/srv/irclogs.ubuntu.com/2012/01/18/#ubuntu-kernel.txt

sconklinsforshee: you around?03:34
Geraldhi05:44
Geraldohsix, are u here?05:45
=== smb` is now known as smb
smbmorning08:56
Kanohi apw , why is CONFIG_INTEL_IOMMU_DEFAULT_ON=y on by default? the option before renaming was NEVER on by default09:30
KanoDMAR_DEFAULT_ON was definitely always off09:31
Kanosame option, just renamed09:31
apwpolicy got applied to new options so it moved on09:31
Kanobut is is no new option09:32
apwrenamed items look like new options09:32
Kanoon one system i have to use intel_iommu=off in order to boot with nv card09:33
Kanobecause of that new default09:33
apwyep there are a few systems with problems as a result09:34
Kanofunnly phoronix mentioned that as regression in 3.2 kernel09:34
Kanobut it was caused just by that config change09:34
apwyeah well well all know how phoronix strive for accuracy09:35
cking'cos phoronix knows best ™09:35
Kanomost likely it never worked09:36
Kanoon those systems09:36
apwyep most likely those systems are broken09:36
apwwe have all sorts of issues with iommus right now, and all sorts of issues with rc6, in some cases related09:37
ckingstill unsure if rc6 + iommu are totally the root cause for some rc6 failures09:38
Kanodid you see a solution for rt2800pci problems with 3.2?09:38
apwunless you enable things in devel you don't find out why they might be off, and what needs fixing09:38
Kanoi googled but did not found anything09:38
ckingit's hard to substantiate facts when facts are rumours09:38
apwperhaps we should check on wikipedia [irony]09:39
ckingLOL09:39
apwKano, anyhow, i see what you are saying about the rename, and will discuss it with leann09:39
Kanohttp://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=commit;h=d3f138106b4b40640dc667f0222fd9f137387b3209:54
=== chrisccoulson_ is now known as chrisccoulson
_rubengotta love all them tweets about kids not being able to do their homework due to wikipedia10:38
Kano_ruben: it is not offline, just disable java script on that page10:38
Kanoit also works when you disable css10:38
Kanothats a block for noobs10:39
_rubeni hardly use wp myself .. so couldnt care less really :)10:39
brendandthat's hardly making a stand11:28
ppisatibrb11:29
apwbrendand, i'd say its blatant self-advertisment11:30
ogra_and its the english page only apparently11:37
ogra_who speaks that anyway 11:37
ogra_:P11:37
apwahh perfect i can redirect all my wp needs via ogra_ 11:38
ogra_only if you also read the pages in german ... i get it as soon as i click on the english version11:38
apwi am using you as translator, you read .de and type it in here for me, purfec11:39
ogra_haha11:39
brendandyou can also just press ESC before the message flashes up. does the same trick as disabling javascript11:50
pgranerapw, ping12:28
apwpgraner, pong12:28
pgranerapw, so the -9 is broke on my box, after every suspend I lock up within seconds12:29
pgranerapw, -8 works fine12:29
apwpgraner, which 8 have you got?  14 or 15 ?12:29
apwsomeone else was mentioning 8s after resume before they hung, hmmm12:30
pgranerapw, pgraner@x220:~$ uname -a12:31
pgranerLinux x220 3.2.0-8-generic #15-Ubuntu SMP Wed Jan 11 13:57:44 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux12:31
apw15 then12:31
apwpgraner, can you test the 3.2.1 mainline kernel and see if that exhibits the same behaviour ... -9 pulled in .112:35
apwnot that i can see anything either way whch would make this occur, hrm12:38
pgranerapw, will do need to send a few emails first12:40
pgranerapw, does this help, this is the what happened during the resume12:45
pgranerhttp://pastebin.ubuntu.com/808528/12:45
apwpgraner, thats just a slower resume, just over the 10s boundary, so probabally not related12:51
apwyou have that disk thats as slow as snot i think12:51
pgranerapw, is an Intel SSD12:51
apwpgraner, as slow as snot when resuming12:52
ckingbut for some reason we get an -EBUSY on the reset, which is peculiar12:52
apwcking, didn't you mention ^^12:52
ckingapw, there were a bunch of issues, including a 3 second delay on that touch screen thingy too12:53
apwyeah, so just hitting the 10s warnign isn't supprising12:53
pgranerapw, I rmmod the touch driver and reinsert via pm-utils12:55
=== mdeslaur is now known as mdeslaur_
=== mdeslaur_ is now known as mdeslaur
ckingpgraner, did you try the "ahci.ignore_sss=1" tweak?12:58
pgranercking, not yet, I'm battling two issues, this suspend resume hang and the slow ssd, any suggestions on which should go first?12:59
ckingah, I missed the earlier comment that S3 resume now fails :-(13:00
ckingpgraner, well I'd sanity check you box with a few S3 tests with a previously known good kernel to see if its a kernel issue first13:05
pgranercking, you mean the fwts tests?13:11
ckingpgraner, just do your normal suspend/resume a coupla times with a previous kernel - if that fails we know it's not necessarily a kernel update issue13:12
pgranercking, I've done that with the -8 kernel and it works fine13:14
ckinglemme see if I can reproduce on my X220i13:15
ckingworking fine with 3.2.0-9-generic #16-Ubuntu SMP Fri Jan 13 20:46:38 UTC 2011 x86_64...13:22
* apw boggles13:23
ckingpgraner, does it behave differently when running on AC or battery?13:27
pgranercking, don't know I'll check that 13:28
=== tgardner is now known as tgardner-afk
=== bladernr_afk is now known as bladernr_
apwwell this isn't very damn helpful: MARC is participating in SOPA Blackout Day. marc.info and lists.kde.org will be dark from midnight US/Eastern January 18th, 2012, until midnight US/Pacific the following day.14:15
apwso one cannot look at mail list archives today either ...14:15
apwall they are doing is pissing me off, and well, frankly, i have no say14:15
egonCALL SENATORS14:17
* apw has no senators, well other than the car and those don't respond to calling14:17
apwhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/917211 <-- smb do you have the h/w to test that ?14:18
ubot2Launchpad bug 917211 in linux "Xen + nouveau + modeset = corrupted console" [Undecided,Confirmed]14:18
egonI like what torbit has done.. popup that you can dismiss14:18
smbapw, by reinstallting my build box maybe... 14:19
smbapw, It has at least a nvidia card in...14:19
apwsmb, bah thats more effort than i wanted expended14:19
smbapw, Also what is this. using pci passthrough to have the card in the guest... Probably in the bug... should have a look14:20
apwyeah ... difficult to say14:21
jsalisburysmb, apw, Is there someone familiar with lxc containers: bug 91766014:30
ubot2Launchpad bug 917660 in linux "Installing qemu-user-static on an i386 lxc container will hose your amd64 host" [Medium,Confirmed] https://launchpad.net/bugs/91766014:30
apwjsalisbury, it is unclear to me that that is a kernel issue14:31
smbFeels like another problem comming from lxc not being a real vm14:31
apwjsalisbury, it feels like the container is not correctly 'closed' and the contents are escaping14:31
smbIf you access proc / sys or what you look at the host ones14:32
jsalisburyapw, smb, could be a qemu issue: qemu: uncaught target signal 11 (Segmentation fault) - core dumped14:32
apwwhether its possible to make it closed or not i am unsure, but i think i'd add a task to lxc for this14:32
jsalisburyapw, will do14:32
apwjsalisbury, i think the complaint is that the request to use qemu for binaries has leaked out into the host14:32
jsalisburyapw, hmm. leaks don't sound good14:34
smbNot sure how the binfmts are updated. Maybe another syscall the containter needs to prevent..?14:34
apwjsalisbury, the kernel does not have a container mechanism, they are faking things by using 100 different bits of protection ... but ... its never been intended for this use so it doesn't work well14:35
* smb tries asking on #server14:35
apwsmb, via /proc, which i am sure is not containerised, it may be possible to remove it though in the container14:35
smbHm, think hallyn is not around yet14:35
* apw hasn't seen hallyn for days14:35
smbright, though that probably completely fails install then14:36
smbproc is not ocntainerized afaik14:36
smbapw, he was on the kernel-team meeting yesteraay14:36
smbbut I think it is too early on that side of the world14:37
apwsmb, then he is hiding from me14:37
apwswine14:37
=== tgardner-afk is now known as tgardner
tgardnerapw, did you tell me I don't have to update CVE tracker states? IIRC you said the shank bot (or its equivalent) will get them set to the right state automatically based on git repo contents ?15:17
apwtgardner, tracker or bug states ?  but in the main both are handled from the trees yes15:18
tgardnerapw, bug states, e.g., fix committed15:18
apwtgardner, there is hourly updates of the tracker from the repo contents, and daily updates of the bugs when jjohansen does his processing15:18
apwtgardner, so yes if you don't fix commit things that will get handled later by jjohansen's scripting15:19
tgardnerapw, cool15:19
tgardnerautomation is king15:19
smbdarn handy15:19
apwavoids me making 100s of errors15:19
* ogasawara back in 2015:51
jsalisburyogasawara, Figured I'd send a heads up.  Lots of duplicates of bug 917962 this morning.16:26
ubot2Launchpad bug 917962 in linux "BUG: scheduling while atomic: swapper/3/0/0x10000100" [High,Confirmed] https://launchpad.net/bugs/91796216:26
ogasawarajsalisbury: you been able to skim through them and determine if it's a recent regression?16:31
jsalisburyogasawara, It appears they are all with 3.2.0-9 and they all were reported today.16:32
ogasawarajsalisbury: would also be good to know if it's easily reproducible16:33
ogasawarajsalisbury: didn't apport kerneloops get turned back on recently?16:33
jsalisburyogasawara, One reporter says it happens every three minutes.16:33
jsalisburyogasawara, I don't think so.  I asked for it to be turned on, but kerneloops.org is still down.16:33
ogasawarajsalisbury: I though bdmurray mentioned he could disable sending it on to kerneloops but still get the reports filed in lp16:34
ogasawarajsalisbury: regardless, might be good to work with the reporter who triggers it every 3 minutes to narrow down the window of regression16:34
jsalisburyogasawara, Ahh, right, he did mention that.  I think he commented out that code in apport.16:34
jsalisburyogasawara, will do.  I'll work on getting the version it was introduced then bisect.16:35
bdmurrayjsalisbury, ogasawara: yes apport and kerneloops have both been uploaded16:35
bdmurrayjsalisbury: that'd be a good bug to write a pattern for ;-)16:36
jsalisburybdmurray, yes indeed.  I'll write one up for this.16:36
apwogasawara, there?  i want to push up an overlayfs update, that ok ?16:57
ogasawaraapw: yep go for it.16:57
ogasawaraapw: also saw your conversation this morning about iommu configs, I've got them fixed up and pending some testing from bug reporters before I push them.16:58
apwas in copying the pre-rename values forward ?16:58
ogasawaraapw: yep16:58
ogasawaraapw: which results in INTEL_IOMMU_DEFAULT_ON to be disabled and will likely resolve a lot of the iommu bugs we're seeing16:59
apwyeah16:59
apwogasawara, sounds good to me16:59
ogasawaraapw: for anyone who really wants it enabled, they can boot with "intel_iommu=on"16:59
ogasawaraapw: I just want to get confirmation on some of the bugs first so I can shove the BugLinks in the commit17:00
apwvery reasonable17:00
=== tgardner is now known as tgardner-afk
doctormonI'm trying to trace a critical bug in 12.04 kernel, but I admit I'm not very good at kernel stuff. Can someone give me a walk through?17:23
apwdoctormon, some background would help17:35
apwogasawara, oh did i mention the new script for adding teh 'rebased to' in the changelog which also finds any LP references?17:39
ogasawaraapw: no, what's it called?17:40
apwdebian/scripts/misc/insert-mainline-changes17:40
apwwhich makes sure we detect and close bugs which come back from upstream17:40
ogasawaraapw: nice, I'll have to use it after the next rebase17:41
doctormonapw: An error in the kernel is causing the pci bus to fail, this causes the video to fail on any driver (vga/nouvou/nvidia) and a black screen to apear.17:43
doctormonSorry for the delay, meeting :-)17:43
apwand we know its the pci bug how ?17:43
apwpci bus17:43
apwdoctormon, and is there a bug open for this behavior17:45
SpamapSSo, I'm on the upstream kernel (3.2.1-030201) on precise, and bcmwl doesn't seem to be working17:46
apwSpamapS, have you installed the headers?  so that the modules get built ?17:46
SpamapSapw: indeed, if I dpkg-reconfigure bcmwl-kernel-source the module is built17:47
apwand how does it fail ?17:47
apwoh i bet it fails to laod17:47
SpamapS-rw-r--r-- 1 root root 3155008 Jan 18 09:39 /lib/modules/3.2.1-030201-generic/updates/dkms/wl.ko17:47
apwif you modprobe wl, what happens, what do you get in dmesg ?17:47
SpamapSwl                   2568210  4294967295 [permanent]17:48
SpamapS[    8.586967] wl: module license 'MIXED/Proprietary' taints kernel.17:48
apwahh so that says, you build wl with the wrong compiler17:48
apwwh17:48
apwwhich is true, as we don't build in precise17:48
apwso i am not supprised it doesn't work17:48
SpamapSwasn't this supposed to be open sourced / upstreamed at some point like.. a year ago?17:48
apwwe really don't expect you to be using kernels in tis way17:48
apwyep they are working on it, there is brcmsmac now which works for my cards17:49
apwwhat device do you have ?17:49
SpamapS02:00.0 Network controller: Broadcom Corporation BCM43224 802.11a/b/g/n (rev 01)17:50
SpamapSmac book air 4,117:50
apwheh mac, you lose ;)17:50
* SpamapS melts like the witch doused with water17:50
apwwhy the heck are you using an upstream kernel anyhow?17:50
SpamapSBecause the precise kernel is unusable for me on this one17:50
apwwhy so?17:50
SpamapSkeyboard cuts in and out, touchpad does not work17:50
apwand 3.2.1 works ?17:51
SpamapSFigured I'd try 3.2.1 before going into bug filing mode17:51
apwtgardner-afk, isn't that one of the ones you have ?17:51
apwENOseth17:52
apwSpamapS, ok to build wl right you need to make a lucid chroot and install the dkms package in there17:52
apwlet it build the modules, and use the result17:52
apw(all very painful)17:52
SpamapSoh *awesome*17:53
apwthey arn't meant for anthying other than a quick test17:53
apwand well binary stuff is always a HUGE pain in the ass, and this is just one example17:53
SpamapSok, well in this case, my quick test is, whatever is wrong with the keyboard and touchpad in the precise kernel is fixed in the upstream kernel17:54
apwthis is why one should not buy machines which need binary stuff ever17:54
apwyou don't need wireless to test that17:54
SpamapSapw: I failed .. the shiny... it sparklezzzz17:54
apwmagpie17:54
* apw suggests the ethernet socket :)17:55
SpamapSapw: so should I make sure to report the bug or just hope that whatever these fixes are will land in the next precise kernel?17:55
apwi'd file something, we won't notice if its on a mac and not one of the 4 we bought17:55
apwSpamapS, specially as we're now onto stable kernel updates so much less will get fixed without action17:56
SpamapSapw: alright, will do... hopefully I can coax the keyboard enough to get a decent report typed in. :-P17:58
apwSpamapS, and go figure out the latest ernel which did work, as i assume it worked in O18:00
apwthey are all in the launchpad librarian for your enjoyment18:00
* SpamapS looks around for the SpamapS-bisect tool that will automatically reboot his machine with each version since 3.0.0-1218:01
doctormonapw: Because going into the root shell mode, lspci shows all the bus ids to be 0000, which is unlikely to happen.18:04
doctormonI've dug up an older bug report18:04
apwdoctormon, and which kernel is exhibiting this behaviour18:05
doctormonhttps://bugs.launchpad.net/bugs/66124818:05
ubot2Launchpad bug 661248 in nvidia-graphics-drivers "PCI Race Condition with COMPAL FL90" [High,Incomplete]18:05
doctormonLinux delen 3.2.0-9-generic #16-Ubuntu SMP Fri Jan 13 22:16:32 UTC 2012 i686 i686 i386 GNU/Linux18:05
apwogasawara, what was the symptoms of that MMCONFIG change that you just applied ?18:07
ogasawaraapw: frequent system hangs/pauses and repeated spewing of error messages to dmesg18:08
apwdoctormon, and is that the old report or your bug ?18:08
ogasawaraapw: bah, I take that back, /me confused my bugs18:08
ogasawaraapw:  the MMCONFIG one is busted usb ports18:09
doctormonapw: it's an old report, and my bug18:09
apwdoctormon, when did this issue appear for you18:09
doctormonwant a new one?18:09
apwi want a bug report from the machine with the issue when it has the issue yes18:10
doctormonKernel 2.6.35, Maverick18:10
apwso it last worked in maverick ?18:10
doctormonapw: No it first appeared in maverick, it work in 2.6.32 lucid.18:11
doctormonworked*18:11
apwok so do file a bug from the machine when running precise if you can18:12
apwas thats a valid update L -> P18:12
doctormonapw: When it's working or when it's failed? I presume failed.18:13
apwif you can when its failed indeed18:13
apwas the pci stuff will be right then18:13
doctormonOK, command line bug tool, that doesn't need a browser right?18:14
doctormonWhy do I get the feeling I might have to do this manually...18:14
apwif it can work then also get a dmesg froma working boot and an lscpi)18:14
apwwell if it can work sometimes you could make the bug in a working moment (and say so)18:14
apwand add the two files above from the broken one manually18:14
apwas in if precise will work sometimes18:15
amitkapw: any of you showing up for ELC in ~1month?18:15
apwunsure as yet18:17
doctormonapw: thanks for your help18:24
apwcking, do we have AGP on arm, i assume not ?18:26
ckingapw, not to my knowledge18:27
apwcking, config seems to agree, good, thanks18:32
ckingsanity exists18:32
=== tgardner-afk is now known as tgardner
* tgardner --> lunch19:03
doctormonapw: Bad news, I went to collect the dmesg/lspci, the recover mode doesn't boot19:17
doctormonStops just after doing ACPI, a message that says to use "pci=usr_crs" which I will try a few times.19:20
=== yofel_ is now known as yofel
thomiHi - Can anyone help give me an update on lp:901386 ? It's turned my brand new laptop into an unusable brick, which isn't so good for work productivity ;)21:22
hallynapw: what's the process you followed to get your kernel.org account back?  did you jsut send an email to hpa?22:27
BenCthomi: running alpha releases isn't so good for work productivity either :)22:28
BenCthomi: if it's too much of an issue, I see there are plenty of working alternatives until it gets fixed22:28
hallynhm, guess i'll try following this old email's instructions22:33
thomiBenC: none of the alternatives seem to work for me (I assume you're referring to acpi=off or intel_iommu=off). Because of that, is it worth filing a new bug?22:35
BenCNo, I mean running a 3.1 kernel (e.g. not running precise, but oneiric)22:35
BenCOr use nouveau(sp?)22:36
thomiahh ok.22:37
thomiThanks22:37
apwhallyn, yeah i followed the original email from hpa23:17
hallyncool, thanks.  lessee what kind of response i get, if that' still the procedure :)23:52

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