[LWN Logo]
[LWN.net]

Sections:
 Main page
 Security
 Kernel
 Distributions
 Development
 Commerce
 Linux in the news
 Announcements
 Back page
All in one big page

See also: last week's Kernel page.

Kernel development


The current development kernel release is still 2.4.0-test8. The 2.4.0-test9 prepatch series is up to 2.4.0-test9-pre7. This patch is said to fix the sound problems that have afflicted recent -test9 prepatches, but some complaints on the subject are still being posted.

The current stable kernel release is 2.2.17. The 2.2.18 prepatch series is currently at 2.2.18pre11. A great many fixes and updates are still going into this series; recently these include a bunch of USB updates and a native Yamaha YMF7xx sound driver ("a result of high-speed collision between ymfpci.c of ALSA and cs46xx.c of Linux.").

So where do the prepatches live, anyway? Every now and then somebody drops us a note asking that question. It makes sense - the announcements for prepatches only rarely say where you actually have to go to pick them up. The answer is:

  • For stable kernel prepatches, and related releases by Alan Cox, go to the directory pub/linux/kernel/people/alan on your favorite kernel.org mirror. The 2.2.18 prepatches live in 2.2.18pre within that directory.

  • For development kernel prepatches produced by Linus, look in the directory pub/linux/kernel/testing.

Your closest kernel.org mirror, of course, will be found at ftp.*.kernel.org, where the * is replaced with your country code. So folks in the U.S. should go to ftp.us.kernel.org, for example.

Bear in mind, of course, that prepatches are exactly that. They are out there for a first round of testing, and have the potential to crash your system, corrupt your files, fry your monitor, drink all your beer, or make you believe that LinuxOne is for real this time. Use with caution.

As an example of what can happen with prepatches, consider the case of the new memory management code. Rik van Riel has been working for some time on an improved memory management scheme; Linus integrated the results of his work in 2.4.0-test9-pre2. It seems that there were still a few glitches still needing to be worked out; quite a few people started reporting system deadlocks.

Some of these problems have proved hard to fix. Low-level deadlocks can be one of the hardest sorts of problems to track down. It also didn't help that Rik van Riel headed off to attend the Linux Kongress in Germany. Things got to the point where Linus threatened to back out the VM patches if the problems didn't get fixed soon.

Progress is being made on that front, and others are counseling patience. The memory update really does seem to make things work much better for a number of people. Removal of these changes would create some discontent.

Meanwhile, Andrea Arcangeli has stated his intent to revive his "classzone" patch as the Real Solution to the VM situation. Classzone got put on hold after the Ottawa Linux Symposium as it was decided to concentrate on Rik's approach; evidently Andrea has changed his mind. It will be interesting to see what sort of results Andrea gets, but the chances of including a completely new, more complex VM solution at this point in the 2.4 process seem pretty small.

Linux on the AlphaServer GS320. Here's the fun boot log of the week. Some folks at Compaq got Linux to boot and run on a 31-processor, 256GB AlphaServer GS320 system. It reports a total of 46,170 BogoMIPS. As they say, "things like kernel builds run really fast."

One more look at kernel patch management. Last week's kernel page asked "why not BitKeeper?" in response to the proposed new patch management system. A few tidbits that have wandered in since that article went up:

  • It turns out that the proposed patch management system is only intended (by Linus) to be used until 2.4.0 comes out. Since 2.4.0 is, hopefully, not too far away, this system will not be around for long.

  • BitKeeper perhaps isn't as hard to understand as some people make it out to be. See, for example, this testimonial from Cort Dougan, who has been using it to maintain the PPC and RTLinux trees for some time.

  • Linus still does, evidently, intend to try BitKeeper. No word as to just when, however.

Those interested in how BitKeeper can be used to track changes in the kernel may want to have a look at the kernel repository browser on the BitMover site.

Other patches and updates released this week include:

  • Jamal Hadi Salim posted a proposed interface change for network drivers that would facilitate the development of a blazingly fast forwarding system for packet routing. He's hoping to see the change go into 2.4.0.

  • LIDS 1.0.1 (the Linux Intrusion Detection System) was released by Xie Huagang. This is the first version of the LIDS patch for 2.4-test kernels.

  • Eric Raymond has released CML2 0.8.0, the latest version of his new kernel configuration and building system.

  • The linux-crypto mailing list was announced by Marc Mutz. This list is intended for discussions of cryptographic code at the kernel level, and will be the forum for developers of the international kernel patch.

  • Keith Owens released kdb v1.5beta1. This is a major rework of the kdb debugger, and isn't recommended for people who prefer not to have to debug the debugger itself.

  • Keith also released modutils 2.3.17. Remember that you need a recent version of modutils to use modules with the 2.4.0-test kernel.

  • A document on debugging early kernel hangs was posted by, well, Keith Owens.

Section Editor: Jonathan Corbet


September 28, 2000

For other kernel news, see:

Other resources:

 

Next: Distributions

 
Eklektix, Inc. Linux powered! Copyright © 2000 Eklektix, Inc., all rights reserved
Linux ® is a registered trademark of Linus Torvalds