PROBLEM: oom killer and swap weirdness on 2.6.3* kernels
in [Kernel]
|
Prev: Dynamic Debug: Introduce global fake module param module.ddebug
Next: arch/mips/kernel/smtc.c: formatting of pointers in printk()
From: dave b on 28 Jul 2010 03:20 On 28 July 2010 15:06, KOSAKI Motohiro <kosaki.motohiro(a)jp.fujitsu.com> wrote: >> On 27 July 2010 21:14, KOSAKI Motohiro <kosaki.motohiro(a)jp.fujitsu.com> wrote: >> >> On 27 July 2010 18:09, dave b <db.pub.mail(a)gmail.com> wrote: >> >> > On 27 July 2010 16:09, KOSAKI Motohiro <kosaki.motohiro(a)jp.fujitsu.com> wrote: >> >> >>> > Do you mean the issue will be gone if disabling intel graphics? >> >> >>> It may be a general issue or it could just be specific :) >> >> > >> >> > I will try with the latest ubuntu and report how that goes (that will >> >> > be using fairly new xorg etc.) it is likely to be hidden issue just >> >> > with the intel graphics driver. However, my concern is that it isn't - >> >> > and it is about how shared graphics memory is handled :) >> >> >> >> >> >> Ok my desktop still stalled and no oom killer was invoked when I added >> >> swap to a live-cd of 10.04 amd64. >> >> >> >> *Without* *swap* *on* - the oom killer was invoked - here is a copy of it. >> > >> > This stack seems similar following bug. can you please try to disable intel graphics >> > driver? >> > >> > https://bugzilla.kernel.org/show_bug.cgi?id=14933 >> >> Ok I am not sure how to do that :) >> I could revert the patch and see if it 'fixes' this :) > > Oops, no, revert is not good action. the patch is correct. > probably my explanation was not clear. sorry. > > I did hope to disable 'driver' (i.e. using vga), not disable the patch. Oh you mean in xorg, I will also blacklist the module. Sure that patch might not it but in 2.6.26 the problem isn't there :) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: dave b on 29 Jul 2010 05:50
On 29 July 2010 19:47, dave b <db.pub.mail(a)gmail.com> wrote: > On 28 July 2010 17:14, dave b <db.pub.mail(a)gmail.com> wrote: >> On 28 July 2010 15:06, KOSAKI Motohiro <kosaki.motohiro(a)jp.fujitsu.com> wrote: >>>> On 27 July 2010 21:14, KOSAKI Motohiro <kosaki.motohiro(a)jp.fujitsu.com> wrote: >>>> >> On 27 July 2010 18:09, dave b <db.pub.mail(a)gmail.com> wrote: >>>> >> > On 27 July 2010 16:09, KOSAKI Motohiro <kosaki.motohiro(a)jp.fujitsu.com> wrote: >>>> >> >>> > Do you mean the issue will be gone if disabling intel graphics? >>>> >> >>> It may be a general issue or it could just be specific :) >>>> >> > >>>> >> > I will try with the latest ubuntu and report how that goes (that will >>>> >> > be using fairly new xorg etc.) it is likely to be hidden issue just >>>> >> > with the intel graphics driver. However, my concern is that it isn't - >>>> >> > and it is about how shared graphics memory is handled :) >>>> >> >>>> >> >>>> >> Ok my desktop still stalled and no oom killer was invoked when I added >>>> >> swap to a live-cd of 10.04 amd64. >>>> >> >>>> >> *Without* *swap* *on* - the oom killer was invoked - here is a copy of it. >>>> > >>>> > This stack seems similar following bug. can you please try to disable intel graphics >>>> > driver? >>>> > >>>> > https://bugzilla.kernel.org/show_bug.cgi?id=14933 >>>> >>>> Ok I am not sure how to do that :) >>>> I could revert the patch and see if it 'fixes' this :) >>> >>> Oops, no, revert is not good action. the patch is correct. >>> probably my explanation was not clear. sorry. >>> >>> I did hope to disable 'driver' (i.e. using vga), not disable the patch. >> >> Oh you mean in xorg, I will also blacklist the module. Sure that patch >> might not it but in 2.6.26 the problem isn't there :) > > Ok I re-tested with 2.6.26 and 2.6.34.1 > So I will describe what happens below: > > 2.6.26 - with xorg running > "Given I have a test file called a.out > And I can see Xorg > And I am using 2.6.26 > And I have swap on > When I run it I run a.out > Then I see the system freeze up slightly > And the hard drive churns( and the cpu is doing something as the > large fan kicks) > And after a while the system unfreezes" > > 2.6.26 - from single mode - before xorg starts and i915 is *not* loaded. > "Given I have a test file called a.out > And I cannot see Xorg > And I am using 2.6.26 > And I have swap on > When I run it I run a.out > Then I see the system freeze up > And the system fan doesn't spin any faster > And the system just sits idle" > > 2.6.34.1 > With and without xorg - WITH spam on the same behaviour as in the > 2.6.26 kernel appears (when xorg is not loaded). > > OOM attached from the 2.6.26 kernel when I used magic keys to invoke > the oom killer :) (this was on the 2.6.26 kernel - before i915 had > loaded and in single mode). s/spam/same/ -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |