See also: last week's Kernel page.

Kernel development

For stable kernel folks, 2.0.36 prepatch 13 is out. Yes, pre12 was supposed to be the release version, but a few things came up. This patch fixes those glitches, and updates a couple of drivers. As it turns out, there will be a pre14 as well, once the final Adaptec driver is available. That should be the end of the 2.0.36pre series.

As of last week's issue, the kernel developers had reached a point of serious anger and recrimination, and work had stopped. This week, the kernel folks are back at work and the furor seems to have mostly died down. It has faded, however, without any sort of real resolution to the underlying problems; at the moment, it seems that it could all happen again.

The basic problem has to do with Linus getting overloaded and dropping patches, and some developers getting frustrated with that. Linus then got grumpy about the form some of the incoming patches took and ignored them, and a somewhat broken kernel release resulted. The basic problem - "Linus does not scale" - remains.

Some have suggested that kernel development should be reorganized around a "core team" which has the authority to apply patches. Ignoring, for the moment, that Linus is unlikely to go for such a scheme, there are other problems. The "core team" approach is what the BSD variants have used all along, and it seems to have a lot to do with why there have been so many BSD splits in the first place. Core teams, successful examples (i.e. Apache) notwithstanding, seem to be susceptible to nasty politics and fragmentation. The drawbacks of this approach were well expressed by Ted Ts'o.

For now, things look set to move forward with people trying harder not to overwhelm Linus (he posted some guidelines for sending patches). Hopefully the Jitterbug patch queue will get going again at some point; it greatly helps communications between patch submitters and Linus. Also simply getting 2.2 out the door - still not as close as one might like - should help a lot.

Version 0.90 of the RAID subsystem has been released, see the announcement for more. This is evidently "yet another complete rewrite" of the RAID system. Included new features include, happily, support for hot adds and removes. It also detects when devices have been renamed and adjusts - something which should be handled by devfs, rather than hacked into a separate subsystem. There are lots of other changes as well, the announcement lists them all (except for a couple more that mingo forgot that were noted later.

A new version of the "gdbstub" patches have been posted. Gdbstub allows the use of GDB to debug a running kernel over a serial line. Linus does not like this approach, so these patches are unlikely to ever appear in an official kernel, but they can be useful for some developers. See the announcement for info on how to get it.

Since we're a weekly publication, chances are we'll be behind a rev or two on the kernel release by the time you read this page. Up-to-the-second information can always be found at LinuxHQ.

Next: Distributions