[00:25] back in a bit === jjohansen is now known as jjohansen[afk] [00:31] rock. Yama is in security-testing-2.6#next === bjf is now known as bjf[afk] [00:35] kees: nice, congrats! [00:35] thanks, now for the next set of crazy patches... [00:36] kees: I saw your pull request, will try to get it applied after I upload the 2.6.35-6.9 kernel [00:37] ogasawara: cool; no big rush, but I wanted to get it in so the /proc/sys paths can get normalized in other packages for when it does appear (I assume next week due to milestone) [01:22] cnd: around? [01:22] jcrigby: yep [01:22] what's up? [01:22] I have a dumb question [01:23] got assigned to do kernel packaging for linaro some time ago [01:23] just started looking into it today [01:23] heh [01:23] the dumb question is what crank to a turn to create the debian source pkg from a checked out kernel git tree [01:24] yeah... [01:24] I just want to understand how the ubuntu kernel works before tweaking for linaro [01:24] our packaging can seem rather obtuse from the outset :) [01:24] we're trying to document stuff on our wiki [01:24] let me find the right page [01:24] I did search the wiki but not much success [01:25] I did try do_source_package=true do_full_source=true full_build=true fakeroot ./debian/rules printenv install-source [01:25] but that just made the kernel source tarball [01:26] jcrigby: https://wiki.ubuntu.com/KernelTeam/KernelMaintenanceStarter [01:26] ahh [01:26] thanks [01:26] oh, do you just need to know how to create the src package? [01:26] that you can upload to a ppa or archive? [01:27] if so, check out: https://wiki.ubuntu.com/KernelTeam/KernelMaintenance#Preparing an upload [01:27] (note the spaces in the url) [01:27] https://wiki.ubuntu.com/KernelTeam/KernelMaintenance#Preparing%20an%20upload [01:27] thats what I happen to be looking into right now yes [01:28] thanks again [01:32] sure === kamal is now known as kamal-away [07:39] yo [07:39] ndiswrapper isn't playing nice with 2.6.35 on ubuntu [07:39] dkms loops [07:41] ogasawara, apw. you around? [07:59] LLStarks: apw is not around [08:00] LLStarks: And I would have thought that ogasawara would still be in bed :) [09:48] Gah! Go *away* vga16fb! [09:49] I've got a perfectly cromulent inteldrmfb providing fb0! Your services are not required! [09:51] Weren't we going to stop binding vga16fb to everything? Is that still happening? [10:06] RAOF, Depends on how quick the other fb driver is. I got an older system where it unfortunately is not [10:57] smb: how to compile a -pae kernel. there is no such target in fdr, binary-generic-pae [10:58] smb: thanks in advance [10:59] cooloney, Hm, must admit i have not selectively tried that. But it get produced with binary as a target on i386 [11:00] cooloney, And the binary-generic-pae works here. Just tried [11:00] cooloney, Are you calling it on amd64? [11:01] * smb pokes cooloney [11:01] oh, it is i386 [11:01] It is [11:02] let me try [11:02] On 64bit you don't need PAE [11:02] $ fdr binary-generic-pae [11:02] make: *** No rule to make target `binary-generic-pae'. Stop. [11:02] cooloney, uname -m ? [11:03] oh, need i build it in chroot? [11:03] * smb slaps cooloney [11:03] Always [11:03] smb: my bad. sorry man [11:03] No problem :) [11:04] smb: thanks man, building now. [11:05] cooloney, You are welcome. just never tell me you are not using a chroot to build again. ;-P [11:07] smb: i remember. hehe, I play cross compiling for a long time [13:24] * smb cries over the concept of history in bzr [13:25] smb, whats the story with the ext4 patchset that bjf proposed last week? Is it gonna make it into pre-proposed anytime soon? [13:28] tgardner, Though proposed is about to be moved kees was asking for security and guess what will win [13:28] smb, hmm, lets work with kees on that. is there a real kitten killer out there? [13:28] tgardner, Unfortunately there is the point release sort of soon and I fear time will run out [13:29] tgardner, I have not looked at the cves itself yet. They seem medium to low [13:29] tgardner, Actually there is another scary series which potentially should go into the point release [13:30] smb, scary series of CVEs ? [13:30] tgardner, I just prepared test kernels that got the writeback patch series (12patches) [13:30] tgardner, No not cve [13:30] Thats about our long time on sync or unmount [13:31] smb, ok. then I think you should ask kees about deferring these CVEs until after the point release so we can get some bug fixes in. [13:32] tgardner, I try to look at the complexity of the cves. If there is change to release them next week, this would be enough time for another proposed baking [13:33] smb, I'd sure like to see your writeback and bjf's ext4 fixes get into a pre-proposed kernel soon. it needs some exposure [13:34] tgardner, Steve and Brad did some test runs which sounded quite promising (no seen regression, against some fails before) [13:35] smb, are they running that XFS test suite? I think that is how some of the issues were originally found. [13:35] And I hope to get feedback on the writeback stuff from kees and some reporters with the test kernels [13:35] tgardner, yes they do [13:57] is "relatime" a defaulted mount option with Lucid? [13:57] cking, I think you have to specifically turn it off. [13:58] tgardner, that came in when? Karmic, Lucid? [13:58] cking, I just remember seeing it as an install option, but don't remember exactly when. I think its been there since at least Hardy [13:59] tgardner: sorry for the confusing. i just assume wanna review the patch from TI's tree. -:) [13:59] cooloney, nope, I want 'em prequalified from you. [13:59] tgardner: as soon as i got updates from sebjan, i will pull them into my tree after i reviewed them. [14:00] tgardner: yeah, understood === kamal-away is now known as kamal [14:02] What is the filename of the source code file associated with the driver file /lib/modules/2.6.32-22-generic/kernel/drivers/net/wireless/ath/ath5k/ath5k.ko for the loadable kernel module ath5k? [14:03] Has ath5ko been compiled from several source code files? [14:04] s/ath5ko/ath5k.ko/ === sconklin-gone is now known as sconklin === bjf[afk] is now known as bjf === ericm|ubuntu is now known as ericm-Zzz [15:46] * bjf is off to the dentist === bjf is now known as bjf[afk] [15:53] hello [15:55] LLStarks: I'm guessing you're seeing bug 590090 [15:55] Launchpad bug 590090 in ndiswrapper (Ubuntu) "package ndiswrapper-dkms 1.56-1 failed to install/upgrade: ndiswrapper kernel module failed to build (affects: 8) (dups: 2) (heat: 181)" [Medium,Confirmed] https://launchpad.net/bugs/590090 [15:55] LLStarks: which is due to the upstream commit: [15:56] commit 22bedad3ce112d5ca1eaf043d4990fa2ed698c87 [15:56] Author: Jiri Pirko [15:56] Date: Thu Apr 1 21:22:57 2010 +0000 [15:56] net: convert multicast list to list_head [15:57] I have followed this guide http://blog.avirtualhome.com/2010/05/05/how-to-compile-a-ubuntu-lucid-kernel/ but , in the record "skipabi=true skipmodule=true fakeroot debian/rules binary-core2" I receved this "make: ** Nessuna regola per creare l'obbiettivo <>. Arresto " [15:58] cnd: ddebs should be trickling in according to pitti [15:58] ogasawara: cool, thanks! [15:59] LLStarks: I'm curious as to why you're installing the dkms package for ndiswrapper though, we already roll it in the Ubuntu kernel [16:02] hello [16:03] I have followed this guide http://blog.avirtualhome.com/2010/05/05/how-to-compile-a-ubuntu-lucid-kernel/ but , in the record "skipabi=true skipmodule=true fakeroot debian/rules binary-core2" I receved this "make: ** Nessuna regola per creare l'obbiettivo <>. Arresto " [16:05] salvarane: what did you name your flavor [16:05] salvarane: did you name it core2 as well? [16:08] config.flavour.rtai [16:09] salvarane: skipabi=true skipmodule=true fakeroot debian/rules binary-rtai [16:09] not found [16:09] I change the name for you [16:09] the real name is skipabi=true skipmodule=true fakeroot debian/rules binary-rtai [16:10] salvarane: i followed those instructions this morning, and everything is building happily [16:10] salvarane: maybe pastebin everything you did to compare [16:11] ok [16:11] I have followed this guide http://wiki.linuxcnc.org/emcinfo.pl?Ubuntu10.04Notes [16:12] step by step [16:20] salvarane, that looks basically correct. I would have run 'fdr updateconfigs' after 'fdr editconfigs' [16:20] there is lots of good stuff at https://wiki.ubuntu.com/Kernel [16:21] ok [16:22] I followed the guide http://blog.avirtualhome.com/2010/05/05/how-to-compile-a-ubuntu-lucid-kernel/ but I changed the name core2 into rtai [16:23] I following your link https://wiki.ubuntu.com/Kernel thanks [16:24] ogasawara, ndiswrapper-common is pulling in the dkms package [16:26] who's the maintainer of that dkms package? I actually didn't even know it existed till yesterday. [16:27] can I use kernel vanilia into ubuntu lucid ? [16:30] hi [16:32] I am running a Ubuntu generic kernel on all the hosts running 8.04 server. is there a way to install server kernel on this host ? [16:32] We have found a bug on the automated install [16:33] Please suggest/guide [16:33] kaushal, there should be a meta package called linux-image-server [16:35] tgardner, ok [16:36] so if i install it i should be having server kernel ? [16:36] yep [16:36] also do i need to reboot it ? [16:36] i think yes [16:36] i think yes [16:37] the issue is that we have this issue on all the 300 production servers [16:37] is there a easy way to do it ? [16:38] kaushal, I suggest you consult with your IT guys. [16:40] tgardner, Thanks === bjf[afk] is now known as bjf [16:53] * bjf is back [17:49] rebooting [17:58] Hello folks [17:59] Just talked to NthDegree in #ubuntu about some strange lockups on one of our servers with raid5 and xfs as the fs. [18:00] I found a thread on the xfs ML about an race window in the xfs code. [18:01] From what I can tell, the Ubuntu Lucid kernel source is based on 2.6.32.11 and I think those xfs fixes were introduced in 2.6.32.12 [18:02] atleast there's a hole bunch of xfs commits there. [18:03] NthDegree told me that there were a PPA for the mainline kernel, however he didn't know if any of the >.11 kerneles were stable enought to give a try. [18:03] What's you oppinion on this? [18:04] The only info I could get from the tech on site (server co-located) were this: Kernel Panic - not syncing: xfs_fs_destroy_inode: cannot reclaim 0xdc50efe4 [18:04] Azoff, the kernel in -proposed is based on 2.6.32.14 [18:04] Azoff, just today a Lucid update went out which brings in ... as tgardner says [18:04] oki, great! [18:05] Sorry, I'm kinda unfamiliar with Ubuntu pakages. How does this -propused package get into main tree? [18:06] Or can I try it without it being in the tree? [18:06] Azoff, Proposed is just an archive pocket which you have to opt in [18:06] ah, I see. [18:06] Packages go there before they are moved to the place where you will get them normally [18:07] how long before it will be moved (approx)? [18:07] Aprox -3 hours [18:07] IOW it has been done [18:08] You should see kernel updates when you check now [18:08] Azoff, http://launchpad.net/ubuntu/+source/linux tells you the current status [18:08] no, I won't :( [18:10] I guess I have to wait a couple of hours... thanks for your help! [18:51] http://news.bbc.co.uk/2/hi/technology/10466317.stm Half a million Vaios 'recalled' for overheating - "But Sony said that this is "not a recall" and that the problem can be rectified with a software patch." [19:00] * tgardner lunches [19:22] ogasawara: I just made a small update to my yama pull-request commits, so if you pulled already I can prepare a separate commit. otherwise, pretend nothing happened, and pull normally whenever you're ready. :) [19:22] kees: I haven't pulled yet, so I'm good [19:53] ogasawara, Julian Andres Klode === sconklin is now known as sconklin-lunch [20:16] * ogasawara lunch [20:16] lunch === sconklin-lunch is now known as sconklin [22:03] kees: for the yama pull request, should the following also be reverted? [22:03] commit 066cf0cf62729b6107d9b6bda195f4346b1607e7 [22:03] Author: John Johansen [22:03] Date: Thu Jun 24 08:12:55 2010 -0700 [22:03] UBUNTU: SAUCE: fs: block hardlinks to non-accessible sources AppArmor portio [22:03] [22:03] This is the AppArmor portion of [22:03] commit 069cb89e17c6dc5b2a1de2469746bc42935850fb [22:03] Author: Kees Cook [22:03] Date: Wed May 12 09:03:08 2010 -0700 [22:03] [22:03] UBUNTU: SAUCE: fs: block hardlinks to non-accessible sources [22:03] [22:03] Applied on top of the current upstreaming version of AppArmor [22:05] ogasawara: if you're doing the reverts manually, yes. if you're just pulling from my repo, no. (I had to handle the "conflict" of that patch already) [22:05] kees: ok cool [22:06] Yama and AppArmor should only be running into eachother in security/Kconfig and security/Makefile now, but I dealt with that in the first Yama commit. everything else should be separate. [22:11] ogasawara: just to make sure we're on the same page, the top of my tree is 9578dd34c5949d41a1237d2ad080bcf438d963e7 [22:11] kees: yep, that's what I have [22:11] okay, rockin' === bjf is now known as bjf[afk] [23:17] ogasawara: do you want me to re-organize my commit messages to include the cherry pick comments, or have you already pulled? [23:18] kees: I haven't pulled yet, will probably do it tomorrow [23:20] kees: I'm too lazy to build across all archs to get the abi's so I'm just gonna wait till I upload within the hour and then grab the abi's in the morning [23:21] ogasawara: okay, so should I re-do all the commit messages to include the cherry pick comment syntax you recommended? [23:22] kees: that'd probably be best [23:22] kees: just for tracking purposes [23:26] ogasawara: okay, I'll do that === kamal is now known as kamal-away [23:38] ogasawara: since Yama adds a new CONFIG option, how should I record that in the git tree? [23:40] kees: usually I just do something like "UBUNTU: [Config] Enable CONFIG_FOO=y" or something similar === sconklin is now known as sconklin-gone [23:45] ogasawara: if this is good, it'll be waiting for you tomorrow too: http://kernel.ubuntu.com/git?p=kees/linux-2.6.git;a=commitdiff;h=12c65e02eed1309d24f6288e6bc7914857942af3 [23:45] kees: yep, looks good [23:52] ogasawara, i'm still waiting for juliank to expain the required ndiswrapper dkms package. [23:52] it doesn't work on the current maverick kernel [23:54] LLStarks: ok thanks, keep me posted [23:57] LLStarks: it actually might be preferred to go the dkms route rather than us maintaining it in the kernel [23:57] LLStarks: we do the same for the broadcom drivers, ie use DKMS [23:58] LLStarks: but that's something I can raise on the kernel-team ml [23:59] would the module installation loops be a kernel problem or ndiswrapper problem?