View previous topic :: View next topic |
Author |
Message |
Perfect Gentleman Veteran
Joined: 18 May 2014 Posts: 1252
|
Posted: Mon May 19, 2014 7:26 am Post subject: |
|
|
aCOSwt, honestly speaking, i've been using ck-kernel for a long time in Arch_Linux. So far my parents' desktop and microserver still run it. |
|
Back to top |
|
|
aCOSwt Bodhisattva
Joined: 19 Oct 2007 Posts: 2537 Location: Hilbert space
|
Posted: Mon May 19, 2014 7:48 pm Post subject: |
|
|
Perfect Gentleman wrote: | please, update to 3.14.4 |
Done.
If you can't wait for the commit, you can download the attached ebuild from the abovementioned link, or there : https://510800.bugs.gentoo.org/attachment.cgi?id=377258 then
repoman manifest and emerge. _________________
|
|
Back to top |
|
|
Perfect Gentleman Veteran
Joined: 18 May 2014 Posts: 1252
|
Posted: Mon Jun 02, 2014 11:46 am Post subject: |
|
|
why do gentoo-sources use old version of BFQ ? |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
|
Back to top |
|
|
thunderrd n00b
Joined: 20 Aug 2010 Posts: 59
|
Posted: Wed Jul 02, 2014 6:19 am Post subject: |
|
|
@aCOSwt, FYI, should you be interested -
This on IRC around 6AM GMT today:
Quote: | conman: hekel, patch seems stable
conman: will release tonight as is
hekel: yep, same here.. no issues on 3.15.2 and 3.15.3 |
Con has pretty much said that BFS is done in development, that he is completely satisfied with it [BFS/ck] at this point. |
|
Back to top |
|
|
_Nomad_ Guru
Joined: 11 Apr 2003 Posts: 571
|
Posted: Thu Sep 11, 2014 12:51 pm Post subject: |
|
|
Any plans on releasing an updated ebuild for 3.16 ? |
|
Back to top |
|
|
Perfect Gentleman Veteran
Joined: 18 May 2014 Posts: 1252
|
Posted: Thu Sep 11, 2014 3:00 pm Post subject: |
|
|
there are another sources with ck-patches like geek/pf/liquorix-sources |
|
Back to top |
|
|
Polyatomic n00b
Joined: 18 May 2014 Posts: 36
|
Posted: Mon Oct 06, 2014 11:07 am Post subject: |
|
|
_Nomad_ wrote: | Any plans on releasing an updated ebuild for 3.16 ? |
Patch up manually
Code: | System uname: Linux-3.16.2-ck2+-x86_64-AMD_FX-tm-9590_Eight-Core_Processor-with-gentoo-2.2
KiB Mem: 16329860 total, 15630228 free
KiB Swap: 0 total, 0 free
Timestamp of tree: Sun, 05 Oct 2014 10:15:01 +0000
|
Flys man ...
Last edited by Polyatomic on Tue Oct 07, 2014 5:27 am; edited 1 time in total |
|
Back to top |
|
|
thunderrd n00b
Joined: 20 Aug 2010 Posts: 59
|
Posted: Mon Oct 06, 2014 3:16 pm Post subject: |
|
|
@polyatomic
Which source package did you use? Vanilla? |
|
Back to top |
|
|
Perfect Gentleman Veteran
Joined: 18 May 2014 Posts: 1252
|
Posted: Mon Oct 06, 2014 3:18 pm Post subject: |
|
|
ck-sources = gentoo-sources + ck-patches |
|
Back to top |
|
|
Polyatomic n00b
Joined: 18 May 2014 Posts: 36
|
Posted: Mon Oct 06, 2014 11:16 pm Post subject: |
|
|
thunderrd wrote: | @polyatomic
Which source package did you use? Vanilla? |
I've cloned Linux-stable git and checked out the 3.16.y branch. |
|
Back to top |
|
|
Chiitoo Administrator
Joined: 28 Feb 2010 Posts: 2581 Location: Here and Away Again
|
Posted: Tue Oct 07, 2014 6:16 pm Post subject: ><)))°€ |
|
|
I've simply edited the version numbers in the latest in-tree ebuild, and it seems to work fine enough here, from what I can tell!
I do hope that wherever aCOSwt has been/gone\taken to, that they're OK. Nay... better than OK! That, and that they may return once more again, or twice, or however many times is required.
Edit: Forgot to add... there might be plans on including these with the experimental USE-flag on gentoo-sources.
(TomWij) from comment #1)
> Might need this fixed when I bring ck to the experimental tarball of
> genpatches. _________________ Kindest of regardses.
Last edited by Chiitoo on Sat Oct 18, 2014 8:50 pm; edited 1 time in total |
|
Back to top |
|
|
_Nomad_ Guru
Joined: 11 Apr 2003 Posts: 571
|
Posted: Sat Oct 18, 2014 4:39 pm Post subject: |
|
|
Polyatomic wrote: | _Nomad_ wrote: | Any plans on releasing an updated ebuild for 3.16 ? |
Patch up manually
|
Already am, just curious |
|
Back to top |
|
|
Cleus Apprentice
Joined: 21 Aug 2004 Posts: 160
|
Posted: Sun Nov 09, 2014 10:17 am Post subject: |
|
|
Very cheat, but it works
Copy ck-sources-3.14.4.ebuild to ck-sources-3.16.5.ebuild
Edit CK_VERSION="1" > CK_VERSION="2", BFS_VERSION="447" > BFS_VERSION="456"
ebuild ck-sources-3.16.5.ebuild digest
Manually download the patch files that are not downloaded from the previous command
emerge ck-sources |
|
Back to top |
|
|
kernelOfTruth Watchman
Joined: 20 Dec 2005 Posts: 6111 Location: Vienna, Austria; Germany; hello world :)
|
|
Back to top |
|
|
yngwin Retired Dev
Joined: 19 Dec 2002 Posts: 4572 Location: Suzhou, China
|
Posted: Sat May 09, 2015 3:18 am Post subject: |
|
|
I have taken on maintainership of ck-sources, and pushed version 3.19.7 and 4.0.2 updates.
Does anyone need updates to older versions, e.g. 3.14.x branch? _________________ "Those who deny freedom to others deserve it not for themselves." - Abraham Lincoln
Free Culture | Defective by Design | EFF |
|
Back to top |
|
|
Myu Apprentice
Joined: 22 Oct 2014 Posts: 164 Location: Belgium
|
Posted: Thu Nov 05, 2015 6:33 am Post subject: |
|
|
Hello everyone,
I'm currently experiencing mouse suttering and overall really poor responsiveness on gentoo-sources while my system is under some mid disk I/O (live imaging the drive with dattobd or copying the image to a NAS, iostat say 13% iowait)
Could ck-sources improve the responsiveness of my system ? I'm running Xfce + Compton, under normal load, everything is buttersmooth as far as I can tell.
I've read elsewhere that ck-sources do not play well with hibernate / suspend, is it still the case ? I kinda like hibernate
Also I'm running the 4.2 series and ck-sources is on 4.1, should I wait for ck 4.2 for an easy migration ?
Thanks in advance to those who answer.
Have a nice day ! |
|
Back to top |
|
|
spectromas n00b
Joined: 13 Jul 2014 Posts: 50
|
Posted: Thu Nov 05, 2015 1:32 pm Post subject: |
|
|
Myu wrote: |
I'm currently experiencing mouse suttering and overall really poor responsiveness on gentoo-sources while my system is under some mid disk I/O (live imaging the drive with dattobd or copying the image to a NAS, iostat say 13% iowait)
Could ck-sources improve the responsiveness of my system ? I'm running Xfce + Compton, under normal load, everything is buttersmooth as far as I can tell.
|
I get terrible performance issues while transferring large files over usb to NTFS and BFS or BFQ (or both together) don't make any difference for me.
Myu wrote: |
I've read elsewhere that ck-sources do not play well with hibernate / suspend, is it still the case ? I kinda like hibernate
|
Never had any problem with that, ck-sources is the same as gentoo-sources but with an extra patch or two. |
|
Back to top |
|
|
Myu Apprentice
Joined: 22 Oct 2014 Posts: 164 Location: Belgium
|
Posted: Thu Nov 05, 2015 2:00 pm Post subject: |
|
|
Hello spectromas, thanks for your reply.
These performance issues are bizarre indeed. What filesize do you transfer usually ? On my end, the file is 38 GB large, I'm sending it though NFS via (g)rsync.
Kinda looks like the OS is allocating or using way too much resources while transferring the data...
I tried to stress another Gentoo I have with stress(1) and a hell lot of I/O workers, it never sutters the way it did on the machine where I have the issue.
I'll try to gather some more metrics. I'll be trying Linux-ck anyway, it looks good |
|
Back to top |
|
|
Chiitoo Administrator
Joined: 28 Feb 2010 Posts: 2581 Location: Here and Away Again
|
Posted: Thu Nov 05, 2015 2:27 pm Post subject: |
|
|
As far as I understand (which isn't saying much), these patches (that implement the BFS) are more for the CPU-side of things, not so much for the I/O (for which you might experiment different IO Schedulers (I have Deadline, CFQ, and BFQ but I only really use BFQ so I can't really compare)).
With ck-sources and the BFS enabled, I can be compiling software with all my 6 CPU-cores in 100% use, and still play games and do whatever else I like while all that is going on. Granted, I haven't tried without the BFS for a while, so I'm not sure how big of a difference it makes right now.
That said, the only time I ever have had slowdowns like mentioned here, stuttering mouse-cursor movement; a very unresponsive system, was when I was running out of RAM. Now I have 16 GiBs of if, so I'll never fill all of it during normal use, even with some virtual machines running around. ^^
I would recommend you to definitely give it a try, as you seem to have already decided to do. You can keep your current/previous kernel(s) around just as well, like I did for a long time, though some time ago I cleaned them all up and stopped emerging gentoo-sources in addition to ck-sources.
Good luck, and welcome to ck! ^^ _________________ Kindest of regardses. |
|
Back to top |
|
|
Myu Apprentice
Joined: 22 Oct 2014 Posts: 164 Location: Belgium
|
Posted: Thu Nov 05, 2015 2:40 pm Post subject: |
|
|
Quote: | As far as I understand (which isn't saying much), these patches (that implement the BFS) are more for the CPU-side of things, not so much for the I/O (for which you might experiment different IO Schedulers (I have Deadline, CFQ, and BFQ but I only really use BFQ so I can't really compare)). |
Indeed, I planned to use both, BFQ for I/O as well as BFS for CPU
Quote: | That said, the only time I ever have had slowdowns like mentioned here, stuttering mouse-cursor movement; a very unresponsive system, was when I was running out of RAM. Now I have 16 GiBs of if, so I'll never fill all of it during normal use, even with some virtual machines running around. ^^ |
I think you have a point ! I'm using a system with only 4 GB of RAM so it might be that when transferring a 38 GB file, I'm filling up whatever buffer too fast and then it starts swapping !
I'll try when I get back to my machine to see if it's indeed the root cause.
Quote: | I would recommend you to definitely give it a try, as you seem to have already decided to do. You can keep your current/previous kernel(s) around just as well, like I did for a long time, though some time ago I cleaned them all up and stopped emerging gentoo-sources in addition to ck-sources. |
Yep, I'm gonna do just that !
Quote: | Good luck, and welcome to ck! ^^ |
Thanks a lot ! |
|
Back to top |
|
|
khayyam Watchman
Joined: 07 Jun 2012 Posts: 6227 Location: Room 101
|
Posted: Thu Nov 05, 2015 11:08 pm Post subject: |
|
|
Myu wrote: | Quote: | That said, the only time I ever have had slowdowns like mentioned here, stuttering mouse-cursor movement; a very unresponsive system, was when I was running out of RAM. Now I have 16 GiBs of if, so I'll never fill all of it during normal use, even with some virtual machines running around. |
I think you have a point ! I'm using a system with only 4 GB of RAM so it might be that when transferring a 38 GB file, I'm filling up whatever buffer too fast and then it starts swapping ! |
Myu ... it doesn't work like that, copying, or moving, a file will cause the file to be cached (see: page cache) but if the memory is needed subsequently that cache will be cleared ... you only access swap if there is no ram available. While additional ram will make the machine more responsive, the fact that its thrashing is mostly down to I/O ... and, perhaps, the filesystem used.
I suspect the reason behind your trashing is that you have an "Integrated Graphics Controller" sharing memory. I have a similar issue (Intel Corporation Mobile 945GM) and there is little that can be done about it other than via scheduling (BFS) and some trickery (ie, using ionice and schedtool).
Perhaps the following will offer a starter in that regard ...
/etc/portage/make.conf
Code: | MAKEOPTS="-j2" # you could also supply '-l 1.00' ... or what-have-you
PORTAGE_NICENESS="15"
PORTAGE_IONICE_COMMAND="ionice -c 3 schedtool -D \${PID}"
EMERGE_DEFAULT_OPTS="${EMERGE_DEFAULT_OPTS} -j1 --load-average=1.00" |
You can also adjust prority of command (ie, in the case of the file cp/mv above), eg:
Code: | % ionice -c3 nice -n 15 cp file /path/
% schedtool -D -e make -j2
% alias vnice='ionice -c3 nice -n 15'
% vnice mv file /path/
% ionice -c 2 nice -n -3 mpv file.mp4 |
Anyhow, I'd suggest enabling both BFS and BFQ and see how it goes ... you should also read the documentation for schedtool (sys-process/schedtool) and ionice (sys-apps/util-linux).
BTW, you can switch between IOSCHED via a kernel parameter ... eg: 'elevator=bfq' or 'elevator=cfq' ... if both are enabled in the kernel.
HTH & best ... khay |
|
Back to top |
|
|
spectromas n00b
Joined: 13 Jul 2014 Posts: 50
|
Posted: Fri Nov 06, 2015 3:10 am Post subject: |
|
|
Myu wrote: |
These performance issues are bizarre indeed. What filesize do you transfer usually ? On my end, the file is 38 GB large, I'm sending it though NFS via (g)rsync.
|
Yes, it's very strange and frustrating, I just can't get to the bottom of it. I tried some live usb environment (Kubuntu) and got the same thing happening but then I tried in Porteus live environment didn't get the problem so I'm very confused about it and almost given up trying to solve it. I usually only transfer file up to around 6-7gb so not enormous. It doesn't seem to happen when transferring to ext4 or xfs so I assumed it's NTFS/FAT issue. The stuttering always appears for me after about 1.5gb has been transferred, before that it's smooth.
Myu wrote: |
I tried to stress another Gentoo I have with stress(1) and a hell lot of I/O workers, it never sutters the way it did on the machine where I have the issue. |
I've not managed to find any other way to reproduce it either. I did find a mention of it on the liquorix kernel forum which mentioned memory compaction.
http://techpatterns.com/forums/about2443.html
I used their patches with those options disabled and things did improve a lot but the problem wasn't solved. So I am guessing it is either a kernel issue, something inherent in the way linux deals with particular filesystems, or some kind of hardware problem.
But the other people are right, this SHOULD be down to the I/O schedulers rather than CPU schedulers. Having said that, there was a very nasty bug around linux 3.18 time with BFS which causes kernel panic with heavy disk I/O so at least sometimes it can be related. I've been very impressed with how much CFS and CFQ have improved, even just over the last year or so. BFS does make things seem a bit snappier but I don't get a huge improvement in performance and BFS has had some serious issues for the last couple of versions so I've stayed away from it. It seems to be more stable since the last release but for me it's simply not worth it. |
|
Back to top |
|
|
mi_unixbird Tux's lil' helper
Joined: 24 Jul 2015 Posts: 118
|
Posted: Tue Nov 17, 2015 8:16 pm Post subject: |
|
|
So since Kolivas' patches for the 4.3.x line have been out for a short week now, any ETA on >ck-sources-4.1.6? _________________ execctl --path exec filectl --current-directory list |
|
Back to top |
|
|
khayyam Watchman
Joined: 07 Jun 2012 Posts: 6227 Location: Room 101
|
Posted: Tue Nov 17, 2015 9:29 pm Post subject: |
|
|
mi_unixbird wrote: | So since Kolivas' patches for the 4.3.x line have been out for a short week now, any ETA on >ck-sources-4.1.6? |
mi_unixbird ... you can bump the package in a local overlay ... it should be sufficent to edit K_GENPATCHES_VER to reflect genpatches from gentoo-source-4.3.0 (so ... ="1") and CK_VERSION to reflect whatever the ck patch version currently is (so ... ="1") ... build the manifest, and emerge.
ck-sources-4.3.0.ebuild: | K_GENPATCHES_VER="1"
[...]
CK_VERSION="1" |
If you're not familiar with local overlays then read the local overlay wiki page.
best ... khay |
|
Back to top |
|
|
|