[08:43] * apw yawns [08:45] good morning apw [08:47] * amitk waves [08:48] .... and amitk :) [08:48] morning ikepanhc [08:48] morning all [08:49] * apw waves [09:13] * cooloney waves at apw amitk cking [09:13] morning guys [09:13] mornign [09:13] Ah, this seems to be a more suitable channel for my question. I am in need to install the 2.6.32.9 kernel. I just need it to compile a certain kernel driver. How would I go and do such a thing? I'm currently doing a git sync of kernel.org, but I doubt I will be able to see the previous versions there. [09:15] Matsy, the 4 number versions are not in linus' tree, they are in the stable trees. we do produce unsupported builds of them however [09:15] https://wiki.ubuntu.com/Kernel/MainlineBuilds [09:17] Hm, okay. I can see it right there. I'm on 10.04 right now. Would I also need the -lucid package from the repository? [09:18] yep [09:23] Interesting enough, then! [09:26] Oh, gah. Need to resize my virtual disk first. I knew I shouldn't have picked 18GB [10:03] I reported, tracked and got kernel regression https://bugs.launchpad.net/ubuntu/+source/linux/+bug/86820 fixed upstream. It will only appear in 2.6.36. Any change we can backport this in 10.10's kernel ? I'am tired of backporting the change in my 10.04 custom kernels... Tanks [10:03] Launchpad bug 86820 in linux-source-2.6.20 (Ubuntu) (and 1 other project) "touchpad doesn't work after suspend (affects: 2) (dups: 2) (heat: 41)" [Medium,Won't fix] [10:39] if we need to give some parameter to one module, can we write the parameter somewhere in /etc? === diwic_ is now known as diwic [10:43] ikepanhc, yep, /etc/moprobe.d/ [10:44] eh... [10:44] add file for myself? [10:45] oh, I see it, write options there === amitk is now known as amitk-afk [10:45] apw: thanks [10:52] yeah you can make new files in there for your purpose [10:57] Hm. [10:58] Okay, the installation of 2.6.32.9 didn't quite work. :( [10:58] Or at least, it's not giving me the result I would like. [10:59] Apparently, it's saying: Kernel configuration is invalid, autoconf.h and auto.conf are missing. Run make olconfig && make prepare to fix it. [10:59] But, those return 'error 2' [11:00] So I'm not quite sure what to do here. [11:02] The error the command make oldconfig && make prepare is giving me is as follows: make: *** No rule to make target `include/config/auto.conf', needed by `include/config/kernel.release'. Stop. === diwic is now known as diwic_afk [11:03] Any ideas? [11:09] Oh, I managed to do it. Now, it's saying it cannot find modpost. Hm. [11:15] Matsy, sorry not ever tried to build a module against those [11:16] apw: Aw, thanks though. I'll just keep on trying. [11:18] Oh no. [11:18] I need 2.6.34.5 [11:18] Darn. [11:19] Flawed information is always nice. [11:19] at least that version is also in the archive [11:21] Yeah, that's true. [11:21] But, no sources available, apparently? [11:30] Gah, this is too complicated. === amitk-afk is now known as amitk [11:34] I might be doing it very much wrong. [11:34] All I want to do is download the 2.6.34.5 kernel, and compile it. [11:45] Oh, gconfig did the job. [11:49] * apw wonders why you need that specific version === ogra_ is now known as ogra [11:50] hi [11:52] hi [11:56] having Ubuntu 10 - I'm running a very time critical task - how can I boot into no-Xwindows and give 99% scheduling to a single task? [11:58] I managed to compile rt/smp and set the highest rt priority for the task but it still gets interrupted too much [12:01] maybe the answer is running DOS but there must be a way to get linux into doing a single task apart from the kernel [12:17] booted into recovery root console does work for now [12:18] dionys, how are you determining it is interrupted too much [12:18] and what is your latency requirement ? [12:19] thinks truly running at RT priority should be able to break the system by starving it of CPU [12:20] for x-less boot you just need to put "text" on the kernel cmdline btw [12:49] sorry I was away for lunch [12:51] as it appears I need to have 20micro seconds [12:52] if a program does a sleep(x) does the kernel come back after x or after x plus the time spend in other tasks? [12:57] I guess after experiencing these interrupts the kernel comes back after x plus the time spend on other tasks [13:04] apw: I determined the interruption by causing CPU load by moving windows and then experiencing more hickups to the program I'm running [13:05] dionys, there is some jitter on how long later it comes back based on the scheduler, you would expect an RT process to have a lower latency there though as it is much higher priority [13:09] as I mentioned by running in a terminal window in recovery mode has taken away the interruptions [13:10] Hm. [13:10] Compiling the kernel is taking a long time. Are there any flags I can set to speed it up? I'm on a quad core system. [13:14] Matsy: a solution would be the break down the compiles into independent work units and give them to one workertask per core - which I don't think is implemented [13:14] Hm. [13:15] What'd the general compilation time be of a Linux kernel? [13:16] On a Q6600 / 4GB ram [13:16] not the quantity of the ram is essential here but the ram timings [13:18] Matsy: -j8? [13:18] but maybe there is a way to make build units run in parallel on each core [13:20] Oh [13:20] Haha, thanks abogani [13:21] Matsy: You are welcome. [13:22] Hm. [13:22] I am building a kernel module [13:22] But, when I try to load it, it can't load, saying Exec format error. [13:22] I'm fairly certain that I am building it against the right kernel [13:22] But, I am not quite sure what the error means [13:23] Could it be because I am compiling it on a 64 bit machine, and that it is going to run on a 32 bit machine? [13:31] Matsy, yes you need to build it in the right environment [13:31] often that is a chroot with a 32 bit install in it [13:33] hm. [13:33] I can't use -m32 to build it?\ [13:36] you may be able, it wouldn't be guarenteed to be the same exact kernel as woudl be built in an chroot though [13:36] Matsy, which kernel are you loading this module into btw ? [13:37] as modules need to built against the kernel they are loaded into [13:37] A custom kernel, it's a smartphone kernel. [13:37] Makefile is pointing to the source of that one kernel. [13:38] My phone is in sync with that kernel. [13:38] But, when I try and do insmod module.ko, it presents me with that errors. [13:38] *error [13:38] you often need the same compiler they used to build the kernel too [13:39] Same.. compiler as well? [13:39] Hm. [13:39] you do for an ubuntu kernel [13:39] I don't know if they're included, let me see. [13:41] Nope, no compilers included. [13:41] Can't I force load this in any way? [13:43] Matsy: Don't you can ask to people have done that smartphone kernel? :-) [13:43] No one has done this before, apparently. [13:43] So it's a bit tricky ;) [13:44] If I'm being a pain for asking here, please tell me, and I'll just look somewhere else. [13:45] I will try #android-dev, though. Might be a bit more relevant indeed. === ivoks_away is now known as ivoks === diwic is now known as diwic_afk [14:58] How do you guys prefer patches? Back in the day, I would just push a git repo to Zinc, but since my affiliations changed, that's no longer possible :) [15:12] hi folks, is anyone aware or working on https://bugs.launchpad.net/ubuntu/+source/linux/+bug/524281 ? [15:12] Launchpad bug 524281 in linux-2.6 (Debian) (and 2 other projects) "Tens of wakes per second in "[kernel scheduler] Load balancing tick" on Core 2 Duo even with only 1 core enabled (affects: 104) (dups: 1) (heat: 638)" [Unknown,Incomplete] [15:13] from searching around mailing lists, there are some reports that it's fixed in 2.6.36, so maybe it merits a backport of whatever patch was applied upstream. [15:25] soren: I suppose that they prefer git format-patch'ed... [15:33] apw, how about a new meta upload for omap4 (so we actually get the new kernel onto the images) [15:33] ogra, can do shortly [15:33] * ogra hugs apw [15:34] thanks a lot ... i just wasted a day trying to find out whats wrong with the new kernel ... rsalveti just pointed out that we still use the old one [15:34] doh [15:34] * ogra thinks with over 40 you need glasses :P [15:36] * rsalveti also hugs apw [15:37] it'll take me a bit as my main dev box just wedged on X and i am trying to figure out why [15:40] cking, heh i can hear you it seems through my X hang === bjf is now known as _bjf === ivoks is now known as ivoks_away === BenC___ is now known as BenC [17:11] soren, email in a format-patch format patch to kernel-team@ [17:11] sidnei, we are aware, though noone has had a chance to go look for that patch yet [17:11] if you find out w [17:12] which patch it is do add it to the bug [17:20] apw, ok, i'll try to find it [17:41] apw, added some links to patches on linus tree that seem related to the bug [17:41] apw, hope that helps [17:41] apw, im candidate for trying out a ppa'd package with the fixes, since i can repro on my laptop [17:46] ogra, hey ti-omap4 meta ... they already seem to be in sync ? [17:48] oh ... hrm perhaps not ... [17:58] ogra, ahh we had a bodge in there, new one uploaded, let me know if that doesn't work for you === diwic_afk is now known as diwic === JanC_ is now known as JanC [19:28] ogasawara, what you doing working [19:43] No more mainline kernels ? [19:45] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/604122 [19:45] Launchpad bug 604122 in linux (Ubuntu) "mmc0: Got command interrupt 0x00030000 even though no command operation was in progress. (affects: 1) (heat: 6)" [Undecided,New] [19:45] seem to occur again :( [21:24] hey guys, brand new laptop, and I can't boot any kernel above 2.6.31.x, ideas as to what to look for? [22:14] hello, the latest maverick kernel wouldn't boot on my toshiba nb205 netbook, what information do you need in the bug report? [22:15] it boots normaly with 2.6.35-17 boots normally, but -19 wouldn't boot [22:55] nxvl, file a bug with the working kernel and whatever output you can get from the broken one [23:16] hey guys, brand new laptop, and I can't boot any kernel above 2.6.31.x, ideas as to what to look for?