=== amitk is now known as amitk-afk === amitk-afk is now known as amitk === robbiew1 is now known as robbiew === lilstevie is now known as lilstevie|ZNC === lilstevie|ZNC is now known as lilstevie === hrw|gone is now known as hrw [08:47] morning [12:53] * ogra twiddles thumbs watching lightspark build [12:58] gah [12:58] the linker fails [12:58] :( [13:01] linking qt took so far 1.5GB of memory and still linking [13:02] heh, but it doesnt fail for you at least [13:02] * ogra curses intel centric code [13:02] ogra: with 10GB of swap it should not [13:02] armv7l/Debian/lib/liblightspark.so.0.4.4: undefined reference to `fastYUV420ChannelsToYUV0Buffer_SSE2Unaligned' [13:02] armv7l/Debian/lib/liblightspark.so.0.4.4: undefined reference to `fastYUV420ChannelsToYUV0Buffer_SSE2Aligned' [13:02] silly ! [13:03] auch [13:03] * hrw -> off [13:04] hrw: I was only able to pass the linking part with 1gb of swap [13:11] rsalveti, static no-neon change for QT re-uploaded [13:12] waiting for approval [13:12] (just FYI) [13:12] ogra: cool, thanks [13:13] We've got Pico Node running on a Beagle-xM with Ubuntu 10.10 now. :) [13:13] congrats [13:13] ogra: what is the package version? will push the same one with with neon on the ppa [13:13] cool :-) [13:13] Thanks. :) [13:13] rsalveti, ubuntu4.1 again [13:14] rsalveti: Once it's there, would you please point me at it, I know some people that will definitely want it. [13:14] i hope it doesnt clash with anything, please wait until i have confirmation [13:14] ScottK: sure [13:14] Ubuntu is proving way more stable and nice than Angstrom, especially for networking. [13:14] ogra: ok [13:14] ogra: Since it was source accepted the version number is used. You'll have to bump it. [13:14] slangasek said he removed 4.1 from proposed but i'm not sure that removes everything [13:14] I also have two different wireless dongles working with Beagle and Ubuntu. :) [13:15] ScottK, hmm, k [13:17] * ogra hasnt seen any wireless dongles that dont work on ubuntu [13:17] ogra: Cool. :) Ubuntu is very wireless friendly. :) [13:18] we try our best :) [13:18] I think moving to Ubuntu for Pico Node was the best choice. :) [13:19] We are working on getting a 2.6.36 kernel up on it now. [13:22] shouldn't be that hard, as maverick's 2.6.35 is quite similar with upstream [13:22] We have a problem with an undefined symbol. [13:22] rsalveti, so ubuntu4.2 it is (and uploaded, waiting for approval) [13:23] ogra: ok, that will be accepted [13:23] thanks [13:24] When I put a beagleboard on my robot, it will have WiFi. :) [13:26] you will even be able to run a unity robot with it *g* [13:26] Unity robot?? [13:27] My robot is W.A.L.T.E.R. [13:33] Right now, I am focused on Pico Node though, and getting it completed. [13:40] hmm, so it needs that for lightspark to build we need some assembler guy to re-write the SSE asm stuff for arm [13:40] to bad [13:43] ogra: since ubuntu uses .35 kernel I do not see such neither [13:43] ogra: lucid ignored one of my wifi dongles [13:43] bad dongle :P [13:44] i just said i havent seen any that doesnt work ... doesnt mean that there arent any [13:44] very good dongle ;D [13:44] :) [13:44] 80211n one [13:45] well, all i have and all my friends have did work :) [13:46] but then i dont buy exotic HW [13:47] ;) [13:47] it was cheapest 802.11n usb dongle [13:47] worked fine with my x86 machines [13:50] * ogra goes for food [14:49] hey guys if I have a kernel with built-in modules like pata and then replace it with a kernel with pata as a module how do I force initramfs-tools to include that module if it cannot detect it [14:52] /etc/initramfs-tools/modules should suffice right? [14:58] Neko_, it should, although initramfs.conf -> MODULES=most should take care of it to, I'd expect [15:02] eureka pretty simple install maverick on beagle did the aptitude install for the full gui but then had to aptitude install for the x11vnc cannot start the x11vnc any ideas? [15:02] Warning: cwillu comes with ABSOLUTELY NO WARRANTY, including warranties of merchantability and fitness for a particular purpose. Do not consume cwillu. If injested, contact a kernel developer immediately. Not suitable for children under the age of 24. May cause addiction. May cause insanity. May cause slight tingling in the fingertips when exposed to sunlight. [15:02] if modules.dep is properly generated it should just be pulled in [15:10] cwillu_at_work, modules=most doesn't seem to cover it and I can't work out where it gets this list [15:14] modules.dep as ogra said [15:14] /etc/initramfs-tools/modules should suffice, it's just nice to have things working properly [15:15] (and by nice, I mean very nearly vital :p) [15:17] modules.dep doesn't really tell anything to initramfs-tools though does it? [15:18] like "I actually need this to boot" [15:18] it will just pull in the ones the ones you list are dependent on too [15:18] how would it know from modules.dep that pata_fsl is required to boot from pata [15:18] that's what the "most" is for though [15:19] how does it tell what to not put in [15:20] is there a list of modules pointless to have in initramfs which it culls out so most != everything? [15:21] dunno - I guess everything except storage and filesystem-related drivers are not useful in there [15:21] You could try booting with rdinit=/bin/sh and see what modules actually end up loaded in the initramfs environment [15:26] you might have toi add the module in question to hook_functions, some module sets are hardcoded in there [15:26] i.e. network card drivers definitely are [15:29] have a look at the auto_add_modules() function in there [15:35] at which point you might just want to add the damn module to the modules file :p [15:36] well, if the module sits in a proper kernel dir it will get auto copied [15:36] i guess its another of these broken platform modules [15:36] I'm trying to remember if pata is the deprecated one [15:37] that dont live in the right kernel subdir [15:37] it should live under kernel/drivers/ata [15:37] i.e., isn't it unified with scsi now? [15:37] then it should live in kernel/drivers/scsi [15:37] or if its more generic it shoudl live in kernel/drivers/block [15:38] * cwillu_at_work waits for his home machine to answer the phone [15:38] using ide on that one, it'll know :p [15:38] ... pick up the syn [15:38] ugh, dynip fail [15:39] * cwillu_at_work retries the ip from his webserver [15:40] there we go [15:40] oh, there is also kernel/drivers/ide :) [15:40] that might be very old pata [15:40] that's the one that's deprecated [15:40] yep [15:41] Neko_, which module are you looking for again? [15:41] well, initramfs-tools cpoies it nontheless [15:41] *copies [15:41] Neko_, there's no "pata" module [15:41] not sure there ever was [15:42] just a bunch of device specific pata_*.ko [15:42] pata_fsl [15:42] i guess thats what he means [15:42] I don't have that driver [15:43] likely living ion some odd platform subdir [15:43] instead of a kernel subsystem dir [15:43] I checked the whole modules folder [15:43] you dont have his kernel :) [15:43] and I'm not looking at an arm machine [15:43] (what arm machines have ide on them, that actually work with recent ubuntu?) [15:43] that too, but he likely uses a special fsl kernel as usual [15:44] none [15:44] dove has SATA [15:44] Neko_, quit using non-existing hardware [15:44] we dont support FSL anymore, but i think their latest babbage boards have a plain PATA port [15:44] so theye might also use pata_fsl [15:45] but i cant tell, since we dont support it anymore :) [15:45] (the releases of FSL babbage we supported didnt have PATA) [15:54] pata_fsl is in 2.6.31-fsl tree [15:54] and it is using libata [15:55] well, the auestion is where the module lives [15:55] *question [15:55] must be a path that initramfs-tools copies [15:55] else its left out [15:55] ./drivers/ata/pata_fsl.o [15:55] and initramfs tools only cares for proper kernel subsystem dirs [15:56] kernel/drivers/ata is copied with MODULES=most [16:02] okay [16:02] boot with brak=premount [16:02] lemme get back with steev and see if he can get some debug [16:02] *break=premount [16:03] that will drop you into the initramfs shell so you can inspect [16:03] ok [16:03] if it really lives in that dir, i would suspect it gets copied but doesnt load [16:03] that is the effect we're assuming [16:03] right [16:04] but before we think, why doesn't it load, I wanted to make sure it wasn't just not there [16:04] same thing as with your audio modules i suspect [16:04] well I modularized those only yesterday otherwise they were built-in [16:04] ata) [16:04] copy_modules_dir kernel/drivers/ata [16:04] ;; [16:04] thats from hook-functions [16:04] one part thats run for MODULES=most [16:05] it means it recursively copies all .ko files below that dir [16:07] yeah I see it I just don't "believe" it :) [16:07] well, it works :) [16:14] ogra: there were rumours that natty wants to get rid of initrd [16:14] hrw, huh ? [16:14] i doubt that [16:14] since lucid you *can* boot without initrd [16:15] but only if you dont use a certain set of packages [16:15] hrw: hehe, you were dreaming probably ;-) [16:15] i guess they might want to get rid of current initrd generation scripts [16:15] I know you don't like it :-) [16:16] well, initrd is supposed to be switched to upstart completely [16:16] so yeah, the scripts will heavily change [16:16] I just hope that natty will still allow to boot without initrd [16:16] but as long as we offer dm-raid or lvm or encryption for any disk stuff, initrd will not be dropped [16:19] sure, but for toys like ac100/efikasb you usually do not do that [16:21] oh, and mountall [16:21] fsck need mountall from initrd before the rootfs is mounted RW [16:21] that would have to be changed [16:47] ogra: wrt the qa testing of the daily images, I hadn't tested them because I didn't see them. They are being posted at http://cdimage.ubuntu.com/ubuntu-netbook/daily-preinstalled/current instead of http://cdimage.ubuntu.com/ubuntu-netbook/ports/daily-preinstalled/current/ (i,e. not in /ports). My mirror scripts didn't see the change, nor did my browser. [16:47] I do check daily. [16:48] have a nice weekend [16:48] GrueMaster, hmm, that might be due to persia's quest of letting ports die === hrw is now known as hrw|gone [16:48] http://cdimage.ubuntu.com/ubuntu-netbook/daily-preinstalled/current/ has the right images though [16:49] Ah. Well, if this is permanent, I'll need to update my mirror scripts. [16:49] no idea if thats permanent or even how they got there [16:49] Nice for people to tell me. [16:49] hrmm. [16:49] but i'd suspect its due to some cdimage reorganization [16:49] there was a BOF for it at UDS [16:49] Also, only omap4 images. [16:49] Yea, I kind of missed UDS. [16:49] yes [16:50] the schedule was online :P [16:50] we dont have omap3 kernels yet [16:50] so no omap3 images either [16:50] ok [16:50] and i dont expect them before A2 [16:50] Yeah, but I failed to publish the announcement about the change. I'll get that done in the next 48 hours (have to fix a couple scripts along the way) [16:50] What's the hold up? Ubuntu vs Linaro kernel? [16:51] GrueMaster, right, see the ML discussion [16:51] waiting for a statement from kernel and security teams [16:51] if it is clear we dont get any support we will build omap3 but only as unofficial images [16:52] That shouldn't stop us from building dailies for testing. They're unsupported anyways. [16:53] The rest is paperwork. [16:53] we dont have a kernel [16:53] i cant build images without one :P [16:53] Can't use the existing maverick kernel? [16:54] its gone [16:54] The documentation for adding kernels is supposed to be published before Alpha-1, but there may not be time to actually use that documentation before then to make a kernel. [16:54] Huh? How so? It is a release package in main. It shouldn't be gone. [16:55] GrueMaster, The "linux" source package no longer provides that binary package. It's not uncommon for the set of binary packages to change between releases. Oh, and "main" doesn't mean anything :) [16:55] it is built from linux [16:55] linux doesnt build that flavour in natty [16:56] we need to move the linaro kernel to main [16:56] which i wont do until the discussion has settled [16:56] Ok, I (kind of) get it. [16:57] building with universe kernel will put a lot of extra work on us which i'd like to justify first [16:57] i.e. by security and kernel team refusing to take the support [16:57] Why? It's a two-line change, and if care is taken with the seeds, doesn't otherwise affect the image. [16:58] Should be independent of who offers to maintain the kernel. [16:58] (or the timing of that offer) [16:59] heh [16:59] two line change ? [16:59] we need a universe mirror on antimony for that [16:59] and way more than two lines [17:00] Isn't that already present for building Xubuntu Desktop, Ubuntu Studio, and Mythbuntu? [17:00] not that i know of [17:00] How do those get built then? I thought it was the same infrastructure. [17:00] in any case i wont bother for omap3 until thats sorted [17:00] We don't currently build those for arm. They may exist for other arch's. [17:01] But I do agree that we shouldn't hold up test image builds for paperwork that can get sorted over time. [17:02] Which one is antimony again? I thought that was separate from the livefs builders.