=== rsalveti_ is now known as rsalveti
=== rsalveti_ is now known as rsalveti
=== MaJian is now known as BruceMa_721
=== smb` is now known as smb
mlankhorstanyone here understands mmaps and stuff in the kernel? :>08:26
ohsixin what respect08:28
smbthere is a lot of stuff in the kernel...08:37
mlankhorstjust mm in general, I'm trying to make forced drm unbinding work, and the part that's missing now is tearing down all mmaps08:38
smbmlankhorst, Hm, not deeply familiar with it but from the interface I'd say the using side is responsible to remember what mappings where done...08:43
mlankhorstofc, but the fun is figuring out how to make the user side kill and free everything :P08:44
ohsixlet them fault in a canary page08:44
* ppisati tries again the install ubuntu on his nexus7...09:28
tseliotapw: what's the point of having rcu_read_unlock_special defined in kernel/rcutiny_plugin.h and defined as a function prototype in include/linux/rcupdate.h? Is it private?10:12
apwtseliot, it may be a left over remnant of some older time, things in that code have moved a lot recently as two of the implementations have merged10:28
tseliotapw: hmm... I see10:29
apwtseliot, it sounds very much like it has moved inline, and not had its prototype removed to me10:38
tseliotapw: that would make sense10:38
apwand i don't think it would cause any issues this way, as essentially it becomes just a prototype, though you might expect to get a warning10:39
tseliotapw: yes, I'm getting a warning but what happens at runtime?10:43
apwtseliot, the right thing as it has only that one definition10:44
apwbut for completeness paste the warning10:44
tseliotapw: WARNING: "rcu_read_unlock_special" [/var/lib/dkms/bcmwl/] undefined!10:46
apwhmmm, that is less what i expected as a warning10:47
tseliotapw: the whole log: http://paste.ubuntu.com/5607386/10:47
apwtseliot, i would be supprised if you can load that module with that warning10:50
tseliotapw: I can't do it from my chroot since modprobe insists on looking for modules.dep.bin in /lib/modules/$(uname -r)10:56
tseliotI'll test it on a real system10:58
lilstevietseliot, there is always insmod if you haven't run depmod :p11:17
tseliotlilstevie: you don't say... ;)11:21
tseliotapw: yes, it errors out: ERROR: could not insert 'wl': Unknown symbol in module, or unknown parameter (see dmesg)11:22
tseliotapw: wl: Unknown symbol rcu_read_unlock_special (err 0) (from dmesg)11:22
apwit actually seems like one of the combinations is just broken right now in that kernel then11:22
apwor is this with you 'transplanting' code11:22
tseliotapw: I simply copied rcu_lock and unlock locally and the unlock function calls rcu_read_unlock_special 11:27
ppisatiogra_: still completely black screen with today's img11:28
tseliotapw: here's the patch http://paste.ubuntu.com/5607457/11:29
apwtseliot, and against which version is it failing ... 3.8 right ?11:30
tseliotapw: yep 3.811:30
apwtseliot, so dispite these being in .h's they are not in .h's really11:35
apwtseliot, they are just being vile.  they are actually just included into rcutiny.c and rcutree.c ...11:35
apwand those in that context are then real externals and not exported at all11:35
apwthis needs to be fixed properly upstream, as it is unreasonable to export the interface without being able to use the locking.  i have discussed an acceptable solutuion with the rcu maintainer11:36
apwbut it won't help existing instances or you ... sigh11:36
tseliotapw: yes, that's what I was afraid of...11:36
tseliotapw: right, so I can't fix it unless I copy a lot of code from rcutiny_plugin.h (i.e. rcu_read_unlock_special and its dependencies)11:38
apwtseliot, which is getting rediculous.  now for us i could spin some accessors, given my discussions i think might be reasonable upstream and we could try those out11:39
apwbut that doesn't help other kernels in the short term11:39
apwtseliot, i assume it is jsut the lowlatency kernel which is in trouble right nwo ?11:40
tseliotapw: yes, it's just that flavour11:40
apwtseliot, and only in raring, so for this monent i think there is little panic at least11:41
tseliotapw: true, at least for now11:41
apwtseliot, zap me the bug number and i'll have a poke at what we might do to fix it kernel side11:42
tseliotapw: let me file one11:42
tseliotapw: I still need to write a proper description but here it is: bug #115403611:54
ubot2Launchpad bug 1154036 in linux-lowlatency (Ubuntu) "rcu_unlock_special is undefined" [Undecided,New] https://launchpad.net/bugs/115403611:54
snadgei updated to 13.04 from 12.10 .. stock kernel wont boot.. but if i select the 3.5 kernel, it will12:02
snadgeintel atom acer aspire one d26012:02
apwsnadge, "won't boot" what actually happens when you attempt to boot that kerenl12:05
apwsnadge, as that is about the most common box around and one we have things which are very similar to in regular use on 13.0412:06
snadgei was seeing kernel panic .. its not on the screen now.. i've installed 3.8.0-12-generic.. and its now booting12:26
snadgebut theres an unrelated broadcom wireless driver problem12:27
snadge!bug 111013912:30
ubot2Launchpad bug 1110139 in linux (Ubuntu) "Broadcom BCM4313 wireless card driver not working" [Medium,Confirmed] https://launchpad.net/bugs/111013912:30
ppisatixnox: btw, did you try my overlayfs kernel?12:39
rtg_ppisati, did you take that over from apw for 3.9-rc2 ?12:41
ppisatirtg_: nope, nexus7 kernel - 3.1.X + overlayfs from precise12:41
xnoxppisati: sorry, not yet, got tangled up in things. Are you pondering to upload?12:43
* henrix -> lunch12:43
ppisatixnox: nope until anyone give it a spin12:43
ppisatimy nexs7 is busted, so can't do myself12:43
xnox /o\12:43
xnoxok. will get to it, as soon as i can. mine was working finish last time i tried.12:44
apwrtg_, i am just playing with it at the moment in fact12:44
rtg_apw, cool12:44
apwrtg_, ... and of course ... there was an overlayfs update over night ... pulling in now13:36
rtg_apw, timing is everything ...13:37
apwi wish i had been even more lax yesterday13:37
rtg_apw, sometimes it is righteous to be a slacker13:40
* smb wished he could use that in the goals...13:42
=== vmesons is now known as vmeson
mlankhorstyou don't say you're a slacker, you have to make it sound like you're putting work into avoiding work :)13:53
rtg_mlankhorst, ok, I am actively pursuing slackerdom :)13:55
mlankhorstagain it's how you bring it :)13:56
ohsixgotta try selling pap smears of famous people13:59
mlankhorstBad way: "I want to slack off". Good way: "increase testability and reduce time needed to isolate problems". Note that in the good way you do some work, to prevent more work. :-)14:05
ohsixmake it your job to make iteration times lower but spend all your time testing14:06
=== kentb-out is now known as kentb
jsalisbury## Kernel team meeting today @ 17:00 UTC14:15
=== ogra_` is now known as ogra_
=== kentb is now known as kentb-afk
chilukHey everyone, the nvidia binary blob has decided to destroy itself all over my logs... should I open the bug against kernel, x.org, or something else?14:57
rtg_chiluk, talk to tseliot14:59
tseliotchiluk: what do you mean?15:01
chiluktseliot let me get a pastebin for you.15:01
chiluktseliot check out http://paste.ubuntu.com/5607938/15:04
chilukthat 20k lines of log is only the first hour of issue...15:06
tseliotchiluk: yes, please file a bug report15:06
tseliotchiluk: against whatever nvidia driver you're using15:07
chilukI'm on latest proposed precise kernel.15:07
chilukand let me check nvidia blob.15:07
chilukjust started happenning, and it only seems to happen in the early morning after the machine has been idle a whiel.15:09
chilukI hit the same issue yesterday, but passed it off to being an power issue from the storm I had.15:10
slangasekjsalisbury: hey, so I just caught bug #1152736 in action here again, this time on 3.5.0-915:22
ubot2Launchpad bug 1152736 in linux (Ubuntu) "system swapping itself to death in raring for no good reason" [Critical,Confirmed] https://launchpad.net/bugs/115273615:22
slangasekjsalisbury: this one wasn't as dire - it sorted itself out in mere seconds, instead of eating my laptop alive for a half hour - but kswap0 was definitely pegging the CPU15:23
jsalisburyslangasek, thanks for the update.  So it doesn't seem to be a regression.  Can you also run on additional test with the latest 3.9-rc2 kernel?  Just to confirm it is in mainline as well.15:24
slangasekjsalisbury: sure.  Remind me where to find the package?15:24
jsalisburyslangasek, http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.9-rc2-raring/15:24
jsalisburyslangasek, I can post it to the bug as well.15:25
roadmrhey folks! I think I found a regression for the bcmwl-kernel-source driver that's in precise-updates (; on a Dell 15R with BCM4312 it fails to connect to any networks. If I downgrade to the version it works fine15:30
chiluktseliot, pad.lv/1154153  has been openned 15:33
tseliotchiluk: thanks15:34
chilukit really might be a power issue, as I just noticed  NVRM: GPU at 0000:01:00.0 has fallen off the bus.15:34
* ppisati needs to reboot, brb15:35
chiluktselliot can you think of any other reason it would "fall off the bus?"15:39
=== kentb-afk is now known as kentb
brendandrtg_, we're doing some SRU testing here and encountering what seems to be an incompatibility issue between the new broadcom driver and the 3.2 kernel15:55
brendandrtg_, doesn't work with 3.215:56
apwretired/CVE-2012-2372: break-fix: 639b321b4d8f4e412bfbb2a4a19bfebc1e68ace4 local-2012-237215:56
brendandrtg_, it works with 3.515:56
rtg_brendand, its possible that it doesn't work with a retrograde kernel. tseliot is the owner.15:57
tseliotbrendand: what's the error?15:58
brendandtseliot, roadmr has more details, i'll hand you over16:00
brendandtseliot, he may take a bit to respond - otp apparently16:01
tseliotbrendand: ok, no hurry16:01
apwrtg_, ok i have pushed an update to overlayfs to the unstable-3.9 branch, i have also tested aufs and overlayfs (minorly) and they seem to at least let me do basic operations16:17
rtg_apw, coolk16:17
roadmrtseliot: ok, I'm here now, sorry! this system does see the wireless networks but is unable to connect to them with the 6.x driver. With the 5.100.x driver, it can connect to b/g networks only, n networks are visible but can't be connected to16:33
tseliotroadmr: it sounds like a regression in the driver16:34
roadmrtseliot: yep! I'm currently testing the proposed kernel, but the 6.x driver also fails with older kernels (i tested 3.2.0-29)16:35
rtg_apw, https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1147678/comments/6 ?16:36
ubot2Launchpad bug 1147678 in linux "kernel 3.5.0-26.40-generic oops immediately when doing schroot w/overlayfs" [Medium,Fix committed]16:36
tseliotroadmr: yes, I don't think it's a kernel issue. 6.x can be buggy on some chipsets16:37
roadmrtseliot: yep. What to do? file a bug? we're also concerned that this will hit people who apply the driver updates and will suddenly find their wireless not working :/16:40
tseliotroadmr: yes please, we should probably revert the upload16:41
roadmrtseliot: ok, I'll start by filing a bug16:41
tseliotroadmr: thanks16:41
ckingkamal, the photo does not show any  < > brackets around the oopsing opcode, which is strange, also I'd like to see the objdump with the opcodes so we can correlate that with the one in the photo16:42
kamalwhere/what is the "oopsing opcode" that you expect to see <> around?16:43
kamalcking: ^^16:44
sforsheekamal, that's the stuff in the line prefixed with "Code:"16:44
sforsheenormally the kernel will put angle brackets around which part of that output corresponds to the oopsing opcodes16:44
* cking wonders if it fell off screen on the right16:45
kamalsforshee, cking: ok, well the lack of <> matches up with my observation that the RIP doesn't line up with the instruction boundaries I guess16:45
ckingmaybe so, yes16:45
kamalcking, sforshee: do you happen to know the objdump flag to make it include the opcodes?16:47
ckingkamal, I normally variations of objdump -d16:48
kamal--show-raw-insn maybe -- I'll try it16:48
jsalisbury##    16:55
jsalisbury## Kernel team meeting in 5 minutes16:55
jsalisbury##  17:00
jsalisbury## Meeting starting now 17:00
jsalisbury##  17:00
* smb is off17:06
=== yofel_ is now known as yofel
kamalcking, sforshee: ok I have added a dump with opcodes (objdump --show-raw-insn) to bug 1136699 ...  aaaaand I'm still confused.  just fyi!17:09
ubot2Launchpad bug 1136699 in linux (Ubuntu) "xserver crash in drm_vblank_count_and_time" [High,Triaged] https://launchpad.net/bugs/113669917:09
=== jsalisbury changed the topic of #ubuntu-kernel to: Home: https://wiki.ubuntu.com/Kernel/ || Ubuntu Kernel Team Meeting - Tues March 19th, 2013 - 17:00 UTC || If you have a question just ask, and do wait around for an answer!
rtg_apw, is this the first request for inclusion ? '[PATCH 00/13] overlay filesystem: request for inclusion (v16)'17:12
apwrtg_, i got it from his git repo ... google wasn't showing that one when i searched17:12
rtg_apw, it was 20 minutes ago17:13
apwoh missread, no i think this is the third time he has asked for inclusion17:13
rtg_apw, has Al formally responded in the past ?17:14
apw"wait wait wait until atomic_open is in"17:14
apwwhich it now is17:14
rtg_well then, maybe it'll happen this time17:14
apwdoubtful... he will have some other objection i am sure17:16
apwbut ... its worth a shot17:16
* rtg_ -> lunch17:18
joshhunti have a question about which git tree i should be pulling from to stay current with the latest quantal kernel updates. when looking at the quantal tree I do not see a tag for 3.5.0-25.39, however looking at my precise tree I do. where tree should i be using?17:21
bjfjoshhunt, one sec checking on that tag, however, you should be using the quantal tree17:22
joshhuntbjf, thanks17:23
bjfjoshhunt, i see the tag in the tree17:24
joshhuntbjf, from git://kernel.ubuntu.com/ubuntu/ubuntu-quantal.git ?17:25
apwjoshhunt, git fetch --tags origin ... perhaps17:25
bjfjoshhunt, yes, i'm in that exact tree right now17:25
joshhuntbjf, ok running git tags now i have: Ubuntu-3.5.0-25.38 Ubuntu-3.5.0-26.40 Ubuntu-3.5.0-26.4117:25
joshhuntapw, i will try that17:25
joshhuntapw, ok that seems to have added the tag17:26
apwjoshhunt, i have that tag in my tree17:26
bjfjoshhunt, and i just did a new clone and it's there17:26
* bjf -> brb17:26
apwbjf, if that tag was somehow not in the main branch you can miss it if you don't pull every day ... though they should be linear in that tree17:26
joshhunti cloned this a while back and have been doing pulls. another person in my group has the same problem. however, the fetch tags thing worked17:26
joshhuntapw, do i need to do 'fetch tags' in addition to periodic pulls?17:28
apwok looking at the history there is some inconstancy in the order, someone has rebased it17:28
apwyou shouldn't in a tree like this, as it should not be being rebased17:28
apwbjf, somehow we rebased .39 off the mainline when making .40 ...17:28
joshhuntapw, ok cool thanks17:30
apwbjf, joshhunt, it may have been embargoed at the time, and that might explain the shenanegins17:30
joshhuntapw, what's that mean?17:30
apwthat .39 carried a very high profile security fix, which may have had unusual handling which may account for the discontinuity17:30
apwwhich means when you pull 'days apart' you don't see .39 as something which is reelvant and you don't get the tag17:31
apwwithout --tags17:31
joshhuntapw, gotcha. thanks for your help.17:32
joshhuntapw, how could you tell that it had been rebased? just so i can better understand when something like this happens in the future.17:32
apwjoshhunt, it is most obvious if you use gitk. ... if you use: gitk origin/master-next Ubuntu-3.5.0-25.3917:33
apwand scroll down till you see .39 you will see that it is off the main branch17:33
joshhuntapw, ahh i should install gitk... :)17:33
apwgitk origin/master-next Ubuntu-3.5.0-25.39 ^v3.517:34
apwjoshhunt, ^^ use that to avoid it making your machine swap17:34
joshhuntapw, thanks17:38
joshhuntapw, really appreciate the help17:38
bjfapw, odd, i remember struggling with it a bit but not rebasing it17:46
bjfapw, we rebase master-next onto it17:46
=== maxb_ is now known as maxb
xnoxppisati: testing overlayfs kernel. seems to work ok, but I am getting some failures with direct write from upstart test-suite.18:15
xnoxwill investigate further.18:15
apwbjf, somehow in that rebase the three commits reposenting it also got rebased18:17
apwbjf, you can see it in the gitk graph: gitk origin/master-next Ubuntu-3.5.0-25.39 ^v3.518:19
bjfapw, it was obviously me18:20
apwbjf, *shrug* it is in the past, and no great shakes as the intent is definatly not lost in the mainline18:22
bjfapw, agreed, though i try to make sure master never gets rebased like that18:22
bjfi'll buy everyone a beer at the next vUDS18:22
apwyeah it is clear you were trying to achieve the opposite, to stack the master-next on top, quite what is different between the two i am unsure18:23
apwnope they are identicle, so it must be metadata somehow18:24
apwbjf, the onlyh difference is the dates, and that is odd as you would expect a rebase even asked to be done from the wrong spot, to note the first few were the same and just leave them be.  very odd18:25
bjfapw, what can i say, it's a gift 18:26
=== bjf is now known as bjf[afk]
ppisatixnox: nice, let me what's the outcome18:52
* ppisati would love to have a working nexus so he could do by himself... :(18:53
* infinity grumbles about someone removing sbsigntool from the kernel PPA.18:53
apwinfinity, i thought that sbsigntool was now in precise (not that i would have removed it)18:53
infinityapw: It's in precise-updates, the PPA builds without pockets, because of security.18:54
apwinfinity, then it should be in -security else the pockets as they are in the release won't build the kernel18:54
apwinfinity, though to get it in the PPA i simply pocket copied it in from -updates in the first place 'with binaries'18:54
apwinfinity, so it is easy to fix18:55
infinityapw: This is a fair point, actually.18:55
* ppisati bails out18:55
infinityapw: Hold off on copying it to the PPA, I *should* copy it to -security so the kernels aren't technically FTBFS.18:55
infinityapw: And your PPA does build against security, so that will fix it.18:56
infinityapw: Thanks for talking sense into me.  Should be fixed shortly.18:57
xnoxppisati: red herring, fails in the same way on amd64, I guess the test-suite is too strict again. So from my point of view it's good to go =)19:00
* xnox should email that.19:00
apwinfinity, np19:03
* rtg_ -> EOD20:02
zequenceI realized it might be smarter to put the kernels into a ~ubuntustudio-kernel owned PPA, so just created one, and uploaded to ppa:ubuntustudio-kernel/linux-lowlatency-sru20:16
zequenceapw: ^20:18
zequenceapw: I could do the metas too from now on20:27
=== bjf[afk] is now known as bjf
bjfinfinity, if someone puts something in *my* ppa without letting me know, i just might be tempted to clean it up some time20:44
bjfjust sayin20:44
bjfinfinity, however, i don't remember deleting it cause i didn't know what it was for20:46
infinitybjf: S'all good, it doesn't need to be there anymore anyway.20:48
infinitybjf: But I suspect it went away when henrix was doing his spring cleaning.20:48
bjfinfinity, it could have been me, i just don't remember doing it20:48
henrixbjf: so, what's missing?20:49
bjfhenrix, heh, the sbsigntool got deleted, but it is no longer an issue20:49
infinityAnyhow, it highlighted a bug anyway, so all good.20:50
* henrix reads backlog20:50
henrixi can't remember deleting anything, but i have fat fingers so...20:51
bryceogasawara, sforshee would one of you mind rolling a kernel with the proposed patch for bug #1153302 if you have a chance?22:30
ubot2Launchpad bug 1153302 in xserver-xorg-video-intel (Ubuntu) "[g33] Non-locking GPU lockup EIR: 0x00000010 PGTBL_ER: 0x00000002 IPEHR: 0x54f00006" [Undecided,New] https://launchpad.net/bugs/115330222:30
bjfbryce, i'll spin you one22:33
bjfbryce, amd64, i386 or both?22:34
brycebjf, thanks.  amd64.22:34
bjfbryce, http://people.canonical.com/~bradf/lp1153302/22:56
brycebjf, thanks22:58

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