[06:07] amitk, ping [07:30] How does updateconfigs work? (i.e., I don't get how the common config is generated ...) [08:37] NCommander: if you add a new Kconfig option, running updateconfigs will prompt you for that config option for all flavours [08:38] amitk, I got that far :-) [08:38] amitk, its more figuring out what the configuration options should be on the new ones :-) [08:39] * NCommander is already mostly done, just finishing sparc config [08:39] you just have to know that. updateconfigs can't read developers minds yet :) [08:39] amitk, Its tedious work [08:39] * amitk nods [08:40] Once I finish 2.6.27, I'll start doing test builds [08:40] * NCommander has access to all the port architectures :-)! [08:40] sweet [08:41] amitk, find out about the kernel.u.c. account? [08:41] * NCommander would like a place to hang his tree [08:45] NCommander: will shoot off email today [08:45] \o\ /o/ [08:46] amitk, generally, if something can be built as a module, it gets built as that. That's the rule of thumb I followed for kernel configuration updating (Linus added at least 30-40 new config options 25 -> 27 ;.;) [08:52] NCommander: probably fine for now, for Jaunty we might turn that rule over its head to improve on boot times. [08:52] amitk, I already put a note that we need to sanity check the config files [08:52] (make sure no insane options for an architecture is there) [08:54] WOOOOOO [08:54] finally [08:54] rebased [08:57] amitk, stupid question, how do I generate a new abi folder? (its pretty clear that 2.6.25 -> 2.6.27 will break an ABI) [09:01] Nafallo: Build process generate it automatically [09:01] Nafallo: Sorry [09:01] I meant NCommander :-) [09:01] ah [09:02] so now I get to see how miserable the ABI broke in two releases [09:02] woow [09:02] ;) [09:02] so whats sauce? [09:21] amitk, ^ [09:27] NCommander: any patch that isn't upstream yet (those that _should_ go upstream + those that will never go upstream) [09:28] amitk, oh, that makes sense, anyway, 2.6.27 is building via debuild, so I haven't completely destoried the world yet ;-) === mdz_ is now known as mdz === _ruben_ is now known as _ruben [10:39] amitk, hola [10:55] it looks like 263059 may have been caused by ftrace === asac_ is now known as asac [11:17] amitk, I just discovered linux-ports-meta doesn't depend on linux-firmware. Care to help fix that? === D2_WON is now known as giovannigiusti [13:39] hey BenC [13:53] * NCommander now has a building linux-port 2.6.27 kernel :-)! [13:56] NCommander: hey [13:56] Morning BenC :-) [13:56] I need some help resolving one last issue with linux-ports-meta before intrepid goes. I discovered this morning that linux-ports-meta doesn't depend on linux-firmware [13:58] BenC, care to help? [14:17] BenC, poke? [14:21] rtg: which ftrace options did you change? [14:23] Keybuk: http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-intrepid.git;a=commit;h=67d9b90a1c844bf1c6daaffd2c60561fc8c445f7 [14:24] NCommander: Ok [14:24] * NCommander has a git patch, but is somewhat confused by the current linux-ports-meta package [14:25] the explanation doesn't really make sense though [14:25] it blames a bug on module _unload_ [14:25] when we're not unloading modules [14:25] (both for e1000 and iwl3945) [14:25] Keybuk: also discarding of init sections [14:25] mjg59: -v ? [14:25] Keybuk: The ftrace issue also arises when the __init sections of modules are discarded after module load [14:27] I guess that certainly fits the pattern of where the iwl3945 hang was occurring [14:29] Keybuk: can you reproduce it anymore? [14:29] nope [14:30] Keybuk: well, I guess that's proof positive. [14:30] something about the way iwl3945 works was causing corruption of ioremapped memory if another module's __init was discarded while it was initialising ?? [14:32] Keybuk: got me. I didn't dig into it nearly that deep. [14:36] me neither, I don't really understand this level of stuff ;) [14:36] I could go and learn, but I'm more inclined to nod sweetly, back away, and stay in my comfort zone :p [14:37] Keybuk: that's pretty much my attitude. I'm buried with other issues. [14:37] there seems to be plenty of confirmation that this fixes it. [14:37] * NCommander needs someone who has no fear to help fix the -ports-meta package [14:38] obviously someone should post to lkml that disabling ftrace has fixed another issue with another Intel driver ;) [14:39] Keybuk: i bugged the wireless folks last week about it, so I'll at least let them know what the solution was. [14:43] NCommander: A Git patch? [14:48] BenC, linux-ports-meta is maintained in git [14:48] (aka git format-patchs) === lamont` is now known as lamont [14:50] NCommander: hmm...not sure why, but ok [14:51] rtg: is there a bug about the iwl3945 kill switch state not changing in userspace? [14:51] Keybuk: yep. hang on... [14:52] Keybuk: probably this one: https://bugs.edge.launchpad.net/ubuntu/intrepid/+source/linux/+bug/193970 [14:52] BenC, I just made a realization; the ports kernel still has the firmwares compiled in as blobs so that change is unnecessary :-) [14:52] Keybuk: I think slangasek uploaded something in pm-utils to accommodate a sysfs change. [14:53] NCommander: that doesn't pull in the linux-firmware stuff [14:53] BenC, I don't think its needed unless I'm mistaken [14:54] NCommander: probably not...what about qla or aic scsi firmware? [14:54] BenC, TBH, I dunno. My powerpc doesn't have any devices that requires firmware [14:55] BenC, I have the patch: http://paste.ubuntu.com/60124/ [15:02] NCommander: You don't want to add it to linux [15:02] NCommander: It needs to be added to linux-image-* if anything [15:02] BenC, that's how it was changed in linux-lpia ... [15:03] NCommander: then linux-lpia did it wrong :) Check linux-meta [15:03] Ah [15:03] I see [15:03] BenC, in that case, care to sponsor two commits, and two sponsors ;-)? [15:06] BenC, http://paste.ubuntu.com/60126/ [15:22] Woo, linux-ports 2.6.27 on PowerPC builds! [15:44] BenC, is my patch ok? [15:49] NCommander: best to ask someone who is actually handling that tree...I don't know enough about it [15:51] * NCommander looks up Stefan Bader's IRC name [15:52] NCommander: smb, but he doesn't appear to be around this morning. [15:53] argh [15:53] * BenC wonders where he is [15:53] * NCommander turns on the smb light [15:54] * NCommander hopes he'll see the Critical bug on linux-ports-meta [16:04] BenC, out of curosity, what are the requirements of getting a kernel.u.c account [16:34] NCommander: Not much...I'll get your account processed today or tomorrow [16:34] NCommander: inbox is flooded from weekend [17:05] i am interested in being part of kernel team. How can i help [17:05] i have some linux internal knowledge on how networking subsystem works to an extent. I have one system and have installed Ibex 8.10 in Ibex [17:13] G__81: start here https://wiki.ubuntu.com/KernelTeam and here https://wiki.ubuntu.com/KernelMaintenance. Also subscribe to the kernel team mailing list. [17:14] here are the list of bugs that need working on: http://people.ubuntu.com/~ogasawara/intrepid-buglist.html [17:18] Yeah thanks i would look into it === thegodfather is now known as fabbione [21:01] if you cherry pick this 2.6.28 patch for intrepid you will enable users with intel G45 motherboards to boot the ibex --> https://bugs.launchpad.net/ubuntu/+source/linux-meta/+bug/285572 [21:02] i've tested this patch applied to intrepid's .27 kernel and it works perfectly (instead of freeze on xorg login I get full compiz with no xorg.conf hacks) [21:02] I tested it on a "Gigabyte GA-EG45M-DS2" board but I'm fairly sure it applies to all the G45 desktop motherboards [21:03] the patch itself is very non-intrusive and doesn't modify any code on non-G45 code paths [21:38] rtg: I think my boot crash is sorted now, just did 31 reboots with 2.6.27-7.12 and it didn't happen [21:40] did the weird 3965 crash disappear as well? [23:52] Do you guys know about the regression in 2.6.24-21-generic? involves snd_hda_intel and snd_hda_codec missing symbols. [23:54] lamont: because it's a bitch to have to recompile the kernel to fix the problem.