Kernel Log: Coming in 3.1 - Part 1 - Networking
by Thorsten Leemhuis
Among the most important advancements of Linux 3.1 are improvements for all major Wi-Fi drivers and virtualisation optimisations.
One week after issuing Linux 3.1-rc1, Linus Torvalds presented the second release candidate of Linux 3.1 late on Sunday night. The new version has been taken as an opportunity to rename the kernel, which had been christened "Sneaky Weasel" at the end of May; its new name is "Wet Seal". Torvalds didn't provide any background reasons for his choice of name – it could be to do with the diving course he attended as a divemaster.
In his release email for the second release candidate, Torvalds was pleased to report that the merge window had been "pretty normal". From now until the final release of Linux 3.1, only corrections and minor improvements that don't present any risk – apart from rare exceptions – will now be accepted into the Linux main development branch.
The Kernel Log is, therefore, already in a position to present a comprehensive overview of the most important new features of the forthcoming Linux version, which is expected to become available in late September or early October. As usual, a number of articles in the "Coming in 3.1" series will gradually cover the kernel's various functional areas. The first part in this mini series is below and describes the most important changes to the network stack and the associated drivers for LAN and Wi-Fi hardware. In the coming weeks, further articles will discuss the kernel's storage support, file systems, architecture code, infrastructure and other hardware drivers.
Many minor, and various major, changes will improve the Wi-Fi support in Linux 3.1. For instance, the kernel developers have added the rtl8192de driver (for example, 1) that supports Realtek's RTL8192DE and RTL8188DE PCIe Wi-Fi chips; Linux 3.0 and two of its predecessors released in the months before had also each included a new Wi-Fi driver co-developed by Realtek. These and other Wi-Fi drivers have since been improved by the developers – for example, the rtl8192cu driver will support additional USB Wi-Fi modules from Linux 3.1.
Ralink's Rt2800 driver now supports the "RT3572/RT3592/RT3592+Bluetooth combo card". The developers have also extended the support for Ralink's series Rt35xx chipsets, and the associated code is no longer considered to be "non-functional and only intended for testers and developers".
From Linux 3.1, the iwlagn driver that covers the Wi-Fi chipsets Intel currently includes in many Core notebooks will support the WoWLAN (Wake on Wireless LAN) standard. The driver now accepts two further module parameters: power_save allows the power-saving features that are disabled by default to be enabled or disabled, and the power_level parameter allows the power-saving level to be set to a value between 1 and 5.
The ath9k Atheros driver now supports the AR9330 component (1, 2). The improved ath6kl driver that is to succeed the identically named staging driver has missed the merge window and will follow in Linux 3.2.
The b43 driver for Wi-Fi chips by Broadcom now supports various 802.11n-compliant Wi-Fi cards that use Broadcom's Advanced Microcontroller Bus Architecture (AMBA) – such as the BCM43224 (PCI-ID 14e4:4353) and BCM43225 (14e4:4357) chipsets. The required improvements were mainly developed by Rafał Miłecki, whose efforts in this area have been sponsored by Poznan University of Technology (1, 2).
Together with the bcma driver this allows the b43 driver, which has been part of the kernel for a long time, to support some of the Broadcom chips that the brcmsmac driver has covered for a few months. The latter is the driver that was originally called brcm80211 and was released as open source by Broadcom in September 2010; however, the driver is kept in the staging area because it does not fulfil the kernel hackers' quality requirements – as a result, it is not included in a few Linux distributions.
Despite the improvements to b43, Broadcom is continuing to work on brcmsmac so the driver can eventually leave the staging area. However, the kernel developers tend to avoid having two drivers in the kernel that support the same hardware; Miłecki has already pointed this out to the Broadcom developers. It is still unclear which of the drivers will make it in the long run, as the Broadcom developers don't appear to have fully understood the notice. Since the b43 driver also supports many other Broadcom chips and is already part of the Wi-Fi subsystem it stands a slightly better chance of success, despite the fact that it currently supports some of the covered range of Wi-Fi chipsets' features less comprehensively than brcmsmac does.