Stable kernel 2.6.20.2 [LWN.net]

Cannot stay at 2.6.16.x because...

Posted Mar 12, 2007 19:32 UTC (Mon) by pr1268 (subscriber, #24648) [Link]

I can't stay at 2.6.16.x with my laptop - Broadcom bcm43xx needed for my wifi card. If I remember correctly, 2.6.17 was the first to include in-kernel driver support for this chipset.

Which brings me to this question: Why is 2.6.16.x getting a bunch of bugfix updates (they're up to 2.6.16.43 now!)? Any reason 2.6.17 didn't get these bugfixes?

bcm43xx

Posted Mar 12, 2007 19:43 UTC (Mon) by ncm (subscriber, #165) [Link]

bcm43xx support is *almost there*. I still have to load bcm43xx_d80211 and unload it, before loading bcm43xx, in order to get any contact with my ddWRT hub. Looking at the patches, it's amazing how much chip-version fiddling the code has to do to make these things work at all. I'd have thought that with Broadcom actively cooperating there would be less churn.

bcm43xx

Posted Mar 13, 2007 16:30 UTC (Tue) by proski (subscriber, #104) [Link]

It would be great if you post the details to the bcm43xx mailing list (bcm43xx-dev at lists.berlios.de).

Cannot stay at 2.6.16.x because...

Posted Mar 13, 2007 5:45 UTC (Tue) by dlang (✭ supporter ✭, #313) [Link]

a maintainer stepped up and decided to commit to patching 2.6.16 long-term.

nobody has done this for later versions

Cannot stay at 2.6.16.x because...

Posted Mar 13, 2007 9:00 UTC (Tue) by tomas2 (guest, #37038) [Link]

"a maintainer stepped up and decided to commit to patching 2.6.16 long-term."

AFAIK it's Adrian Bunk.
http://lkml.org/lkml/2005/12/3/55
http://lwn.net/Articles/194555/
http://kerneltrap.org/node/6386

I thought back then when I first read about this that Debian would go with this kernel for Etch, but they didn't. Anyone knows why?

Tomas

Stable kernel 2.6.20.2

Posted Mar 10, 2007 12:39 UTC (Sat) by Duncan (guest, #6647) [Link]

Not here! While I had an issue with the rcs (the md/RAID-6 unaligned I/O
log spamming due to an update in that system not updating the logging as
well), I filed a bug on it, and it was fixed. Also, partly due to the
consequences of failure and because it hadn't been working several kernels
back when I last tried it, I hadn't tried hibernate for quite awhile, but
with .20 it was working great, even from with KDE/Xorg. That's no small
milestone when I'm on dual CPU and multiple RAID and LVM, plus the usual
video complications, among other things.

As a result, I stayed on .20 for rather longer than normal, only switching
to .21-rc3 instead of trying -rc1 or 2 as usual. And that was only
because twice now, I've found and filed bugs on rcs, with the result being
that they have been fixed and I've had a working release that otherwise
likely would have been bugged for me, so I /do/ try to test the rcs before
they get /too/ far along, to ensure releases continue to work and be
relatively bug free on my hardware. Otherwise, I think I'd be sticking
with 2.6.20 for some time, as it really was /very/ stable for me.

(FWIW, I've had no problems with 2.6.21-rc3 so far, either, but I didn't
expect to as there wasn't anything in the changelog that looked to be a
major update for my hardware. Now .22 there likely WILL be such major
changes, and with them potential bugs, as at least one major change
confined to x86 this round is planned to expand to amd64/x86_64 and other
archs for .22.)

Duncan

Stable kernel 2.6.20.2

Posted Mar 13, 2007 8:46 UTC (Tue) by malor (subscriber, #2973) [Link]

You know, I remember when stable Linux kernels were actually stable.

Stable kernel 2.6.20.2

Posted Mar 13, 2007 15:20 UTC (Tue) by nix (subscriber, #2304) [Link]

They still are. Serious regression fixes or trivially obvious fixes only, is the rule.

The churn is many orders of magnitude less than on the trunk.