[00:55] ogasawara: is there any new kernel for me to test on this Geode LX? :) [00:56] Q-FUNK: building it right now [00:56] ah, ok :) [00:56] I would give it a spin before going to bed [00:56] Q-FUNK: sorry for the delay. I'll post a link in the bug when it's ready [00:56] alright :) [01:05] ogasawara: I dunno if this is relevant but, just for the heck, I tried using kernel-package on a 2.6.30 tarball with the 2.6.31-rc5 patch and that LKML guy's config and it failed to build on Karmic. GCC kept on complaining about truncated files during build. [01:06] it built fine using the GCC on Jaunty, though. === spO is now known as joejc [09:15] * ogra scratches head, so why are all udebs built for linux-fsl ? [09:16] i thought tim disabled them [11:00] ogra i thought there was much upset that they were missing [11:00] i think they got reenabled cause cjwatson kindly fixed up kernel-wedge to make getting them to work possible in one lifetime === lexical is now known as lexical_afk [11:10] apw, yeah, it helps a lot, else i would have had to hack around it in the builder scripts, now i'm only missing meta and am done :) [11:10] i assume that'll be a simple one [11:10] now that the names are acceptable [11:10] i assume the kernel works? [11:13] havent tested yet [11:13] i'll do so after i rolled a rootfs/bootloader for amitk [11:14] too late to change it a assume now anyhow? [11:22] well, its broken anyway [11:23] no USB and no NIC [11:23] ya [11:23] but it should boot [11:23] as long as it does that i'm happy [11:40] its probabally to late to do anything about it even if it doesn't boot [11:40] i assume the archive is basically frozen from today? [11:41] later today, yes [11:41] but our kernel is in universe ;) [11:41] i hacked up the builder scripts, so freeze doesnt affect us ;) [11:41] we could upload changes until last minute if needed [11:42] but i'm confident it boots, its amits config which i tested already and i dont belive the packaging trashed anything [12:00] the lange boards don't have ethernet :/ [12:00] * amitk looks for usb-ethernet adapters [12:01] apw: did unison work for you? [12:02] amitk, yeah did thank you, only used it for a small shared directory syncing back the accumulated sprint delta to it,but it seemed to work for me [12:02] i'll need a bit more time to work out how i want to use it in my existing framework, but its looking useful [12:05] ogasawara: seems that we found our winner :) [12:07] apw: I have a .AMIT dir in $HOME. I move all my configs into that dir and symlink it from $HOME. e.g. .muttrc, .xemacs, etc. Then I sync .AMIT to all my machines. [12:07] ugly hack, but works. [12:07] yeah i have something similar going on in general [12:08] just the sync was mostly one sided there, and this makes it more bi directionally capable [12:10] amitk, did the redboot work for you ? [12:11] yes [12:12] good to know [12:12] i was told the lange requires a different redboot [12:12] but lool was wrong (or wrongly informed) then [12:12] it may be documented to require a different one [12:13] yes, though it might differ between lange 5.1 and 5.2 [12:13] indeed [12:18] confusing [12:19] it gets funnier if you think about the fact that 5.1 is newer than 5.2 :) [12:22] ogra: Hardware might not work if you're not using Lange's redboot with Lange boards [12:23] lool, ah, well, i think the 5.2 is still close enough to the babbage to work with TO2 [12:23] It might mostly work [12:23] 5.1 had more changes afaik [12:23] 5.1 and 5.2 share the same lange52 tree in the sources we got [12:23] as long as it gets the kernel to run it should be fine [12:24] No [12:24] not in redboot [12:24] In RedBoot too [12:24] the redboot i have seen had two binary trees [12:24] not sure how much they differ though [12:24] Yeah one for lange31 and the other for lange 5.x named lange52 [12:24] * ogra admits he didnt look very close [12:26] Ok don't state that I was wrong then [12:26] (Sorry but it's a bit annoying) [12:27] sorry [12:27] amitk: Lange 5.1 does have Ethernet on the board but it's wired over the USB bus [12:27] i.e. it's an USB Ethernet adpater soldered on the board [12:27] he only has 5.2 [12:28] and no babbage anymore [12:28] 5.2 only has wireless and no driver [12:30] yeah, it is a PITA to develop one since I can't get my kernel over http [12:30] s/one/on [12:31] thats why i set up the rootfs to be able to use flash-kernel [12:31] true, but it requires me to build .debs I guess. [12:31] no [12:32] ohh? [12:32] just copy it to /boot and run flash-kernel, then? [12:32] just cp your zImage over the /boot/vmlinuz-2.6.31-0-babbage [12:32] cool [12:32] then run flash-kernel and reboot [12:32] i forgot to install linux-firmware, you might need that if you want to use a usb NIC [12:33] ogra: I was responding to the comment that "< amitk> the lange boards don't have ethernet :/" [12:33] lool, yeah, i know [12:34] amitk: You might want to script some fis based script to update your SD card and take it out to write the kernel [12:34] (In case your kernel doesn't boot and you can't run flash-kernel) [12:35] That's shorter than serial port pushes [12:35] amitk: These are old notes I have to do that by hand: http://paste.ubuntu.com/251307/ [12:35] amitk, http://paste.ubuntu.com/251308/ [12:36] You need the redboot-tools package for the fis command [12:36] thats what i use with the same setup i gave you [12:36] from my laptop [12:36] amitk: You don't need the padding [14:31] rtg, my babbage images will build from universe, that should give us some extra hours for the meta (we can slip the freeze a bit) [14:31] bjf, you probably ought to update the meeting page https://wiki.ubuntu.com/KernelTeam/Meeting [14:32] ogra, I'll get the fsl meta package done today [14:32] cool [14:33] I live to serve [14:38] rtg, done, I just had the date of the next meeting hosed, the rest of the info is up to date [14:38] bjf, 'tanks [14:43] rtg: Hey Marvell currently pushed kernel trees to kernel.u.c; do you think it would be an issue if they pushed u-boot as well? [14:44] lool, we've plenty of disk. [14:44] Ok thanks [14:45] lool, in general, anything they push should be gplv2. other then that I don't much care. [14:45] rtg: I'm not sure it's GPL *v2* [14:45] lool, its some open source license, right? [14:45] I rather suspect it's a mixture of GPL versions and perhaps LGPL along; but it's as opensource as u-boot is [14:45] It is [14:46] A mix of GPL and LGPL is GPL [14:46] mjg59: A mix of GPL and LGPL is a mix of GPL and LGPL :) [14:46] You might mean the binaries are effectively GPL [14:46] No, I mean the work is GPL. Sections of it may have additional permissions. [14:48] rtg, btw, i changed the buildscripts already to use linux-babbage, if you pick any other binary name for meta, please let me know before the freeze is in effect (that has to go to min) [14:48] *main [14:49] ogra, in fact, I was just researching that. [14:49] well, if you use another name, tell me what it will be, livecd-rootfs needs to know the name of the binary for building [14:50] ogra, what about updates from Jaunty? the previous meta package binary was linux-image-imx51, etc. [14:50] you will need to add conflicts and replaces line in your new control file for the old names [14:51] *lines [14:51] rtg: I personally have objections to the fsl-imx51/babbage names but I think we should fix that post a4 [14:51] lool, ogra: well, I'm doing the meta package right now. I might as well choose the names that you guys want. [14:51] rtg: I'd rather we change the linux image name back to imx51 after a4, but because that was already uploaded and we're in a hurry, it's best to defer to post a4 I think [14:52] rtg: The linux-image names are incorrect; I prefer if both match at all times and we'll fix it after A4 [14:52] rtg, well linux-imx51 would indeed be best and cause least hassle [14:52] I mean the ABI versionned names [14:52] I'm not sure it's a good idea to go for a mismatch at this point [14:52] lool, lets not confuse this conversation to much, its currently only about meta [14:53] We might not be seeing the consequences if the ABI versionned name and the meta name dont match [14:53] because meta will go in the archive now, while the actual binaries are in already [14:53] ok, but I'm getting different input from amitk. How about you guys get together and decide on the final names? In the meantime I'll just name the meta packages *imx51* [14:53] amitk: What's your input? [14:53] ok, i'll change livecd-rootfs back [14:54] ogra: ? [14:54] ogra, just wait until you actually see what gets produced. [14:54] rtg, i need to do it before freeze [14:54] ogra: You dont know what the name will be; just leave it alone until you do? [14:54] it affects other arches if i make a typo or something, livecd-rootfs is used by all live images across the board (and in main) [14:55] ogra, then tell me the exact package names upon which your livecd build depends. [14:55] It does not affect other arches if your build fails [14:55] lool, i had to add universe today already [14:55] lool: send me an email in writing that all imx51 flavours will run on a single kernel and I'll agree to your proposal. IMHO, the distro should only concentrate on the dev boards, not OEM ones. [14:55] But it does if you do a syntax error yes [14:55] amitk: Ok [14:55] lool, so i changed to what the description of the linux image package says (linux-babbage) with a note that i will change back to what it actually will be [14:56] amitk: I think the way you present it makes it highly political instead of being simpl technical but I have no problem in putting my views in an email [14:56] i just dont want to change livecd-rootfs after the freeze [14:57] lool: it _is_ political. Someone is going to come in a few months wanting to enable an OEM board that doesn't work off the same binary. [14:57] then we do the naming dance again. [14:57] amitk, how can all imx51 flavours run from one kernel? We've already got 2 ARM kernel trees. [14:58] aren't they both imx51 variants? [14:58] rtg: you are confusing single SoC, multiple boards vs. a branch for each SoC. I think. [14:58] in theory, multiple boards _can_ run from a single kernel binary. If the code is written such. [14:59] amitk, you are right in that I'm confused :) [14:59] rtg, currently all babbage boards run from the imx51 tree ... we would like to add support for lange as well as its the same arch effectively [14:59] where does dove fit into that mess? [14:59] the current naming scheme would men to have a -babbage flavour and a -lange flavour [14:59] rtg: dove == separate SoC from marvell [14:59] dove is a different thing [15:00] its not imx51 compatible? [15:00] ogra: or a -lange flavour aliases to -babbage if that is possible. [15:00] having a -babbage and a -lange flavour would effectively enforce us to build two images for each of them (rootf images i mean) [15:01] having simply a binary that supports all of imx51 would not cause such probs [15:01] and currently it doesnt look like we would actually need any -lange flavour at all, the patches all apply to the imx51 tree without touching babbage stuff afaik [15:02] having that two name setup effectively binds the double amount of ressources, costs the double space for images and twice the amount of maintenance [15:02] rtg: please don't get distracted by this naming issue for A4. This conversation should _really_ have been brought up after A4. [15:02] right [15:03] it's nothing to have now, binary kernels are up now and wont change atm [15:03] amitk, ogra: ok, I'm gonna email the meta naming convention so that you have it in writing. [15:03] thanks [15:04] amitk: There you go [15:04] amitk: I agree this should all be post A4 [15:04] Hi guys, sorry to interrupt your discusssin, just simple question - Please explain why ubuntu kernel ftp site lists separate mdules, like virtio-modules-2.6.31-5-generic-di_2.6.31-5.24_i386.udeb . I thought these modules should be a product of kernel compilation? [15:04] amitk: Do you agree we should just use babbage in meta as well now for a$? [15:04] a$ [15:04] grr [15:05] a4 [15:05] * ogra doesnt care as long as he gets told the right name :) [15:05] lool: agreed. It will be easier to rename everything if we decide to. [15:06] rtg, so call it linux-babbage for now and we'll adjust post A4 [15:07] rtg, and dont care about upgrade paths for now, its only important that we have solved that for final, A4 is only first shot [15:12] ogra, email sent [15:13] rtg, agreed for A4 ... [15:14] ogra, cool. when you want the names changed post A4 please start an LP bug so that we're all reading the same page. [15:14] yeah, good idea [15:14] lool, mind to file it since you are the driving power here ? [15:15] ogra, lool: make sure you assign it to me lest I not see it in the blizzard of bug reports. [15:15] indeed :) [15:16] Sorry I'm in a call [15:16] File a bug on the renaming? [15:16] Sure can do [15:16] great, thanks [15:16] * ogra would do it but you have the best arguments :) [15:19] levonshe, those .udeb files are the result of kernel compilation [15:20] rtg: What's the package name for meta? [15:20] levonshe, they are not used in normal systems though [15:20] rtg: I mean same meta? [15:20] lool, linux-babbage says the mail [15:20] https://bugs.launchpad.net/ubuntu/+source/linux-fsl-imx51/+bug/411968 [15:20] Malone bug 411968 in linux-fsl-imx51 "Please rename babbage to imx51, just like in jaunty" [Undecided,New] [15:20] lool, meta isnt uploaded yet [15:20] ogra: I mean the source [15:25] apw, re: bug #350789. I though we fixed it? [15:25] Malone bug 350789 in linux "AppArmor Debug: Hook being called from interrupt context" [Medium,Triaged] https://launchpad.net/bugs/350789 [15:26] rtg: we did [15:26] yeah its only on our lists cause its in an odd state [15:26] rtg: status is updated [15:27] jjohansen, ta [15:27] jjohansen, is the fix applicable to Karmic, or already applied there? [15:27] apw: already applied [15:27] and released? [15:28] apw: yeah, it actually has a slightly alternate fix in it from the start [15:28] so its already uploaded in karmic? then the linux(ubuntu) task should be Fix Released [15:28] ah okay [15:29] and i don't think it can be assigned to a jaunty milestone [15:32] TheMuso: it appears that those changes in alsa-utils and pulseaudio to rely upon udev features have bombed majorly. how can we fix them? === ivoks_ is now known as ivoks [16:26] ogra, linux-meta-fsl-imx51 uploaded. perhaps you could annoy the various mobile team archive admins in order to get this package processed through NEW ? I'm off onto the next big thing... [16:26] will do, i'll yank your chain if it fails to build though ;) [16:27] ogra, I would expect no less. [16:59] >> [16:59] >> Ubuntu Kernel Team Meeting - Today @ 17:00 UTC - #ubuntu-meeting [16:59] >> === pgraner changed the topic of #ubuntu-kernel to: "Home: https://wiki.ubuntu.com/KernelTeam/ || Karmic Kernel Version: 2.6.31 || Ubuntu Kernel Team Meeting - Tuesday, 11 August - 17:00 UTC" [17:48] rtg, james_w had some complaints aout meta but left it through [17:48] ogra: I don't think there's a need for it to explain the history of linux in debian/copyright given that it's just a metapackage [17:48] plus, Vcs-Git is invalid [17:48] (for the next round :) ) [17:49] ogra, what's wrong with 'Vcs-Git: http://kernel.ubuntu.com/git-repos/ubuntu/ubuntu-karmic-meta.git fsl-imx51' ? [17:50] no idea, i havent complained [17:50] rtg, i've got a couple of karmic updates dropping stuff from ubuntu as per discussions at sprint... is the tree up to date? [17:51] apw, nail it [17:52] apw, and while you're at it, make sure I haven't wrecked anything. there are a few packaging changes [17:53] will do ... i'll be uploading it to my preview ppa so that'll be a good test [17:53] ogra, the copyright does seem a bit over the top, but then its been there since Feisty AFAIK [17:54] well, lets ask james for the actual final meta, i dont think we need to care for that one at all === JanC_ is now known as JanC [18:42] rtg ok i've just uploaded karmic tip to my daily PPA [18:43] we'll know shortly if its all spagetti [18:46] apw, well, I _did_ build it here before I pushed :) [18:50] iwlagn is currently using 101% of CPU on my system. this has to be bug. have any of you run into it? [18:52] maco, seems like a lot. have you contacted anyone on the wireless mailing list? [18:53] no, id thought it stopped. i dont think it was doing it with -3 but its there in -5 again [18:53] sconklin, I'll see your comment, and raise you by one. https://launchpad.net/bugs/412039 [18:53] Malone bug 412039 in linux-backports-modules-2.6.28 "ath5k: Add LED support for Acer device" [Medium,In progress] [18:54] maco, does it still happen with Karmic LBM ? [18:55] rtg: its nondeterministic. if i modprobe -r and then modprobe, itll stop [18:56] "i dont think it happened on -3" means "i dont recall getting annoyed at my keystrokes not being registered in the last couple weeks...uh...." [18:56] (when it pegs my cpu, and i try to type, some keystrokes just get dropped) [18:57] maco, that kind of implies an interrupt storm [18:57] i can try installing lbm, but since i dont have any nice reproducible way to get it to happen... [18:57] im behind a nat, so i doubt anyone's trying to DoS my laptop and actually reaching it [18:58] if thats what you mean by interrupt storm....that the interface is being hammered [18:58] maco, no, I mean that the firmware might be misbehaving and generating too many interrupts. Hence, your loss of keyboard inputs [18:59] ah. i was surprised to see iwlagn being a process of its own [18:59] would lbm have new firmware? i didnt think it did [18:59] maco, I think iwlagn still uses their own rate selection process, so perhaps its that thread running amok [19:00] I'm checking on firmware versions 'cause I can't remember [19:01] maco, i4965 is one minor rev newer. everything else looks the same. [19:01] alright, ill give it a try then [19:02] doh, i4965 is the same. I'm dislexic. [19:02] oh [19:02] the biggest difference with LBM will be in the driver itself. [19:02] its worth a shot [19:03] ok any suggestions on why Xorg is using 58% of CPU (now.... before iwlagn started acting up it was using 98%) when ive got compositing off? or should i ask #ubuntu-x that one? [19:04] maco, definitely an X question [19:04] (yay dual core! actually makes it possible for wireless to use 100%, graphics to use 60% and my mail client to use 40%) [19:04] maco, you're kind of hell on battery life :) [19:06] rtg: well none of the above SHOULD use as much cpu as they do [19:06] maco, hence my amusement. [19:07] if i wanted to compile anything, id be screwed [19:07] itd take a week [19:08] actually, i find my cpu spins up all hot and crazy when i compile things on a remote system over ssh and dont even have that ssh screen visible (so drawing the compile output isnt the issue) [19:08] rtg: maybe iwlagn goes stupid when there's lots of small amounts of data going through? [19:09] maco, it still generates plenty of network traffic (lots of small packets) [19:09] maco, try using screen on the remote, then disconnect [19:09] so maybe IRC is the ame way? [19:09] *same [19:09] hmm, IRC generates such a small amount of traffic. [19:09] when in 35 channels? [19:10] even so, unless someone is pasting a boatload of stuff, there really isn't that much. [19:10] im trying to think...maybe the combination of IRC + KMail syncing + apt = iwlagn :( [19:11] maco, possibly. I'd drop a not to the wireless folks. perhaps they've already seen something similar. [19:11] note* [19:12] ok [19:59] ogasawara: seems that I got bingo on that Bug #396286 [19:59] Malone bug 396286 in linux "2.6.31-generic: kernel panic near the end of initramfs" [High,In progress] https://launchpad.net/bugs/396286 [20:00] Q-FUNK: I'm building the next test kernel, just have to give it a while to finish building [20:00] Q-FUNK: as usual, I'll post to the bug when it's ready [20:01] ogasawara: ok. I was just curious to hear about possible causes I should watch for. [20:01] Q-FUNK: that's what we're narrowing down [20:01] Q-FUNK: there's 71 commits left to bisect [20:05] oh [20:06] between 999.200908071658 and 999.200908110142 you meant? [20:08] as 2.6.30.999.200908110142 crashes during boot while 2.6.30.999.200908071658 didn't [20:09] Q-FUNK: right, narrowing it down between those two [20:27] hi [20:28] there is a problem with the configuration of the ubuntu kernel which lets a system freeze completely [20:32] it concerns the configure symbol CONFIG_SATA_PMP which has some bug that will freeze some systems under heavy IO Load [21:01] apw, I have a config patch for SPARC for you (it just built off the latest head as of a few hours ago) [21:01] (I fear if I send it to the list, I'll hit the moderation queue again) [21:40] Hi, I've got a MSI notebook with UNR installed. Some kid told me that using the 2.6.30 kernel would improve speed. How do I go about installing it? [22:10] speed of what?