/srv/irclogs.ubuntu.com/2010/07/30/#ubuntu-kernel.txt

=== bjf is now known as bjf[afk]
jjohansendaishadar: most things will still work00:09
jjohansendaishadar: the ubuntu specific patches provide features, drivers, and fixes that are not upstream yet, and in some cases will never be upstream but are useful in their current state for our users00:10
jjohanseneg. aufs is used to provide union mounts for the livecds and schroots00:11
jjohansenndiswrapper lets users fall back to windows wireless drivers00:12
=== sconklin is now known as sconklin-gone
daishadari see... so they may be things i notice but they should be relatively minor, depending on what features i'm using00:13
jjohansenAppArmor provides the default MAC, but if you boot without things will work you just won't have the extra restrictions on some applications00:13
daishadarthe other problem is that i will lose binary drivers (e.g. nvidia), right?  but i should be able to download those from the manufacturer's website...00:14
jjohansenright you loose prepackaged binary drivers, but with some digging you can find the source stubs and blobs and compile them00:15
jjohansenwhere you get them just depends00:15
jjohansenusually from the manufactures website or out of ubuntu DKMS packages00:15
daishadarok, great, so it sounds like it's reasonable.  i'm thinking about buying a thinkpad t410s laptop but apparently there are major issues if you aren't running 2.6.3400:18
ogasawarajjohansen: \o/, just saw your email00:21
jjohansen500:24
jjohansenhehe, lets try that again00:24
jjohansen500:24
jjohansen  \o00:24
ogasawarawill definitely have to highlight that in the release meeting tomorrow00:25
jjohansenogasawara: I'll get you a request pull out as soon as I rebase and verify00:26
ogasawarajjohansen: cool, thanks00:26
jjohansenogasawara: I have a couple kernel related work items that won't make tomorrows freeze but may make feature freeze deadline, should I just push those items out now?00:36
ogasawarajjohansen: yah, go ahead and push them out to be safe00:37
jjohansenokay00:37
bjf[afk]jjohansen, that's fantastic, good job00:45
jjohansenthanks :)00:45
=== jeff_ is now known as Guest27050
* bjf[afk] wants to scream about the build scripts01:56
xsaiddxhello guys 02:31
xsaiddxi got this on start up02:31
xsaiddxhttp://pastebin.com/kgSrGqe502:31
xsaiddxanyone ca help me pls02:31
jjohansenxsaiddx: ubuntu kernel? or other?02:33
jjohansenwas this a clean install or an upgrade?02:33
xsaiddxjjohansen, sorry02:33
xsaiddxi didnt get you02:34
jjohansenbasically there is not enough information to figure out what is happening02:35
xsaiddxno i removed plymouth and it takes off mostly all of the packges02:35
jjohansenugh, uhmm yeah that is a problem02:35
xsaiddxso i had to install an image kernel  and follwin that tip in the end of it it says 02:36
xsaiddxunlink sbin/init so i got this message i showed you02:36
xsaiddxand b4 i had acces only to my console02:36
xsaiddxbut like this xsaiddx@none02:36
jjohansenhrmm, what do you still have access too?02:37
jjohansenfor ex. can you reinstall plymouth02:37
xsaiddxno02:37
xsaiddxnow only acces to see that nice message and thats it02:38
xsaiddx: D02:38
xsaiddxi can go live via a cd and do link sbin/init ?02:38
jjohansenhrmm can you boot a livecd, or usbstick?02:38
xsaiddxwud it work02:38
xsaiddxyeh02:38
jk-xsaiddx: you could try booting with init=/bin/sh02:38
xsaiddxuhm how cus i cant type anythin02:39
xsaiddxor you mean from the live cd02:39
jk-xsaiddx: at the grub prompt02:39
xsaiddxtheres no grub now02:39
jjohansenhold down the left shift key at boot02:39
jk-shift to get the grub menu02:39
jjohansenhas to be left shift02:39
jk-select the right boot option, press 'e' to edit it02:39
jk-https://wiki.ubuntu.com/DebuggingKernelBoot02:40
jk-instead of step 3, add 'init=/bin/bash'02:40
jk-(probably remove 'splash' and 'quiet' too actually)02:40
jk-which 'tip' were you following that told you to remove init?02:42
xsaiddxjk ill link you02:43
xsaiddxhttp://forum.phillw.net/viewtopic.php?f=4&t=35, this02:43
* jk- looks02:43
xsaiddxjk btw where ishud type that init02:43
jk-xsaiddx: what is on your screen at the moment?02:44
xsaiddxjk??02:45
xsaiddxsory didnt get you 02:45
jk-what do you have on your screen now?02:45
jk-are you in grub?02:45
xsaiddxno im on xp 02:45
xsaiddxnow02:46
xsaiddxwhen i start lubuntu theres no grub02:46
jk-xsaiddx: instructions are here: https://wiki.ubuntu.com/DebuggingKernelBoot02:46
xsaiddxjsut that message i shpwed you02:46
jk-jjohansen: hm, wiki says 'escape' for some reason02:47
jjohansen????, its always been left shift for me02:47
jk-jjohansen: i blame manjo :)02:47
xsaiddxjk well hit esc then what02:47
jjohansenyeah its got to be manjo02:48
jk-xsaiddx: then read the next step02:48
jk-xsaiddx: if escape doesn't work, try left-shift instead.02:48
xsaiddxyeh but still didnt get it sorry02:48
xsaiddxcan you tell me what to do 02:49
jk-xsaiddx: do you have important stuff on that machine? maybe a reinstall is besst02:49
xsaiddxyeh 02:49
xsaiddxand i  dont like the idea to reinstall all02:50
jk-1) reboot the computer02:50
xsaiddxyeh ten02:51
xsaiddxthen02:51
jk-2) after the BIOS screen goes away, press left shift02:51
jk-and hold it down02:51
jk-you should get the grub screen02:51
xsaiddxthen02:51
jk-is this a separate computer? can you do this as we talk?02:52
xsaiddxno i have dual boot02:52
jk-so you're going to remember this stuff?02:53
xsaiddxyou can tell me im takin notes tho02:53
xsaiddxhello jk02:54
xsaiddxjk-, hello02:56
jk-xsaiddx: just a sec02:56
xsaiddxok02:56
xsaiddxsure02:57
jk-xsaiddx: print this out: https://wiki.ubuntu.com/DebuggingKernelBoot02:57
jk-xsaiddx: before you press 'b', add "init=/bin/bash"02:57
xsaiddxdude my english sux and i didnt get that doc so thats why iwas askin your help clearly02:58
jk-I've just re-written the doc for you02:58
xsaiddxwhere i can add that init.... ?02:59
jk-xsaiddx: however, you will need to do a few more things after you've done that boot, and I can't tell you what you need to do there02:59
xsaiddxwhy i have to tell wht im goin to get so you can kno is it02:59
xsaiddxi can go and see now and come bk n tell you02:59
jk-sorry, i don't understand either of those lines. could you rephrase?03:00
xsaiddxjk-, well im goin to try okies brb please and tnx for ur help03:00
xsaiddxi said03:01
xsaiddxyou cant tell me wht to do after that 03:01
jk-i would suggest these commands, but I can't be sure:03:01
xsaiddxcus you need to see wht im goin to have as an error so you can kno the issue03:01
xsaiddxso i said ican go and come back to tell you03:02
xsaiddxokies ?03:02
xsaiddxwell im goin now03:02
xsaiddxbut i jst wnana kno when gettin grub when ishall add tat init=...03:02
jk-apt-get install --reinstall linux-image-generic-$(uname -r)03:02
jk-xsaiddx: it says where to add it in that document03:02
jk-sorry, make that:03:03
jk-apt-get install --reinstall linux-image-$(uname -r)03:03
jk-i have updated the document with more details. please refresh the page to see the latest version03:04
xsaiddxok03:04
jk-also, apt-get install --reinstall upstart03:05
xsaiddxok ill get back to you tnx03:06
xsaiddxwish me luck 03:06
cooloneyericm: how can i build the upstream kernel for omap without defconfig file?05:13
jk-cooloney: the defconfigs should still be there, no?05:31
cooloneyjk-: no omap4_panda_defconfig05:35
jk-ah05:35
jk-menuconfig? :)05:35
cooloneyand i tried to build the omap kernel from Tony's linux-omap tree05:35
cooloneyit failed05:35
cooloneyweird05:35
cooloneyjk-: yeah, i chose TI OMAP system, it will select omap2/3/4 and build for Panda board as well05:36
cooloneyjk-: my bad. i just turned on ARCH_OMAP. forget to turn off OMAP2 and OMAP305:46
cooloneynow it is building 05:46
jk-ericm, cooloney: pull request sent, for the debug macro changes05:58
rippsIt looks like l-b-m and linux-meta were merged together, and now linux-meta is held up in NEW. How long before it gets into the repos?06:08
ericmcooloney, I don't think omap4_panda_defconfig is gonna squeeze into mainline recently07:02
cooloneyericm: i know, it was enabled in omap_4430sdp_defconfig07:23
cooloneyericm: i can build the kernel, but doesn't boot on my board. serial port driver is not set to the right one07:24
ericmcooloney, that always happen, maybe you should change the command line in u-boot07:32
cooloneyericm: still checking the right config, will let you know the result.07:43
cooloneyericm: btw, i am trying tony's linux-omap tree instead of linus mainline07:43
ericmcool07:43
trapmaxanyone experiencing "INFO: task x blocked for more than 120 seconds." messages. In our system it gets triggered when using lvcreate. Started after updating to 2.6.32-24-server -kernel. Workaround?08:58
RAOFtrapmax: That'll happen when something bogs down so much that forward progress doesn't get made fast enough.  I've managed to get almost all my tasks to spit that message when doing a bunch of I/O in btrfs.08:59
trapmaxRAOF: this did not happen before updating to newest kernel09:00
RAOFI'd guess that what's happening is lvcreate is blocking I/O, and is chomping on long enough to trigger that messsage for your other tasks to trigger the message.09:00
smbUsually there would be a backtrace about it. To offer more info09:00
trapmaxhttp://pastebin.com/KUzVGZSQ09:01
smbMost of the -24 update was ext4 which would not have an impact on lvcreate though09:01
smbOh, actually it is ext4 related...09:01
trapmaxwe have ext4 environment though09:01
smbI was just wondering because lvcreate itself does no filesystem09:02
smbIt could be related to the general writeback regression that we are currently seeing with 2.6.32 kernels. Maybe the ext4 update in some cases just makes it happen more often09:03
trapmaxthis makes our lvm-snapshot -backup virtually impossible09:04
smbtrapmax, I got some test kernels to try if you want to.09:04
trapmaxproduction server. not really confident with that option09:05
smbtrapmax, Hm, hard to offer a workaround that way. The ext4 updates are verified to fix some data integrity issues and seem to cause long(er) IO times in your case09:09
smbthe bug in question would be bug 58509209:09
ubot2Launchpad bug 585092 in linux (Ubuntu) "giant IO delays (affects: 1) (heat: 20)" [Medium,Triaged] https://launchpad.net/bugs/58509209:09
smbAs this relates to some syncs done synchronously, I wonder whether issuing manual sync's from the command line change anything09:10
trapmaxwe have manual sync runnin with 5s sleep09:11
smbHm, ok09:12
smbtrapmax, All the backtraces in your log somehow go through writeback. I am currently trying to get some feedback on the backport of the writeback patches which are in the test kernels. But this will take a bit.09:15
smbUntil then the only thing to play with seems to be the interval of the manual sync. The previous kernels somehow worked but from the tests that were done for the ext4 patches I would not really recommend going back to those old kernel.09:16
=== lag is now known as lag-day-off
trapmaxsmb: ack09:19
jjohansenrebooting09:43
* smb blinks09:43
jjohansenurghh, don't update!!!10:00
smbjjohansen, Maverick kernel? Or something else?10:01
jjohansenmany things10:01
jjohansenseems kernel is broken for networking10:01
* smb thinks jjohansen should be sleeping instead of updating anyway ;-=10:01
jjohansentheme is broken10:01
jjohansengnome panel seems to be broken10:02
smbSeems everything is ready to alpha310:02
smbHave we been doing something in the networking area?10:02
jjohansenI have a high load almost 50% each cpu, but nothing is showing up via, top, htop etc10:02
jjohansenthe load isn't the kernel as I had to reboot into an older kernel, and it still remains10:03
smbjjohansen, would perf give more hints?10:03
jjohansensmb maybe10:03
RAOFYou might need to re-run apt; they're discussing the gtk update which is probably what has caused your “everything broken”10:04
jjohansentheme is comletely broken too, I had to revert to human :)10:05
* smb likes human. :)10:06
jjohansenRAOF: already have no dice10:06
RAOFThat, plus a reboot, fixed it for me.  Although I didn't have a high load.10:06
jjohansenI'm sure it will get sorted out, soon just putting out the warning that now is not a good time to update10:07
jjohansensmb: do we have a wiki on perf yet10:08
smbI guess an indication that its not kernel or is would be good to know for Leann on the release meeting10:08
smbjjohansen, I am not sure. Maybe not10:09
smbI usually try "perf top" first, that would give an overview on calls10:09
smbjjohansen, Other than that maybe in some presentation slides from chase10:10
smbII think the other was a "sudo perf record" paired with a "perf report"10:11
jjohansensmb: right, but that records a command, I was looking for a way to just get a sampling of all commands and kernel fns10:12
jjohansenafter all what I am seeing is not showing up10:12
smbWould "sudo perf record -a -f -c 5000" be what you are looking for10:13
smbThe 5000 might be a tad long10:14
jjohansenmaybe10:14
smbI was stabbing in the dark with the count10:14
smbSeems already quite long with 5010:15
jjohansensomething is really hammering my system load average 3.28 2.10 1.1810:16
smbStrangely you would expect it to show up in top if it gets accounted in the load10:16
jjohansencpus are varying between 30- 100%, but htop isn't showing10:17
jjohansenthat is it shows the cpu load and the processes are like 7%, 9% etc just small little figures10:17
smbI thought that one patch from Chase about accounting small runtimes went upstream as well10:18
jjohansenhrmm, so did I10:20
smbjjohansen, btw, -c# is not really needed. And i have no clue what it relates to10:20
jjohansenI think its something to do with panel icons, as they have been slowly showing up, 1 by 110:20
jjohansenI now have a wifi icon, and a screen orientation icon, which I didn't before10:21
smbI see no such thing here. Oh, you mean your load10:21
jjohansenyeah10:21
smbSo you could try "sudo perf record -f -a", let it run a bit then ctrl-c10:22
smbThen do "sudo perf report"10:22
jjohansenyeah, I'll give that a try, as soon as I get perf installed on this machine, the install is crawling10:23
smbjjohansen, Maybe start it and get some rest? :) I see your continent mostly in the dark...10:25
jjohansenhehe, no I am going to finish getting a pv-ops kernel together and fix my AA request-pull10:26
smbjjohansen, *sigh* There is always the next thing to do. ;-)10:28
jjohansenhehe, yeah but I want this to make kernel freeze for alpha-3 so ...10:29
jjohansenrebooting10:29
=== ericm|ubuntu is now known as ericm-away
=== yofel_ is now known as yofel
=== cndougla is now known as cnd
=== TeTeT_ is now known as TeTeT
smoserhi. i'm guessing this is known, but just want to make sure if its not already resolved that someone is looking at it:13:28
smoserThe following packages have unmet dependencies:13:28
smoser  linux-virtual: Depends: linux-image-virtual (= 2.6.35.11.12) but it is not going to be13:28
smoserinstalled13:28
smoser, stderr: E: Broken packages13:28
smbsmoser, Hm, it feels like ringing a bell. Though I am not sure whether this was back in a previous release. Could you send this hint to the kernel-team mailing list as a more permanent reminder?13:30
smoserthis is last night maverick13:30
smoserthe uec build fails due to that.13:30
smoseri'll send mail.13:30
smbOk, thanks. The kernel looked like a recent Maverick. It just sounded familiar but usually I am not so much looking at the bleeding edge. :)13:32
smoserit seems to me that its fixed now.13:36
smbHm, maybe that was just a delay in the archive...13:38
smbI would not think that someone had a chance to change anything, yet. Though there are some early birds...13:39
=== bjf[afk] is now known as bjf
smbtgardner, Re: bug to look at: ok. I have also seen some reports of problems with lvm snapshot since the ext4 patches went into updates. I have not yet looked at the other bug but it seems some changes in ext4 might trigger the bad writeback behaviour or even hangs there. Just fyi14:43
tgardnersmb, ack. you should verify the kernel version they are running to see if there is a strong correlation14:46
diwicbjf, ping14:48
bjfdiwic, go14:57
diwicbjf, just wondering what time takashi releases his daily and if your c-o-d's are synced to that14:58
diwicbjf, but it matters less as launchpad is three days behind. :-(14:59
bjfdiwic, don't know and no14:59
bjfdiwic, let me check14:59
bjfdiwic, i just pushed a source package, i'll look at my cron job to see what's up with that15:18
jjohansenogasawara: you about?15:38
ogasawarajjohansen: yep15:39
ogasawarajjohansen: saw both your apparmor and pv-ops pull requests, will review and apply momentarily15:40
jjohansenI am updating specs, and pushing out the pv-on-HVM part of pvops out to next week15:40
ogasawarajjohansen: ack, sounds good15:40
jjohansenogasawara: so the pv-ops one doesn't include pv-on-HVM drivers which is what I am pushing out.15:41
jjohansenI am putting in an explanation of them15:41
ogasawarajjohansen: perfect, thanks15:42
jjohansenogasawara: check your mail15:48
ogasawarajjohansen: ack15:49
ogasawarajjohansen: have you slept yet?15:53
jjohansenno15:53
ogasawarajjohansen: go take a nap15:53
jjohansensoon15:54
jjohansen-> nap17:08
manjonap ? 17:09
manjojjohansen, you ?17:10
jjohansenmanjo: no its my evil twin17:11
manjojjohansen, hmmm that sound more possible :) 17:12
* manjo have never seen jjohansen sleep17:12
tgardnerogasawara, I've a couple of issues with jj's pv-ops pull request. I'll fix 'em up and send you an alternate pull request17:15
ogasawaratgardner: cool thanks, I haven't had a chance to review yet anyways17:16
evWould anyone be willing to add the fix for bug 599569 to the next 10.04 update?  It works perfectly on the two 10.04 systems I've tested it with.17:16
ubot2Launchpad bug 599569 in linux (Ubuntu) "No kernel support for Sierra Wireless 250U 3G/4G USB Dongle (affects: 4) (heat: 24)" [Undecided,Triaged] https://launchpad.net/bugs/59956917:16
tgardnerev, I think the stable team accepts bribes17:17
ogasawaralooks like stable was CC'd on the upstream patch17:17
ogasawaraso should be coming down the queue already?17:18
evname your price and I'll smuggle it it across the border when I go to Debconf tomorrow17:18
smbIf thats true it will come sooner or later17:18
smbI would need to check the queue17:18
smb...17:18
evoh, cool17:18
smbev, There is a USB: Add PID for Sierra 250U to drivers/usb/serial/sierra.c queued for the next (2.6.32.17) release from upstream stable17:21
evhooray17:21
tgardnerev, don't hold your breath. it could still take weeks.17:22
evwell, I've solved the problem for myself (well, my dad really)17:22
evI'd just like to make sure it gets solved for others17:23
evwhich seems to be happing, glacial pace or otherwise17:23
ev:)17:23
=== sconklin-gone is now known as sconklin
cwillu_at_workiotop reports CONFIG_TASK_DELAY_ACCT not enabled in kernel, cannot determine SWAPIN and IO %17:52
cwillu_at_workit's one thing to not know I'm missing out, it's quite another when a program tells me I'm missing out :)17:53
jpdscwillu_at_work: bug #493156.17:55
ubot2Launchpad bug 493156 in linux (Ubuntu) (and 1 other project) "Please enable CONFIG_TASK_DELAY_ACCT (affects: 55) (dups: 3) (heat: 232)" [Medium,Fix released] https://launchpad.net/bugs/49315617:55
tgardnercwillu_at_work, use maverick, or the LTS backport kernel on Lucid17:56
cwillu_at_workokay, thanks17:56
* cwillu_at_work reads the relevant threads17:56
cwillu_at_workalso, is there any way to make swap act... sane? by which I mean:  at the end of my day, everything is running smoothly:  gedit, inkscape, firefox, they all run without hiccups.  However, at the start of my day, I have to wait 5-15 seconds at fairly random intervals as those applications hang, presumably swapping back in on demand18:16
cwillu_at_workcase in point:  inkscape just spent the last thirty seconds paging in at around 1mb/s;  however there was no disk write traffic at the same time:  to me, this implies that memory is available, but the allocation policy is buggered18:19
tgardnerogasawara, I reset jj18:26
tgardnerI resent jj's pull request18:26
ogasawaratgardner: ack.  my tree is somehow corrupted at the moment so am getting it fixed back up first18:27
bjfanyone else notice the daily live isos have not changed for 3 days?18:27
tgardneroh, bummer. bad disk, or fat fingers?18:28
ogasawaratgardner: I have a feeling it might be the disk :(18:28
tgardnerogasawara, back to the trusty, but ancient, laptop18:29
sconklinI'm bailing for the rest of the day, I feel terrible. See you Monday!18:36
=== sconklin is now known as sconklin-gone
* tgardner lunches18:53
* ogasawara runs errands, back in an hour19:40
* bjf taking a headache break21:10
=== bjf is now known as bjf[afk]
jjohansenogasawara: hrmm, what happend that my patches needed to be reset?21:48
ogasawarajjohansen: looked like just minor commit log changes that tim made a small tweak with updateconfigs21:49
jjohansenhrrm, I'll poke at what he did, I did do a rebase that did a 3 way merge, perhaps that did something?21:50
jjohansenogasawara: is there anything you would like help on, or should I dive back into pv-on-HVM drivers?22:00
ogasawarajjohansen: nope, looking good on my side.  just waiting for test builds on ports to finish before I push all the bits22:00
jjohansenokay22:01
manjojjohansen, you did not sleep enough 22:10
jjohansenmanjo: ? really I thought almost 5 hours was pretty good22:11
jjohansenwell 4.5h maybe22:11
manjojjohansen, good for 4.5 days atleast! 22:12
manjo:)22:12
jjohansen:)22:12
manjoadios amigos .. have a good weekend22:21
cwillu_at_workWhat's the trick to run make-kpkg in a freshly clone'd git repo?23:26
cwillu_at_workIt's complaining about a missing files list file, which I can't imagine I have to actually build by hand23:26

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