kernel26 out of date

xikyu46 · 2009-10-06 08:43:29

when i searching new kernel26 package, i found that in core repo only version 2.6.30.6-1, and it's release 2009-09-10. Doesn't this package too obsolete ? I know there are testing repo that release version 2.6.31.2, but does it too danger to use this repo ..?

davvil · 2009-10-06 09:47:02

I use the kernel 2.6.31 from testing without problems (I had issues with wireless with 2.6.30). If you want to install only the kernel you can update your system, activate the testing repository, install the kernel package and afterwards deactivate the testing repository again. There is no guarantee that this will work, I think, but I didn't have any problems.

Allan · 2009-10-06 09:51:36

2.6.31.2 will be making its way to [core] once it gets signed off by some devs.

.:B:. · 2009-10-06 10:07:59

I'm running testing, and still alive.

Of course I can't vouch for everyone .

xikyu46 · 2009-10-06 15:53:47

hm.. i think i will wait till release stable one. I work in Internet center and security is very important for me.. i hope this kernel version still stable again attack. thanks to all of you..

ugkbunb · 2009-10-06 16:01:56

I have been running [testing] since I initially installed Arch... and have yet have any *major* breakages... the few small hiccups were quickly fixed by a new package pushed to the repo or searching these forums/google for a fix.

dahankzter · 2009-10-06 17:37:59

I know its probably hard to get around to it what with the big blob that the kernel is but i kinda really need it asap.

You see I really enjoy making fun of the Ubuntu people at work for running that "old and busted" software!
Now its getting really close though...

Still if it was safer to use testing I would commit more testing time myself too.
As soon as the new kernel is out i think i will put my root partition on btrfs so i can use snapshots.
I guess trying out testing would be safe then.

ataraxia · 2009-10-06 19:21:49

2.6.31.1 caused some people on the mailing lists some trouble, and it took long enough to work it out that 2.6.31.2 came out first. As Allan says, I expect it will be in core very soon (within the next day or two).

brando56894 · 2009-10-06 22:27:32

If you guys really want 2.6.31 compile it yourself The easiest way to do it is to get kernel26parallel from the AUR and build it. It will download, patch and compile the kernel for you (custom configuration is optional, but recommended.) then install it along side the default kernel.

The most you have to do is configure the kernel options (unless you want to tweak the pkgbuild, Ive also found that xconfig wont let you integrate some features into the kernel for some reason so you'll have to manually specify 'y' in the .config file before you build the kernel) and it goes on automatically from there.

Heres a small tip I found that significantly decreases the build time needed for large thing such as the kernel, window managers, etc.. but it's only really useful if you have a lot of ram (2gb or more) and/or a fast processor (more than one core, or above 3ghz for a single core) as Ive found that each job (simultaneous compilation) takes up about 20mb of ram.
1. To do this on a per compilation basis add --jobs [any number other than one] to make if youre compiling manually or add it to the makeflags array in the makefile.
2. To set it so aurbuild or abs automatically uses this option add --jobs [#] to the makeflags array in /etc/makepkg.conf. for both of these options make sure not to add the [ ], the flag should look like -j64 or --jobs 64

I have mine set to 64 simultaneous jobs and it takes up a lot of ram (about 1gb or so in addition to what I was already using, i have a total of 2gb of ddr2 800) and really slows down everything else (unless I increase the priority of a certain process or decrease the priority of the build jobs)

Blind · 2009-10-06 22:36:58

Hmmm, I wouldn't think you would have too much of a gain when setting the number of jobs higher than the number of cores/cpus you have. Actually, unless you have 64 Cores, a setting like this is probably even counter productive, because the kernel scheduler will have to reschedule 64 tasks running at full speed...

Wintervenom · 2009-10-06 22:40:18

It is recommended to set the number of concurrent jobs to one more than the number of CPUs that are in your machines.

graysky · 2009-10-06 23:57:24

Back when I was running Debian, I did some experimenting on my quad core machine measuring kernel compiling time with different -jx values. For compiling the kernel, I found that I got the best times with the same number of cores (i.e. 4). You mileage may vary

http://forums.debian.net/viewtopic.php?t=33960

xikyu46 · 2009-10-07 03:54:23

brando56894 wrote:

If you guys really want 2.6.31 compile it yourself The easiest way to do it is to get kernel26parallel from the AUR and build it. It will download, patch and compile the kernel for you (custom configuration is optional, but recommended.) then install it along side the default kernel.
The most you have to do is configure the kernel options (unless you want to tweak the pkgbuild, Ive also found that xconfig wont let you integrate some features into the kernel for some reason so you'll have to manually specify 'y' in the .config file before you build the kernel) and it goes on automatically from there.
Heres a small tip I found that significantly decreases the build time needed for large thing such as the kernel, window managers, etc.. but it's only really useful if you have a lot of ram (2gb or more) and/or a fast processor (more than one core, or above 3ghz for a single core) as Ive found that each job (simultaneous compilation) takes up about 20mb of ram.
1. To do this on a per compilation basis add --jobs [any number other than one] to make if youre compiling manually or add it to the makeflags array in the makefile.
2. To set it so aurbuild or abs automatically uses this option add --jobs [#] to the makeflags array in /etc/makepkg.conf. for both of these options make sure not to add the [ ], the flag should look like -j64 or --jobs 64
I have mine set to 64 simultaneous jobs and it takes up a lot of ram (about 1gb or so in addition to what I was already using, i have a total of 2gb of ddr2 800) and really slows down everything else (unless I increase the priority of a certain process or decrease the priority of the build jobs)

ok... if i try compile kernel myself.. what the different with testing one..?

ngoonee · 2009-10-07 04:38:30

xikyu46 wrote:

brando56894 wrote:
If you guys really want 2.6.31 compile it yourself The easiest way to do it is to get kernel26parallel from the AUR and build it. It will download, patch and compile the kernel for you (custom configuration is optional, but recommended.) then install it along side the default kernel.
The most you have to do is configure the kernel options (unless you want to tweak the pkgbuild, Ive also found that xconfig wont let you integrate some features into the kernel for some reason so you'll have to manually specify 'y' in the .config file before you build the kernel) and it goes on automatically from there.
Heres a small tip I found that significantly decreases the build time needed for large thing such as the kernel, window managers, etc.. but it's only really useful if you have a lot of ram (2gb or more) and/or a fast processor (more than one core, or above 3ghz for a single core) as Ive found that each job (simultaneous compilation) takes up about 20mb of ram.
1. To do this on a per compilation basis add --jobs [any number other than one] to make if youre compiling manually or add it to the makeflags array in the makefile.
2. To set it so aurbuild or abs automatically uses this option add --jobs [#] to the makeflags array in /etc/makepkg.conf. for both of these options make sure not to add the [ ], the flag should look like -j64 or --jobs 64
I have mine set to 64 simultaneous jobs and it takes up a lot of ram (about 1gb or so in addition to what I was already using, i have a total of 2gb of ddr2 800) and really slows down everything else (unless I increase the priority of a certain process or decrease the priority of the build jobs)
ok... if i try compile kernel myself.. what the different with testing one..?

The difference is what you make it. What config you use.

I don't want to seem snarky here, but from your questions and responses I would recommend you do not try to compile the kernel yourself. If you really want to, then please read the wiki on it, and make sure never to overwrite the stock kernel so you have a fallback. Also make sure your menu.lst lists the stock kernel. Google is your friend, of course.

If you didn't understand much of that previous paragraph, then I would REALLY recommend not compiling the kernel yourself.

zenlord · 2009-10-07 06:39:31

ngoonee wrote:

If you didn't understand much of that previous paragraph, then I would REALLY recommend not compiling the kernel yourself.

Nah, IMHO you should never try to discourage someone into compiling there own kernel. You learn a lot by doing so, because you just know the first time around you'll screw up something and that you'll have to reinstall the whole system because you don't know yet how to chroot into your borked system and repair it.

I have tried 5 times IIRC and only the last attempt was succesful (and necessary on Debian stable). Everytime you fail, there's something else you learn to do better :-)

ngoonee · 2009-10-07 08:38:59

zenlord wrote:

ngoonee wrote:
If you didn't understand much of that previous paragraph, then I would REALLY recommend not compiling the kernel yourself.
Nah, IMHO you should never try to discourage someone into compiling there own kernel. You learn a lot by doing so, because you just know the first time around you'll screw up something and that you'll have to reinstall the whole system because you don't know yet how to chroot into your borked system and repair it.
I have tried 5 times IIRC and only the last attempt was succesful (and necessary on Debian stable). Everytime you fail, there's something else you learn to do better :-)

Some people don't want to learn, some think they do but don't realize how much work it is. Those who do tend to try first before asking

xikyu46 · 2009-10-08 05:31:48

ngoonee wrote:

Some people don't want to learn, some think they do but don't realize how much work it is. Those who do tend to try first before asking

ok thanks for your suggestion. i know a little bit about kernel. and i will try it

ngoonee · 2009-10-08 05:50:56

xikyu46 wrote:

ngoonee wrote:
Some people don't want to learn, some think they do but don't realize how much work it is. Those who do tend to try first before asking
ok thanks for your suggestion. i know a little bit about kernel. and i will try it

Have fun

tomk · 2009-10-08 06:17:26

xikyu46 - you mentioned your work, which implies that you use your Arch installation at work. You also have not detailed any problems with the 2.6.30.6 kernel, apart from your opinion that it may be "too obsolete" (it's not, btw). So by all means investigate building your own kernel, but make sure you do it in a way that keeps your current system usable for your job.

xikyu46 · 2009-10-08 08:41:03

tomk wrote:

xikyu46 - you mentioned your work, which implies that you use your Arch installation at work. You also have not detailed any problems with the 2.6.30.6 kernel, apart from your opinion that it may be "too obsolete" (it's not, btw). So by all means investigate building your own kernel, but make sure you do it in a way that keeps your current system usable for your job.

ok thanks tomk. i appreciate your oppinion. i will build my own kernel first to make it up to date with kernel.org.

about problem that i have with this kernel, yup, till now i don't have any problem with this kernel.. . so actually everything is fine. i just wondering why our (archlinux) kernel is a little bit obsolete

oli · 2009-10-08 08:47:38

Well, kernel 2.6.31 fixes a severe problem for _some_ users: "libata: implement and use HORKAGE_NOSETXFER, take#2". So I can use my dvd rom again. But I'm able to compile the kernel myself ;-)

xikyu46 · 2009-10-10 13:21:07

many thanks to Tobias Powalowski (tpowa) now our kernel has been up to date.. salut.. for arch developer

graysky · 2009-10-10 13:49:22

@xikyu46 - cool. you must have been hanging out over the mirrors to see that it just posted! I ran an update earlier this morning but missed it.

Arch Linux

#1 2009-10-06 08:43:29

kernel26 out of date

#2 2009-10-06 09:47:02

Re: kernel26 out of date

#3 2009-10-06 09:51:36

Re: kernel26 out of date

#4 2009-10-06 10:07:59

Re: kernel26 out of date

#5 2009-10-06 15:53:47

Re: kernel26 out of date

#6 2009-10-06 16:01:56

Re: kernel26 out of date

#7 2009-10-06 17:37:59

Re: kernel26 out of date

#8 2009-10-06 19:21:49

Re: kernel26 out of date

#9 2009-10-06 22:27:32

Re: kernel26 out of date

#10 2009-10-06 22:36:58

Re: kernel26 out of date

#11 2009-10-06 22:40:18

Re: kernel26 out of date

#12 2009-10-06 23:57:24

Re: kernel26 out of date

#13 2009-10-07 03:54:23

Re: kernel26 out of date

#14 2009-10-07 04:38:30

Re: kernel26 out of date

#15 2009-10-07 06:39:31

Re: kernel26 out of date

#16 2009-10-07 08:38:59

Re: kernel26 out of date

#17 2009-10-08 05:31:48

Re: kernel26 out of date

#18 2009-10-08 05:50:56

Re: kernel26 out of date

#19 2009-10-08 06:17:26

Re: kernel26 out of date

#20 2009-10-08 08:41:03

Re: kernel26 out of date

#21 2009-10-08 08:47:38

Re: kernel26 out of date

#22 2009-10-10 13:21:07

Re: kernel26 out of date

#23 2009-10-10 13:49:22

Re: kernel26 out of date

Board footer