LKML Summary Podcast

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091129.mp3

For the US Thanksgiving holiday and weekend of November 29th 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: Early platform driver buffer, Intel processor performance, Kbuild, LIRC, PCI, scheduler, and tracing.

Early platform driver buffer. Magnus Damm posted a patch implementing support for an early platform driver buffer in which such drivers can store pieces of the kernel command line that they will need to later process.

Intel processor performance. Ling Ma (Intel) posted to let everyone know that Intel has performed some compilation comparisons between using the GCC -Os and GCC -O2 flags. Until now, there has been an assumption that optimizing the kernel image (on x86 systems) for size would result in greater performance because of improved cache utilization, but as Ling points out, modern Intel processors have excellent pre-fetch support and benchmarks show they perform better when the GCC optimizations for performance are used instead. The difference can be as much as an 8% performance improvement.

Kbuild. Sam Ravnborg replied to a posting of a git tree from Michal Marek asking Stephen Rothwell to replace the existing linux-next kbuild tree with Michal’s one, suggesting that Michal is about to own that.

LIRC. Jon Smirl took the opportunity of a discussion along the lines of reconciling existing LIRC code into some kind of input layer extension to repost some existing patches he had written that implement in-kernel IR support using evdev. He includes example use cases. Later, Jon posted an RFC intended to stir a debate about what an in-kernel IR system should do.

PCI. Stefan Assmann posted a patch intended to bring the kernel into conformance with the PCI-SIG’s guidelines on the usage of PCI trademarks and logos, by changing occurances of PCI-X, PCIe, Gb/s, and so forth. Separately, Rafael J. Wysocki posted an initial round of patches intended to provided runtime power management for PCI devices through ACPI/native PCIe PME.

Scheduler. Christian Ehrhardt posted to let everyone know that the current kernel isn’t recalculating various scheduler tunables on cpu hot add/remove. Consequently, sysctl_sched_min_granularity, sysctl_sched_latency, and sysctl_sched_wakeup_granularity are calculated in relation to the number of CPUs online at boot time (sched_init_granulatrity). Peter Zijlstra asked if this was “virt junk that’s playing dumb games with hotplug isn’t it?”. In fact it was an s390x feature in which CPUs come and go according to utilization.

Tracing. Li Zefan posted a patch converting a number of trace events over to DEFINE_TRACE, reducing code size and memory footprint.

The latest kernel release is 2.6.32-rc8.

Jan Engelhardt posted to let everyone know that increasing max_map_count to too large a value causes programs to fail to run (for example, attempting to write 2GB-1 to the /proc/sys/vm entry). This suggests some range checking bug exists in that particular sysfs interface file. Amerigo Wang posted a patch.

Stephen Rothwell posted a linux-next tree for November 26th. Since Wednesday, the m68k, trivial, and percpu trees lost conflicts, while the m68knommu, sparc, acpi, net, backlight, and workqueues gained issues. The total sub-tree count remains steady at 154 trees.

Stephen Rothwell posted a linux-next tree for November 27th. Since Thursday, there was a new tree (logfs), a new kbuild maintainer, and a removed tree (kbuild-current, since the maintainer has changed). The sparc, rr, and osd trees had issues, while the workqueues tree lost its build failure. The total sub-tree count remained at 154 trees with the add/remove tree dance.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091125.mp3

For Wednesday, November 25th, 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: GFS2, Kconfig, PATA, and perf.

GFS2. Steven Whitehouse posted an “extra early pre-pull patch” due to the “larger than usual” number of changes in the latest version. The changes to GFS2 include the usual bug fixes, but also a cleaned up ACL implementation (including support for ACL caching), XFS-style quota support (with netlink quota notification over a modified dquot netlink interface), and much more.

Kconfig. Michal Marek posted a patch implementing support for custom one-line comments surrounding configuration entires in the .config file. This allows users to add comments such as “# some remark” around “CONFIG_FOO=y”. Separately, Nir Tzachar posted version 7 of his “nconfig” patches that implement an ncurses based replacement for menuconfig.

PATA. Bartlomiej Zolnierkiewicz posted an 86(!) part patch series implementing a large number of cleanups to the PATA drivers, intending to bring any last straglers onto par with their older IDE counterparts (the intention is to eventually kill off the legacy “IDE” layer).

Perf. Tom Zanussi posted version 2 of a 7 part patch series implmenting general purpose scripting support for perf trace. The patches create a generic interface through which one can expose the binary output ofr “perf trace” without having to parse the human readable ASCII version – i.e. a non-binary more easily parsed ASCII output alternative for perf. A perl example is included, for those who aren’t scared by the word “perl”.

The latest kernel release is 2.6.32-rc8.

Stephen Rothwell posted a linux-next tree for November 25th. Since Tuesday, there was a new tree (spi), the microblaze, ubifs, mfd, and alacrity trees lost their conflicts, while the ext3, rr, voltage, trivial, and percpu gained issues. The total sub-tree count increased to 154 with the new tree.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091124.mp3

For Tuesday, November 24th, 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: 1394, KSM, LIRC, and perf.

1394. Stefan Richter posted to let everyone know about forthcoming changes in the latest linux1394-2.6 git tree. These include support for the userspace FFADO drivers for firewire audio. He would like to see these fixes in 2.6.32, which seems unlikely at this stage in -rc.

KSM. Hugh Dickins posted a 9 part patch series intended to “at last mak[e] KSM’s shared pages swappable”. KSM is the Linux kernel technology that can scan through physical memory for identical pages, replacing duplicates with copy-on-write references to a single page. Previously, such reconciled pages couldn’t be swapped out directly without splitting them back out again.

LIRC. Somewhat unsurprisingly, someone objected to the idea of doing all IR decoding in the input layer. Maxim Levitsky pointed out that many people seem to believe that all infra-red devices send in some universal code that can easily be decoded in such a fashion. The reality (as anyone who has built an LIRC decoder can tell you) is that this is not the case and that the wide variety of possible IR devices suggests this is not easily mapped to input layer events for all possible kinds of IR device.

Perf. Various patches were posted for performance events (an area that is still undergoing a lot of active development). Amongst them were some patches from John Kacur to prevent bitrot in perf-annotate by having it share more code with perf-report and futex benchmarking in perf bench support from Mitoke Hitoshi.

Finally today, Pavel Machek asked everyone to please keep the LKML English only, in response to an email that had contained non-English text. Although the community is of course open to all, list traffic has traditionally been in English only. This may change one day, but for now it is the “rule”. Some had suggested that posting in English-only was overly limiting for those seeking responses from those who understand other languages.

The latest kernel release is 2.6.32-rc8.

Stephen Rothwell posted a linux-next tree for November 24th. Since Monday, there were two new trees added (samsung, and bjdooks-i2c), the microblaze, mips, ubifs, net, wireless, trivial, and alacrity trees gained issues, while the mfd tree lost its build failures but gained another (so a previous version was usued, as it was for other trees, such as alacrity). The total sub-tree count increases to 153 with the additional two trees.

Alan Cox noted that the linux-next tree no longer boots on KVM in 32-bit mode.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091123.mp3

For Monday, Novemember 23rd, 2009 , I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: Ftrace for Microblaze, kexec, kprobes, KVM, LIRC, and rwlocks.

Ftrace for Microblaze. Michal Simek posted a five part patch series implementing ftrace support for Xilinx Microblaze soft-core CPUs. On a tangent, your author is interested in reviving his ML403 board and running the latest Microblaze – anyone wanting to help out can drop me a line.

Kexec. Amerigo Wang reposted a patch series intended to provide support for reducing the reserved memory used for a crash kernel if it is already more than sufficient to store the crash kernel image. The patch implements a new sysfs interface through which one can reduce the reservation size.

Kprobes. Masami Hiramatsu posted version 5 of a 10 part patch series implementing optimized jumps (aka “Djprobe”) for probe points. Kprobes is the dynamic probe infrastructure intended to allow one to insert probe functions at certain points within the kernel. Jason Baron and others have also recently been working on jump optimization with labels, and so might benefit from working with Masami on a common approach to such patching.

KVM. Gleb Natapov posted the second version of a 12 part patch series implementing asynchronous page fault support for paravirtualized KVM guests. With appropriate support in the guest kernel, and the right communication between it and the host KVM hypervisor, the guest kernel can receive a hint that the page it needs is currently not available and will be faulted in, allowing it to switch to another task in the interim.

LIRC. Mauro Carvalho Chehab, and Jarod Wilson debated the future of the until now out-of-tree LIRC Infra-Red drivers vs. implementing such support through the standard kernel input layer. Doing IR through the standard input interface would mean that the kernel could handle a lot of the heavy lifting, but it would be based upon an assumption that the majority of IR users are really controlling input devices over that protocol, and not weird IR gadgets.

RW Locks. Nick Piggin raised the issue of fairness for rwlocks. The current implementation of rwlocks in the Linux kernel is (intentionally) unfair with respect to writers, which can be starved for some arbitrary period of time. As Nick points out, there are few users of rwlocks in the current kernel, but he would nonetheless like to have a “fair” alternative.

Finally today, Luis R. Rodriguez asked about bug reporting for open firmware (not the OpenFirmware system level firmware, but other firmware for which the source is open), now that there is at least one (ar9170) device firmware for which source is available. He wonders whether it makes sense to discuss on LKML or whether there is a more appropriate answer.

The latest kernel release is 2.6.32-rc8.

Stephen Rothwell posted a linux-next tree for November 23rd. Since Friday, the omap, infiniband, input, alacrity, and staging trees lost conflicts, while the microblaze, mips, net, mfd, and tip trees had issues. The total subtree count remains steady at 151 trees.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091122.mp3

For the weekend of November 22nd 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: kbuild, kFIFO, and LogFS.

kbuild. Sam Ravnborg announced that he is stepping down as the kbuild maintainer, citing a lack of time. As he notes, he has done this soley on a hobbyist basis (while managing a busy schedule of work and a family life), and the job was becoming “a duty and not that fun suddenly”. Rather than uproot himeself and work for a Linux company, he has decided to step down and persue other interests instead. He posted a patch to the MAINTAINERS file also. I am sure many will be sad to see Sam go, as his work has been excellent.

kFIFO. Stefani Seibold posted version 0.7 of a new kfifo implementation. This has been discussed before, and this 7th iteration includes various typo fixes, and other trivial cleanups, suggesting that it is nearing completion. As Stefani points out, the existing API had very few (17 files only) users, due to an overly simple API that could not handle various use cases, could not be used dynamically, and required a spinlock even in cases where it would otherwise be unncessary. The new patches implement a more robust API.

LogFS. Joern Engel surprised folks with an announcement that various changes to the format of LogFS had been made and that it was now ready for inclusion. Apparently, the last time it was discussed, Linus had said something along the lines of “go and don’t come back until all format changes are done”. Joern is seeking testing and comments ahead of a potential merge to mainline.

In today’s announcements: LTTng 0.179. Mathieu Desnoyers announced yet another release of LTTng.

util-linux-ng version 2.17-rc1. Karel Zak posted to let everyone know the the first rc of util-linux-ng 2.17 is now available. It includes a typically large number of changes, including changes to libblkid support for exposing the block layer topology, and various new commands such as “unshare” that allow a process to be created with some of its namespares not shared by its parent.

The latest kernel release is 2.6.32-rc8.

Rafael J Wysocki posted a list of regressions between 2.6.31 and 2.6.32-rc8-git1, following Linus’ previous 2.6.32-rc8 announcement.

Stephen Rothwell posted a linux-next tree for November 20th. Since Thursday, the microblaze, and infiniband trees had issues, with the net and edac-amd trees lost theirs. The total sub-tree count remains steady at 151 trees.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091119.mp3

For Thursday, November 19th, 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: 4096+ CPUs, KVM, licensing, performance, procfs symlinks, zombie processes, and a new kernel -rc release.

4096+ cpus. Rusty Russell posted a 6 part patch series intended to complete the conversion of cpumask support over to supporting dynamic allocation for greater than 4096 CPUs. Now, using a configuration option and accessor functions, one can choose to support CONFIG_CPUMASK_OFFSTACK or not. Legacy on-stack uses (CONFIG_CPUMASK_OFFSTACK=n users don’t notice much except that the mm_cpumask and tsk_cpumask accessors are now required).

KVM. Avi Kivity posted 35 patches comprising batch 2 of a 2 part batch of patches intended to hit the forthcoming 2.6.33 merge window. Highlights were covered previously and include improved kernel context switching speed, better interoperation with other users of virtualization extensions, improved IRQ scaling, nested SVM improvements and tracing, improved cpufrequ integration, and spin loop detection on newer hardware. Your author continues to be scared by the productivity of the KVM team.

Licensing. Rusty Russell brought up the GPLv3 in asking Tim Abbott (of ksplice) about the license on the binary search library within lib/bsearch.c. He wanted to use it for SAMBA, which is GPLv3, and noted that this is one reason why all of his kernel code is explicitly licensed as “v2 or later”. He requested a small license change on that code so that it can be used in Samba without being re-implemented.

Performance. Ajay Patel posted a series of lmbench runs showing a comparison between 2.6.18 and 2.6.27 in which he showed what appeared to be a significant performance degradation between the two kernels. The kernels were actually not really an Apples to Apples comparison though, because one was a modified “Enterprise” kernel and the other was a Fedora kernel, which has a very different set of compilation options. In both cases, those kernels are now very old and so the results are hard to work with.

Procfs symlinks. Jeff Layton posted a 4th attempt at a patch to /proc symlink support, this time causing such symlinks to behave like regular symlinks. The existing code treated such symlinks in a special fashion and resulted in a failure of the path walking code to revalidate such dentries (so if the state of the associated file permissions changed during the running task, it could result in certain contrived security issues), which lead to various issues. The patch results in one user-visible change in the form of unlinked files no longer appearing in procfs for opening – so we can say goodbye to admin hacks that restore deleted critical system libraries on production systems.

Zombies. Nick Piggin posted to let everyone know that he experienced a zombie process during various ptracing on a recent kernel. He also seemed to lose job control around the same time. The task was last in proc_clear_tty.

In today’s announcements: Linux 2.6.32-rc8. Linus Torvalds announced release 2.6.32-rc8 of the Linux kernel at 2:56pm Best Coast Time (PST). As he puts it, “the way things are going, this will likely be the last -rc”. He points to the usual need for eyeballs on regressions, and notes that he’s away for the entire US Thanksgiving week and won’t be working on a release then.

The latest kernel release is 2.6.32-rc8.

Stephen Rothwell posted a linux-next tree for November 19th. Since Wednesday, the new, sound, and cpufreq trees lost conflicts, while the input and edac-amd trees gained conflicts. The total subtree count remains stready at 151.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091118.mp3

For Wednesday, November 18th, 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: BKL, Jump Labels, and VMWare.

BKL. Jan Blunck posted version 3 of his latest round of BKL removal by pushing use down to the filesystems. His plan is to later remove the BKL from the VFS entirely, once it is entirely contained within individual filesystems. The latest patches push uses within do_new_mount down to the filesystems, removes the BKL entirely from EXt2 and doesn’t cause filesystems that weren’t using it to acquire a dependency in the cause of the cleanup. Separately, Alan Cox posted a first round of RFC patches doing similar in the tty layer.

Jump labels. Jason Baron posted the latest version of his jump labeling patches. These are intended to introduce a low-overhead conditionally enabled jump mechanism that can be used, for example, in implementing tracing of functions through a selectively enable tracepoint jump label. Matheiu Desnoyers noted that Jason could work around the requirement of disabling lockdep through the use of his “notifier atomic call chain notrace” patch.

VMWare. Dan Merillat started a thread entitled “Linux 2.6.31 – very swap-happy with plenty of free RAM”, in which he detailed how his 64-bit x86 system with 4GB of RAM would start swapping “like crazy” within a few seconds of being given access to a swapfile, despite having 1.7GB of disk cache and 750MB of free RAM. He was running several VMWare VMs, occupying a total of 1GB, which wasn’t a good sign – their kernel module could have been a culprit.

Finally today, is anyone at Coverity listening? Several kernel developers were trying to get in touch with Coverity via their info@, and linux@ addresses and were getting 550 errors, implying that those are no longer valid contacts. But David Maxwell followed up confirming that Coverity continue to perform scans of the kernel source against their source code checker (detecting common bugs from source level inspection), adding that some glitch had caused the most recent run to be on September 14th. He is short of time to keep the scan site updated, which it was noted by others had not been updated in “years”. David also provided some feedback concerning how the results of scans are shared with a select group of people, including various maintainers.

In today’s announcements: LTTng version 0.174. Mathieu Desnoyers posted yet another version of LTTng. The 0.714 release fixes a missing del_timer and buggy mips32 clock tracing support. This release was quickly followed by the 0.175 release. The second day in a row with multiple releases.

The latest kernel release was 2.6.32-rc7.

Stephen Rothwell posted a linux-next tree for November 18th. Since Tuesday, the net tree lost a conflict (but gained a build failure for which Stephen applied a patch), and the trivial, omap_dss2, and sysctl trees lost their conflicts. The total sub-tree count remained at 151 trees.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091117.mp3

For Tuesday, November 17th, 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: CFS hard limits, MadWiFi, and noatime in procfs.

CFS hard limits. Bharata B Rao posted version 4 of his patch series implementing “hard limits” for processor utilization by specific task groups in the CFS. This latest version adds cpu hotplug support for CFS runtime balancing in the form of a patch that “reclaim[s] runtimes lent to other cpus when a cpu goes offline”.

MadWiFi. Luis R. Rodriguez replied to a thread on the madwifi development list, copying the LKML, stating that madwifi is essentially dead and has been for some time. According to Luis, most of the devices previously covered by MadWiFi now have upstream support, with the exception of some ath5k users who might be holding out “until ath5k gets feature-parred”. Luis says those using MadWifi are mainly doing so out of nostalgia (due to “romantic experiences with it with extensive features and its history”), and that any real future for it other than going into maintainance mode is “just wasting time and electrons”. It’s nice to see Linux wireless has moved along so nicely.

Noatime. Bernd Petrovitsch posted a question concerning the opening of files contained with the /proc pseudofilesystem using the O_NOATIME flag. He wondered why this would not succeed for files such as /proc/uptime and /proc/cpuinfo. He also wondered why specifying “noatime” as a mount option for procfs made no difference either. Although it seems odd that someone would want to affect (or otherwise) the timestamp on dynamically generated files, perhaps he was concerned about existing software file open behavior.

In today’s announcements: GIT version 1.6.5.3. Junio C Hamano announced version 1.6.5.3 of the git SCM as used for Linux kernel development. The latest version contains a number of fixes, including a fix for “git blame” when the specified input file ends in an incomplete line, a fix for “git cvs import”, and a git for color use on “git diff”. Various git-email and gitweb fixes were also included, along with many other improvements.

LTTng version 0.171 and LTTV 0.12.21. Mathieu Desnoyers announced version 0.171 of LTTng and version 0.12.21 (trace format 2.4) of LTTV. The newer versions, as is implied, use a newer trace format that is not compatible with the older versions. The reason behind these changes stems from a desire for correct some alignment concerns with structures such that they always have 64-bit alignment, even on a 32-bit architecture. Later, Mathieu posted LTTng 0.173 and LTTV 0.12.22, which are “required to push the trace format forward to 2.5″.

The latest kernel release was 2.6.32-rc7.

Stephen Rothwell posted a linux-next tree for November 17th. Since Monday, there was a new edac-amd tree, the mips, kvm, and tip trees lost conflicts and a build error, while the net, trivial, and sysctl trees gained conflicts. The total subtree count increased to 151 trees in the latest compose.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091116.mp3

For Monday, November 16th 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: Console noise, KVM, and thread renaming.

Console noise. Mike Travis posted a 6 part patch series intended to reduce excessive console noise by removing “repetitious messages”. This was now in its third version.

KVM. Avi Kivity posted a 42 part patch series comprising part 1 of a 2 part batch of patches intended to hit the forthcoming 2.6.33 merge window. Avi mentions various “highlights” including improved kernel context switching speed, better interoperation with other users of virtualization extensions, improved IRQ scaling, nested SVM improvements and tracing, improved cpufreq integration, and spin loop detection on newer hardware. The latter is an implementation of the support for a modern CPU feature in which looping spinlocks can be automatically detected and the processor yielded.

Thread renaming. John Stulz, having not heard objections from his previous RFC postings, requested that Andrew Morton pull his patches enabling threads to rename their siblings through /proc/pid/tasks/tid/comm into -mm.

Finally today, Andreas Mohr got annoyed enough with the existing init code to add a documentation reference (and the documentation) for the case of “no init found” during boot.

In today’s announcements: LTTng version 0.170. Mathieu Desnoyers released version 0.170 of LTTng, which contains a fix for a concurrency issue between cpu hotplug, trace start/stop, and marker arm/disarm.

Userspace RCU 0.3.1. Mathieu Desnoyers announced version 0.3.1 of his userspace RCU library. This includes a fix to the build system to support cross-compilation.

The latest kernel release was 2.6.32-rc7.

Rafael J. Wysocki posted a list of regressions between 2.6.31 and 2.6.32-rc7-git1. Most were drivers, but there was also an oops starting udev on boot, an IDE issue, an mm/page_alloc warning, and an ext3/jbd oops amongst various other corruptions and serious enough sounding regressions. He followed up with individual messages, copying the original bug reporting CC list. There was also another list posted in which Rafael itemized regressions between 2.6.30 and 2.6.31 that have not yet got fixes in current mainline.

Stephen Rothwell posted a linux-next tree for November 16th. Since Friday, there were two new trees (omap_dss2, and workqueues), the sound tree lost its conflicts, the omap_dss2 tree gained a conflict against the omap tree, a warning introduced by the tip tree became an error in the sparc build, and the percpu tree lost a conflict. The total sub-tree count increased to 148.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.

Audio: http://media.libsyn.com/media/jcm/linux_kernel_podcast_20091115.mp3

For the weekend of November 15th, 2009, I’m Jon Masters with a summary of today’s LKML traffic.

In today’s issue: compat-ioctl, cpuinfo, ftrace, locking, perf, and s390x.

Compat-ioctl. Arnd Bergmann posted a series of RFC patches intended to put the compat-ioctl support on a resource diet. He implements this through inlining some functions, simplifying various structures, and passing pointers directly rather than through a layer of indirection.

Cpuinfo. Ingo Molnar had previously suggested clearing the “ht” flag within /proc/cpuinfo for a given CPU core if it only had one sibling since, as Ingo pointed out, “ht” is meaningless in such a context. Peter Anvin was inclined to agree, although was initially confused by another proposal to rename the “ht” flag entirely.

Ftrace. Zhangjin Wu posted a 16 part patch series implementing ftrace support for MIPS systems. This was version 8, and incorporated various feedback as well as removal of needless use of -mlong-calls GCC flags.

Locking. Frederic Weisbecker and Hitoshi Mitake had a dialogue about plans for a full “perf lock” tool that will be used with the locking tracepoints to capture runtime performance data for kernel locks. On a side note, the “lockdep” events were renamed to simply “lock” since they pertain solely to locking and unlocking (rather than dependencies).

Perf. Mitake Hitoshi posted a 4 part patch series implementing a new “mem” subsystem within the perf bench utility, intended to evaluate memory performance, as well as support for comparing various different memcpy algorithms on as applied to specific CPUs. This was timely, given a discussion of prefered memcpy algorithms for Intel Core2 vs. Intel Nehalem CPUs. Separately, Peter Zijlstra noted that performance counters are limited supply on physical hardware and are multiplexed in the case that more are asked for that are present, which results in a warning that results have been scaled. He also reminded everyone that the granularity of the round-robin performance counter switching is fixed at the timer tick, which can result in some zeroed results for some counters if they don’t have chance to get some CPU time.

s390x. Martin Schwidefsky sent a series of 52 patches intended for the next merge window (2.6.33). 27 of those patches rework the common-io-layer, 6 patches are for fault handler optimization, and 6 add support for cex3 crypto cards and “some other stuff”.

Finally today, Stephen Hemminger thought he had found the faulty commit that was causing synchronous ext2 mounted floppy disks to trigger an IO error on remove. The problem disappears when a seemingly unrelated tracing patch is reverted, suggesting that there may in fact be a timing bug. Andrew Morton felt that actually, the warning was fairly harmless noise from ext2 (which was unable to write back metadata changes). The discussion continued.

In today’s announcements: LTTng 0.168. Mathieu Desnoyers announced LTTng 0.168 had been released for kernel 2.6.31.6. It includes a bunch of optimizations and bug fixes, including a new periodic flush scheme.

The latest kernel release was 2.6.32-rc7.

Stephen Rothwell posted a linux-next tree for November 13th. Since Thursday, Linus’ tree lost its build failure, along with Rusty’s ‘rr’ tree, and the sysctl tree too. Meanwhile, the sound tree gained a conflict against the omap tree. Stephen remerged at the end to get Linus’ -rc7 tag. The total sub-tree count remained steady at 148 trees.

That’s a summary of today’s Linux Kernel Mailing List traffic, for further information visit www.kernel.org. I’m Jon Masters.