[00:00] Heh [00:00] machine freezing upon gdm start [00:00] I _think_ that's fixed. [00:02] But I can see why you might not want to play around with it :) === bjf is now known as bjf-afk [00:13] dtchen: Have I managed to send my mail to ubuntu-x@ & kernel-team@ using the right email address this time? :) [00:15] both posts came through to kernel-team@ [00:15] I'm no longer subbed to ubuntu-x@, can't comment on that without looking at mailman archives [00:15] Urgh. Sorry. [00:15] Well, kernel-team@ will do. [01:10] I know everything! [01:10] Ban me! [01:11] !ops [01:11] Help! lamont, zul, T-Bone, mdz, or jdub [01:12] !ops [01:12] Help! lamont, zul, T-Bone, mdz, or jdub === MTecknology is now known as ALearing === ALearing is now known as ALearing_ === ALearing_ is now known as AdamLearing === AdamLearing is now known as MTecknology [03:15] i am on lucid,why it not run update-grub automatic? [03:25] wzssyqa: /etc/kernel-img.conf probably lacks these lines? [03:25] postinst_hook = update-grub [03:25] postrm_hook = update-grub [03:25] if so, there's a bug about that [03:26] lamont: do_symlinks = yes [03:26] relative_links = yes [03:26] do_bootloader = no [03:26] do_bootfloppy = no [03:26] do_initrd = yes [03:26] link_in_boot = no [03:26] yeah, so workaround is to add the postinst and postrm hook lines... [03:27] lamont: o,ok,need report a bug? [03:28] https://bugs.edge.launchpad.net/ubuntu/+source/grub/+bug/470265 [03:28] Malone bug 470265 in grub "jaunty to karmic upgrade failed to update menu.lst (update-grub missing from kernel-img.conf)" [High,New] [03:28] feel free to add more detail if you can [03:29] lamont: o ,it happens when update from 2.6.32-6 to 2.6.32-7 [03:31] well, the issue is that the hooks aren't there, so installing a kernel isn't told to run update-grub [03:32] i am having a problem with karmic (also same problem with jaunty) where kde freezes and on the serial console and syslog there are various Call Traces. Caps/Lock lights/unlights when pressed and I can do the magic sequence to reboot. while its initially frozen I can still ping from another machine and usually at some point afterwards the serial console and ssh becomes responsive so i can log in and then do a clean reboot. i have s [03:33] lamont: when install a new kernel,it will update /etc/kernel-img.conf? [03:35] I don't believe so [03:35] as in you'll want to manually add those lines to kernel-img.conf [03:35] and then installing a kernel will run update-grub [03:35] the bug is around "why did this go away?" [03:35] lamont: why it work well before? [03:35] I haven't been following it that closely, other than being affected by it. [03:35] *shrug* [03:36] (if we knew, then the bug would be fixed already) [03:36] lamont: haha [03:37] lamont: i run dpkg -S /etc/kernel-img.conf,it display noting [03:38] no. edit the file /etc/kernel-img.conf as root, add those two lines at the end of it [03:39] lamont: i want to ask ,how the file produce [03:45] I haven't dug into it at all [03:46] lamont: o ,thx === saispo_ is now known as saispo === ghostcube_ is now known as ghostcube [09:53] ogasawara: I went ahead and fixed the bug reporting instructions for linux [09:57] Keybuk, there are two DEVTMPFS options, which ones do you need? [09:58] DEVTMPFS & DEVTMPFS_MOUNT [12:48] apw: I'm about to switch my dev box kernel to lucid. Any gotchas you know about? [12:53] it may eat your kitten [12:54] nothing serious [12:54] artir: kittens are overrated. Good riddance! [13:00] :D [13:01] i'd wait till alpha1 [13:01] which is this thursday [13:39] amitk, i updated one of my crashies to it yesterday and its in one piece so far [13:39] the kernel seems fine, am on it on here. though i have no sensors [13:42] apw: thats good enough for me. karmic userspace/lucid kernel on its way. [13:43] i have a couple of boxes in that configuration, the only oddy so far is the loss of sensors on one lappy [13:43] i am suspicious i have i915 hangs in there, yet to be proven [15:19] apw: both options are sane [15:20] DEVTMPFS_MOUNT only affects people who don't use an initramfs and gives them a ready-populated /dev [15:20] it makes init=/sbin/sulogin work particularly well [15:23] Keybuk, ok will enable both then :) [15:23] Keybuk, pgraner is saying that his machine loses KMS if he doesn't disable vga16fb [15:23] (disable == blacklist) [15:23] yeah, I found it odd that vga16fb stays loaded [15:24] I had expected it to fail to load as the driver was already taken [15:24] Keybuk, any suggestions as to resolution? [15:26] Keybuk: bug 494062 [15:26] Malone bug 494062 in linux "KMS Intel Fails with Lucid Kernel" [Undecided,New] https://launchpad.net/bugs/494062 [15:30] apw: find out why you get a framebuffer even when vga16fb fails to bind to a device [15:31] Keybuk, what i don't quite get is i don't get the same behaviour no karmic userspace (even with vga16fb) disabled [15:31] enabled eve [15:33] Keybuk, it makes no attempt to try and confirm it can get the device [15:33] weird [15:33] maybe it's because the vga16fb doesn't claim devices [15:33] but writes to some special vga bios thing? [15:34] i think it just used the vgastate module to handle things [15:34] its entirly possible if you opened it it would work even with kms enabled [15:35] I guess we'll have to go with userspace trickery for this one then [15:35] Keybuk, what i need to understand is why it works fine in karmic userspace [15:35] for me its completely consistant [15:35] hmm [15:35] lucid works fine for me to btw === bjf-afk is now known as bjf [15:35] am updating to there to test there now [15:36] hrm ... what i don't see on karmic userspace is the double fb loaded thing [15:36] if i have modeset i just get the kms one [15:36] apw: bug 494066 for the resume issue [15:36] Malone bug 494066 in linux "Resume from Suspend has no videowith Intel graphics on Lucid" [Undecided,New] https://launchpad.net/bugs/494066 [15:36] if i don't i get the vga16fb as well [15:36] pgraner, thanks ... [15:36] apw: I'll get the rest of the data in the bug, apport-collect just crashed and won't upload anything right now [15:36] heh nice [15:37] Keybuk, what userspace trickery would you use ... [15:37] something based on the uevent still? [15:37] hmm [15:37] on lucid, if I resume from suspend, I get a black screen [15:37] (backlight doesn't switch on) [15:38] and that's without vga16fb loaded [15:38] Keybuk, that sounds like other people's whinge, and pgraner confirmed blacklist of the vga16fb didn't help either [15:38] Keybuk, the same kernel seems to work for me on karmic userspace [15:38] Keybuk: just filed 494066 on that issue [15:38] right, so not the same thing [15:38] huh [15:38] that's odd [15:38] yep, i think so too [15:38] wonder whether some old karmic userspace was doing something [15:39] well quirks have potentially just stopped working haven't they [15:39] as part of pitti's halsectomy [15:39] yes... but shouldn't *have* quirks [15:40] we're in a world of hurt :) [15:40] yay upgrade wants to remove libc6, that should be fun [15:41] so I guess a plan B will be to load vga16fb manually in userspace [15:42] rather than relying on udev auto-loading [15:42] :-( [15:42] Keybuk, can you trigger that from the uevent still ... [15:42] hello,my kernel boot arguments are CONFIG_CMDLINE="console=ttyS0,115200 mem=64M root=1F02 rw".My question is,what does the "root=1f02" mean? [15:42] apw: yes, I think so [15:42] anyone help me?please [15:43] I was thinking, for pci 0x03 devices (graphics cards), if we don't load anything by modprobe then load vga16fb [15:43] i thought root= was your root paritition, its possible its allowed to be a major,minor encoded as hex [15:43] yep, thats what the modalias does [15:43] match wise [15:47] Keybuk, so is that a 'drop that vga16fb modalias patch?' [15:49] if it's causing problems, yes [15:49] I think we should attempt a few hours of honest debugging first [15:49] but we can do this a different way if needs be === thunderstruck is now known as gnomefreak [16:00] Keybuk, it can be worked round it seems with i915.modeset=1 so i don't think its a disaster, any more than the suspend not working is [16:00] its possible today to backport lucid kernel to karmic (or use the same package) or its better to use http://kernel.ubuntu.com/~kernel-ppa/mainline/ ? [16:00] leleobhz, depends what you are trying to achieve, whats in the lucid kernel that you need [16:01] apw: i generally dont use mainstream kernels, but i want to test energy management and intel performance increase in my netbook and notebook [16:02] i have a dell inspiron 9 and a intel centrino notebook [16:03] ive read these things have improved a lot in 2.6.32 [16:06] apw: what you think about? [16:10] leleobhz, depends what you are trying to achieve, whats in the lucid kernel that you need [16:11] apw: not specifically to lucid kernel, as kernel-ppa have the mainline packages [16:12] i just want to know the best way to have the kernel 2.6.32 and if possible, be updated about new build of .32 [16:13] there is nothing out there right now that will get you updates like that no [16:13] apw: i thing i ill get a little speedup and a better energy control (i have a lot of issues with power consuption) [16:13] mmmm [16:13] so make no difference use lucid kernel in karmic and use mainline from kernel-ppa? [16:17] ** [16:17] ** Ubuntu Kernel Team Meeting - Today @ 17:00 UTC - #ubuntu-meeting [16:17] ** [16:19] mdz: the '-p' option is there since you need it Hardy - I was planning on removing it in April. [16:25] apw: ? === bjf is now known as ___bjf [16:26] leleobhz, the lucid kernel in karmic is going to be more likely to have ubuntu things enabled yes, though its in a lot of development instigated instability right now as we head into alphas [16:34] apw: ok, but comparing http://kernel.ubuntu.com/~kernel-ppa/mainline/ to actual lucid kernel package, what of them can be more stable / less broken? [16:34] or is the same thing [16:34] well the 2.6.32 package there will never change, its a one time deal [16:35] (the idea is: i dont know the diference from 2.6.32 from lucid and from mainline kernel) [16:35] our lucid kernel changes as i change it to add features etc for the release [16:35] the lucid kernel tracks with the lucid userspace for which its designed [16:35] apw: so mainline is more approach to vanilla kernel? [16:35] it may work with karmic, and indeed many of the kenrel team run it that way, but its a lucky dip [16:36] yes the mainline builds are literally mainline commits built with our config [16:36] i know its designed for lucid, but i think i should get a chance, because improvements in power saving, ssd handle and intel improvements is a very interesting thing [16:36] hmmm [16:37] understood [16:37] i think so ill try lucid kernel [16:37] neither will be idea with karmic userspace [16:37] ? === hypera1r is now known as hyperair [16:38] apw: but at point to broke things and kill my dog? [16:40] s/idea/ideal/ [16:40] its not something we plan for, so if it does break you get to keep both pieces :) [16:41] :] [16:41] apw: i think i shoud give a try [16:41] its you that was published in wiki some tests with dell mini 10v? [16:41] i am running it here right now, its 'ok' considering the time of the cycle [16:42] a number of us test on that as a 'reference platform' for comparison [16:42] but performance and energy consumption is visible from .31 to .32? [16:42] apw: because as i said, ill test it also in my inspiron mini 9 [16:43] not something i am focused on directly, interested in boot performance mostly at this point [16:43] and getting new features such as radeon kms enabled [16:58] apw: boot performance is good too. and i use ssd, so i think i can get a great improvement [16:58] but the idea is make things with intel vga faster and reduce energy comsumption [16:58] will test lucid kernel [16:58] apw: thanks a lot for you time [18:05] Does John Johansen lurk these parts? [18:06] yes he does, as 'jj' [18:07] Good good [18:07] I believe as jjohansen, but does not seem to be around [18:07] quest linux-lucid% sudo ARCH=i386 pbuilder --build ../linux_2.6.32-7.11.dsc [18:07] I think I'll hang here till I happen to see him... I found a bug in his mainline apparmor patch [18:07] ... I wonder whether that will work [18:25] I'd like to set apparmor=0 for the live CD kernel command line moving forward, given that the network-interface-security upstart job causes apparmor to run, which we definitely don't want on the live CD. [18:25] Does anyone have an objection to that? [18:26] I should note that while I could just remove that upstart job inside casper, I'd ideally like to guard against any job on the live CD from causing apparmor to run. [18:26] jjohansen, Ah, just the man I was looking for. :P [18:27] hi [18:27] jjohansen, I was the one who just sent you that email re: apparmor mainline [18:28] okay, just give me a few seconds while I bring up my mail [18:28] No worries. [18:28] I'm gonna go have a smoke anyway. brb [18:29] syn-ack: when did you pull the kernel? [18:29] jjohansen, first time about 12 hrs ago and just an hour ago [18:30] okay, just an hour ago should be good, give me a few minutes to look at this [18:31] 12 hrs ago was probably bad as it has been really unstable with lots of updates being pushed in [18:34] Will do. [18:49] Ah, the update just showed up on gitweb, jjohansen... lemme give this one a shot [18:49] syn-ack: just wait a minute there is something wrong with that update [18:49] ack. :/ [18:52] syn-ack: give it a try now [18:52] ko [18:59] jjohansen, Give me a bit as I'm dealing with my baby and this resync is kinda slow going ATM [19:00] syn-ack: np [19:09] jjohansen, Building now. [19:15] jjohansen, whats the status of the AA update, the meeting implied it was ready? [19:16] apw: I am having success building it, and ran through testing. [19:16] syn-ack: reported a problem with his building and he is trying again right now [19:16] jjohansen, No go on my end, it left me with this nice error again: http://pastebin.com/d60d5e17f [19:17] hrmm, strange give me a minute [19:17] k [19:18] apw: OOI, which magic do you use to cut the generated kernel packages down? [19:18] apw: so apparently there is some build breakage under some circumstances [19:18] I know about flavours=generic, but how do you get rid of things like source, libc-dev, doc, udebs, etc. [19:18] Keybuk, cut them down? [19:18] so the build doesn't [19:18] jjohansen, this is what I'm using to build: CONCURRENCY_LEVEL=3 fakeroot make-kpkg --initrd --append-to-version=-laptop-custom-5 kernel_image kernel_headers [19:19] oh when i build i generally use like fakeroot debian/rules binary-generic [19:19] ah [19:20] yeah that doesn't work with pbuilder ;) [19:20] syn-ack: got it and replicated [19:20] ah ... that i can imaging ... [19:20] jjohansen, Good deal [19:21] jjohansen, so this is likely only going to affect 'mainline' users not me? [19:22] apw: this should only take a couple minutes to fix [19:22] its just an include statement not finding its file [19:22] Thats what I figured [19:23] strange that it works building one way and not another [19:27] What are you using to build with? [19:46] apw: I think it's funny that I instinctively bump the kernel ABI every time now [19:46] JUST IN CASE [19:48] syn-ack: give it a try now [19:49] standby [19:49] syn-ack: I had just been using make O= [19:49] ah [19:49] That is kinda funny that it wouldnt build then [19:49] it was a stupid little bug [19:50] yeah, strange that it would work with make O= but not make [19:50] jjohansen, "ITS ALWAYS SOME MUNDANE DETAIL" [19:50] heh [19:51] yeah mundane love to trip me up [19:51] [19:52] man, I'm still getting used to these Blinkenlights on my wifi killswitch [19:53] See, the whole reason I went to roll my own was to include some of the HP WMI BS so I could use of my hotkeys and my status indicator on my killswitch working right... well now it works, it just flashes with my TX/RX [19:58] Keybuk, heh its a way forward for sure [20:46] jjohansen, alright, thanks again, completely built with no errors [21:50] Keybuk, i assume our initramfs will wait for the root filesystem source device to appear [21:59] apw: you removed nfs kernel support for lucid? [21:59] amitk, nope [22:00] there is a bug in the userspace, the fix was literally just uploaded about 1m ago [22:00] bug # [22:00] bug #493145 [22:00] Malone bug 493145 in nfs-utils "[Lucid] NFS kernel server doesn't work anymore with 2.6.32" [Medium,Fix released] https://launchpad.net/bugs/493145 [22:01] apw: yes, that's basically what it does [22:01] give it an hour for the publisher to run and you should ahve the fix [22:01] Keybuk, ok, just reviewing some patches we have heaped up for x to test to see what may help, now i have a nice way to compare [22:02] apw: but I don't run lucid userspace, or is this a bug due to lucid kernel and karmic userspace? [22:02] yes, its a bug in all userspace with 2.6.31 -> 2.6.32 kernel upgrade [22:03] I'm sorry I couldn't be here at 4am for the kernel meeting; I'm interested in understanding what's happening wrt nouveau so I can work out what to do with the DDX. [22:04] apw: will this be sru'ed to karmic (probably not) [22:04] RAOF we are still reviewing the heap of nouveau delta to work out what we can do safely with it [22:04] likely it will be offered as an optional addon [22:04] amitk, no one has asked, but its a trivial one liner you can apply by hand, the patch is on the bug [22:05] apw: Ok. Did you notice my recent mail to the kernel-team list, whereby just dropping in the nouveau directories & exporting 4 additional symbols from ttm works on the existing lucid kernel? [22:05] (ie: the only change to shared code is exporting 4 symbols from ttm.ko) [22:06] RAOF, sorry not seen it as yet, that sounds like a damn fine step forward. that might make it possible to merge them which would be more approariate [22:06] what was the subject line? [22:06] Re: [ubuntu-x] Nouveau information and proposed plan [22:07] I've also got a git branch at http://cooperteam.net/kernel-lucid-nouveau.git with a nouveau-scratchpad branch that's got all the work to build nouveau in the lucid kernel. [22:07] (That's in the email) [22:08] RAOF, that is excellent [22:10] RAOF, its very late here, i'll have a look at it in the morning, but that you for the heads up [22:10] No problem. [22:36] jjohansen, If you have any new additions and stuff like that you'd like me to test and you don't see me in here, just hit me up on the email since you have it now. [22:37] On the mainline that is [23:00] syn-ack: will do [23:03] woohoo jury duty's been rescheduled [23:03] God Love the courts