Linux creator Linus Torvalds released the 2.6.0-test10 Linux kernel today, tentatively calling it the “stoned beaver” release. Linus plans to hand the kernel over to Andrew Morton in a few weeks, and then it will be up to Andrew to decide when we see the final 2.6.0 stable kernel. Download it from a mirror.
The 2.6.0-test10 Linux Kernel Released
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
…sure we will, but when??? Prolly good testing is important, but then again we are all wayting…
for this year
Not looking like it now. I would like to upgrade to 2.6.0, but I want a final stable release.
Maybe a Febuarary birthday present isn’t so bad.
For all intents and purposes, this is quite stable. I’ve been using it since -test3 and it’s been much more stable than 2.4.20. I would recommend switching as soon as possible, especially if you’re a desktop user.
How is it more stable? Or are you just making an uneducated opinion that since it seems faster for desktop usage, Greg?
Well I am using 2.4.22 on slackware right now, what would I gain from 2.6?
I have been hearing a lot of alsa problems with slack + 2.6.0 lately.
I wonder if that means that Linus will be taking a few months off from Kernel hacking and then start up on 2.7 by say march? if so, I think the Linux kernel development is going to be shifting into high gear!!!
or, it could mean that Linux is going to use the time he spent normally in the stable branch in areas such as cleaning up the dark recesses of the code and other fine tuning that you do not get a chance to do in regular development cycles, and while he does this, spend some time with coming up with a well thought out Driver interface and other things that user space would welcome.
???! What’s the reasoning behind complaining about the release date? “Good testing” is not just important, it’s the only thing right now. 2.6 is available for you to toy around with now if you want it. I haven’t had any problems with it besides the fact that some of my third party drivers weren’t compiled for it. If you really want something rock solid and compatible with everything, wait for 2.6.8 (or so) comes out.
in my second stanza I meant Linus, not Linux in the first Pseudo sentence.
that is what they said about 2.4 🙂 what was the patch level of 2.4 when it was finally all sorted out? 2.4.14 or something? ahh, those were the days, I remember the anticipation for 2.4 like it was yesterday.
Better late than half-baked.
Let’s not forget the mess that was 2.4.0. It was a maddening series of upgrades waiting for the thing to become what we had all hoped the 2.4.x series would be. It’s one of the things that pushed me to being a more or less full-time FreeBSD user. Linux 2.6 does look promising though.
ive heard alot of people saying that kernel 2.6 is more stable then 2.4 ever wsa or for them etc.. i would also like to know how they tested the “stability” factor ?
not trying to start a debate just curious when do u decide when something is stable ? you cant say when all bugs are taken out because then you will be waiting for ever (as its almost impossible to take out every bug)
but then releasing to early would make too many SERIOUS bug get through :S
anyway just to note i have tested kernel to 2.6 from test 7 on my dual amd mp system i compiled with the same config etc… it compiled fine all the time just when booting it wouldnt work, i tried fixing probs but then thought forget it. test 8 worked like a charm but then test 9 didnt ?? but now it does so i have no idea ??
but i would say it improves desktop performance
even on dual amd mp 1800+ 1gig ram u160 scsi raid 0 
but overall the system feels faster smoother just things you notice when you change between kernels
anyway just my 2p (uk 2p that is so about 3.5 cents :p)
worth
Snake
I wouldn’t recommend that. It’s not because it’s stable on your machine that it’s more stable than 2.4.20… It might be for you, but it might not be for him. I’ve encountered many problems with the early -test kernels. -test3 and -test4 weren’t really stable for me. They were even slower than 2.4.20! The 2.6 branch is useable on my computer only since -test7.
I wouldn’t recommend people to use the beta kernels unless they want to help developers or if they like to live on the bleeding edge. Some software don’t support it, anyway.
well, if you follow proper scientific based software engineering techniques, you would have almost no bugs and perhaps you might get ZERO bugs. why you ask then are there so many buggy software products from professional companies? because they do not follow the all the software engineering techniques because the cost would be to high and it would take a while to complete a project that the PHB says should take 6 months.
seriously, those techniques work very well. it is unfortunate that they are only followed in high stakes computing in equipment where life and death are at stake.
How would one install this on fedora
make sure you have all the QT libs installed for xconfig to work. download the sources and configure it. then install it.
I have had weird problems installing my own kernels though (due to the initial ram disk requirements and it for some reason not seeing my boot partition) so YMMV.
also, be prepare to par down the kernel since EVERYTHING is marked as yes or module….hey, they want to test all the parts 🙂
oh, the build process is also weird looking now…it is the same commands as before, but the python script being used prints out nice and orderly messages and not the big mess of stuff that the old process printed.
Preemtive isn’t stable yet. Without it, it is stable according to Linus.
The fact that there are bugfixes (and because i personally haven’t found it stable, _with_ preemtive strike enabled) it isn’t stable for everyone. Want stable? Use preemtive. I’m actually just copying what Linus said, and it is just logic imo. OTOH, if you’d like to see preemtive more stable or would like to use it, try it.
You can just check it out, add the packages and directories which are needed for 2.6, create a 2.6 entry in your bootloader, and move back to 2.4.x when you like to.
I’m gonna check out test10 with preemtive, see if it is stable, if not file a bug (before searching wether it is already reported) then try test10 without preemtive and see what happens then.
The most common reason for needing a bigger initial RAM disk is from having a debug build of modutils getting copied to it.
Want stable? Use preemtive. -> Don’t use preemtive then.
Btw the installer is fairly easy imo. The tree has been spitted to a better hierarchy. Just do a make help and check that out. You can use either QT or GTK for graphical frontend (seems TK got dropped). I prefer the ncurses one. Fedora (and other distro’s) will probly provide packages for test10 Soon. At least in unstable branches.
why not? was LL just to much or something?
LL made the system great for sound work, od did tehy just merge all the aspects of LL into the kernel code so no need to turn it off or on, it just is?
yep i know what your on about m8 i did functional programming along with logic programming at uni 2 years ago.
but with a project this scale and with so many different people around the world working on it wouldnt be easy (dont have to mention cost
)
Snake
I don’t know about ‘for you’ or ‘on your machine.’ I run it successfully on my desktop, as well as an old portable I was fixing up for someone else, and also my home router/firewall/proxy. The latter two run Debian and the former runs Gentoo. It definitely does seem faster for desktop usage. However, the router is working much better than with 2.4 kernels, and hasn’t shown any instability.
I wouldn’t recommend using it if you’re short on memory. My (admittedly rather monolithic) kernel is appx. 2M, which is insane. If you have alsa problems, compile your sound card mdule in–desktops are usually not too concerned about a few more K of kernel.
Again, draw your own conclusions. 2.6 is also far more user-friendly to configure and compile, but that’s ancillary.
I’m not sure how any one user could give a well-educated opinion about the overall stability of the kernel. You’d have to see it run on a rather extensive range of hardware. I’d guess, therefore, that when people say it seems stable to them, they probably mean that it is stable on their hardware.
It seems to be stable on my (fairly boring) hardware: a Dell Inspiron and a reasonably new, somewhat modified Compaq desktop with fairly standard components. (I’m not certain exactly what model number the Compaq is — it was “refurbished” (read:factory remainder/overstock) and doesn’t have too many identifying marks.) The most noticeable difference has been desktop responsiveness. I say ‘stable’ because it hasn’t exhibited any unstable behavior whatsoever, which is what I take to be the best measure of stability.
Due to box-death reasons I haven’t yet run it on a router, so I can’t comment on how well it does that.
I use test 9 and I’m impressed with it. Lets not make a big deal out of kernel releases. It’s just a kernel that is in the process of being perfected. And as open source proponents and enthusiasts we are informally obligated to test this them. 🙂
If the new version doesn’t work well for you, you can always switch to an older version. I have 5 kernels on my system that I can switch to at will. If one kernel starts acting up, I switch to another. The power of Linux. Tell me you can do that on Apple or Windows. 😛
Download the new kernel, compile it, install it and test it. It shouldn’t affect your old kernel or settings, except your distribution is truly pathetic, utterly screwed and just useless. If it works for you, use it. If you have problems, report them and switch back to the older more reliable kernel.
From my experience, 2.6.0 is a gazillion times more responsive and overall better than its predecessors. I’m downloading test 10 at the moment. I anxiously look forward to it.
..what a superb release name “stoned beaver” is.
Actually, the driver API was just overhauled in 2.6, so we’re not going to see a big change for awhile. Anyway, I’m running test9 (with the NVIDIA drivers no less). I’m on a laptop, so I reboot once every week or two, but it hasn’t crashed yet. Most likely, things will just not work rather than being unstable. This is unavoidable — the official 2.6 release tends to get driver maintainers in high gear to fix anything that might have broken. Is it more stable than 2.4? Really can’t say — the only time I’ve crashed 2.4 is when I fed my CD burner bad CDRs*. You might notice if you’re one of those people with 100 day uptimes
*> Yes, those CDs nuked XP too. In the course of saving a buck (literally, the CDRs were $1 cheaper than the ones next to them on the shelf) I managed to hose two different filesystems
The .config questions are plentiful and hard. Anyone know of a repository of some up-to-date standard or typical minimum .config templates that can be used as a starting point for make oldconfig and xconfig?
For test9 on SuSE9 I used as starting point the .config for test5 included on SuSE’s DVD. But there are many new parameters since then and I think it also has far more enabled than I need. For example, mine is a desktop so I don’t think I need anything for support of laptops.
Okay, I just installed test10. I think it is dam fast. Can anyone confirm this, or is it just me? I wasn’t expecting a huge improvement between test9 and test10.
I moved from 2.4.22 to 2.6.0-test9-bk3, and the mm subsystem is where I saw blazing speedup. It’s only a 1 GB RAM box, though.
Caveat: very, very, very subjective “measurement”:
2.4.22: 20 seconds to parse Rhythmbox 0.6.1’s music library
2.6.0-test9-bk3: 2 seconds to parse said library.
This is 954 songs, 5.1 days’ worth, 33.3 GB. The majority of the files are FLACs that reside on an ext3 partition. The rest are mixed FLACs, mp3s, vorbis on XFS. Both extremely subjective and ill-proofed times were obtained from cold boot directly into X Windows (XFree86 4.3.0.1) -> Enlightenment 0.16.6 -> Rhythmbox 0.6.1.
Allow me to say once again that is an extremely bad “measurement,” but it should give you an idea of just how good the new vm is in 2.6.0-test.
Hi All,
I have a small problem… If I turn on ACPI and APIC (In the kernel configuration) on my notebook (Sony Vaio FX505, Athlon4 1.2GHz, 256MB, ViaKT133?)and then want to switch to the external Display (<Fn>+<F7>) the kernel freezes! If I turn APIC off the system doesn’t freeze, but it doesn’t switch to the external display either! Does someone have the same problem, or know a patch?
Thanks
Golum at work dot ork
> I have been hearing a lot of alsa problems with slack + 2.6.0 lately.
My emu10k1 (SB Live! Platinum) works perfectly with 2.6 on Slack; YMMV.
I dont think that a new kernel can speedup parsing operation from rhythmbox.
The 2.6 kernel has better preemptive feature – so it looks like it’s faster, but i even don’t think it’s really faster . Threads implementation could bring some improvement.
It’s known that XFS drivers for 2.6.x is MUCH faster than for 2.4.x series. It could be the point why parsing where faster. And how it is ‘it looks like it’s faster, but it’s not?’
You feel it or not – that’s the case, I think. And mostly people feel it when testing 2.6.0-test series.
it stabel on my compaq presario 900 with ati igp320 grafics.
i would say that this machine was built with black magic. bu i got everything running with 2.6-test9mm sources and a patched version of xfree 4.3.99-r14. even accelerated 3d
and not a singel crasch for moths.
I have used 2.6-test* since test6, and I am really looking forward to the stable version. All the tests I’ve tried have been stable on _my_ system, but I am waiting for such nice patches as the graphical boot to come. 8)
CC fs/proc/array.o
fs/proc/array.c: In function `proc_pid_stat’:
fs/proc/array.c:398: Unrecognizable insn:
(insn/i 1335 1670 1664 (parallel[
(set (reg:SI 0 eax)
(asm_operands (“”) (“=a”) 0[
(reg:DI 1 edx)
]
[
(asm_input:DI (“A”))
] (“include/linux/times.h”) 37))
(set (reg:SI 1 edx)
(asm_operands (“”) (“=d”) 1[
(reg:DI 1 edx)
]
[
(asm_input:DI (“A”))
] (“include/linux/times.h”) 37))
(clobber (reg:QI 19 dirflag))
(clobber (reg:QI 18 fpsr))
(clobber (reg:QI 17 flags))
] ) -1 (insn_list 1329 (nil))
(nil))
fs/proc/array.c:398: confused by earlier errors, bailing out
make[2]: *** [fs/proc/array.o] Error 1
make[1]: *** [fs/proc] Error 2
make: *** [fs] Error 2