See also: last week's Kernel page.

Kernel development

One of the more interesting things that did happen was the release of an alpha NFSv3 server for the 2.2 kernel series. This release fills in a longstanding gap in Linux's NFS capabilities, and will be highly welcome to a lot of users. Now all that remains is to get people testing the patch and get the inevitable problems ironed out. Even if it does not become an official part of the 2.2 series, there will eventually be a full NFSv3 implementation that can be used with the 2.2 kernel. Congratulations are due to G. Allen Morris and all the others who have worked on this code. See the announcement for a pointer to the patch.

Sound drivers were a subject of conversation this week, spurred by Creative Laboratories' release of a beta driver for their SB Live! card. The presence of a driver is a nice thing, but it is currently a binary-only release, with all of the problems that can bring. There is currently no statement in place as to whether source will eventually be available. The driver can be found on the Creative Linux page.

In response to requests for suggestions on a good Linux sound card, a few people suggested the use of Trident 4DWave-based cards. The word is that these are economical, capable cards with full support via the ALSA sound system.

And, speaking of ALSA (the Advanced Linux Sound Architecture), the project has a new domain. ALSA's web pages (and other resources) are now available at www.alsa-project.org.

There are now two separate sites dedicated to Linux performance tuning information. One has its own domain at tunelinux.com, the other is hosted out of nl.linux.org. Neither is all that long on information at this point. It seems like these folks should get together and make a single, coordinated site.

Universal Serial Bus (USB) support. Last week we mentioned the Linux USB web page as the source for information on USB development for Linux. Two things ganged up on us with this one...

• This project shut down at the end of last week. In the announcement of the closure, Iñaky Pérez González (the project's founder) cited being tired of USB, wanting to play with other parts of the kernel, and also...

• Linus slipped a separate USB implementation into the 2.2.7 release. He acknowledges Iñaky's work in the credits in the source, but this is a separate effort. Given its source, this appears to be the "real" USB implementation that will become part of the kernel.

There is a new site devoted to Linux support for token ring networking at www.linuxtr.net. They have started things off with a new alpha driver for IBM PCI token ring cards.

Various patches that came out this week:

• Vojtech Pavlik announced a first testing release of the "Linux generic input device layer." It is supposed to handle input devices (keyboards, mice, joysticks, ...) in more general way than is done now, and includes some support for USB devices.

• Pavel Machek released an alpha capabilities patch. For those who have been following the capabilities discussion here, this patch supports the "capabilities in the ELF header" model. There has been no word from Linus as to which approach to capabilities he supports, still.

• Andre Hedrick released a massive patch to the IDE subsystem which tries to unify IDE access across the various architectures supported by Linux. This is a far-reaching patch, and Andre would like a lot of people to test it.

• Version 98 of the devfs patch was announced by Richard Gooch. He also updated the Model-Specific Registers (MSR) patch.

A couple of critical articles in the press stirred up discussion this week. For those who have not seen them:

• Linux and the enterprise in Windows NT Magazine levies some specific attacks on the design of the Linux scheduler and I/O mechanisms. The kernel developers tend not to agree...

• Ken Thompson, co-inventor of Unix and all-around highly-respected guy, came out with a strange attack in this interview in IEEE Computer. "My experience and some of my friends' experience is that Linux is quite unreliable. Microsoft is really unreliable but Linux is worse." It is not clear just where all this was coming from.

Section Editor: Jon Corbet

For other kernel news, see: LinuxHQ Kernel traffic

Next: Distributions