Stable kernel update 2.6.29.1 [LWN.net]

Stable kernel update 2.6.29.1

Posted Apr 3, 2009 15:51 UTC (Fri) by nix (subscriber, #2304) [Link]

Give that networking basically doesn't work on older cards without it
(well, it works for a few minutes and then disappears), yes, I'd say it
matters.

(Only anyone who can be bitten by this bug has already been bitten and has
patched it or downgraded again.)

Stable kernel update 2.6.29.1

Posted Apr 4, 2009 2:36 UTC (Sat) by jengelh (subscriber, #33263) [Link]

I observed no networking problem on 2.6.29-rc6 with sis900.ko, which I run all day. Do you perhaps have a SHA for the first-bad-commit?

Stable kernel update 2.6.29.1

Posted Apr 4, 2009 7:35 UTC (Sat) by Duncan (guest, #6647) [Link]

I happened across the LKML 2.6.29 announcement thread about a day and a
half after release, and this bug formed one of two subthreads below the
announcement. The commit was hours before the release and people were
reporting problems hours (perhaps minutes) after the release. No -rcX
version had the commit. I don't believe even the last -gitX daily
snapshot had it tho I'm not sure on that. The commit was a fix for other
broken NICs but the fix isn't quite as easy as a simple revert as there
are other issues involved. Linus' intuition definitely failed him this
time, as he decided it was OK to release without another rc.

That said, I don't know the actual commit number, but as I said, it's one
of the major followup subthreads to the LKML 2.6.29 announcement, so it
should be easy enough to look up if you want more than I mention above.

I know various folks on the gentoo/amd64 list which I subscribe to had the
bug and I was able to tell them about it, recommending they revert to
2.6.28.x or late .29 rc or snapshot versions until 2.6.29.1. I didn't
have the problem myself altho I run git kernels directly, as I hadn't git
remote updated since a couple days after -rc8 (uname -r gives me
2.6.29-rc8-241-g65c2449), and the offending commit was days later. What I
had was stable and I don't upgrade to GIT_HEAD between release and rc1 as
there's just too much chance of stuff going wrong to be worth trying it,
so the release announcement effectively signaled me to hold off. Then I
came across that subthread and discovered I did NOT want to upgrade to
even the tagged 2.6.29 release. I may try 2.6.29.x, now that I'm sure
that's fixed, or wait until after 2.6.30-rc1 and git remote update and git
pull to mainline HEAD once again. If I do the latter and find a
regression, I'd file a bug of course, and switch branches to the 2.6.29.x
stable series while I wait for 2.6.30-rc1+ to get usable again.

... OK, I decided to go drag it up again after all. Here's the first post
of that subthread, courtesy GMANE.

http://article.gmane.org/gmane.linux.kernel/811417

<rant>

(The other subthread was "That Massive Filesystem Thread", as covered by
the recent subscriber-only article I can't read yet since I refuse on
principle to pay for a subscription to a site about open source that can't
practice what it preaches, along with other kernel folks, demanding OTHERS
open their code, while years later, LWN itself still says "soon", the
same "soon" it said years ago. I subscribed for awhile and hope I can
again someday, but my conscience was mocking me for my own hypocrisy so I
could not and did not renew, instead emailing my reasons. Yes, there are
excuses, but come on! It's been years! ATI and and SUN with Java had
excuses too. They came around, yet LWN hasn't. Even that letter was
years ago now, /still/ no change, LWN still claims it'll release the
code "soon". Would that it were so and I could actually subscribe in good
conscience again, and read and perhaps comment on stories such as that
before they became old news!)

</rant>

Code

Posted Apr 4, 2009 14:48 UTC (Sat) by corbet (editor, #1) [Link]

I refuse on principle to pay for a subscription to a site about open source that can't practice what it preaches, along with other kernel folks, demanding OTHERS open their code, while years later, LWN itself still says "soon", the same "soon" it said years ago.

Please, Duncan, point out where we "demand" that others open their code. I believe you've misread our message. Choosing to, say, avoid hardware which lacks free software support is very different from a placing a demand on other peoples' code.

I do want to post our code; there is no value to us at all in keeping it internal. But it will take work to do that. When I am no longer concerned with the basic problem of keeping LWN alive and keeping my family going on an income that is, shall we say, rather smaller than I could be making elsewhere, I'll do my best to find a week to put the source in order and get it out there.

In the mean time, I honestly do not feel the need to justify my contributions to the free software community to anybody.

(FWIW, we tried to get a summer of code student to do the release work last year, but the GSoC Powers That Be decided that the project was unworthy. Oh well.)

Code

Posted Apr 4, 2009 20:34 UTC (Sat) by nlucas (subscriber, #33793) [Link]

Please ignore him. This is the only online subscription I ever had and feel any penny is worth it. The quality of the articles here surpass any idea I could have of reading anything else.

Code

Posted Apr 6, 2009 11:30 UTC (Mon) by aleXXX (subscriber, #2742) [Link]

Same here.

Alex

Stable kernel update 2.6.29.1

Posted Apr 5, 2009 7:30 UTC (Sun) by man_ls (subscriber, #15091) [Link]

I subscribed for awhile and hope I can again someday, but my conscience was mocking me for my own hypocrisy so I could not and did not renew, instead emailing my reasons.

Having source code for LWN would be nice, but it would not change fundamentally the uses we make of the site. That is probably why even Stallman does not make too much noise about server code (unlike with drivers, userspace programs and firmware).

I for one prefer to have the weekly edition on time than having our editors spend one week cleaning up the code. I suspect we outnumber people of your opinion.

Stable kernel update 2.6.29.1

Posted Apr 8, 2009 0:06 UTC (Wed) by clump (subscriber, #27801) [Link]

The subscription options are very reasonable. I've been under the most expensive subscription option for years and still find the amount I pay quite fair. Perhaps you could pay the least now, and pay more when source is available to show your satisfaction. You've mentioned that you do want to read subscriber content...

Not having source is a "known issue", if you will. I might suggest if you feel strongly enough that you engage in a conversation with LWN, and offer suggestions for opening the code. Maybe you could contribute, get corporate subscriptions, etc.

In short, get involved.

Stable kernel update 2.6.29.1

Posted Apr 4, 2009 11:08 UTC (Sat) by nix (subscriber, #2304) [Link]

The bug was introduced by 303c6a0251852ecbdc5c15e466dcaff5971f7517,
committed on Mar 17.

Fixed by reversion: 8f1ead2d1a626ed0c85b3d2c2046a49081d5933f upstream,
ed421a64825501e0bdfe848c9decf05d270a9adb in -stable.

Stable kernel update 2.6.29.1

Posted Apr 6, 2009 13:38 UTC (Mon) by Trou.fr (subscriber, #26289) [Link]

Even if it seems there are some security bugs, as usual ...
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2....

Stable kernel update 2.6.29.1

Posted Apr 14, 2009 4:23 UTC (Tue) by eteo (subscriber, #36711) [Link]

I wrote about this on full-disclosure. Check it out:
http://marc.info/?l=full-disclosure&m=123951062415557...