[02:04] back on later [02:39] hey folks, i'm trying to track down an acpi regression from maverick to natty. i'm a bit clueless with the ubuntu kernel tree - if i clone the natty repo, will it have all the older maverick stuff too? [02:57] The upgrade from 2.6.32-31 to -32 (lucid, x86-64) broke my LXC subsystem. Containers didn't start at all; IIRC lxc-start(8) failed to mkdir in /var/cgroups (which was mounted). [02:58] Before I wade through all the LP notes in the changelog.Debian, anybody know offhand what's going on? [02:58] Unfortunately I can't give you exact details because the machine in question is mission-critical and I already failed back over to -31 when the issue occured on Sunday. [03:01] twb: first I've heard of it but I would very much like for you to file a bug if you can. Especially if you can do it from a machine with the same hardware running that particular kernel [03:01] Looks like you disabling CONFIG_NET_NS completely breaks containers with different network ranges. So, uh, good job. [03:02] Not exactly the way in which we like to overachieve [03:03] And it was done to fix a bloody FTPd. I hate FTP (http://mywiki.wooledge.org/FtpMustDie). [03:03] twb: if you file bugs (and please do), then please also add the tag "regression-updates" [03:05] OK, I'm doing that now [03:06] What is the package name to report against? [03:06] thinking [03:08] file against linux (the kernel) Even if that's wrong it'll get the logs we need. [03:08] and post the bug number(s) here. I'll check back a little later, I'm in the middle of cooking dinner . . . [03:08] Will do [03:18] Am I allowed to set "importance critical" ? [03:20] Too late, I did [03:22] When malone emails me back, I'll cite the bug number [03:42] It hasn't emailed me yet :-/// [03:43] You haven't set the “don't email me for stuff that I do” switch, have you? :) [03:43] AFAIK I haven't touched that setting [03:43] I have used the lp web UI long enough to tell it my GPG key, so I can use email like the Goddess intended [03:55] update-initramfs: Generating /boot/initrd.img-2.6.32-32-server [03:55] Running postinst hook script /usr/sbin/update-grub. [03:55] Generating grub.cfg ... [03:55] /usr/sbin/grub-probe: error: raid5rec is not loaded. [03:55] User postinst hook script [/usr/sbin/update-grub] exited with value 1 [03:55] why does the kernel hate this VM so? [03:56] lamont: is the VM using md RAID? [03:56] why yes, yes it is [03:56] more to the point, halp how make better? [03:57] I have no idea. I am so sick of grub that I use syslinux for my (md RAID1 /boot, md RAID5 LVM root) systems now. [03:58] Last time I used grub2 it didn't support auto-detecting (grub-probe) anything inside the md RAID segment at all, and I had to manually tell it which modules to compile into the MBR. But that was around 2007. [03:59] There is also a #grub which might help [03:59] cool [03:59] thanks [04:04] j04 [04:04] 11 [04:04] sconklin, RAOF: FYI #launchpad are looking into malone boo-boo now. [04:05] twb: sweet, Cascading failures [04:06] I was going to tell you about an even better one, but I've repressed it and can't remember [04:14] twb: I have to knock off so I can be all fresh and rested and ready to play whack-a-mole with regressions tomorrow. please subscribe me to the bugs when you get them created, or email me the numbers. Or ping me tomorrow during the day central time. [04:14] What's your email address? [04:15] Thanks, I'll email you if I can't subscribe you directly. [04:15] Good night. [04:15] twb: in PM (although it's not a secret) [04:15] I actually probably signed that kernel that failed [04:16] sconklin: before you go, could you look up at my git question? [04:16] looking [04:16] thanks :D [04:16] hypatia: Hi! [04:17] hiya! [04:17] so - good question [04:17] No [04:17] i've been bugging poor matt garrett about this for 2 weeks [04:17] and finally pinned it on a kernel regression [04:18] by which i mean "works in maverick, not natty" [04:18] We start each new series from whatever the upstream kernel is at the beginning of the cycle. So while it will have a lot of the same history, Maverick (in this case) will have had selective patches applied for fixes and from upstream stable, and so will differ from the new upstream [04:19] do you have an idea which file or subsystem it's in? [04:19] (I'm asking so I can steer you to the right git usage) [04:19] sconklin: it only boots with acpi=off [04:20] that's about all i've got :/ [04:20] oh, that's pretty generic [04:20] hp laptop, insyde bios [04:20] same deal with hdd or ssd [04:20] 10.10 x86 and 64bit both boot fine with acpi on [04:20] url? [04:21] I hate to do this to both you and Colin, but cking is absolutely the most knowledgeable person on our team about this sort of thing [04:21] ohsix: for the laptop? [04:21] for whatever report on it, i've got some stuff with insyde around [04:21] and Colin has been working on a firmware test suite that smokes out a lot of BIOS issues [04:21] sconklin: no skin off my back, i'm happy to learn something new and help out if i can :) [04:22] ohsix: i'll ping you when i file an issue on it tomorrow [04:22] must do CS homework tonight, was just hoping to get some kernels building :) [04:23] hypatia: there's also a lot of interesting info on Colin's blog [04:23] have you done any bios updates recently? even if you haven't tried resetting the bios options to the default (fine if you make your changes over after that) [04:23] http://smackerelofopinion.blogspot.com/ [04:24] it's fixed the weirdest stuff on mine; theres a binary option block that can be changed by all sorts of things you can't actually see with the bios setup, but the reset will still reset them all [04:24] ok, g'night, gotta fall over now. [04:24] ohsix: i am running the latest bios which is from jan 2011... but it's been modded to remove the pci-e whitelist for wifi cards [04:24] stupid hp consumer crap [04:24] ohsix: i'll give the resetting thing a try [04:24] you might start looking there since it's modified [04:24] * hypatia nods [04:25] i can try reloading the old bios too, i have it somewhere [04:25] i don't like those whitelists either; but they do it for testing w/the fcc from what i understand [04:27] yeah, i kinda think it's bull - like how broadcom etc. used to refuse to make Free drivers on the grounds that they would be modifiable and thus allow the user to get around fcc approvals [04:27] sconklin: #796993 [04:27] [04:29] broadcom is kind of special; they have a lot of customers and it's all proprietary stuff, any part of the drivers being opened for some products can compromise their market position too [04:33] and then my laptop died, because i have no acpi and i forgot i was unplugged [04:33] ;_; [04:33] ^^ motivation [04:59] Does the latest Oneiric kernel work well with the nvidia driver? It seems to not build when it encounters the version number 3.0.0 [05:01] the version change is still percolating through just about everything that touched the kernel [05:01] aside from that i don't know :] [05:02] ohsix: so it should work well? It failed to build when I tried using the kernel from the mainline ppa 2 weeks ago [05:03] couldn't say; i don't use it, its probable that there are things other than the version number that are keeping it from being built, binary drivers are like that [05:05] * cdbs takes a ball of garbage and throws it on the nVidia banner hanging on the neighbour's house [05:08] cdbs: ahaha, third-party FTL [05:08] you could try the nouveau driver :] [05:08] It's a pity Intel moved their GPUs from the motherboard to the CPU, because now I can't buy high-end servers with intel GPUs [05:08] (The on-CPU GPU is only available on low-end CPUs) [05:10] ohsix: Nouveau doesn't work with a new GPU released last month :@ [05:10] the disadvantage of buying stuff that's too new [05:14] ohsix: btw, it built with the new kernel [05:14] in dkms, of course [05:14] but that's maybe because I'm using xorg-edgers? [05:15] maybe, it tends to get updates for new stuff pretty quick [05:24] ITYM the disadvantage of buying hw from a FOSS-hostile vendor === jk-- is now known as jk- === fairuz_ is now known as fairuz [07:59] * apw yawns [08:32] * smb joins yawning [08:57] smb, moin [08:58] bjf, Morning visitor [09:00] ## [09:00] ## Ubuntu Kernel Team Meeting - Today @ 17:00 UTC - #ubuntu-meeting [09:00] ## agenda: https://wiki.ubuntu.com/KernelTeam/Meeting [09:00] ## [09:01] ## [09:01] ## Ubuntu Kernel Team Meeting - Today @ 17:00 UTC - #ubuntu-meeting [09:01] ## agenda: https://wiki.ubuntu.com/KernelTeam/Meeting [09:01] ## [09:03] sconklin: Yes, assign all linux-mvl-dove, linux-fsl-imx51, linux-ti-omap4 to me, or rather my new team, ubuntu-armel-qa. [09:03] thanks [09:07] bjf, Wow a very early early warning... [09:42] * apw watches the load on people hit 30, i suspect its underpowered [09:43] * smb tries to parse above sentence and fails... [09:43] people.c.c machine load [09:43] ahhh [09:43] long term load ave >15 [09:43] Now it makes sense [09:44] I was just not prepared to look at people as the machine, but as people... :) [09:49] doesn't pitti run scripts there? [09:56] everyone and his dog does, which is why its cying in the corner [10:32] apw, May I point you to another mail in my watch list? "Re: problems on CAN bitrate" This is about a config option that is apparently needed for that driver, which is not experimental, and still disabled in Oneiric. [10:43] Hello, is there a utility (like PCITree for windows www.pcitree.de), that would allow me to set configuration registers of a PCI device ? [10:45] pciutils ! [10:45] sorry for the noise [11:12] apw, the flavours page has nothing for oneiric [13:07] ogasawara: apw, bugs with two digit problems are tagged "kernel-3.0" [13:11] bjf awsome [13:11] ogasawara, about ? [13:11] ogasawara, i am thinking i'd push a config change and wanted to make sure you not rebasing atm === Quintasan_ is now known as Quintasan [13:26] ogasawara, ok all done [13:26] smb, changed [13:27] apw, CAN_BITRATE? [13:27] smb, yep [13:27] smb, email in your inbox [13:28] apw, Cool thanks. Well hopefully through mailing list. :) They had been querying about earlier releases too, but as they lack any response they seem not have that problem urgently enough [13:29] someone just raised a bug against 3.0.0 kernel, should it be Invalid? [13:30] brendand, bug #? [13:30] bjf - i was asking more in principle, but it's https://bugs.launchpad.net/ubuntu/+bug/797141 [13:30] Ubuntu bug 797141 in ubuntu "Can't mount NFS shares under kernel 3.0.0" [Undecided,New] [13:31] brendand, they are valid [13:31] bjf - okay [14:00] hey folks, any advice on doing a bisect between maverick and natty kernels? i have them both checked out [14:00] the bug was introduced somewhere after Linux butler 2.6.37-020637rc2-generic #201011160905 SMP Tue Nov 16 09:08:47 UTC 2010 x86_64 x86_64 x86_64 GNU/Linux [14:46] hypatia, if the problem is exposed in the mainline builds as well i would use those to narrow it down [14:48] i'm bisecting from Ubuntu-2.6.36-0.1 [14:48] compiling right now, woot [14:48] * hypatia hasn't built a kernel in a dog's age [14:49] brendand, if you find any could you add kernel-3.0 as a tag please, so we can monitor them [14:50] apw - will do [14:50] hypatia, all the uploads we ever did are available in the launchpad librarian [14:52] hypatia: the reason apw is suggesting that is because the mainline builds have a more linear history, and are easier to bisect, or at least to narrow it down to two adjacent versions. [14:52] and because the all exist in one place [14:53] * ogasawara back in 20 [14:56] apw, I've completely forgotten the rune to build a kernel .ddeb - what is it again? [14:57] apw: oh nifty, thanks [14:58] cking_, I believe you look for fdr skipdbg=false [14:58] you need to have full_build=true [14:58] or skipdbg=false ... [14:59] oops, even more options... [14:59] ah - thanks, I used the latter last time and couldn't recall it today [14:59] smb, more? === cking_ is now known as cking [15:00] cking, I forgot/did not remember full_build [15:00] so many ways to skin a cat [15:00] * smb tried so say there are even more options to do the same [15:01] apw: do we have a wiki page describing lts backport kernels, why they exist, and how to apply them? [15:02] sconklin, i am not sure we have that ... let me looks [15:18] apw: any suggestions on how to use launchpad librarian? i've not used it before [15:19] hypatia, you can find all the versions ever here: https://launchpad.net/ubuntu/+source/linux/+publishinghistory [15:19] oh rock on [15:19] thanks! [15:19] click through for the binaries [15:21] bjf: I'm only seeing one bug tagged kernel-3.0, does that match what you're seeing? [15:26] will have to wait til i get to school and off of spotty 3G [15:36] ogasawara: yes, it was the first [15:37] bjf: ok cool. didn't know if you had some crazy script parsing logs to auto-tag in which I would expect to see more bugs [15:38] ogasawara: finger script === charlie-tca is now known as charlie-tca1 === zul_ is now known as zul [16:21] hi === fairuz is now known as fairuz_ [16:24] since I'm using 11.04 on my laptop, my computer freezes sometimes when it's going to the screensaver and when it's doing that, i can only reboot or restart gdm. Do you know how I can fix that ? it's really annoying to have to restart my computer each time i want to make a break :] [16:27] YoBoY: Are you using one of the closed video drivers? [16:27] abogani: no [16:28] YoBoY: What video driver are you using? [16:29] I have an intel chipset, I think it's the i915 [16:30] YoBoY: Ok try to disabling effects (at GDM login time on Session widget) [16:30] ## [16:30] ## Kernel team meeting today @ 17:00 UTC [16:30] ## [16:30] (my laptop have an hybrid graphic cards intel+nvidia, currently using the intel graphic card) [16:31] abogani: starting with a classic desktop ? [16:32] abogani: done... I have to wait now ^^" [16:32] YoBoY: good luck [16:34] ogasawara: i think the meeting is yours :-) [16:34] bjf: ok cool [16:41] * ogasawara stares at -rc3 build.log and wonders why PERF_VERSION is not getting set properly now [17:09] ogasawara, what is it getting set to [17:09] there were changes in -rc3 which changed that versioning [17:09] ie. how it propogates down [17:10] it was lone of the first commits puled in since the previous -rc [17:10] apw: I flipped on some debugging, lemme pastebin [17:10] apw: http://pastebin.ubuntu.com/626658/ [17:12] ogasawara, so perf-version looks ok in your output doesn't it [17:13] following a suggestion I got here yesterday, I am attempting to perform a kernel bisection to further debug bug 793437 [17:13] Launchpad bug 793437 in linux "Unusable Slowness In 2.6.38-8" [Undecided,Confirmed] https://launchpad.net/bugs/793437 [17:13] PERF-VERSION-FILE:2: *** missing separator. Stop. [17:13] ogasawara, that is a Make error right ? [17:13] apw: yep [17:14] is the branch pushed ? [17:14] apw: not yet, lemme push [17:14] anywhere is fine [17:14] I know that my bug is not present in the stock 2.6.35-28 kernel that is current in maverick, but I am having through figuring out how to turn that into a tag from which to perform bisection [17:16] apw: git://kernel.ubuntu.com/ogasawara/ubuntu-oneiric.git master-next [17:16] s/through/trouble/ [17:22] ogasawara, oh i think i see whats broken there [17:22] ogasawara, its assuming its within the kernel tree to allow it to work out the kernel version ... [17:22] which it isn't ... let me poke at it for a sec ? [17:22] apw: sure [17:26] ogasawara, ok so i think for now revert this one: 5d61b9fd19d9f3cf653dbba615876e7792eea5ea [17:26] and see if that sorts you out for now, and i'll see what i can do to sort it out [17:26] apw: ack, /me tests [17:32] apw: cool, working now [17:42] Has the discussions ended? [17:45] Is the Power usage still an issue? Is it been looked into? [17:48] a lucid uname gives me: 2.6.32-24-jump1.01 - does it look familiar to someone? [17:48] db260179: yes it is still an issue. There has been a patch that reduces the problem but it isn't entirely solved yet [17:48] pmatulis, Looks like a self compiled kernel (you can place a lot there) [17:50] jjohansen: Thanks for the reply. What is causing the issue? [17:50] db260179: if we knew that for sure it we be fixed [17:51] db260179: though it isn't just a single patch, it seems to be an interaction of at least 2 making it much harder to track down [17:52] apw: do you know what I can use instead of (un)lock_kernel using Linux 3.0? [17:53] tseliot, add a driver specific mutex [17:55] apw: the main problem is that I'm dealing with fglrx [17:55] tseliot, hee [17:56] tseliot, make lock/unlock take a my_bkl_replacement mutex ? [17:56] tseliot, and pray it only wants to exclude itself [17:56] apw: right, but I was wondering what the replacement could be [17:57] tseliot, there is nothing, its gone [17:57] eveyrthing was converted to driver specific mutexes [17:58] apw: oh, so are you suggesting that I point that to an empty function? [17:58] tseliot, no make a new 'global' mutext called 'ati_lock' and make lock/unlock_kernel take and release it and hope its enough [18:00] ## [18:00] ## Meeting starting now - #ubuntu-meeting [18:00] ## agenda: https://wiki.ubuntu.com/KernelTeam/Meeting [18:00] ## [18:00] as in make local functions called that [18:00] apw: ok, thanks [18:01] tseliot: if it isn't you could look at the most recent patches to remove the bkl, and try taking a subset of those locks [18:02] jjohansen: yes, I saw that some drivers now use lock_sock() but it won't work in my case. I'll keep looking [18:03] tseliot: right, but there may be some other lock, or small set of locks that will [18:04] jjohansen: from a quick search I've noticed that other drivers are simply using mutex_unlock() [18:04] ogasawara: patch review done, basically keep all the patches. Though the AA patches are going to get an update, but that will happen when I finish up and get the pull request together. I'll update the wi [18:05] jjohansen: ack, thanks === jburkholder1 is now known as jburkholder [19:12] apw, jjohansen: using a mutex seems to work here. I even got Unity to work :) . Thanks for your help [19:14] tseliot: I'm not surprised it works at some level, its whether it has some odd failures when interacting with the console, suspend, etc, that are worrisome [19:14] tseliot: none the less, its good to hear you are having success [19:15] jjohansen: yes, but it should be better than having no working driver at all (when there's no alternative) in oneiric [19:15] tseliot: indeed :) [19:15] hopefully AMD will work on a proper fix soon [19:15] yeah === yofel_ is now known as yofel [20:45] bjf, sconklin - can you paste me the link to the calendar you're using to track SRU spins? === PascalFR is now known as boxerjaune === Quintasan_ is now known as Quintasan [23:48] following advice from diwic, I've spent all day bisecting the ubuntu-natty kernel to debug 793437, and feel that I have accomplished extremely little [23:48] (the last 6 revisions that I have tried all failed to build, for a variety of reasons) [23:48] what is the likelihood that continuing with this is just a waste of my time?