[LWN Logo]
[LWN.net]

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

See also: last week's Kernel page.

Kernel development


The current development kernel release is 2.3.11, which came out just before LWN went to "press." There is no announcement for this release yet. It contains a number of architecture-specific changes, the new resource manager, a bunch of USB tweaks, and lots of little fixes.

The current stable kernel is still 2.2.10. There are tremendous numbers of patches stacked up for the stable series at this point; see the 2.2.10ac12 announcement for the rather impressive list. A subset of those have been packaged together into 2.2.11-pre2 - there was simply too much to try to put into one release and hope that it would all work well. Linus is said to have already agreed to the 2.2.11 prepatch, so, barring major problems, this one may go out sometime soon.

The new resource management code generated some discussion this week. This code handles allocation of hardware resources (such as I/O ports); it was completely rewritten by Linus last week. There is general satisfaction with the new scheme, which cleans up handling of resources considerably. Not everybody is entirely happy, however.

David Hinds, creator of the PCMCIA subsystem, has been working on better support for dynamic hardware - such as PCMCIA cards. This support is intended to extend to other, newer "hot-plug" architectures, such as the upcoming "hot plug PCI" and USB. David's work had involved making changes to the older resource manager to facilitate the tracking of hardware separately from the resources used by that hardware. Things were beginning to come together on that front.

Linus's new resource manager then appeared out of the blue, without any (public, at least) advance discussion. Changes from Linus tend to show up in this sort of fait accompli mode. David's changes are gone, and his hardware tracking no longer works.

David has asked for some small changes so that he can make things work again; Linus, so far, has been reluctant to go along. There appears to be a fairly strong disagreement on how things should be done, and it could well end up retarding the development of some needed capabilities. One can only hope that some sort of solution gets worked out shortly.

The Linux Storage Management Workshop will be held in Darmstadt, Germany, on September 6 and 7, 1999. It thus adjoins the sixth Linux-Kongress, being held in Augsburg on the 8th through the 10th. The workshop will look at the current status of Linux development in journaled file systems, logical volume managers, RAID, backup and restore, etc. Their hope is to attract many of the hackers working in that area, along with companies which are interested in storage management. It looks most interesting. See the announcement for more information.

Various patches and updates released this week:

  • The latest RAID patches for both 2.2 and 2.0 kernels have been announced. These should probably be applied by anybody wanting to do serious Linux RAID work. With luck, they will make it into the 2.2.12 kernel.

  • A new version of the IEEE 194 (FireWire) patch has been announced by Emanuel Pirker. At this point the two competing FireWire initiatives seems to have been thoroughly merged, and development is proceeding at a higher pace. Some ground has yet to be covered, however, before it is ready for real use.

  • As usual, H.J. Lu has a knfsd patch out. These patches are still required for serious NFS servers running on the 2.2 kernel - especially in heterogeneous environments. They are not going to make it into 2.2.11 (too much other stuff in an already huge patch), but will likely be given serious attention for 2.2.12.

  • It wouldn't be a week of kernel development without a devfs patch from Richard Gooch.

  • Logical Volume Manager 0.7 has been released by Heinz Mauelshagen.

  • A TCP Vegas patch for both 2.2.10 and 2.3.10 has been announced by Neal Cardwell. Vegas is a congestion avoidance algorithm which is meant to make TCP play better (and perform better) in situations with a lot of traffic - i.e. on the Internet.

Section Editor: Jon Corbet


July 22, 1999

For other kernel news, see:

 

Next: Distributions

 
Eklektix, Inc. Linux powered! Copyright © 1999 Eklektix, Inc., all rights reserved
Linux ® is a registered trademark of Linus Torvalds