• Immutable Page
  • Info
  • Attachments

KernelPodcast20090526

Coming to you from Cambridge, Massachusetts, for Tuesday, May 26th 2009, I'm Jon Masters with a summary of today's LKML traffic.

In today's issue: Tracepoints, modules, Machine Check Exceptions, and IO scheduling.

Tracepoints. A number of patches were posted implementing new tracepoints. These included block IO tracepoints (converting block trace points over to use TRACE_EVENT()) from Li Zefan, timer and tasklet tracepoints from Anton Blanchard. It would seem that a lot of people are interested in tracepoints at the moment, if recent traffic is seen as such an indication.

Modules. Amerigo Wang posted a patch series that merges module_32.c and module_64.c into a unified x86 module.c. Michal Marek posted a patch that generates a modules.builtin file containing a list of modules that were built into the kernel. As Michal notes, this will make it easier for tools like mkinitrd to detect whether a needed module is missing or whether it is compiled into the kernel. It also will enable tools like modprobe to do similar, and provides a patch both I and Kay were pondering recently.

MCE merge. Also on the subject of 32/64-bit x86 merging, Andi Kleen posted a patch series intended to merge the 32/64-bit implementations of Machine Check Exception support for x86 CPUs. In his opinion, this allows one to use the "much superior" 64-bit MCE code on 32-bit systems too. It also contains the basic MCE "injector" test harness because it was used in testing. While he was at it, Andi also posted version 2 of an MCE improvements patch series that adds support for MCE recovery and some new status bits.

IO scheduler based IO controller. Vivek Goyal posted version 3 of the IO Controller patches, rebased to 2.6.30-rc7. The latest revisions includes per-device per-cgroup rules, more disk time and sector statistics, and additional debug parameters.

In today's announcements: Eric Paris announced 'sandbox'. This is a small program that exploits SELinux capabilities in order to attempt to confine untrusted binaries. The idea is to make it easier for administrators to lock down tightly untrusted applications in a sandbox where they can not use the network and open/create any file this is not handed to the sandbox process.

The latest kernel release remains 2.6.30-rc7, which was released by Linus on Saturday evening. Although that tree contains very little churn by comparison with precious RCs, it is highly likely that an rc8 will follow this week.

Stephen Rothwell posted a linux-next tree for May 26th. Since Monday, the x86 tree gained a conflict against Linus' tree (for which an email was sent), and the net tree lost 2 conflicts. The tree still fails to build for powerpc allyesconfig, and still pulls from a total of 140 trees.

Are you interested in getting more involved in educating the next generation of kernel hacker? We're looking for an experienced kernel developer to help us with a book project. Drop me a line if you're interested at jcm@jonmasters.org.

That's a summary of today's LKML traffic. For further information visit kernel.org. I'm Jon Masters.

Tell others about this page:

last edited 2009-05-27 14:07:14 by JonMasters