/srv/irclogs.ubuntu.com/2012/01/27/#ubuntu-kernel.txt

jwijsalisbury: for the i915 edp regression in oneiric/precise, you might want to look into https://lkml.org/lkml/2012/1/25/20303:05
eQuiNoX__hey guys07:41
larsk103I have a weird problem with harddrives under the mpt2sas driver. When i do anything but 4k writes, the kernel reads from the drives before writing. These are 512 bytes sector drives, and the kernel seems to know this. Where can I report this?07:44
=== smb` is now known as smb
apwppisati, i see you found the MMC issue, nice10:41
ppisatiapw: yep10:41
apwppisati, did it take all night ?10:41
ppisatiapw: nope, i fnished yesterday around 2110:42
apwppisati, phew, i was worried you'd have to bisect and it'd take ages10:42
ppisatiapw: i prefer to do it manually10:42
ppisatiapw: there were modules that were more suspects than others10:43
ppisatiand i keep forgetting mumble in the morning...10:43
apwt'is easy to do, i sometimes think i am using it when i am not10:43
ppisati(EE) module ABI major version (10) doesn't match the server's version (11)10:54
ppisatiuhm10:55
RAOFppisati: What driver?10:55
RAOFI thought I'd caught them all, but sufficiently badly-packaged drivers might have escaped me :)10:55
ppisatiRAOF: panda board10:56
ppisatiRAOF: let me upload the X log10:56
RAOFThat should use omapfb, right?  I thought that got a rebuild?10:56
RAOFThanks for the log.10:56
ppisatiRAOF: it's a dkms, the hw accelerated driver11:04
RAOFThat's the kernel driver; your error comes from the X server trying to load the DDX, which is different.11:05
ppisatiah ok11:05
* ppisati is X ignorant...11:05
RAOFAre your logs anywhere?11:06
* RAOF will go soon; it's 10pm11:06
ppisatihttp://people.canonical.com/~ppisati/Xorg.0.log11:06
RAOFWhere does that come from?11:07
ppisatiRAOF: from my panda, dist-upgraded this morning11:07
ppisatiRAOF: http://paste.ubuntu.com/818693/11:08
RAOFWhere does that driver come from?  It doesn't appear to be in the archives.11:09
ppisatiRAOF: http://paste.ubuntu.com/818694/11:09
apwRAOF, could it be a partner job ?11:10
RAOFThat's broken from the Xserver 1.11 transition.  It won't work until you rebuild it against the newer server.  Or, if it's closed-source, it won't work until PowerVR rebuild it against the new server.11:10
ppisatiwell, it says: "pool/main/p/pvr-omap4/pvr-omap4_1.7.10.0.1.7-1_armel.deb"11:11
ppisatido we have closed/outside world stuff in main?11:11
apwppisati, does this machine have any odd ppas on it11:12
ppisatiapw: let me check11:12
RAOF“main” is the default component for PPAs.11:12
apwits unlikely powervr is something we have source for, me thinks11:13
RAOFThis does indeed seem likely.11:13
apwPowerVR SGX540 kernel driver for OMAP4 in DKMS11:13
RAOFAlso, is it really so hard to package X drivers?11:13
ppisatiyes, it comes from TI ppa11:13
apwand the kernel driver isn't ... in the kernel11:13
apwppisati, ok so you need to get the arm folks to poke them to rebuild it11:13
ppisatideb http://ppa.launchpad.net/tiomap-dev/release/ubuntu oneiric main #Added by so11:13
RAOFWe've got all this fancy dpkg-helper stuff!11:13
ppisatiftware-center11:13
RAOFppisati: Yeah.  That driver's not going to work until it gets rebuilt against the new Xserver ABI.11:14
ppisatigot it11:14
ppisatii'll poke them then11:14
ppisatiogra_: ^^^11:14
RAOFConsider it a tax on out-of-archive drivers ;)11:14
ppisatirsalveti: ^^^11:14
* apw notes A2 is ... SOON11:14
apwRAOF, i take it you have never been told there is a ti* wibble X driver to worry about11:15
ogra_TI hasnt done an upload yet, not much we can do about this but pinging them (which happened)11:15
RAOFapw: Indeed not.11:15
RAOFWell, maybe I have, but I'd forgotten entirely.11:15
apwogra_, heres hoping you arn't gated on them for your A2 images11:15
ogra_they are also working on an armhf version of the driver... 11:15
ogra_apw, its a PPA driver ...11:16
apwso presumably ppisati can remove it and get something working11:16
ppisaticool, so i don't have to worry about it11:16
ogra_indeed we arent depending on PPA stuff for oour images :)11:16
RAOFYeah.  I *did* check all the drivers in the archive.11:16
apwi presume just his performance will suck11:16
RAOFAlthough I would have missed a driver packaged as badly as that one - it doesn't even have a dependency on the X server!11:16
ppisatiapw: i personally prefer the server images11:17
ogra_Ti doesnt manage to stick to our release schedule (their ubuntu team is to small) for that reason their stuff lives in the PPA11:17
ppisatiall the X/light/unity/$fancyferretgraphicstuff just slow down my board11:17
ogra_as long as the xfbdev drver is working all is fine 11:17
ppisatiogra_: that's working11:18
ogra_(thats what we ship by default)11:18
ogra_plan is actually to remove the PPA icon from the desktop as well and eventually get the pvr driver into the archive, but that hasnt happened yet and will need agreement from the TI side to stick to the release schedule (which is being discussed)11:19
RAOFogra_: Once it wanders archiveward, please give me a ping so I can fix up its packaging a little.11:19
ogra_RAOF, i will ! thanks for the offer ... it will be multiverse anyway though (package has an EULA) 11:20
RAOFThat's fine.  I'd like to make sure it actually turns up on my sweep for drivers I'll break when changing Xserver ABI though :)11:21
ogra_ah, indeed :)11:22
diwicapw, still "Pending publication"...hmm...11:27
diwicfor 3.2.0-11.1911:27
apwdiwic, we were waiting on powerpc last night, i asked for it to be rescored11:27
apwdiwic, yay they are all there, will hastle someone11:27
apwdiwic, hopefully it'll be in shortly11:28
RAOFHeh.  Poor powerpc11:28
apwshe is lagging and no mistake, even arm is getting done before her now11:28
diwicapw, thanks a lot11:28
apwdiwic, ok ... they are accepted now, i'll go check meta11:29
ppisatippc hw is quite cheap nowadays, can't we throw more hw at it?11:29
ppisatiand, btw, why do we support ppc? :)11:29
apwppisati, i hear we have a box in house, waiting to be installed, but ... is is on the case11:29
apwogasawara, 3.2.0-11 now accepted, i've uploaded (your) meta11:34
_jmp_ppisati: maybe one day Ubuntu will run on Playstation (ppc for now) as Game-OS :)11:34
apwdiwic, ok meta is now in the queue11:36
apwdiwic, so it should all be in the archive in about an hour11:36
diwicapw, excellent11:37
ppisati_jmp_: with only 256mb of ram? neeehhh... :)11:38
_jmp_ppisati: PS4 maybe :D anyway... hi * :)11:39
ppisatiogra_: on the beagle ml, people are reporting issue with beaglebone + armhf: kernel doens't boot, while O/armel was working fine11:52
ppisatiogra_: http://groups.google.com/group/beagleboard/browse_thread/thread/370de0a4734bdd24/6a98b1623bf7a39d11:53
ogra_ppisati, thats talking about images from elinux !11:54
ogra_he rolls his very own kernel11:54
ogra_talk to robert nelson ;)11:55
ogra_(rcn_ee is his irc nick)11:55
ogra_see the last mail in that thread11:55
ppisatiogra_: acutally he says "BeagleBoardUbuntu#Precise_12.04_armhf_testing"11:55
ogra_yes, that means he uses our userspace for his u-boot and kernel hackery11:56
ppisatiogra_: he said he tried even upstream11:56
ogra_i havent heard about issues from tobin with armhf on beagle yet, does he have different HW from yours ?11:57
ppisatiogra_: don't think so11:58
ogra_also note it doesnt talk about the beagle*board* at all11:59
ogra_that thread is abotu the beagle*bone* (successor)11:59
ppisatiogra_: yep i know11:59
ogra_i dont think any of us has that HW12:00
ogra_unless someone sends us some boes we wont be able to support it 12:00
ogra_dont worry about it until we got some HW i'd say, we cant QA or test it at all without HW12:01
ppisatik12:02
* herton -> errand+early lunch, back in +- 2h12:35
=== yofel_ is now known as yofel
aquariuscking, heya; any chance you have a bit of time to think about my u300s not suspending? (No problem if not, of course. :))12:54
davmor2aquarius: it just hates you ;)12:56
aquariusis a possibility, I admit it.12:56
ckingaquarius, quite frankly, i'm stumped by it - and our bios engineer is off on chinese new year at the mo12:56
aquariuscking, there's a certain satisfaction that it's actually a real genuine problem and not just me, I suppose!12:57
aquariushibernate works (although I can't set the machine to hibernate from the gui, for reasons I do not understand)12:57
apw(i am sure i asked this before, but i like to make sure) you don't have powernap installed do you12:58
aquariusI doubt it, since I don't know what it is12:58
* aquarius checks12:58
aquariusapw, I do not have it installed :)12:58
apwand you don't take CPUs offline12:59
ckingapw, it's one of those mystery "hang" issues which exists outside of the kernel context12:59
smbapw,  ogasawara, tgardner head up on bug 92181613:04
ubot2`Launchpad bug 921816 in linux "BUG: unable to handle kernel NULL pointer dereference at 00000030" [Medium,Confirmed] https://launchpad.net/bugs/92181613:04
smbLooks like seccomp patches are at fault13:04
eQuiNoX__hey guys13:04
apwsmb, ok so not something thats critical, as nothing uses them ?13:04
smbapw, No clue what is using them, but at least one thing from the seccomp qa regression tests does13:05
apwsmb, as far as i know the only thing that every _might_ use it is chromium-browser, but in the thread from the submitter there was an indication that it was not in use13:10
smbapw, Yeah, seem we never heard anything before and it seems right now it may beonly triggered by the qa regression suite and maybe only on vms13:11
smbWill test bare metal now13:11
apwsmb, sounds good, thanks13:12
tgardnersmb, will drewery had a new patch set proposed on the list awhile ago, but the upstream discussion was so lengthy that I chose to wait a bit.13:12
smbtgardner, right, I have gone forward and made a kernel with all old seccomp patches reverted and his new version applied. That one seems not to cause a crash but some subtests to fail13:13
smb(which could be because its not implemented or different or. ...)13:13
apwdefinatly should ask kees and will about that, as we want to be on whatever he is upstreaming13:14
tgardnerapw, right. they ought to be online in a couple of hours13:14
smbtgardner, apw I'd reply to will's mail to us. Just want to have the info about the old patcheset and the tests verified before I do13:16
apwsounds ike a plan to me13:16
rsalvetiogra_: ppisati: that abi incompatibility was expected 13:18
rsalvetiTI was just waiting the upload to happen to release a newer driver13:18
rsalvetilet me send an email to xavier so he can push it to the PPA13:18
ogra_do you think they will do it before A2 ?13:18
rsalvetiand also request the armhf one13:18
ogra_(would be helpful indeed)13:18
ogra_i pinged nicolas about it but have no answer yet13:19
rsalvetiogra_: generally xavier is quite fast on uploading a newer version13:19
ogra_ah, great13:19
rsalvetiwe'll see13:19
ogra_yep13:19
rsalvetilet me send the email :-)13:20
ogra_thanks !13:20
ogra_:)13:20
eQuiNoX__hey guys, it would be great if you could spend a minute to help me out here => http://askubuntu.com/questions/98995/system-programming-in-ubuntu. thanks for your time!13:34
ogasawarasmb: thanks for the note.  I believe that jjohansen was looking into the seccomp qa regression test failure.14:02
smbogasawara, Ah ok. Right now it seems to be reproducible on 64bit vm and bare metal and causing an oops. Just gathering more date about i386 and comparing to oneiric14:03
smbogasawara, Right, so oneiric seems to be ok. There is one failure of the tests for libcap but that is likely rather the test incorrect (/me hopes)14:05
jjohansensmb: yeah hggdh opened Bug #921816 for it14:05
ubot2`Launchpad bug 921816 in linux "BUG: unable to handle kernel NULL pointer dereference at 00000030" [Medium,Confirmed] https://launchpad.net/bugs/92181614:05
jjohansenI haven't looked at oneiric yet14:05
smbjjohansen, right. jsalisbury pinged me about that14:05
smbjjohansen, Just did a run14:05
jjohansenI had been holding off on the new version of seccomp waiting on the discussion14:06
* jjohansen doesn't actually expect we will carry the older version14:06
smbjjohansen, Right now it looks like someting in precise together with the seccom patch is bad14:06
smbjjohansen, Yep, I build a test kernel with the new version14:06
jjohansenyeah14:06
smbjjohansen, looks better but fails some subtests14:06
* smb is about to fire of a mail thread14:07
jjohansenyeah, thanks for doing that I keep meaning to but ..14:07
smbjjohansen, no worries. mail is out. btw, running test-kernel-security.py on oneiric I got one fail in the last test that tries to check for libcap (oneiric would have libcap2) and fails. It is less of a worry but probably should be looked after14:11
jjohansenright14:11
* smb sent mail without waiting for slooooow i386 to finish update all other stuff...14:12
jjohansenhehe, understandable14:14
ogasawaratgardner: are you updating chroots on gomeisa by chance?14:31
tgardnerogasawara, rebuilding them since they seem to have gotten borked. lemme go check on 'em14:31
ogasawaratgardner: ok cool.  got some random error when trying to chroot into precise-amd64, so I'll just wait.14:32
tgardnerogasawara, you might want to use tangerine. gomeisa seems to have a file system issue.14:33
ogasawaratgardner: ack14:33
apwtgardner, oh that sounds bad14:34
tgardnerapw, 20copyfiles: cp: not writing through dangling symlink `/var/lib/schroot/mount/precise-armhf-5b681eba-8a44-4335-a08e-f089dc799848/etc/resolv.conf14:35
apwtgardner, isn't that a security thing, one of kees14:35
apwtgardner, as in something valid, not broken FS14:35
tgardnerapw, a change in behavior then? this is a 3.0 kernel14:36
tgardnerand what is a dangling symlink ?14:36
* apw looks for the thing he is thinking of14:37
apwtgardner, ok, i think i am thinking of hardlink things14:40
apwtgardner, but a dangling symlink is just a symlink that points to a non-existant place14:41
tgardnerapw, yeah, I don't see anything in yama14:41
apwand that directory there no longer exists cause its an schroot place14:41
tgardnerapw, the failure was on armel, so perhaps I should check that the x86'en chroots recreated OK.14:42
apwi suspect that that may be being triggered by the resolvconf transition.  that changes your resolv.conf into a link i think14:42
tgardnerapw, ah, 'cause i386 has the same issue.14:43
apwsee if the resolv.conf is a symlink inside the chroot14:43
tgardnerapw, you mean precise is now a symlink? hmm14:44
tgardnerapw, well, its a link on my desktop. resolv.conf -> /run/resolvconf/resolv.conf14:45
=== bladernr_afk is now known as bladernr_
tgardnerapw, I think this is actually a schroot problem with /etc/schroot/sbuild/copyfiles15:09
apwtgardner, hmmm, well that just lists the files to copy15:15
tgardnerapw, right. its actually /etc/schroot/default/copyfiles15:15
apweven that is just a list of files15:16
tgardnerwhich is used by /etc/schroot/setup.d/20copyfiles15:16
* ppisati -> the dove kernel is backing on tangerine, back in a bit15:17
tgardnerI think I'll ask on #ubuntu-devel. surely they have encountered this issue already.15:17
apwetc/resolv.conf -> /run/resolvconf/resolv.conf15:17
apwno i think this is the bug15:17
apwthat should not be an absolute link15:17
apwit should be etc/resolv.conf -> ../run/resolv.conf15:17
apwit should be etc/resolv.conf -> ../run/resolvconf/resolv.conf15:18
apwthe second one ...15:18
apwso that it works in chroots from outside ... as the copy (20copyfiles) has to work from outside to see both files15:18
tgardnerapw, well, only if /run is mounted within the choot15:18
apwrun should be unique to the chroot right?  even if its not adding one, the directory should work15:19
tgardnerapw, so you think the ultimate bug is in the resolvconf package ?15:19
apwi think its odd to use an absolute link like that ever, so i'd query if thats legal for this exact reason15:20
tgardnerapw, ack15:20
apwif it is, then 20copyfiles needs to read all the links it finds and map / to the chroot15:20
apwor use something more like 'cat source | chroot FOO cat - >destination15:21
apwto do the copy15:21
* ogasawara back in 2015:35
=== bladernr_ is now known as bladernr_afk
tgardnerogasawara, you're making tangerine squeal.16:03
ogasawara:)16:03
apwogasawara, 420 ... ouch16:08
apw"420:1 and falling" "344:1 and falling"16:09
ogasawarabjf: there's complaints that rls-p-tracking-bugs.html is broken...16:10
smbapw, Guess when its one anything else is our own problem16:11
apwsmb, :) indeed16:11
bjfogasawara: naturally16:13
apwbjf, let me guess this is the first you have heard of it16:13
bjfapw, yup16:13
ogasawaraI think this if the first people were aware (ie have looked at the page)16:14
bjfogasawara: let whomever know i'm looking into it16:14
ogasawarabjf: will do16:14
ogasawarabjf: it was pointed out in the release meeting16:14
bjfogasawara: figured as much16:15
ppisatiherton: what's the status of the lucid/master? is it ok? can i go ahead and rebase?16:15
hertonppisati, yep, you can go ahead16:16
bjfogasawara: the report is back16:31
ogasawarabjf: thanks16:31
brendandtgardner - hi16:51
tgardnerbrendand, hmm?16:52
apwogasawara, i just pushed an ipv6 config change onto your tree16:53
brendandtgardner - do you think it would be important to test different wireless frequencies as well as different bands (e.g. 2.4Ghz and 5Ghz)?16:53
tgardnerbrendand, not particularly, no.16:54
brendandtgardner - cool16:54
tgardnerogasawara, gomeisa should be back to normal in 10 mins or so16:55
ogasawaraapw: ack, thanks16:57
ogasawaratgardner: k, I'll hammer on it next16:57
apwogasawara, it doesn't need to go in for A2 if you have already tested, it can happily wait for upload#1 after16:58
ogasawaraapw: cool, I think I'll let it wait then since I've just finished my test builds.16:58
apwin fact not putting it in given all the other instability we have ...16:59
=== JanC_ is now known as JanC
=== bladernr_afk is now known as bladernr_
tgardnerogasawara, gomeisa is ready to bend to your will.17:21
ogasawaratgardner: thanks!17:21
* ogasawara debates ripping out all the seccomp patches for the Alpha-2 upload18:40
tgardnerogasawara, it can't be any worse then the current situation.18:41
tgardnerogasawara, and having just read the email list, kees agrees.18:42
argeshey. i'm getting this odd error on my t420 where when I resume from suspend, it then re-suspends. is there a wiki on how to get verbose suspend info? or should I just file a bug and go from there. 18:43
argesits somewhat rare for me. but annoying when it happens.18:43
tgardnerarges, you just missed cking, but I remember that the double suspend used to be a problem.18:44
tgardnerI can't remember what was causing it.18:44
* apw calls it a day ...18:44
ogasawaratgardner: was that the same issue cking and him were debugging in budapest?18:44
tgardnerogasawara, I don't recall if that was it.18:45
tgardnerseems like the double suspend issue was in Oneiric18:45
* tgardner -> lunch18:46
=== cking_ is now known as cking
ckinggah, my AP is a pile of unreliable do do18:46
smbcking, nah that is the new "this is the weekend" feature. :)18:47
argescking, hey! having some wierd double-suspend issues with my t420. are there some bugs I should look at? or verbose info I can get to see what the problem is?18:47
ckingarges, double -suspend? what's that18:48
argescking, sorry, so when I wake my laptop up. it resumes suspend. I see the login, but then it immediately goes back into suspend 18:49
argesits not every time, so i occasionally hit this18:49
argesbut I want to get better info on it, so I can file a good bug / start looking into it18:49
ckingarges,  file a bug - then when it happens attach the dmesg output and the /var/log/pm-suspend.log so it can be eye-balled18:51
argescking, ok. will do18:51
ogasawaraarges: this is a wiki I found, but not the one I was thinking of...https://wiki.ubuntu.com/DebuggingKernelSuspendHibernateResume#Debugging_Suspend18:53
argesogasawara, ok cool i'll read though this one too! thanks18:53
hertonarges, I remembered/found another one too, this may be useful, an excellent instrumentation cking did using systemtap, for debugging suspend: https://wiki.ubuntu.com/Kernel/Reference/S3SystemTapDebug18:55
ogasawaratgardner: I'm ripping out the seccomp patches.  I've still got enough time to build, test, and upload.18:55
hertonbut it would be more intended for the case of a hang18:56
ckingherton, this is kinda useful for hangs, but I'm unsure how useful my S3 scripts will be for a bounce18:56
hertonindeed18:56
argesherton, ok that might be good for some deep debugging. i'll try to get a bug filed with dmesg / log info when I hit it again. thanks18:57
ckingarges, does google show anything useful, you can't be other other user running ubuntu on this machine?18:57
argescking, not sure what you mean by that.  i'm the only user on this machine 18:58
ckingarges, i meant, there must be loads of other people with this machine on the planet running ubuntu, so if it's a common issue you may find it with google ;-)18:59
argescking, yea i'll do that first18:59
ckingsearch is your friend ;-)19:00
ckingsomebody make bzr run faster please!19:20
tgardnerogasawara, ack19:20
jcastroHi, I'm looking for the generic backported compat-wireless package for oneiric, I think it's this one but it looks too specific to be the generic metapackage: http://packages.ubuntu.com/oneiric-updates/linux-backports-modules-cw-3.1-3.0.0-15-generic22:18
bjfjcastro: i think you are looking for linux-backports-modules-3.0.022:35
=== bladernr_ is now known as bladernr_afk
chrxnHail, Sages!22:56
chrxnI am trying to compile with 1000hz timing in EC2 for use with VOIP. I'm using some instructions that I found for doing this, but I have hit a wall, and I don't know how to continue exactly.22:58
chrxnThis is the error ( http://pastebin.ca/2106891 ) I've encountered after following these instructions ( http://tinyurl.com/875emvt )23:01
ohsixif the ec2 vm's can do hrtimers HZ is going to be somewhat moot unless you end up depending on a driver (loool virtual machine) that still relies on jiffies23:01
chrxnSo, even if this could be acheived,  and it could most likely by a more experienced person than I, it would be of little consequence?23:03
ohsixyep23:04
chrxnThank you, oh sage.23:04
ohsixyou can see what sort of timers are being used by userspace software and kernel threads in /proc/timer_list23:05
chrxnThis is way over my head. It only took a jiffy of talking to you to know that my efforts are futile in doing this thing.23:08
chrxnI will keep the kernel as-is.23:08
chrxnThanks for your guidance ohsix 23:08
ohsixno problem23:11
=== Eimann_ is now known as Eimann

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