=== kamalm is now known as kamalm-away === MTeck is now known as MTecknology [03:17] What's changed from 9.10 to 10.04 that would negatively impact iwlagn? [03:19] The kernel. [03:19] I'm just saying. [03:20] soren: :P [03:20] soren: I get to be on the sucky side of bug 544254 [03:20] Malone bug 544254 in linux "iwlagn (i4965AGN) continually drops and reconnects to access point " [Undecided,Confirmed] https://launchpad.net/bugs/544254 [03:24] soren: I was just wondering if anybody knew off the top of their head what may have decided to break it [03:24] found something on lkml... [03:25] http://lkml.org/lkml/2010/1/16/126 - close but not quite [03:28] MTecknology: A friendly word of advice: You're not going to make any friends with that kind of rhetoric. [03:29] soren: hm? [03:29] MTecknology: The whole "decided to break it" nonsense. [03:30] soren: something's making it not work. I'm not smart enough to figure out where the iwlagn driver might have changed. I didn't see anything bad about saying that [03:31] MTecknology: When I get that sort of attitude from someone, I'm likely to ignore them. I'm just saying. [03:31] soren: I didn't know there was attitude in that [03:31] MTecknology: I'm not speaking on behalf of the kernel team. As I said: Just a friendly piece of advice. [03:31] how would you phrase it? [03:32] In a way that doesn't imply that someone is out to get you and has purposely broken things just to make your life suck. [03:33] soren: we must be reading two different lines [03:34] bugs happens, things break, it's the way of life. sometimes you can do one thing that fixes things for thousands of users; but it'll break things for two other users. [03:35] I understand the nature of bugs. [03:37] soren: so why is what i said bad? [03:38] MTecknology: I'm just saying that implying that people have gone out of their way to wilfully break things is not going to get you better feedback, that's all. [03:38] I never implied that [03:38] "I was just wondering if anybody knew off the top of their head what may have decided to break it" [03:39] ^ I asked about a change breaking it [03:39] whatever, if that's the way you see it........ [03:39] I'll just try to go through the bug report... [03:40] "decided to break it" in my book more than implies that a conscious decision was made to break something. [03:41] Anyhow, this is silly. I'm not speaking on behalf of the kernel team. Heck, they might even enjoy that sort of rhetoric. I'm just saying how I'd react, that's all. [03:41] And with those words, I'll wander off and do vaguely useful things. [03:43] soren: OTOH what doesn't imply a person :P [03:44] hurry for uploading 14.6MB [04:27] any ideas when we'll have linux-backports-modules-wireless-2.6.32-17-generic available? === sconklin-gone is now known as sconklin [13:10] lifeless: fair [13:58] Hey all [13:58] which is the kernel for lucid-lynx ? [13:59] I'm following the instructions on https://wiki.ubuntu.com/KernelTeam/GitKernelBuild but I'm not sure which kernel is going to be used for lucid [13:59] git://kernel.ubuntu.com/ubuntu/ubuntu-lucid.git [14:00] smb: do I need to clone git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git as well ? [14:00] smb: get both? [14:00] JohnFlux, If you do more than one kernel it helps. Though its not mandatory [14:00] Its just that you can use the linux-2.6.git tree as --reference [14:00] smb: hmm, i'm getting horribly confused by these instructions then [14:01] which saves you disk space for copies [14:01] linux-2.6 is a copy of Linus upstream tree [14:01] the link I gave says I need to have both, and use an 'overlay' [14:01] smb [14:01] smb: [14:01] overlay? [14:01] smb: that what it says [14:02] JohnFlux, Let me have a look [14:02] smb: I'm not too sure either, other than guessing from the name [14:02] smb: so, once I have that ubuntu-lucid.git cloned, how do I build it? Where do I get the .config for it from? [14:03] JohnFlux, That git tree contains the config and all [14:03] smb: do I need to copy the config to .config or something though? [14:03] If I'm having a problem in the new kernel and using the backports fixes that issue. How can I apply the backport headers to a custom kernel? [14:04] smb: how do I build? [14:04] You should have either a build chroot if lucid is not installed on the machine [14:04] smb: I've just upgraded to lucid [14:04] smb: so I should have everything that I need [14:04] fakeroot debian/rules binary-generic [14:04] That builds only a generic kernel [14:04] I always do 'make all modules_install install' for a local system; long ugly command for a pretty deb [14:05] smb: will it include staging? [14:05] The package will land in the dir above [14:05] It will include the same drivers as our builds [14:05] So all of staging that is enabled normally [14:06] smb: okay good. Basically I'm trying to get my hardware to work - it needs a minor change [14:06] smb: when I make the change, how can i rebuild but give the package a different name? [14:07] so that I can try different changes [14:07] JohnFlux, you edit the debian-master/changelog [14:07] The first line contains the version number [14:07] You can add something like +myversion1 [14:08] This is what I use for the pretty deb AUTOBUILD=1 CONCURRENCY_LEVEL=${CONCURRENCY_LEVEL:-3} ionice -c 3 schedtool -D -e make-kpkg --initrd --rootcmd=fakeroot --append-to-version=-hyper${KERNELREV:-1} kernel_{headers,image} [14:08] Just beware of '-' in the version number [14:08] how can I build with make -j 4 or so [14:08] to speed up the compile? [14:08] MTecknology, That looks pretty much overdone for a simple build [14:09] JohnFlux, Usually it uses -j [14:09] smb: right, I have 4 cores [14:09] JohnFlux, If you want to override this you set CONCURRENCY_LEVEL=x [14:09] smb: oh, it does by default? [14:09] cool [14:09] JohnFlux, yep [14:09] I built a gentoo system once with -j instead of -j# .... very very bad idea - bad things happen :P [14:10] yes, top is confirming that I have about 3 or 4 cc1's at any one time [14:10] thanks [14:10] smb: binary-generic is what I want? :) [14:10] smb: I'm after the default kernel for i386 [14:11] or 686 or whatever it's called [14:11] JohnFlux, The default kernel there depends on amount of RAM. For more than 4GB you might want binary-generic-pae [14:11] but otherwise its binary-generic [14:12] smb: hmm, I've never had much luck with pae in the past [14:13] never boots for me - I'll leave that for now [14:13] JohnFlux, Depends on cpu as well. There are some via c7 which just don't have pae capability [14:14] If cat /proc/cpuinfo does not have pae in flags it will fail [14:21] So.. backports are upstream changes that may break for some users but backported because they fix things for other users? [14:22] MTecknology, If you speak about linux-backports-modules (the package), yes. [14:22] smb: yup, i was - thanks :) [14:40] anyone got time to give me a few pointers on solving a kernel panic issue i've got? [14:55] ngregory: ya, pastebin anything you can see so some smart person here can point you in the right direction [14:58] i raised a launchpad bug - but unfortunately its not getting much (i guess everyone is pretty busy) . last stack track from a mainline kernel http://launchpadlibrarian.net/41648154/panic5-2.3.33.txt [14:59] ideally want to know its its xfs / lack of memory or some random scheduler bug [15:06] smb: dh_testdir: cannot read debian/control: No such file or directory [15:06] smb: it aborted with that error [15:07] JohnFlux, Ah, fakeroot debian/rules clean [15:07] JohnFlux, You need to run clean once to generate some files [15:07] /bin/bash: kernel-wedge: command not found [15:07] JohnFlux, sudo apt-get install kernel-wedge [15:08] smb: shall I add that as a prerequisite to the wiki? [15:09] JohnFlux, Would make sense. There might be more. I think you would get a list by running dpkg-buildpackage -D ... [15:10] JohnFlux, On the other hand most of the wiki seems to address building a standard Linux tree from upstream with a config close to the one used for ubuntu builds [15:13] smb: indeed, I think I'm following the wrong guide [15:15] JohnFlux, maybe this contains parts that would help you better https://wiki.ubuntu.com/KernelTeam/KernelMaintenanceStarter [15:19] what would prevent a module from loading automatically whereas the same module can be loaded manually with modprobe? on 8.04 server i have such a situation with a 10 Gb NetXen/HP card (NC522SFP). === kamalm-away is now known as kamalm [16:11] hi, how can someone get a patch into the ubuntu lucid kernel? [16:16] mrec, send an e-mail with the patch info to kernel-team@lists.ubuntu.com [16:22] thanks, hopefully someone will pick it up [16:29] JFo: git format-patch <- just fyi [16:30] MTecknology, I don't deal in the patching, just the bugs [16:30] :) [16:31] s/JFo/mrec/ [16:35] MTecknology: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=7152b592593b9d48b33f8997b1dfd6df9143f7ec [16:35] this one you mean? [16:36] mrec: what about it? [16:37] you wrote git format-patch [16:37] there you can pick it :) [16:38] I'm too tired to follow :P [16:38] time to reboot into a new kernel - hopefully this one doesn't carry the same bug [16:41] huh, that bug could be the cause of my deadlock [16:42] ali1234: my issue is wifi [16:42] i have an issue with wifi too [16:42] separate issue tho, i think [16:43] i'm currently more bothered about the fact that the kernel is deadlocking when using gadgetfs [16:44] ali1234: wlagn locking up or really slow? [16:44] my wireless issue is random memory corruption [16:45] oh [16:45] it does not lock up or go slow, it just corrupts packets silently [16:45] interesting [16:45] if i turn on debugging i get "poison overwritten" errors [16:45] anyway, i sent that one to ath5k developers [16:48] ali1234: g'luck [16:48] hrm - time for me to run [19:34] JFo, do u recall ? do we have a ton of bugs where HPmini does not boot when splash enanbled ? [19:34] JFo, I just get a screen that says Glib-Warning getpuuid_r() failed to do unknown user id (0) [19:34] and can't switch console [19:41] no idea [19:43] ok [22:15] I'm having issues with uio on a dove kernel, and I'm not getting any useful debug messages [22:15] anyone around to help with debugg? [22:15] ^ing [22:46] NCommander: is it dove specific? [22:46] or arm specific? [22:46] (I'm guessing so, and I can't help if that's the case...) [22:55] cnd: dove, but I think I have an idea [22:55] uio modules don't seem to work as modules [23:24] Ok. We need a way to ensure that vga16fb does *not* bind when there's a perfectly good kms driver loading that's just about to claim fb0. [23:28] RAOF: I was interested in tackling that too [23:28] what needs to be done? [23:28] I don't know much about this area [23:28] I'm not entirely sure. Possibly vga16fb could be taught about vgaarb? [23:29] what's vgaarb? [23:29] vga arbitration. [23:29] I claim thee, VGA card! [23:29] * cnd goes to lxr for research [23:30] I'm by no means an expert here, either. [23:30] http://lxr.linux.no/linux+v2.6.33/Documentation/vgaarbiter.txt [23:31] But nouveau devs are paranoid about vga16fb messing with their state (even though I've seen no bad effects with vga16fb binding fb1), and in rare cases it seems that vga16fb claims fb0 before novueau has fully initialised, resulting in tears. [23:32] Hm. Maybe vgaarb is not what I meant. [23:33] RAOF, yeah, vgaarb doesn't look quite right here [23:34] RAOF: do you know what the difference is between vag16fb and vesafb? [23:34] No, I don't. [23:34] Anything I said would be based simply from educated-guessing on the module names. [23:35] I'm guessing vga16fb is for really old stuff, vesafb for slightly newer [23:36] vesa is a newer (but still old) video interface. It's kinda vga+ [23:36] yeah [23:36] I'd guess that the big difference between vga16fb and vesafb is that vesafb would be doing some modesetting. [23:40] RAOF: doing some research I found bug 530451 [23:40] Malone bug 530451 in linux-backports-modules-2.6.32 "Nouveau interacts badly with vga16fb - blank screen before X" [High,Fix released] https://launchpad.net/bugs/530451 [23:40] it seems to say that as long as the correct driver is above vga16fb in modules.order, things should be fine [23:40] is that not the case? [23:40] vga16fb on fb1 when you have a KMS fb on fb0 also makes lshw completely screw up the screen [23:40] Right. [23:41] ahhh [23:41] cnd: I thought that was the case, and in all *my* testing it was, but there are a couple of bugs where vga16fb still seems to claim fb0 [23:41] so we need to prevent vga16fb from doing anything [23:41] Yeah. [23:44] it shouldn't claim fb0 anymore ever if theres a KMS driver being loaded, have you seen any reports that it was that aren't lbm-nouveau related RAOF? it still loads and screws things up on fb1 though all the time [23:46] Sarvatt: bug #546359 seems to have vga16fb claim fb0 without lbm-nouveau. That's a dual-gpu case, so is somewhat of a corner case. [23:46] Malone bug 546359 in xserver-xorg-video-nouveau "nouveau does not work with nVidia G94 [Quadro FX 1800]" [Undecided,New] https://launchpad.net/bugs/546359 [23:46] That said, there are going to be quite a large number of dual-gpu laptops running Lucid. [23:46] well, it would if say intel radeon or nouveau loaded *before* the agp module causing it to fail which is a pretty nasty problem some people are hitting [23:51] hmm yeah thats worth looking into and definitely multi-gpu specific