[00:01] silly question, but i'm not a kernel hacker. if i add stuff to debian/config/$arch/config, do these needs corresponding entries in ubuntu/aufs/Kconfig? [00:16] laga: yes [00:29] hello is the kernel source in hardy include patch rt ? to compil [00:39] benje: there's already a linux-image-rt package you can install in hardy [00:39] BenC, this version of kernel doesn't work for amilo [00:40] but the patch is in source trying apply new tellit ;) [00:41] BenC, amilo don't get dvd working cause of pciide share irq options when get it off dvd working now i try to get sound card i get problem with compiling alsa-sources i retrying [00:41] with new fresh compil [00:42] after we solve compilation of alsa with this kernel i post it until fix is commited [01:11] hi, did somebody tell linus that paravirt enabled does not let fglrx + nvidia compile without a tricky hack? [01:11] with 2.6.27 [01:11] gpl only error... [01:12] for exports: pv_cpu_ops and pv_lock_ops may not be gpl only [01:14] now i know a trick that can compile any driver with gpl only symbols, but i dont know if that is ment to be used ;) [01:15] think about it... [03:31] Kano is so smart...none of us could have figured out how to make a module fake being gpl compatible to circumvent in-kernel license enforcement [03:33] * lifeless snorts === rikai_ is now known as rikai [08:08] log === thunderstruck is now known as gnomefreak [14:38] hi rtg [14:38] did you try binary 3d drivers with 2.6.27 yet? [14:40] Kano: I'm not working on 2.6.27, try BenC [14:40] why are you always the last with a commit then ;) [14:40] Kano: I don't use them [14:40] BenC: you should,becauce it is critical currently when you want to use em [14:41] That makes no sense [14:41] both fglrx and nvidia run into gpl-only problem [14:41] when paravirt is enabled [14:42] Well, I'm not disabling paravirt [14:42] the changes needed to make the drivers compile are small [14:42] then 2 symbols my not exported as gpl only [14:42] And I'm not changing upstreams decision on EXPORT_SYM_GPL() [14:42] may [14:42] these are wrong [14:42] Then get upstream to change them [14:42] tell upstream to do so [14:43] and acknowledge that [14:43] if you want to be able to use binary drivers without extra hacks [14:43] You're the one worried about it, don't boss me around to take care of the things you care about [14:43] well YOU should be worried about because YOU dont want to patch it. and ia sure ubuntu users want to use those drivers! [14:44] As far as I know, nvidia binary driver is working fine right now [14:44] i know how to compile em and get around the problem [14:45] Kano: does this only affect 2.6.27? [14:45] yes [14:45] Then I don't have time for it right now [14:45] a 2 line change [14:45] Kano: A two line change the breaks the license of the GPL [14:46] pv_cpu_ops and pv_lock_ops are needed [14:46] Email linux-kernel@vger.kernel.org [14:46] BenC: the 2.6.27-rc2 kernel you pointed me at yesterday seems fine, audio works as well as it was from my own build of alsa (everything except the digital mic, which the alsa guy didn't manage to get working) :) [14:46] explain this to them... [14:46] Ng: excellent, thanks [14:46] np, I hope this helps the quest for latest and greatest alsa in intrepid ;) [14:46] BenC: whre in the gpl it is broken when you change the code? you only have to publish the changes! [14:47] Kano: if you change a GPL-only export for the sake of a non-GPL compliant module to use, that is definitely breaking the license [14:47] BenC: show me the line in the licence [14:48] Kano: Changing the export is legal. However, it indicates that the third party module is illegal. [14:49] If you're doing it in order to distribute the third party module along with the kernel, that's almost certainly illegal. [14:49] how about compile tricks to avoid the problem? [14:50] Kano: end-users are free to do what they want to get it to compile...however, we (as in Canonical) are not willing to put ourselves at risk of just the community backlash that it would incur [14:50] No. If a closed module needs a symbol that's under EXPORT_SYMBOL_GPL, that indicates that the closed module is a derivative work of the kernel. [14:51] Therefore there is no way to distribute that module unless you also distribute source. [14:51] mjg59: well ti was not when the module was introduced [14:51] Kano: So? [14:51] EXPORT_SYMBOL_GPL is a hint, not a technical protection mechanism [14:51] mjg59: so when you distribute it as source like you use dkms then it is legal? [14:52] If the module is available under license terms compatible with the GPL, then that's fine [14:52] Kano: the entire kernel has been GPL'd since it started...anything compiled against it is generally considered a derivative work [14:52] In which case it should simply have a MODULE_LICENSE header that indicates it's under a GPL-compatible license [14:52] BenC: well but linus never wanted that you can not use 3d drivers [14:53] and never forced nvidia to opensource the driver [14:53] Kano: He doesn't care whether it's a 3d driver or not [14:53] Linus didn't think that the nvidia driver was a derivative work of the kernel [14:53] That's mainly because the binary blob was not written for Linux [14:53] But Linus isn't the only copyright holder of the kernel [14:54] well binary drivers usually are binary blobs + a wrapper to get integrated into the kernel [14:54] Kano: which is just re-exporting functions, which is even more devious, and doesn't shield from being a derivative [14:55] http://www.nvnews.net/vbulletin/showthread.php?t=117209&page=2 [14:56] 17+23 [14:57] take a deep look into 23 [14:59] BenC: it seems you removed some files from the aufs tree when you integrated it, like br_nfs.c. was that intentional? [14:59] BenC: besides that, did you notice that rt2x00 does not compile? === superm1 is now known as superm1|away [15:21] BenC: and one last thing: why do you need 686 optimizsation,you gain no speed with it, you only make it impossible to boot the system with amd k6-(2/3) which are fast enough when they have got enough ram [15:22] 586 is really enough [15:23] Kano: rt2x00 compiles for me locally [15:32] compiling again your git, when i tried last it did not work nor did it work with an addional git pull from master [15:32] in master there is ath9k btw [15:33] laga: I probably removed it because we weren't compiling it in [15:38] BenC: i'll add it back then. [15:38] i'm getting an error which prevents me from building the intrepid kernel: http://pastebin.ca/1168076 [15:38] how can i resolve this? [15:38] Kano: if there are problems with the nvidia driver you should talk to me, since I'm the new maintainer. If something doesn't work I can write a patch for it. Any links to the issue? [15:39] http://www.nvnews.net/vbulletin/showthread.php?t=117209&page=2 [15:39] the hack at 23 is only needed because the exports are gpl only [15:40] the hack works perfectly, maybe too good ;) [15:40] i patched the legacy nvidia drivers myself and fglrx 8-7 too [15:40] all work now [15:40] but you need that hack which should not be the case [15:41] correctly it has to work when the official patch at 17 is applied [15:43] Kano: ok, thanks, I'll have a look at it [15:43] if you want the patch for the legacy drivers just tell me [15:44] Kano: thanks but I'm waiting for NVIDIA to make them compatible with the new Xorg first ;) [15:44] legacy 3 aka 173 is compatible [15:45] so 2/4 [15:45] laga: that pastebin is just a warning [15:46] tseliot: try my script with -v 3 option [15:46] Kano: I don't consider 173 legacy yet since 177 is still beta [15:46] it is legacy because new cards dont run with 173 [15:46] it is for geforce 5 series only needed [15:47] geforce 6+ run with 177 [15:47] Kano: I know ;) [15:47] also the legacy definition is from nvidia [15:48] I know this too [15:48] then change you mind [15:50] Kano: I know it's a legacy driver but I still find it funny that NVIDIA has no non-legacy stable driver. That's all. [15:50] BenC: i dont know what you have got locally but when i check it out it get this: http://paste.debian.net/14596 [15:50] tseliot: will 177 supercede 173, or does it drop support for things that are in 173? [15:51] BenC: it adds the support for some cards and drops a series [15:51] Kano: I don't know what you have, but it isn't the latest git...I reverted the patch that causes that [15:51] tseliot: nvidia is too quick to drop support, IMO [15:51] BenC: i checked it out 30 min ago! === superm1|away is now known as superm1 [15:52] Kano: Then you need to fix something...line 1127 in my tree is an empty line [15:53] i can not fix anything, i only did a checkout of the official ubuntu-next git [15:54] maybe you did not publish your change? [15:54] Kano: confirmed at http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-next.git;a=blob;f=drivers/net/wireless/rt2x00/rt2x00dev.c;h=f42283ad7b023b697e4aa9375b5c01557ae54f2b;hb=HEAD [15:54] BenC: I see your point but fortunately (for users, not for me...) they keep supporting legacy drivers [15:54] Kano: no, it's definitely there...the problem is on your end [15:55] BenC: i wiped out the whole tree, it is a new checkout. maybe git is broken [15:55] cant you do a checkout on your system and try it [15:56] i am sure i did not change a line myself in that driver [15:56] Kano: I did a checkout already [15:57] to confirm [15:57] Kano: it's a problem on your end [15:57] Kano: git-pull and see what happens [15:57] always at the same file for every checkout? [15:59] BenC: you were right about the warning ;) i scrolled up and it showed me the real error [16:20] BenC: it looks like the kernel still needs patching for aufs. but that's just to make some functions public [16:20] but the symbols don't have to be exported anymore [16:20] laga: CONFIG_FS_AUFS=y [16:21] yes. it still needs patches, or it won't compile. [16:21] laga: Ah, if we have to patch s/static// anyway, then we might as well export [16:21] Only reason for doing =y was to avoid patching [16:21] that'd be good because it would allow aufs to work as a module, too. [16:22] well i made already live iso images with 2.6.27 ;) [16:22] is there any reason not to have a full-blown aufs module like in hardy? [16:23] BenC: what do you think about 586 not 686 like hardy? === asac_ is now known as asac [16:24] laga: Other than not liking to hack the hell out of the kernel, no [16:24] i dont think that 2.6.27 should require a new system, that will break many old systems during upgrade [16:29] BenC: okay, then i will get a complete copy of the checkout you had back then, apply the kernel patches and submit it to you? [16:32] BenC: now you pushed some new things which introduced a new error, could you please update your system clock, you always changes things in the past [16:32] you added gfs which included asm/semaphore.h, but the new position is linux/semaphore.h [16:33] you definitely changed things at max 1 h ago and still it shows last change was 21h ago [16:34] BenC: http://paste.debian.net/14600 [16:36] Kano: it's not system clock, it's called a rebase [16:37] sure and then the error is on my end, clear... [16:38] Kano: let me make this perfectly clear...2.6.27 is not my primary work...I do it to make things easier in the future [16:38] Kano: if you are depending on it to produce some whack off-shoot of Ubuntu, then you'll just have to live with my delays [16:44] BenC: so would you be OK with including a complete copy of the source? that'd also make upstream happier i suppose [16:44] err, complete copy of the source == everything that normally lives in aufs/fs/aufs/ in the upstream source. [16:45] laga: sure [17:05] BenC: could you fix that new error? [17:09] Kano: yes, I will fix it [17:10] can i wait for it [17:22] You sure can [17:23] 1 h or less [17:23] Kano: it probably happened already [17:24] 22h ago? [17:24] but since the clock is off.. it probably hasn't :D [17:24] it depends how fast and in what direction you are travelling atm [17:24] i thinks so... === mkrufky is now known as mkrufky-away [18:34] BenC, which linux-restricted-modules-common should i download in order to get the iwlwifi-4965-2.ucode firmware? [18:34] hyperair: 2.6.26-5.12 [18:34] thanks [18:35] i suppose i'd need to install he 2.6 kernel with it? [18:35] sorry [18:35] i meant 2.6.26 [18:36] No, it will work with 2.6.24 as well [18:36] i see. okay. [18:36] but since iwlagn is only in 2.6.27, it doesn't make much sense to download it for anything else [18:36] hahah [18:36] well. [18:37] i did download the 2.6.27 kernel from kernel.ubuntu.com [18:47] that will work then === mkrufky-away is now known as mkrufky [21:20] BenC: http://paste.debian.net/14616 === rikai_ is now known as rikai [21:42] BenC: when you added aufs, did you use fs/aufs or fs/aufs25 from the upstream source tree? [21:44] BenC: i got a checkout from sourceforge from the same date as you (according to BOM) and i'm seeing lots of changes. [21:45] BenC: aaah, your checkout was probably too early for that monday release [21:48] well 2 extre LM flags are missing... [21:49] i dont know why you need gfs [22:49] just disabled it and it compiled... === superm1 is now known as superm1|away