Linux Kernel 2.6.7 Released

conrausch writes "German Heise News reports among others that the new Linux Kernel 2.6.7 was just released, and that it fixes the previously mentioned bug in the floating point exception handling. Whether or not you offer shell access to other people, get it now from kernel.org or one of the mirrors."

Comments

Got it

[pcmanjon]

I just compiled and installed it. It's not that bad.. or good... orr... how the hell should I know?

System doesn't seem to run much different, I haven't read the changelog

but for those of you who want to read the changelog it can be found HERE:

http://www.kernel.org/pub/linux/kernel/v2.6/Chan ge Log-2.6.7

English: Linux Today has human redable changelog

[VC]

Linux 2.6.7

Re:what about 2.4?

[egreB]

Out of curiosity, what would prevent someone from being able to switch to kernal 2.6?

The driver architecture in Linux kernel 2.6 changed somewhat from 2.4. Drivers will have to be patched or rewritten to work with 2.6. This is being worked on, but lots of unofficial patches to the kernel haven't caught up yet. My laptop, for instance, was unable to get X up at adequate resolutions with 2.6 (albeit this was around christmas - I might give it another shot with this release).

Then there's low-level userspace programs (stuff not running as a part of the kernel itself) that needs some change. Examples are the PCMCIA-suite.

Re:Problems with JFS?

[flex941]

This e-mail might help you out.

Re:Is it just me,

[Sunspire]

The 2.6.xx revisions have no bearing at all on when the 3.0.0 or 2.7.0 trees will get created. The quick turn around times are due to many factors; the new versioning and source control procedures put in place for 2.6 naturally encourage a more rapid pace while elimating the "did my patch make it into Linus's tree?" problems of yesteryear, which in turn has people submitting more, perhaps smaller, patches in a very rapid fashion. The 2.6 kernel is also right now being developed by more developers than ever, until the 2.7 branch gets spun all the efforts are basically focused on this single tree, timely releases keep code divergence down and hopefully prevents 20kloc ALSA merges from happening.

What, are you afraid they're suddenly going to run out of numbers for the 2.6.xx branch? ;) Hint: after 2.6.99 comes 2.6.100. With vendor kernels you can't say where in the 2.6 branch you are anyway, when you're running 2.6.6-1.423 it's can be anywhere between 2.6.6 and 2.6.10 feature and security wise.

Re:Just curious

[Shaman]

I have approximately 20 machines using 2.6 since last fall and with the exception of one (an AMD-64 box), they have all been exemplary. That machine became stable with 2.6.6 though its BIOS seems flakey (hardware problems.. ugh)

In particular, my HT machines seem to perform very well with 2.6.3 and up.

Re:Woohoo, another kernel compile.

[Taurim]

Kernel 2.6.7 is the solution for this situation. It integrates the very useful following patch from Con Kolivas :

[PATCH] sched: SMT niceness handling

From: Con Kolivas

This patch provides full per-package priority support for SMT processors (aka pentium4 hyperthreading) when combined with CONFIG_SCHED_SMT.

It maintains cpu percentage distribution within each physical cpu package by limiting the time a lower priority task can run on a sibling cpu concurrently with a higher priority task.

It introduces a new flag into the scheduler domain
unsigned int per_cpu_gain; /* CPU % gained by adding domain cpus */

This is empirically set to 15% for pentium4 at the moment and can be modified to support different values dynamically as newer processors come out with improved SMT performance. It should not matter how many siblings there are.

How it works is it compares tasks running on sibling cpus and when a lower static priority task is running it will delay it till high_priority_timeslice * (100 - per_cpu_gain) / 100
eg. a nice 19 task (Note : Seti@Home for example :-) ) timeslice is 10ms and nice 0 timeslice is 102ms On vanilla the nice 0 task runs on one logical cpu while the nice 19 task runs unabated on the other logical cpu. With smtnice the nice 0 runs on one logical cpu for 102ms and the nice 19 sleeps till the nice 0 task has 12ms remaining and then will schedule.

Supermount

[Bralkein]

I just thought I would post a brief message about supermount. If anyone wants to upgrade to 2.6.7 and still use supermount, I don't think vanilla kernels have it in there (yet, I'm sure it'll get in there sooner or later). I'm pretty sure the Mandrake and Gentoo kernels have support for it (gentoo-dev-sources do, anyway), but I just looked at gentoo-dev-sources and it is at version 2.6.5, dunno about Mandrake, but I'm sure it will take a few days for all the distros to catch up.

If you want to upgrade for security reasons, but you also want supermount in your kernel (as I do), this guy seems to have a patch for 2.6.7, which might come in handy if you don't want to wait for your distro to catch up. I am going to use this patch myself, but I cannot guarantee that it won't bone your system so to speak. The patch is not just supermount, it looks like it has some other stuff in it too, so decide for yourself!

Seeing as how I'm posting this, I may as well give a little background for those not "in the know". Supermount is a sort of filesystem, you mount your CD-ROM and floppy drives (or even USB sticks) with it, and it will automatically mount and unmount the media when you insert or remove it, kind of like on Windows. Personally, I think it is great, and it is hard to live without it now I have it.

You can learn more about it at the project website. Jeez, if it turns out the vanilla kernel does have supermount after all, I am going to look a right idiot... *presses Submit*