« previous 1 2 3 4 5 6 next »

Virtualisation and Memory management

KVM now supports the "unrestricted guest" mode of Intel's next generation of desktop and notebook processors (Westmere), which is scheduled for release under product names such as Core i3 or Core i5 in early 2010. The main Git-Pull request by KVM developer Avi Kivity discusses further improvements to KVM – including improved tracing facilities and the eventfd-based irqfd and ioeventfd mechanisms for integrating user software and kernel software with guest systems.

Another new addition is KSM, which originates from the KVM developer circle. The acronym is short for "Kernel Shared Memory" or "Kernel SamePage Merging" and describes a framework which scans the memory of multiple userland processes for identical areas; if it finds matching areas, it combines them and reduces the memory load by deallocating the redundant copies. This is, for instance, interesting for KVM virtualisations where multiple similar guest operating systems that share the same software libraries and programs run on a computer, causing large areas of the data stored in the guests' memory to be identical. An article about the Linux Symposiums 2009 in The H Open and the text version of an OLS 2009 presentation it links to, describe the technology in detail.

The new HWPOISON, which was introduced by Intel developer Andi Kleen, adds some techniques for handling and avoiding memory errors. These techniques are planned to be included in Intel's Nehalem EX series of server processors, which is expected in early 2010.

X86 systems, Tracing

A new addition to the kernel is the support of Intel's Trusted Execution Technology (TXT), formerly known as LaGrande Technology. Together with the components of the Trusted Boot (tboot) project, TXT systems can ensure that a kernel hasn't been attacked and compromised before executing it.

The Linux kernel can now be optimised for Atom CPUs during compilation. The kernel hackers have also integrated code to support Intel's Simple Firmware Interface (SFI) – an alternative to ACPI developed by Intel and intended for use with the Moorestown platform Intel plans to introduce next year. Moorestown is heavily geared towards Linux use and is intended for smartphones, Mobile Internet Devices (MIDs) and embedded environments.

The emerging Performance Counters have been renamed Performance Events, because this apparently better describes this technology, which has seen major advancements in the past few months. Also new is the "perf sched" sub-program, which adds convenience to the task of analysing the process scheduler. Another new addition is the Timechart tool mainly developed by Arjan van de Ven, which allows traces recorded via "perf record" to be visualised as SVGs for easy analysis – in his blog, van de Ven explains the whole procedure in detail and gives various examples of use.

Power management and Scheduling

Various patches add support for ACPI 4.0 to the Linux kernel. Several major changes to the power management code create the basis for making better use of the runtime power saving features on modern I/O devices – details can be found in this LWN.net article. An enhancement to the Cpuidle framework, also explained on LWN.net, should increase I/O throughput on larger servers in particular – in some cases significantly.

Changes to the process scheduler eliminate two weaknesses found in speed comparisons made using Con Kolivas' recently published and independently developed BFS ("Brain Fuck Scheduler"). One of these problems resulted in x264 encoding being significantly slowed on multi-core systems. One of the x264 developers explains the background to this problem on a blog and also takes the opportunity to make some recommendations on improving collaboration between developers.

Next: Configuration, Storage, Audio and Video

« previous 1 2 3 4 5 6 next »