You are not logged in.
is there a way to know what has changed between 2.6.26-1 and -2 (seems an Arch dev change, not a source change)
is there somewhere that talks about these changes?
Offline
Offline
is there a way to know what has changed between 2.6.26-1 and -2 (seems an Arch dev change, not a source change)
is there somewhere that talks about these changes?
Offline
thanks for the links guys
this one in particular mentions what i was looking for (in case anyone else is interested)
http://archlinux.org/pipermail/arch-dev … 07266.html
Offline
Does anyone know why the group scheduler has been re-enabled in 2.6.26? If I'm not mistaken, it's now off by default in mainline because it causes some bad latencies. In Arch it was already turned off in 2.6.25 because of this problem. So why now that it's off in mainline it has been turned on in Arch? Anyone knows?
Thanks.
Offline
If I'm not mistaken, it's now off by default in mainline because it causes some bad latencies
I am not sure, but I was about to drop back to 2.6.25 due to some serious latency issues.
2.6.26 seems to be very slow for me compared to 2.6.25, makes my graphics in games and what not very choppy.
Offline
Does anyone know why the group scheduler has been re-enabled in 2.6.26? If I'm not mistaken, it's now off by default in mainline because it causes some bad latencies. In Arch it was already turned off in 2.6.25 because of this problem. So why now that it's off in mainline it has been turned on in Arch? Anyone knows?
Thanks.
Maybe because of this :
http://bugs.archlinux.org/task/10512
pacman roulette : pacman -S $(pacman -Slq | LANG=C sort -R | head -n $((RANDOM % 10)))
Offline
Maybe because of this :
http://bugs.archlinux.org/task/10512
Yes, maybe. But I think the problem is that the config used in 2.6.26 also has CONFIG_FAIR_GROUP_SCHED=y and that's the option causing troubles with latencies. I think the group scheduler is a bit messy with options now and it's difficult to know which options need to be enabled/disabled to have the pre-group scheduler (2.6.24) behavior. I hope the devs can find out and set the right options.
Offline
shining wrote:Maybe because of this :
http://bugs.archlinux.org/task/10512Yes, maybe. But I think the problem is that the config used in 2.6.26 also has CONFIG_FAIR_GROUP_SCHED=y and that's the option causing troubles with latencies. I think the group scheduler is a bit messy with options now and it's difficult to know which options need to be enabled/disabled to have the pre-group scheduler (2.6.24) behavior. I hope the devs can find out and set the right options.
I will try and rebuild the kernel package tonight with that option disabled. Will let you know how it works out for me.
Offline
Bogart wrote:shining wrote:Maybe because of this :
http://bugs.archlinux.org/task/10512Yes, maybe. But I think the problem is that the config used in 2.6.26 also has CONFIG_FAIR_GROUP_SCHED=y and that's the option causing troubles with latencies. I think the group scheduler is a bit messy with options now and it's difficult to know which options need to be enabled/disabled to have the pre-group scheduler (2.6.24) behavior. I hope the devs can find out and set the right options.
I will try and rebuild the kernel package tonight with that option disabled. Will let you know how it works out for me.
Did you have a chance to test it? Maybe enabling CONFIG_FAIR_CGROUP_SCHED instead of CONFIG_FAIR_GROUP_SCHED solves the performance/latency problem while still solving the I/O problem in that report?
Offline
This might be of interest for you:
http://www.archlinux.org/pipermail/arch … 07304.html
pacman roulette : pacman -S $(pacman -Slq | LANG=C sort -R | head -n $((RANDOM % 10)))
Offline
I took the 2.6.26 sources and re-compiled with the 2.6.25 config.x86_64. Specifically changing back all the things changed from 2.6.25 to 2.6.26 due to the one bug report. http://bugs.archlinux.org/task/10512
compiling using pre 2.6.25 config solved the problems I was having with my computer being sluggish and stuttery
I see the bug report has to do with some hardware issues, Or so the original poster thought so. Well, I am on a abit MB with nforce chipset running an AMD dual core 4200+. However, I do use nvidia 8600 GT as video card. I thought the fixes applied in 2.6.26 were supposed to fix issues with AMD hardware and in fact, for me, it's the total opposite.
If anyone has any idea's on what I should now enable in the config, I can try to re-compile the kernel and tell you if the settings are working correctly or not. Till then, I will just wait for another kernel upgrade and see how things are then.
Last edited by rooloo (2008-07-28 09:39:11)
Offline
I have lost the ability to use "sudo pm-suspend" and "sudo pm-hibernate" since the new kernel 2.6.26.
Both of those commands worked perfectly on kernel 2.6.25...
... now all that happens is my screen dims to black, but all of the LED lights stay on, and nothing powers down. Then I can't resume by touching the touchpad... I am stuck on a blank screen and am forced to use the power button to restart.
Other than that, everything seems to be working as usual.
Offline