noticeable pause using the kernel26-ck package (bfs) - anyone else?

graysky · 2010-06-10 21:43:31

I'm using the kernel26-ck package using the BFS patch for kernel 2.6.34. I have been experiencing a noticeable pause or a "freeze" of the mouse cursor/system using this kernel. It seems to happen at irregular time intervals - often while the system is under load (x264 encoding or running mprime), but it will also happen when idle. If I boot into the kernel26 package (official package) I get no such behavior.

Feedback is welcomed.

graysky · 2010-06-17 20:35:10

Also happens on my i686 system

demian · 2010-06-17 21:46:37

I'm not using the AUR package but i patch my kernel with -CK, too and have no such issues on my two machines.
Does your configuration use 1000 HZ and preempt voluntary?

Last edited by demian (2010-06-17 21:53:04)

SanskritFritz · 2010-06-17 21:51:36

I'm using zen kernel, which includes BFS and autoiso patches, no problems here.
Quite the contrary with the stock arch kernel, a simple updatedb grinds my system to halt, so much so, that I cant even use my mouse, it just jumps around every 5 secs or so.

Last edited by SanskritFritz (2010-06-17 22:09:53)

graysky · 2010-06-17 22:23:34

@demian - no, on the 1000 Hz it's using tickless which is the default. No, on the preempt voluntary; it's preemptible.

[*] Tickless System (Dynamic Ticks)
(X) Preemptible Kernel (Low-Latency Desktop)

These aren't my own custom settings - they are the Arch defaults. If I copy over /usr/src/linux-2.6.34-ARCH/.config to the PKGBUILD directory and do a diff on it vs. the one that the package comes with, only the date differs:

$ diff dot_config config.x86_64 
4c4
< # Mon May 17 09:16:10 2010
---
> # Mon May 17 07:56:01 2010

Con recommends not modifying anything, just patching. To my knowledge, this is the right way to do it.

Con's FAQ wrote:

How do you recommend I use this?
It's designed so that you just patch it in and use it. You shouldn't need to do anything at all. But since people still want to know every last thing...
THESE ARE OPTIONAL FOR LOWEST LATENCY. YOU DO NOT NEED THESE!
Configure your kernel with 1000Hz, preempt ON and disable dynamic ticks.
You shouldn't need to tune BFS virtually ever. The only tunable for the scheduler itself is the rr_interval value (see documentation). Try 3ms if
latency is everything to you. When compiling software, do not use more jobs than you have CPUs! So make -j2 on dual core, -j4 on quad core and so on. Nice levels are strictly obeyed so if you nice your compiles they'll be virtually unnoticeable. (nice -n 19 make -j2). Run your distributed computing clients SCHED_IDLEPRIO (eg folding at home, mprime etc): schedtool -D -e mprime
This will make your distributed computing client *never* cause slowdowns in your other userspace applications, at the cost of slightly slower progress of the client. Run your audio and video apps SCHED_ISO:
schedtool -I -e amarok
This will run amarok as an unprivileged real-time task. Note that if you start an application that tries to get real-time scheduling (eg jackd) and you are not starting it as root, BFS will automatically elevate it to SCHED_ISO for you to give it the next best thing.

Last edited by graysky (2010-06-17 22:28:35)

broch · 2010-06-18 14:01:42

1) you did not compile kernel26-ck with default config
2) you did not provide any valid information:
- top
- iotop
- massages.log, errors.log
3) diff between config26 and config26-ck should be more than one I think, so probably your config is screwed up (as config is dynamic and always checks during make process errors can occur)

4) differences between config26 and config26-ck are limited to -ck patches so I don't understand your reasoning for replacing one by the other. This may potentially lead to the above problems if done "as is".

graysky · 2010-06-18 20:56:19

@broch - the PKGBUILD actually patches the config26-ck but it starts from the arch standard.

hokasch · 2010-06-18 23:01:34

have experienced the exactly same thing, but no time for tinkering at the moment so switched back to arch kernel. to be exact, my cursor doesn't freeze, but input "lags".

P.S.: not sure if there was a change in aur PKGBUILD since the last kernel26-ck version...

Last edited by hokasch (2010-06-18 23:03:44)

brisbin33 · 2010-06-18 23:52:27

x86_64, up to date ([testing]), nvidia mobo, intel core2quad.

no issues, runs better than stock.

aur/kernel26-ck built with the optional suggestions.

hokasch · 2010-06-19 07:39:41

Just booted into ck again, and the lag is definitely noticeable, booting arch kernel and it is gone. From what I recall, I should just have used the default options as well, but as said before, haven't looked into it / rebuild it since.
Acer EX 5635ZG, nvidia 105g gpu / core2duo / ICH9M

graysky · 2010-06-19 12:46:10

brisbin33 wrote:

x86_64, up to date ([testing]), nvidia mobo, intel core2quad.
no issues, runs better than stock.
aur/kernel26-ck built with the optional suggestions.

...which optional suggestions? From ck's FAQ or?

graysky · 2010-06-19 14:40:56

broch wrote:

2) you did not provide any valid information:
- top
- iotop
- massages.log, errors.log

k. just booted into the -ck package. Just firefox (browsing) and wget are running and I'm seeing the behavior now at about 10:34 AM.

There isn't anything relevant in either messages.log nor the errors.log even close to that time stamp. iotop didn't show anything unusual nor did top.

brisbin33 · 2010-06-19 17:32:36

graysky wrote:

brisbin33 wrote:
x86_64, up to date ([testing]), nvidia mobo, intel core2quad.
no issues, runs better than stock.
aur/kernel26-ck built with the optional suggestions.
...which optional suggestions? From ck's FAQ or?

sorry i thought it'd be clear b/c you had quoted it a few posts back: Configure your kernel with 1000Hz, preempt ON and disable dynamic ticks.

good luck getting it sorted out.

leadghost · 2010-06-19 19:19:58

I have an old dual zeon machine that runs properly w/ CK patches unless the hyperthreading option is enabled in bios. If it is I get massive lags. If I push it really hard one CPU will spike badly and maybe even lock the box up if process is not killed. Stock kernel runs proper. Maybe this is off base but hope it might be of help.

Arch Linux

#1 2010-06-10 21:43:31

noticeable pause using the kernel26-ck package (bfs) - anyone else?

#2 2010-06-17 20:35:10

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#3 2010-06-17 21:46:37

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#4 2010-06-17 21:51:36

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#5 2010-06-17 22:23:34

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#6 2010-06-18 14:01:42

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#7 2010-06-18 20:56:19

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#8 2010-06-18 23:01:34

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#9 2010-06-18 23:52:27

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#10 2010-06-19 07:39:41

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#11 2010-06-19 12:46:10

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#12 2010-06-19 14:40:56

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#13 2010-06-19 17:32:36

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

#14 2010-06-19 19:19:58

Re: noticeable pause using the kernel26-ck package (bfs) - anyone else?

Board footer