started a series reworking the config around CONFIG_AUTO_ZRELADDR. The goal is to get rid of zreladdr in arch/arm/mach-*/Makefile.boot, or even the whole file. Making the transition transparent is not obvious, especially in the presence of deployed broken (by design) u-Boot versions in the field (they all are).
== Linaro Kernel Maintenance ==
* Merged the following in linux-linaro-3.1:
- sched_mc optimization from Vincent Guittot
- Fix for mmap greater than 2GB from Rob Herring (this is not final version, but we need to expose user space to the associated changes in memory layout to make sure nothing regresses.)
* Tagged linux-linaro-3.1-2011.11-0.
== Miscellaneous ==
* Reviews of the draft "FOSS howto" document from Paul McKenney
== Patch Review ==
* Big Endian (including BE8) fixes from Junxiao Bi
* Optimised swahb32() from Dave Martin
* Some missed patches in RMK's reset rework series.
== Upstream Contributions ==
* Pull request for my patch series reworking static IO mappings and removing mach/vmalloc.h.
* Investigation about some regressions introduced by the above series. The issue turned up to be simple, however I ended up reworking one debugging blackout area in the kernel for that i.e. one function from which it was impossible to get any debugging information out before. The option of having a backup page table to keep the mapping for the debug serial UART is now possible.
* Started looking at the work necessary to remove mach/io.h files.
== Linaro Kernel Maintenance ==
* None.
== Patch Review ==
* kprobes fixes from Tixy * Kirkwood MPP config fix from Arnaud Patard * unwinding for XIP kernel from Uwe Kleine-König
== Upstream Contributions ==
* Fix default value for PHYS_OFFSET Kconfig setting (ARM patch 7186/1)
* Rework the early boot code register assignment to facilitate debugging. Almost ready for submission.
* More look at the debugging code that can help diagnose boot problems with the series moving static mappings inside the vmalloc region. The patch removing the serial debugging blackout is causing problems of its own while trying to debug the caused regression on Snowball.
== Linaro Kernel Maintenance ==
* Merged the following in linaro-3.1: * fault.c: Port OOM changes into do_page_fault, from Kautuk Consul * stable v3.1.2, v3.1.3 and v3.1.4 from Greg KH
* Merged the following in linaro-3.0: * stable v3.0.8 through v3.0.12 from Greg KH
== Miscellaneous ==
* Made some noise about the pending fix for the u-Boot uImage situation which finally resulted in the required patches to be merged in the u-Boot repository. Yay!
== Patch Review ==
* Orion/Kirkwood/etc mbus data rework in preparation for DT support from Andrew Lunn.
* Kirkwood GPIO implementation using atomic set/clear_bit from Holger Brunck. Patch NAKed.
* Generic instruction opcode manipulation helpers from Dave Martin.
* Fix for kernel boot failure on NSLU2 from Arnaud Patard.
* Optimization for process_sdio_pending_irq from Sangwook Lee. I asked for further justification as the optimization may not be so.
* Allow sched_clock to be selected at runtime from Marc Zyngier.
* More assorted small patches from random people.
== Upstream Contributions ==
* Rework the early boot code register assignment to facilitate debugging. Posted for comments. Need to be updated after other patches touching the same area were merged in RMK's tree.
* Resumed work on the arch_idle() removal patch series.
== Linaro Kernel Maintenance ==
* None.
== Miscellaneous ==
* Discussion about removal of NO_IRQ on ARM.
* Helped Daniel Lezcano with regressions on ux500 from my vmalloc series.
* Looked at alleged problems with the dynamic phys-virt patching code, reported by Michael Walle.
== Vacations ==
* Away from December 23rd, back on January 3rd.
== Upstream Contributions ==
* Investigation of a problem with the dynamic v2p/p2v patching reported by Michael Walle. The cause might be some cache issues on Kirkwood in relation with the u-Boot version used.
* Updated my early boot code register assignment rework to RMK's latest devel-stable branch and sent to the patch system.
* Git pull request for updates to the Marvell Orion SOC family.
* Completed work on the arch_idle() removal patch series. Posted patches on the lak mailing list for comments.
== Linaro Kernel Maintenance ==
* Merged the following into linaro-3.1: * Merged branch for thermal CPU cooling from Amit Daniel Kachhap. * Perf fixes from Will Deacon. * Fixed compilation issue involving oprofile_arch_exit from myself.
* Tagged linux-linaro-3.1-2011.12-0, then linux-linaro-3.1-2011.12-1.
== Vacations ==
* Away from December 23rd, back on January 3rd.
== Patch Review ==
* Review of the ARM JIT compiler for packet filters from Mircea Gherzan.
== Upstream Contributions ==
* Integrated comments and review tags to the arch_idle() removal patch series in preparation for RMK to pull. Had to redo it again because of a snafu with some other Samsung patches this ended up depending on.
* Sent shmobile fix for the vmalloc series to RMK.
* Re-updated my early boot code register assignment rework onto mainline as requested by RMK.
* Started studying the big.LITTLE switcher.
== Linaro Kernel Maintenance ==
* Nothing much on that front, beside some discussions with Alexey who is going to take care of the Linaro kernel from now on.
On Mon, 9 Jan 2012, Nicolas Pitre wrote:
[...]
== Linaro Kernel Maintenance ==
- Nothing much on that front, beside some discussions with Alexey
who is going to take care of the Linaro kernel from now on.
s/Alexey/Andrey/
Nicolas
== Patch Review ==
* Patch fixing the omap_barriers_init for generic ioremap changes by Tony. RMK introduced a stricter (and more correct) interface for early memory reservation which breaks it again though.
* Patch series removing NR_IRQS by Rob Herring.
* Patch by Will deacon fixing a serious problem with the clear_refs interface in /proc on ARM. A quick fix, and a more permanent solution involving gate_vma were reviewed.
* Earlier initialization of vectors page by RMK. This could make a patch I produced a couple weeks ago to help with early debugging output during page table initialization much more effective.
== Upstream Contributions ==
* Rebased my patch series removing arch_idle() and most instances of mach/system.h from the source tree onto v3.3-rc1. Sent a pull request.
== big.LITTLE switcher ==
* Helped Dave Martin with the port to the GNU build tools by adding some of the needed library functions.
* Immersed myself into the Cortex A15 and the virtualization extensions documentation.
* Started identifying the needed code to support the b.L switcher from the Linux kernel environment.
Nicolas
linaro-kernel@lists.linaro.org
-
Nicolas Pitre