Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
MuQSS - The Multiple Queue Skiplist Scheduler
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3  Next  
Reply to topic    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
Chiitoo
Administrator
Administrator


Joined: 28 Feb 2010
Posts: 1740
Location: Here and Away Again

PostPosted: Sat Oct 22, 2016 9:21 am    Post subject: ><)))°€ Reply with quote

Staying with BFS for a while more, but will need to give MuQSS a go soon.

Hmmm... “mux”, eh? All I can think of is “mucus”, heh.

Ant P. wrote:
Wanted to get some MuQSS numbers now that zen has it, but the machine I got those BFS ones on just gets stuck in a no-output reboot loop with it. I really don't have the patience to go debugging stuff like this right now so looks like I'm going to be stuck running 4.8.1 for a few weeks...

Might be worth a shot again with this? :]

ck wrote:
linux-4.8-ck3, MuQSS version 0.115

This is mainly a bugfix release for those who had boot failures, TOI patched
failures, and warnings. Otherwise it only has minor changes.

As a sidey-note...

hnaparst wrote:
I was able to contact github by using OpenDNS.

A friend was having issues with several sites, too, using Google DNS. I've been experimenting with OpenNIC, which did allow me to load affected sites without any issues at the time.
_________________
Kind Regards,
~ The Noob Unlimited ~

Sore wa sore, kore wa kore.
Back to top
View user's profile Send private message
Juippisi
Guru
Guru


Joined: 30 Sep 2005
Posts: 339
Location: /home

PostPosted: Sat Oct 22, 2016 10:05 am    Post subject: Reply with quote

i4dnf wrote:
x11-drivers/nvidia-drivers-370.28 works just fine with 4.8.x kernel (no extra patching needed)


Apparently it did! Thanks for the tip.

Code:

[    3.294361] BFQ I/O-scheduler: v8r4 (with cgroups support)
[    3.453974] MuQSS CPU scheduler v0.114 by Con Kolivas.


And thanks hnaparst for the ebuild!
Back to top
View user's profile Send private message
Perfect Gentleman
l33t
l33t


Joined: 18 May 2014
Posts: 781

PostPosted: Sat Oct 22, 2016 11:20 am    Post subject: Reply with quote

Code:
$ dmesg | grep -i sched
[    0.550413] io scheduler noop registered
[    0.550433] io scheduler bfq registered (default)
[    0.550435] BFQ I/O-scheduler: v8r4 (with cgroups support)
[    1.454175] MuQSS CPU scheduler v0.115 by Con Kolivas
Back to top
View user's profile Send private message
ycUygB1
Apprentice
Apprentice


Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon

PostPosted: Sat Oct 22, 2016 3:03 pm    Post subject: Reply with quote

ck wrote:
Quote:
linux-4.8-ck3, MuQSS version 0.115

This is mainly a bugfix release for those who had boot failures, TOI patched
failures, and warnings. Otherwise it only has minor changes.


I have added ck-sources-4.8.3-r1 to my overlay and removed all versions before 4.8.3, which are vulnerable to the Dirty Cow exploit. Users should upgrade ASAP.
Back to top
View user's profile Send private message
thunderrd
n00b
n00b


Joined: 20 Aug 2010
Posts: 58

PostPosted: Sat Oct 22, 2016 3:25 pm    Post subject: Reply with quote

Heh, you beat me to it; I just edited your 4.8.3 ebuild for CK_VERSION="3" as well, for my own overlay.

I was about to let you know here, but apparently you're on top of things!

Thanks for the effort, hnaparst.
Back to top
View user's profile Send private message
ycUygB1
Apprentice
Apprentice


Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon

PostPosted: Sat Oct 22, 2016 4:37 pm    Post subject: Reply with quote

Uploaded ck-sources-4.8.4 to my overlay. Hopefully the pace will slow down...
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5733

PostPosted: Sat Oct 22, 2016 8:24 pm    Post subject: Reply with quote

Tried MuQSS 0.115 but it's still making my desktop barf... which is weird. It's "only" an AMD K10, nothing but regular SMP and ACPI to worry about. The slightly newer E-350 I have worked first try.

I might have to go back to CFS for the time being, don't want to be running a vulnerable box...
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6108
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Sat Oct 22, 2016 9:58 pm    Post subject: Reply with quote

Well,

post it then on Con's Blog,

seems like he's not aware of that - perhaps there are other users with an AMD K10 that could test and give feedback to resolve this
_________________
https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa

Hardcore Gentoo Linux user since 2004 :D
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5733

PostPosted: Sun Oct 23, 2016 12:48 am    Post subject: Reply with quote

Reading the comments on there it seems like someone else brought up crashes in 0.115 and a last-minute fix appeared. I don't know if the "0.115" I tried today is the same code, but now I might as well wait for the next update. The same thing happened when schedutil appeared and it took a week or two to get back to a booting kernel.
Back to top
View user's profile Send private message
thunderrd
n00b
n00b


Joined: 20 Aug 2010
Posts: 58

PostPosted: Sun Oct 23, 2016 3:52 am    Post subject: Reply with quote

@AntP: Deadline or BFQ?
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5733

PostPosted: Sun Oct 23, 2016 2:44 pm    Post subject: Reply with quote

BFQ...
Back to top
View user's profile Send private message
mir3x
Guru
Guru


Joined: 02 Jun 2012
Posts: 427

PostPosted: Sun Oct 23, 2016 2:52 pm    Post subject: Reply with quote

hnaparst
something wrong with your ebuild

Code:
>>> Emerging (1 of 1) sys-kernel/ck-sources-4.8.4::hnaparst
>>> Failed to emerge sys-kernel/ck-sources-4.8.4, Log file:
>>>  '/var/tmp/portage/sys-kernel/ck-sources-4.8.4/temp/build.log'
>>> Jobs: 0 of 1 complete, 1 failed                 Load avg: 2.73, 2.14, 1.34
 * linux-4.8.tar.xz SHA256 SHA512 WHIRLPOOL size ;-) ...                 [ ok ]
!!! Fetched file: genpatches-4.8-5.base.tar.xz VERIFY FAILED!
!!! Reason: Insufficient data for checksum verification
!!! Got:     
!!! Expected: MD5 RMD160 SHA1 SHA256 SHA512 WHIRLPOOL

_________________
Installation aborted to prevent system self-destruction
Back to top
View user's profile Send private message
ycUygB1
Apprentice
Apprentice


Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon

PostPosted: Sun Oct 23, 2016 3:14 pm    Post subject: Reply with quote

Thanks, I forgot to upload the Manifest file. Give it a try now.
Back to top
View user's profile Send private message
mir3x
Guru
Guru


Joined: 02 Jun 2012
Posts: 427

PostPosted: Sun Oct 23, 2016 3:33 pm    Post subject: Reply with quote

That was fast :D
Thx, it works.
_________________
Installation aborted to prevent system self-destruction
Back to top
View user's profile Send private message
thunderrd
n00b
n00b


Joined: 20 Aug 2010
Posts: 58

PostPosted: Mon Oct 24, 2016 4:15 am    Post subject: Reply with quote

@Ant P:
Try again without BFQ. Use Deadline, and see if the problem persists. I had something similar, and that solved it for me.
Back to top
View user's profile Send private message
thunderrd
n00b
n00b


Joined: 20 Aug 2010
Posts: 58

PostPosted: Mon Oct 24, 2016 10:02 am    Post subject: Reply with quote

New version -ck4 + MuQss .116 available. Blog page and patches here:

http://ck-hack.blogspot.com.au/2016/10/linux-48-ck4-muqss-cpu-scheduler-v0116.html

I've patched to this one because I need the better/more permanent SCHED_YIELD fix for the Folding@Home client, and it's running fine as we speak.
Back to top
View user's profile Send private message
ycUygB1
Apprentice
Apprentice


Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon

PostPosted: Mon Oct 24, 2016 6:54 pm    Post subject: Reply with quote

I have added this to my overlay sys-kernel/ck-sources-4.8.4-r1
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5733

PostPosted: Wed Oct 26, 2016 3:55 pm    Post subject: Reply with quote

thunderrd wrote:
@Ant P:
Try again without BFQ. Use Deadline, and see if the problem persists. I had something similar, and that solved it for me.

Thanks, that was it. Running 0.116 now. I don't understand why it's fine on one system and an early-boot crash on this one, but whatever. Maybe because I'm still using spinning rust?

Here's some new emerge timing numbers to compare to those BFS 0.512 ones from a few days ago. Completely changing the disk scheduler shouldn't affect it too much, all my stuff is on NFS anyway:
Code:
66.54user 1.10system 1:00.06elapsed 112%CPU (0avgtext+0avgdata 230312maxresident)k
0inputs+0outputs (0major+122272minor)pagefaults 0swaps

It's a tiny bit faster in realtime, even though the CPU usage is up about 30%. I don't think that's a mistake, the changelog does say CPU time accounting was made much better for short running processes.
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6108
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Wed Oct 26, 2016 4:01 pm    Post subject: Reply with quote

Make sure you use the additional 4 patches on top of 0.116 from https://github.com/ckolivas/linux/commits/4.8-muqss

I've had horrible slow downs and stalls with vanilla 0.116

Now it seems to act fine with those additional 4 patches.

Btrfs + ZFS also seem to be stable with it
_________________
https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa

Hardcore Gentoo Linux user since 2004 :D
Back to top
View user's profile Send private message
ycUygB1
Apprentice
Apprentice


Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon

PostPosted: Wed Oct 26, 2016 4:04 pm    Post subject: Reply with quote

The ebuild in my overlay uses what I hope is a combined release provided by Con Kolivas

http://ck.kolivas.org/patches/4.0/4.8/4.8-ck4/patch-4.8-ck4.xz
Back to top
View user's profile Send private message
Chiitoo
Administrator
Administrator


Joined: 28 Feb 2010
Posts: 1740
Location: Here and Away Again

PostPosted: Thu Oct 27, 2016 2:32 pm    Post subject: ><)))°€ Reply with quote

kernelOfTruth wrote:
Make sure you use the additional 4 patches on top of 0.116 from https://github.com/ckolivas/linux/commits/4.8-muqss

I've had horrible slow downs and stalls with vanilla 0.116

Now it seems to act fine with those additional 4 patches.

I may have seen what you described as well.

Coming to the machine, that had been running for a couple of days maybe, I quickly noticed there was something odd going on.

The mouse cursor was “sticky” when moved around, application windows switched from one to another rather slowly, and all in all, things felt like the machine was running out of memory. Nothing was using the CPU (using BFS, even if the CPU cores were all 100% used, there usually was no noticeable difference anyblue), and I had about 15 GiBs of RAM free (1 GiB in use).

As I was closing applications in preparation for reboot, things became slower and slower. It was getting so slow, that when I only had the KDE(4) System Monitor running, I ended up killing X via ctrl_alt_backspace since it was taking so long. I then unloaded the nvidia proprietary modules, as it was the other thing I suspected being the cause, but so far it made no difference. Even in the console, something like 'dmesg' output froze at times for a while.

After issuing 'reboot', I waited for over 10 minutes at “* Caching service dependencies ...”, before going for 'REISUB' (which was very responsive, but I'm not entirely sure that it actually did its things (I did see the messages of it supposedly doing the things, though).

Currently running without the 'mucusss' scheduler as a test (was running 'ck4', with BFQ enabled). Perhaps it was nvidia-drivers after all (it is my first time on 4.8 series here, after 4.7).
_________________
Kind Regards,
~ The Noob Unlimited ~

Sore wa sore, kore wa kore.
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5733

PostPosted: Thu Oct 27, 2016 9:45 pm    Post subject: Re: ><)))°€ Reply with quote

Chiitoo wrote:
kernelOfTruth wrote:
Make sure you use the additional 4 patches on top of 0.116 from https://github.com/ckolivas/linux/commits/4.8-muqss

I've had horrible slow downs and stalls with vanilla 0.116

Now it seems to act fine with those additional 4 patches.

I may have seen what you described as well.

Coming to the machine, that had been running for a couple of days maybe, I quickly noticed there was something odd going on.

The mouse cursor was “sticky” when moved around, application windows switched from one to another rather slowly, and all in all, things felt like the machine was running out of memory. Nothing was using the CPU (using BFS, even if the CPU cores were all 100% used, there usually was no noticeable difference anyblue), and I had about 15 GiBs of RAM free (1 GiB in use).

My desktop seemed fine for a day without those patches but I'm feeling the pain right now.

The machine's not exactly idle: I'm in the middle of an `emerge chromium` and running distccd for other stuff, loadavg of about 1.5x `nproc`, but things are only usable if I schedtool -D -n 19 both those process trees. Which causes them to stop running entirely.

Time to update again...
Back to top
View user's profile Send private message
thunderrd
n00b
n00b


Joined: 20 Aug 2010
Posts: 58

PostPosted: Fri Oct 28, 2016 1:14 pm    Post subject: Reply with quote

Looks like new version due out tomorrow, some changes involving improved use of turbo mode, and more interestingly, CONFIG_HZ=100 default.

Patches are in the git repo now, but inclusive -ck5 version surely due soon.

Maybe timing is good, with 4.8.5 announced today:

https://lkml.org/lkml/2016/10/28/133
Back to top
View user's profile Send private message
thunderrd
n00b
n00b


Joined: 20 Aug 2010
Posts: 58

PostPosted: Sat Oct 29, 2016 1:33 am    Post subject: Reply with quote

OK, MuQSS .120, -ck5 version is available.

http://ck-hack.blogspot.com.au/2016/10
Back to top
View user's profile Send private message
ycUygB1
Apprentice
Apprentice


Joined: 27 Jul 2005
Posts: 276
Location: Portland, Oregon

PostPosted: Sat Oct 29, 2016 4:18 am    Post subject: Reply with quote

I have added ck-sources-4.8.5 to my overlay, updated as thunderrd specified in the previous post with MuQSS 0.120 using the ck5 patches and the Gentoo patchset.

Please note that I have removed all the experimental Gentoo patches. The Gentoo experimental patchset includes BFQ, which is provided by Con Kolivas, and also a patch to enable optimizations for gcc>4.9 for multiple cpus. If time permits, I may look into re-enabling this patch. For the time being, I just disabled all the Gentoo experimental patches.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Unsupported Software All times are GMT
Goto page Previous  1, 2, 3  Next
Page 2 of 3

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum