Manjaro LTS kernel with ck1(BFS)+BFQ patchsets (64bit/32bit/32bit-PAE)

[Rob]

You need to have virtualbox-host-modules to run virtualbox.
As I see there is not a package for it currently, you can download the PKGBUILD from here-
http://git.manjaro.org/package-sources/extra/blob/master/linux312-extramodules/virtualbox-modules/PKGBUILD
You would need to modify it according to Korrode's kernel, then compile, install, reboot.
(need the kernel headers to build the extra-modules)
http://www.paradoxcomputers.com.au/manjaro/repo/linux-ck1bfq/

You don't need to manually make a package, just make sure the -headers package for your ck1bfq kernel is installed and then install:

Code: [Select]

pacman -S virtualbox-host-dkmsand then reboot and it will build and load the module for you automatically.

[aaditya]

You don't need to manually make a package, just make sure the -headers package for your ck1bfq kernel is installed and then install:

Code: [Select]

pacman -S virtualbox-host-dkmsand then reboot and it will build and load the module for you automatically.

Ah, I didnt know that, thx

[mips]

You don't need to manually make a package, just make sure the -headers package for your ck1bfq kernel is installed and then install:

Code: [Select]

pacman -S virtualbox-host-dkmsand then reboot and it will build and load the module for you automatically.

Thanks you kindly sir! I will reboot shortly.

[kremator]

can it be installed somehow using mhwd-kernel? why not?

[Rob]

can it be installed somehow using mhwd-kernel? why not?

It's not up to me to adjust mhwd to account for available custom kernels.

I would think anyone enthused enough to be trying custom kernels could handle installing some packages with pacman/pamac/etc.

[kremator]

All works OK so far. Philm if You by any chance read this thread: Please be so kind and make a patch to mhwd so we can use "mhwd-kernel -i linux312-ck1bfq" . After all linux is all about choice, right?

[Rob]

A new version of BFS has been released that may fix suspend/resume issues for users that experience them (although I personally never have).
http://lkml.indiana.edu/hypermail/linux/kernel/1312.0/01538.html
That update will be reflected in these kernels soon.

-------------

Sadly, since the last update i did i've noticed VLC playback jolting when there is other heavy disk activity EDIT: actually it's only particular circumstances, not any old activity, jolting that doesn't even happen on the normal kernel. So, (i presume) somewhere in recent upstream updates that were applied across 3.10, 3.11 and 3.12 series kernels, something was changed that's negatively impacting BFQ. If you experience this my suggestion is to just not use these kernels for now. I'm hoping I can find time in the next few days to provide upstream with proper test results and debug data, but I am very busy with work at the minute.

EDIT:
On the other hand, BFQ is still performing as expected on my usual 'Play video in VLC off same hard drive that stress --hdd 1 is hammering and watch it jolt on CFQ, but not on BFQ' test. So i dunno.

[kremator]

severe audio player glitches when compiling from source (both cpus 100%). same when mp3gaining files. setting higher priority won't help. seem to be disk usage related somehow even if listening to online radio (no disk cache). stock kernel all OK

[Rob]

severe audio player glitches when compiling from source (both cpus 100%). same when mp3gaining files. setting higher priority won't help. seem to be disk usage related somehow even if listening to online radio (no disk cache). stock kernel all OK

As I said in the last post, I can't deny i've seen some odd behaviour since the last update (although my usual test still yielded the result i expect for BFQ vs CFQ) ...although not what you describe.

It would be good if you can try using the stress command and then doing other tasks to see if the problem is replicated.
For example, for me on my quad-core i run stress --cpu 4 --hdd 1 which starts 4 processes hogging CPU cores and one smashing the current drive's I/O. This is usually a way to show how much better BFS+BFQ is. Even now my system does still open apps, play media and respond to user input like it's idle, where as with that stress command running on the normal kernel my system becomes sluggish... however, as I said, i do have some oddities since the last update, like the first time pacman is run after a reboot (the initial database load) will cause playing media to stutter on BFS+BFQ but not on normal.

*shrugs*
All I can say is don't run this for now if it's causing you problems, wait for these things to be fixed upstream.
Despite some currently oddities, It still does provide massive benefits for me, on my main system, with the type of workloads i have.

[kremator]

I've run "stress --cpu 2 --io 1 --hdd 2" and this kernel really works better than stock, BUT that's just a benchmark: in real life, in chrome dev (in full gpu mode) mouse cursor hoovers from link to link in very jumpy/jerky way unlike stock kernel.

[philm]

Both kernel-series seems to be outdated. linux310 will be at 3.10.25 this Saturday. linux311 is at version 3.11.10.1

[Rob]

Both kernel-series seems to be outdated. linux310 will be at 3.10.25 this Saturday. linux311 is at version 3.11.10.1

Yep, I don't update it regularly enough, I can't be bothered maintaining 9 kernels.
To be honest, people may as well just use 'repo-ck'. Many Manjaro users already do without issue, and the nvidia module versions in that repo VS Manjaro don't seem to have that much time where they're different.

I'll start another repo soon with only linux310-pae, linuxNNN-netbook and the BFS+BFQ+AMD-K8 optimized 64bit kernel I run myself on my main workstation.

Could someone please un-sticky this thread.

[pavbaranov]

To be honest, people may as well just use 'repo-ck'. Many Manjaro users already do without issue, and the nvidia module versions in that repo VS Manjaro don't seem to have that much time where they're different.

Unfortunetly, after last update of glibc (v.2.8-11), grub (v.1:2.00.1282.g5ae5c54-2) and linux-api-headers (v.3.12.4-1) /all of them are in our testing and unstable repository/, linux-ck kernels are no longer useable. Attempt to boot creates kernel panic. The same applies to other non-manjaro kernels like pfkernel or almost any builds from AUR.
Only yours kernels are usable with testing and unstable branch.

I'll start another repo soon with only linux310-pae, linuxNNN-netbook and the BFS+BFQ+AMD-K8 optimized 64bit kernel I run myself on my main workstation.

I'll be waiting...

[mips]

I'll start another repo soon with only linux310-pae, linuxNNN-netbook and the BFS+BFQ+AMD-K8 optimized 64bit kernel I run myself on my main workstation.

Please include a non-pae 32-bit one as well, it's what makes my old laptop tick

[hoffe]

Unfortunetly, after last update of glibc (v.2.8-11), grub (v.1:2.00.1282.g5ae5c54-2) and linux-api-headers (v.3.12.4-1) /all of them are in our testing and unstable repository/, linux-ck kernels are no longer useable. Attempt to boot creates kernel panic.

Yep, I can confirm that. However, this obviously only has to do with grub, which meanwhile is back to version 2.00.5086-1 in the repos. Just downgrade grub and the ck-kernel will boot again!