=== fabbione-vac is now known as fabbione
cwilluokay, so it looks like some recent xorg changes have fixed up the intel immediate crashing on resume from suspend, so now I can actually sensibly bug people about my delayed slow-crash after resume-from-suspend that I've had from early 2.6.30rc's through to present 2.6.31rc408:00
cwillunow if only I could remember who I was talking to about that :p08:01
cwilluapw, I think that might have been you08:01
cwilluapw, yep, that was you:  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/38080708:07
ubot3Malone bug 380807 in linux "[karmic, intel] Laptop locks up moments after resuming from suspend" [Medium,In progress] 08:07
cwilluapw, the other bug that was confounding things is fixed and closed, so it might be a little easier to get something done now :p08:07
jjohansencwillu: with this bug does your filesystem still require the fsck after reboot using reisub?08:20
cwillujjohansen, you know, I can't remember.  The last couple times I just rebooted and came back a minute or two later08:21
jjohansencwiilu: if so, is it possible to test with an alternate fs, say ext308:21
cwilluwas thinking about that08:21
cwillus/was thinking/was just thinking/08:22
cwillulooks like I've got enough room on my backup disk to take an image and reinstall fresh08:23
jjohansencwillu: that would be really helpful08:24
cwilluokay, so I'm going to just do an rsync, recreate an ext3 filesystem and rsync back (so, not actually a fresh install)08:24
cwilluis that still useful?08:24
cwilluI could do a complete reinstall and just restore /home, but that'll be more... annoying, than I'd like08:25
jjohansenyeah the rsync to ext3 is useful08:26
cwilluk, waiting for the backup to finish... will probably take a little while as it hasn't been home for the nightly backup in a couple weeks08:26
* cwillu looks around08:27
cwilluwhat, doesn't everyone do an automatic nightly backup? :p08:27
jjohansenbasically with reisub not working (ie. having fsck) the first place I would look is the fs layer08:27
jjohansencwillu: I think most people around here use rsync, though I have played with simple backup config and it is easy and not too bad08:28
cwilludtchen, also, I remember poking you about some pulseaudio instability and drop-outs, which was blamed on a bad driver.  Indeed, 2.6.31 on this machine (different from the laptop I'm talking about to jjohansen) hasn't had any audio issues that I've noticed.08:51
apwcwillu, heh sounds good09:04
lesshastethe touchpad on my toshiba tecra a9 fails intermittently and I see this in dmesg psmouse.c: DualPoint TouchPad at isa0060/serio1/input0 lost sync at byte 110:06
lesshastehttp://bugzilla.kernel.org/show_bug.cgi?id=12577 appears to fix my problem10:23
ubot3bugzilla.kernel.org bug 12577 in Input Devices "DualPoint TouchPad at isa0060/serio1/input0 lost sync" [Normal,New] 10:23
apwlesshaste, is there an ubuntu bug for that one?11:27
lesshasteapw, yes https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-input-synaptics/+bug/36596811:27
ubot3Malone bug 365968 in xserver-xorg-input-synaptics "touchpad randomly failing" [Undecided,New] 11:27
lesshasteactually there seem to be a few11:27
lesshastelots of people are reporting something similar11:27
lesshasteI emailed the kernel mailing list too11:27
apwno wonder we have not seen it, its not got a linux task, nor does it have a link to the upstream bug11:28
lesshasteapw, that's bad... in general the triaging of kernel related bugs on launchpad is really weak11:38
apwwe do our best given our team is small, we can only triage those bugs marked for the kernel after all11:38
apwnormally the x group is pretty good at triaging those and pushing them to linux for us11:39
apwthis one sounds like its fallen through the holes.11:39
apwi've done the admin on it so at least we can see it11:39
lesshasteand yes the problem is that they are not even marked for the kernel usually11:42
lesshasteso the problem happens before you :)11:42
lesshasteI should say that it's a pretty fatal bug11:42
lesshasteI have to reboot to recover11:42
apwlesshaste, so you can trigger this?11:43
lesshasteapw, actually I don't know how to trigger it but it happens after I use X for a bit11:44
lesshasteevery day I would say11:44
apwlesshaste, what arch you using i386 or amd6411:48
lesshasteerr.. i38611:49
lesshasteapw, although this is slightly OT, some teams don't seem to exist at all.. I had a grub bug that I fixed myself but no one from the grub team ever contributed to it as far as I can see11:54
lesshasteis there anyone on that team?11:54
apwthere is probabally no specific team for grub.  foundations tend to care as its a critical package, we care as it impacts us directly as in interfaces directly with us11:55
apwthough i would say we are concentrating on grub2 this cycle11:55
lesshastethat was the fix in fact11:56
apwmove to grub2?11:56
ubot3Malone bug 389930 in grub "grub menu does not display after cold boot on Toshiba laptops" [Low,Confirmed] 11:56
lesshasteapw, yep11:56
lesshastescott howard was very friendly but I don't feel this was his expertise11:57
apwnoone upon noone really wants to work on grub, upstream has moved on to grub2 and so should we11:57
lesshasteapw, ok.. does anyone want to work on grub2?12:08
apwwe are taking an interest in it from a bug point of view, not planning any dev for it here12:10
apwas a distro we are blazing a trail being one of the first mainstream distros to jump12:11
lesshasteapw, :) yes12:23
ibrarCan Any body tell me how to set ALL_PATCH_DIR12:58
ibrarI am compiling kernel using "fakeroot make-kpkg --initrd --append-to-version=-my --added-patches equal.patch   kernel-image kernel-headers"12:58
ibrarand always getting error 12:58
ibrardebian/ruleset/misc/patches.mk:108: *** Could not find patch for equal.patch.  Stop.12:58
apwibrar, can't say i've ever used it sorry13:08
apwlesshaste, ok am pushing some packages with that 9-byte format patch applied, will update the bug when they are up13:08
ibrarapw: I have made changes to kernel! I only want to incremental build that 13:11
ibrarapw: Can you guide me how 13:12
apwi generally build using either in a ppa, or using the direct build interfaces myself13:12
ibrarapw: this fakeroot make-kpkg --initrd --append-to-version=-my --added-patches equal.patch   kernel-image kernel-headers command does not pick my changes so I have to clean and rebuild the whole kernel which waste a lot of time 13:12
apwgenerally you have to remove the build stamps from debian/stamps to get a partial rebuild, iirc.13:13
ibrarapw: These are really higher level timestamp and removing this cause almost compiling whole kernel again13:15
apwgenerally if the build one is still there it cleans the whole thing when using the debian/rules targets13:16
apwgenerally i am building using fakeroot debian/rules binary-generic13:17
apwand removing the build specific stamp lets me redo that without a full rebuidl13:17
rtg_ibrar, apw: how about a command of this form 'make -C debian/build/build-generic M=/my-module-directory-here' ?13:18
apwone doesn't end up with a whole kernel but if all you need is the .ko that can work13:22
apwrtg_, you prolly noticed i pushed a startnewrelease13:28
rtg_apw, yep13:28
apwi am also testing an aufs update, seems to be a few small bug fixes upstream so pulling them into karmic13:28
rtg_cool. Have our Live CD users noticed anything? Was A3 even built using aufs?13:29
apwyep the live cds are pretty magic, they use aufs if it exists so they switch automatically when we uploaded.  just before A3 the images were using it, and direct reports of goodness in testing13:30
apwlesshaste, pushed ... all yours13:31
rtg_apw, I might get back on LTS backports this week.13:32
apwthat sounds cool.  its one of the laggers though we should separate it out as its not a release goal persee13:33
lesshasteapw, thanks!  how would I find them?13:33
apwdetails in the bug13:33
lesshasteapw, ah yes.. thanks13:34
rtg_apw, I'm pretty close to having the first batch of uploads. Just need to review the -meta details.13:34
apwso we should be able to get -server to test on it and let us know if it might work for them13:34
rtg_apw, yep, they are the only audience I'm interested in.13:34
* cwillu proposes renaming livecd's to magiccd's13:53
cwilluoh, incidently, is it expected that wireless devices default to off in 2.6.31?14:07
cwilluI always have to hit the killswitch to turn it back on now14:07
rtg_cwillu, what laptop do you have?14:08
cwilluacer aspire 369014:09
cwilluwireless has worked fine since edgy14:09
rtg_cwillu, its likely the acer WMI or RFKILL driver that is causing the problem. the rfkill subsystem was completely rewritten14:10
gnarlThere has been an rfkill rework in 2.6.30/1 which still could have some problems14:11
gnarlrtg_, Yeah, apw was looking on some cases iirc14:11
* cwillu pokes apw with the laptop-from-hell :p14:11
cwillubut it should or shouldn't be defaulting to off?14:11
rtg_cwillu, no, it should default to off14:12
gnarlIt should be defaulting to the state it had last imo14:12
cwilluI can file a bug once I know if its actually a bug and not just something finally working as expected ;p14:12
rtg_should not*14:12
cwillu(k, in the 'misread that to mean what you intended' sense :p)14:12
=== You're now known as ubuntulog
apwcwillu, do you find hitting the h/w switchy thing twice sorts it out?  generally where the default is inverted we see that occuring14:13
cwilluapw, I'll have to try that14:13
cwillujust heading off to work, but I'll let you know14:13
cwillubut... after startup, hitting it once brings it back.  after resume from suspend, I have to hit it once, and then restart networkmanager before it'll kick in14:14
cwilluthe latter may just be my usual delayed and gradual crashing of everything14:14
cwillu(on resume)14:14
=== fdd is now known as taoup
=== taoup is now known as fdd
Q-FUNKhowdy!  is there any 2.6.31 without DRM support that I could test?  I'm starting to believe that bug #396286 might be caused by the kernel and/or udev insisting on loading DRM to get KMS, even on hardware where it's not ported.15:45
ubot3Malone bug 396286 in linux "kernel 2.6.31-generic oops after loading initramfs" [Undecided,New] https://launchpad.net/bugs/39628615:45
mjg59udev will load the drm module even without KMS, yes15:47
mjg59But there's no drm for the Geode15:48
mjg59So that's not what's happening15:49
Q-FUNKmjg59: indeed, there isn't any ported yet AFAIK15:50
mjg59So it's not a drm issue15:50
Q-FUNKthe drm-related modules are the only difference I can spot between what 2.6.30 and 2.6.31 loads whenever dpkg-reconfiguring the kernel15:52
mjg59If there's no driver with a PCI ID that'll bind to the Geode then nothing will end up running15:52
Q-FUNKI'm just running out of options as to what could cause this kernel crash during bootup.  I've checked every possible issue I could, short of learning how to read kernel core dumps15:55
Nghow's 802.11n support in karmic? (I was in a room that had a/b/g/n yesterday and I only seemed to connect to g)16:33
rtg_Ng, 4965? I'm using an xps1330 to an 802.11n AP16:36
rtg_Ng, I have one of those somewhere....16:38
Ngrtg_: it could just be that NetworkManager wasn't trying to use the 11n in preference, I was just idly curious if we'd generally expect it to work16:38
rtg_Ng, yes, in general 802.11n _should_ work.16:39
rtg_apw, push it17:57
apwdoh, will push the aufs update ...17:58
apwrtg pushed17:59
apwrtg that dell rfkill switch thing i guess thats testable on anything with a h/w kill switch 18:01
apwKeybuk, just fyi the next kernel update will include a few aufs bug fixes18:05
apwKeybuk, its applied in the kernels here: https://edge.launchpad.net/~apw/+archive/green if you want to play, passed basic testing here18:10
apwKeybuk, bah will take  a few mins to sort, PPA copy jut did something mad18:11
apwKeybuk, now they are there18:28
rtg_apw, do you have anything in the pipe for Karmic? I'm thinking about getting the aufs and dell-laptop changes uploaded.18:39
apwno nothing ...18:40
rtg_apw, ok, I'll wrap a bow on it in a bit18:40
apwrtg_, ack18:40
apwrtg the nice thing is that would also test sparc build fixes18:42
lesshasteapw, is shm.c: mmap() failed: Cannot allocate memory an  application bug?19:21
apwwill generally be something that side though nothing is guarenteed in this world19:22
lesshasteok :)19:22
lesshasteexcept death and taxes of course19:23
lesshastepeople seem to see it all over the place so I doubt its actually a specific app issue19:24
ivoksrtg_: you said that you've removed drbd from karmic kernel, but it's still here?20:01
sbehhi, 1. is there an option for debian/rules that does not clean the kernel-build? 2. how do i get linux-headers-2.6.28-14.deb because the file linux-headers-2.6.28-14-generic.deb depends on it?20:03
rtg_ivoks, I'm not sure where or when I said that. the git log shows a couple of maintenance touches since your January update.20:04
ivoksrtg_: mail sent to ivoks@grad.hr, 24/06/200920:05
ivoksrtg_: anyway, could you remove it or should i report a bug about it?20:05
rtg_ivoks, start by reporting the bug. That'll give me time to remember what in hell drbd does.20:07
ivoksrtg_: it's network mirror, we moved kernel part to drbd8-source package which uses dkms20:07
rtg_sbeh,  there are several meta packages tha you can use to kkep your headers packages up to date; linux-headers-generic, linux-headers-server, etc.20:08
sbehwhy is building kernels on debian that terrible complicated20:08
sbehon gentoo i unpack the source and make menuconfig && make && cp .. /boot, thats it20:08
sbehrtg_: I build git to boot from nfs20:08
ivoksyou can do that on debian/ubuntu too20:08
rtg_sbeh, install devscripts, then type 'debuild -b'. 20:09
sbehivoks: but then i loose the ability to use apt for kernel-modules20:09
ivokssbeh: which you had on gentoo? :D20:11
sbehrtg_: this does create deb-packages, i don't know which files linux-headers-2.6.28-14-generic is awaiting in linux-headers-2.6.28-14.deb20:12
rtg_sbeh, 'debuild -b' creates everything, including the headers debs20:13
sbehoh reset, i read the KernelCompile-Wikipage and I already got linux-image*.deb and linux-header*.deb, but the linux-header*.deb depends on a deb that did not get build while following the instructions on the wikipage20:14
sbehwhat do you mean by 'everything'?20:14
rtg_sbeh, the linux-headers packages depend on linux-headers-*_all.deb. 'debuild -b' will build that package as well.20:15
sbehdoes it take hours too?20:16
sbehlike debian/rules binary-generic?20:16
rtg_sbeh, that depends on your build machine.20:16
sbehlike ...20:18
sbehargh, i will just try it20:18
rtg_sbeh,  read the man page on debuild first. by default it cleans first20:19
sbehdpkg-checkbuilddeps: Unmet build dependencies: xmlto docbook-utils transfig sharutils20:22
sbehthe hole texlive-distribution, wonderful!20:22
rtg_sbeh, have you tried satisfying the build dependencies by installing those packages?20:23
sbehi dont want to20:23
sbehlinux can be build without texlive *g20:23
rtg_sbeh, this ain't linux, this is debian.20:23
sbehhm, ok, I will build 2.6.28-13 than, so I can use the already build packages, thanks20:25
lerotdo most distros such as ubuntu  use the same system calls which UNIX has as mentioned by POSIX?21:15
praveenlerot: I think they do21:38
sbeheven gentoo with freebsd-kernel does ;)21:42
Kernel_n00bAnyone here ?21:43
sbehdon't ask to ask, just ask!21:43
Kernel_n00bI need to measure the time consume by the Page-Fault process.21:56
Kernel_n00bI'm trying to measure the software's overhead (no including the HDD overhead).21:56
Kernel_n00bI'm using Ubuntu 9.04 running Linux Kernel 2.6.2821:56
Kernel_n00bI have a small application which generates Page-Fault and then calculate the time (user mode) it is handled.21:56
Kernel_n00bIn order to deduce the IO (HDD) Time, I try to wrap the following section:21:56
Kernel_n00bfile : linux/mm/page_io.c21:56
Kernel_n00bfunction: int swap_readpage()21:56
Kernel_n00bbio = get_swap_bio(GFP_KERNEL, page_private(page), page,21:56
Kernel_n00bIs this the correct location ?21:56
Kernel_n00bIn addition, in order to time the process correctly, I couldn't decide whether I should use "gettimeoftheday()" , RDTSC (on Intel's CPU) or maybe the new Intel's HPET.21:56
Kernel_n00bAny Suggestion ?21:56
Kernel_n00bThank you for your help.21:56
Kernel_n00bNo one ?22:00
Kernel_n00bsbeh ?22:02
manjoKernel_n00b, have you seen this http://ww2.cs.fsu.edu/~hines/present/timing_linux.pdf22:07
Kernel_n00bNope.. I'll check it out.22:09
manjoKernel_n00b, looks like that is what you need to do instead of adding kernel code 22:09
Kernel_n00bHow did you find it ?22:11
Kernel_n00bI think I'm back to the same question :-(22:15
Kernel_n00bgetinfo() is using RDTSC22:16
Kernel_n00bas in my question....22:16
Kernel_n00bthe problem is that I still neeed to measure the HDD Access time, and deduce it from the PageFault time.22:16
Kernel_n00bAnyone ?22:27
Kernel_n00bSomeone ?22:40
=== Kernel_n00b is now known as Kernel_n00b__
jjohansenKernel_n00b__: all of the timing sources have problems, so its a pick your poison type of thing.22:42
Kernel_n00b__I was hopping someone could tell me which function should I measure (in order to deduce the HDD overhead)22:44
Kernel_n00b__I know that timing is a tricky issue. it is always tricky if you're running an multitask OS22:44
jjohansenKernel_n00b__: I haven't played in the vm code lately so it would require digging22:44
jjohansenKernel_n00b__: how do plan on accounting for everything else that can happen while your process is waiting for its page fault22:45
jjohansenKernel_n00b__: just because you start a page fault doesn't mean its the only io going on, and there could be other interrupts, cpu use etc22:47
Kernel_n00b__I plan to kill all process that I don't need (including network).22:48
=== MT- is now known as probuntu
Kernel_n00b__I though of installing a RAM-drive and swap to it. this is how I can measure the HDD access VS memory access (which is much faster)22:49
jjohansenKernel_n00b__: measuring at swap_readpage (I haven't looked) will probably work for a rough value as long as you get a good sampling get rid of the outliers 22:49
Kernel_n00b__I think I'll used RDTSC to measure timing. I need uSec accuracy22:50
jjohansenKernel_n00b__: I would go ask on kernelnewbies and see if anyone more familure with current vm is hanging out22:50
jjohansenKernel_n00b__: make sure to set it so the processor doesn't do any speed changes, or C state changes then, as those will affect your RDTSC22:51
Kernel_n00b__I'll also use CPUID in order to ensure no "Out-Of_Order" execution on the measuring section :-)22:53
=== probuntu is now known as MT-

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!