=== kamal-afk is now known as kamal [00:34] kirkland, alrighty I'll start tracking people down [00:44] zul, ping per kirkland he said to see if you could maybe lead the server Q&A tomorrow since he can't? [01:31] hi, i'm having some issues with 10.04 and the MCP89 chipset on a MacBook Pro trying to boot 10.04 LiveCD === kamal is now known as kamal-away [04:27] Can anyone tell me what 'invalid opcode: 0000 [#1] SMP' in dmesg means? [04:41] max: something tried to execute an invalid instruction [04:41] can you pastebin the whole dmesg? [06:22] '~$ ps -ef | grep flush; root 252 2 0 May05 ? 00:00:00 [flush-8:0]' What process spawns this process? Where is described the function of this process? [06:26] I'm pretty sure that no userspace process spawns that process, and it's a kernel thread to handle the write-queue to one of your block devices. [06:31] What IO scheduler does ubuntu 10.4 use? [06:38] RAOF As the [] clearly states, it is a kernel process. [06:39] RAOF: Where can I find additional information about it? [06:40] Didn't we have this coversation before? [06:40] Ah, yes “look for a definition of stuct file_operations in the filesystem implementation, it should have a line like '.flush = ext4_flush'” [06:58] RAOF: I havew found a generic definition of the structure in /usr/src/linux-source-2.6.32/Documentation/filesystems/vfs.txt but not yet the specific definition in the ext3 filesystem implementation. [06:59] grep ext3_flush **/*.c might help. :) [07:21] good morning! [07:22] hey psurbhi [07:23] psurbhi: why the change from csurbhi? === cmug is now known as Guest57205 [07:33] jk, no such reason.. just made it aligned to my name :) [07:33] * psurbhi compiles the patch for 543617 [07:50] jk, when are you traveling to brussels? [07:52] psurbhi: 2 days time [07:56] ok === Guest57205 is now known as cmug [08:08] * apw waves to jk- [08:08] o/ all [08:13] heya apw [08:14] long time no see ... hows DT going [08:14] yeah, good! looking forward to some intresting discussions at UDS :) [08:15] still getting my head around how to best handle the crazy ARM clocks though [08:15] i think the clock configuration on imx51 is turing-complete [08:16] heh now that is an interesting thought [08:16] yeah, maybe I can calculate a fibonacci sequence using only the clocks [08:16] sounds like a nightmare [08:17] naw, it's ok. just trying to figure out a way to represent it that isn't going to bite us later [08:17] bjf, you 'there' already ? [08:17] apw, yup [08:17] bjf they run out of beer yet? [08:18] apw, the first run at them was tried last not, but they held out [08:18] s/not/night/ [08:18] have the breweries been informed that the kernel team is coming to town? [08:19] bjf, i have low expectations sadly [08:19] jk-, we always try and tell them to order 3x, they always laugh, until the end of the first night [08:19] apw, low for what? [08:19] aha [08:19] ah [08:19] bjf, the beer holding out [08:20] lots of good hiking/walking trails in the forest [08:20] the showers are _AWSOME_ [08:20] need to figure out the laundry setup [08:21] Anyone happen to have a link to best-practice for reporting fail-to-boot bugs? Someone in #ubuntu-bugs wants to report one, and I'm not sure `apport-bug linux` is the right answer. [08:24] Oh well, normal bug then :) [08:24] apw: i am trying to cross build ti-omap kernel, but failed [08:24] pls help me take a look here [08:24] http://pastebin.ubuntu.com/428760/ [08:25] even with skipabi=true and skipmodule=true [08:25] yep you have to have validly formed abi/module info even with skips enabled [08:26] skip just says "if the result is bad, ignore it" it doesn't mean "don't do the checks" [08:26] apw: right, i know that. [08:27] apw: i just fresh git clone the tree [08:27] which one? [08:27] and checkout ti-omap branch [08:27] and cross build it [08:27] then failed like this [08:27] looks like the build scripts corrupt [08:27] ok ... i'll try it here [08:27] apw: ok, thanks. [08:28] cooloney, looks more like the abi info is corrupt to my eye, but let me try here [08:30] apw: thanks, dude [08:32] cooloney, building now, will be about 10 mins [08:32] cooloney, what is the sha1 of your tip [08:32] or the base you are using which ever [08:32] apw: cool, your build machine is powerful enough [08:32] arm cross builds are about 2x the speed of native ones, go figure [08:33] apw: oh, you mean the ti-omap branch of main? [08:33] apw: it is 841b9542344b64c7d7607404ca9fd54a8ce72a74 [08:33] yeah just wanted to know where you are working from, to confirm its the same [08:34] apw: the default thing, i did not change anything before cross building [08:34] morning [08:34] cking, apw, morning :) [08:34] cking: morning, [08:43] cooloney, ok this happens here for me as well, will investigate [08:47] apw: you are the man [08:47] thanks [08:49] np ... [09:14] cooloney, ok i think i have this sorted out ... give me a couple to confirm [09:16] bjf, whats the network like there? you able to work on it? [09:17] * psurbhi breaks === psurbhi is now known as psurbhi-afk === cmug is now known as Guest48926 [09:23] apw, asking bjf if the network works is a little premature - once we have a gazallion developers on the network it's gonna suck [09:24] cking, indeed was more interested as he is there for 3 days before [09:27] apw, i'm in the conference area now, on the wireless and it's working nicely right now [09:28] * apw wonders where smb is [09:28] apw, don't know what it's going to be like when the riff-raff show up :-) === psurbhi-afk is now known as psurbhi [09:28] bjf, we'll kill it in the face in 10s of arriving i am sure [09:29] apw, there is a design team sprint happening right now and somehands going one [09:29] bjf, yeah so not so many people i guess [09:29] apw, nope, not yet [09:30] smb, yo ... you got anything pending for the ti-omap lucid branch? i've just found a packaging bug in it thats stopping all builds and i want to slam it in [09:30] apw, we've taken over the hotel wireless as usuall, it's our folks managing it [09:30] thats something at least [09:30] apw, Nope nothing for that as far as i can remember [09:30] apw, kind of have the run of the place to myself, very nice actually [09:30] ok, you ok with me pushing it? right now its not possible to build ti-omap at all [09:31] bjf, must be very odd [09:31] apw, sure go for it [09:31] apw, it is a bit [09:31] smb: yeah, that also affects me as apw said [09:31] which floor shall i work from today [09:31] cking, you mentioned someplace for me to do my conference writeup? [09:32] cooloney, you also running ti-omap stuff? [09:32] apw, they have a very nice coffee bar room with lots of 'snacks' [09:32] bjf: https://wiki.canonical.com/KernelTeam/Conferences/ [09:32] bjf, sounds like a a disaster [09:32] smb: right, i try to cross build it today [09:32] and met such issue which apw should slove it now [09:32] cooloney, thanks [09:33] smb, cooloney, ok pushed ti-omap ... cooloney could you rebase on that and test [09:33] apw: no problem, [09:33] if it builds ok for you as well then i'll call it good and send out the patches [09:34] apw: synced and building now on emerald [09:35] apw, Any other known disasters. I partially try to recover from upgrading to lucid on my main netbook [09:35] smb, na thats the only one i know about at least [09:36] * smb wants his high resolution back on the external lcd [09:36] smb, so very picky [09:36] apw: so i am just wandering what's the CONFIG_UNUSED_SYMBOLS? [09:37] it makes EXPORT_UNSED_SYMBOLS produce things which are meaningful in the ABI [09:37] apw: it will cause the abi checking fail, interesting.. [09:37] apw, I am not even starting about thunderbird loosing all of lightning in that process. 'cause I know you don't care. :-P [09:37] only cause the abi-checker is itself broken [09:46] * psurbhi tests the patch for lp543617 [09:55] apw and smb, the new CONFIG_UNUSED_SYMBOLS patch fixed this building issue of ti-omap. [09:55] thanks [09:56] apw, what do you mention in a patch taken from kernel bugzilla? (which is not yet upstream) [09:56] as in, instead of commit upstream [09:56] OriginalLocation: perhaps [09:56] ok [09:56] thanks [09:58] psurbhi, In case it is in linux-next you could take that sha and linux-next [09:58] psurbhi, Or i would say cherry-picked from [10:01] smb, thanks :) [10:15] smb, apw, there seems to be a second patch which has come in just now from Jens Axboe [10:15] https://bugzilla.kernel.org/show_bug.cgi?id=15906 [10:15] I tested Dmitrys patch and that works fine [10:16] psurbhi: Error: Could not parse XML returned by bugzilla.kernel.org: The connect operation timed out [10:16] psurbhi, interesting [10:16] Hm someone mentioned another patch yesterday [10:16] smb, its just come in today [10:16] infact i could not see this posting till morning [10:17] When kees was asking yesterday, someone pointed to a mail thread [10:17] * apw still can't see it cause its taking so long [10:17] http://www.spinics.net/lists/linux-ext4/msg18780.html [10:17] apw, yeah.. its taking a long while..to open for me too [10:18] smb, ok.. it was posted on kernel bugzilla today.. anyway.. i think we will wait for a while? [10:18] I would suggest you make a kernel with both of them and give it to kees for testing [10:18] and meanwhile also test this new patch [10:18] psurbhi, well i guess if the patch looks reasonable we want to test both together, and get them to kees [10:18] for testing [10:18] smb, apw ok [10:19] where do i keep the kernels? i have one on emerald now [10:19] psurbhi, people.canonical.com ? [10:19] apw, ok [10:20] psurbhi, i normally make like lpNNNN-lucid in mine and put them there [10:24] ok [10:27] Dmitrys patch is a very small change.. and is tested ..whereas the patch by Jen Axboe introduces some new status "WB_SYNC_NONE" to fix this one..and its untested by anyone so far [10:27] apw, smp ^^ [10:27] psurbhi, so is jens' patch a replace ment for dmitry's ? [10:28] apw, yes, his version fixes the bug in the proper way.. whereas dmitrys patch is sort of a hack [10:28] damn, so it depends if jens is intending on CC: stable his [10:29] how about taking dmitrys patch as of now and then later reverting that to jens one.. once it goes to stable? [10:29] i guess we could consider taking Dmitries with the revert of the hack we have and then move to jens' if we get it via stable [10:29] ok :) [10:29] * cking wonders if electioneering allowed on voting day [10:29] so do you think, i should send the email out on u-k mailing list? [10:30] cking, what does that mean? [10:30] psurbhi, apw Hm, yeah. We want to move as quickly to a solution that does not regress other cases and solves the long umount [10:30] and then later probably revert that in favour of Jens' patch when it is in stable [10:30] smb, yep...so do i send out the email with dmitrys patch? [10:30] ok..cool.. so i will do it right away.. [10:30] seems like a plan then ... so yes get it out on the list [10:31] * smb hopefully remembers untils then [10:31] apw, just got a phone call from Labour giving me some push to vote for them. That's not allowed is it? [10:31] so, then i will put a note in the patch email saying we wait for Jens patch and later revert to it [10:31] cking, it seems rather unethical to me, but i suspect they are just calling all not yet voted people to encourage them which is techinically allowed [10:31] cking, "Thanks you for calling. I would have voted for you, but now that you call..." [10:31] psurbhi, yeah please do put a note [10:32] heh.. [10:32] smb, my response exactly [10:33] psurbhi, And clearly make it a SAUCE patch [10:45] cking, you have it soo easy, you can't believe all the calls we get when it's election "season" here, and it goes on and on and on and ... [10:45] cking, what's nice in oregon is that we can vote by mail, and once you vote, it's marked in the roles and they stop calling you :-) [10:46] cking, the rule is "vote early, vote often" [10:46] bjf, 6 weeks is enough for me - didn't see one doorstep visit at all and got just 2 phone calls [10:46] cking, can't remember how many people i had at the door [10:47] cking, what's funny/annoying is i'm registered democrat, the wife is registered republican, we get it from both [10:48] cking, she doesn't vote republican, just too lazy to change her registration [10:50] bjf, I think because I did a postal vote I got zero hassle from the parties this week [11:01] apw: the CONFIG_UNUSED_SYMBOLS on mvl-dove isn't set, does it need to be, including any trees based on it? === cnd_mini is now known as cnd [11:02] cnd, it should be, but it clearly doesn't break anything so you don't need to worry [11:02] it will get fixed naturally as we move forward [11:02] ok, thanks [11:02] cnd is it not very early where you are? [11:03] apw: I'm transitioning to brussels time [11:03] so it's not as terrible on the first day or so [11:03] are you mad :) [11:03] ? [11:03] that sounds a bit mad, waking up early voluntarily [11:04] that, and my wife is in ob/gyn rotation [11:04] meaning she gets up at 4:15 am [11:04] nnng [11:04] she just put in a 30 hr day non stop yesterday... [11:04] so safe for her patients [11:04] got up at 4:15 on tuesday, came home at 8 am on wed [11:05] mental, they do that here, one does not want to be in a hospital i cannot do a thing that tired [11:05] she's just a med student so she doesn't really DO too much yet [11:05] I don't know if attendings and residents work off the same schedule [11:05] but they may [11:06] heh i bet she is giving drugs and the like ... its all stupid [11:06] apw, it's only people's lives at stake. apw, you need to be alert so you don't break millions of PCs [11:06] nah, doesn't have that power yet === cooloney is now known as cooloney-afk [11:06] is that what caused the EC breakage? [11:06] apw, do you need more caffeine? [11:07] :) [11:07] i always need more of that [11:07] * apw got up at 7 to vote, and i am already useless ... [11:07] apw, when do you normally get up? [11:07] keenie [11:07] lass was going and made me go too [11:08] more like 8 [11:08] my stomach gets all discombobulated when I get up any more than an hour earlier than usual [11:08] so I'm hoping to head that off by doing a slower transition of sorts [11:09] this is really just a test to see how well this works [11:09] apparently beer softens the blow [11:09] cnd, sounds bad ... though getting up at 4 for a week before you get there sounds bad too [11:09] I just remember going to a conference in greece a few years ago, and I was dead tired all the time [11:10] yeah throw in beer till any sickness can be attributed to that [11:10] apw, I've actually been inching my way [11:10] yeah figured you might [11:10] an hour earlier to bed and to rise each day [11:10] still mad [11:11] cnd, you will get used to it after a few years of travel around the globe [11:11] perhaps [11:11] I'm starting to become very jealous of sabdfl's private jet :) [11:12] and I haven't even done any international travel yet! [11:13] even his jet doesn't get rid of the timezone change [11:14] yeah, but at least he isn't packed like a sardine [11:14] maybe he can get some sleep? [11:14] I never am able to [11:16] argh, marvell hal driver build infrastructure is utter crap [11:16] it breaks on anything higher than -j1 [11:16] anyone know of a way to force one specific subdir of the kernel to be -j1? [11:17] they do this stupid thing where they have a dummy.o target at the front of all the real targets, and dummy.o depends on their rule to make all the makefiles for the rest of the targets [11:18] which breaks horrendously as soon as you do -j2 and it tries to compile the first real target before it has a Makefile [11:21] cnd, hrm ... t [11:21] there must be a way [11:21] Hello & good morning [11:22] hi [11:22] hi manjo [11:22] manjo, good bet for bug 416487 [11:22] apw, what I tried is generating the Makefiles myself, and then committing that result [11:22] Malone bug 416487 in bluez "Bluetooth Doesnt Work - Eee PC 1005HA-P" [Undecided,Incomplete] https://launchpad.net/bugs/416487 [11:22] cking, hey checkout google.co.uk graphic [11:22] of course the generated makefiles use absolute directories in their -I statements... [11:22] gah [11:22] though it may just be one Makefile [11:23] so I might be able to patch that to $(src) [11:23] apw, nice google image of No.10 [11:24] cnd can't you just fix the makefile so that all of the files depend on dummy.o's depenacies [11:25] apw: seems logical, and I've tried to think of ways, but I can't figure it out [11:25] apw: it's like: obj-y := dummy.o dir1/ dir2/ dir3/ dummy_end.o [11:25] cnd, how come our mvl-dove trees do not have this issue [11:25] where dummy_end.o goes and deletes all the Makefiles! [11:26] apw, perhaps they don't build these drivers? [11:26] they definatly make new makefiles and delete them [11:26] I'm not sure [11:26] bjf, was it you who did mvl-dove originally ... does this ring a bell [11:26] apw, just catching up with the discussion [11:27] cnd, are you starting from mvl-dove or making your own tree ? [11:27] apw, starting from mvl-dove, and applying some patches on top [11:28] i ask because mvl-dove is configured differently as it cannot build 'out of tree' [11:28] apw, cnd, yes i remember that there were some big issues with mvl-dove builds [11:28] ok so you should have that [11:28] apw, cnd, one of the things is it wouldn't build out-of-tree [11:28] in fact, it's all mergeable into mvl-dove except that we need to build uImages instead of zImages [11:28] bjf: well, I'm trying to build in-tree [11:28] bjf, that's a tad unpleasant [11:28] cnd, well thats normal, all arm use those [11:28] apw, cnd, that's why you see at the beginning of the build, we rsync the code over to the build directory [11:29] bjf yep, if cnd is based of mvl-dove that should be turned on anyhow [11:29] apw, only if they use uboot [11:29] does anyone use redboot? [11:29] if not, then we should just switch over to making uImages [11:29] i mean we generate the uImages as they are flashed as i recall ... [11:29] cnd, it sounds like you started with their tree and are bringing it back into ours [11:30] cause our current build process doesn't allow building different outputs for different flavours [11:30] cnd, how about starting with our tree and applying their changes ontop of ours [11:30] ogra, am i right in thinking the kernel-flash extracts the uImage and we fake install zImage for arm generally ? [11:30] Lots of folks use redboot [11:30] apw: Yes, kernel-flash should mask the platform-specific kernel installation hackery when it doesn't boot from /boot [11:31] bjf, that's what I'm doing, starting with mvl-dove and applying patches [11:31] cnd, switching from uImages to zImages or vice versa is easy [11:31] cnd ... so we probabally should do the same for this [11:31] apw, not really, zImage is the source for generating the uImage [11:31] apw, so its not really a fake [11:31] bjf, yeah, but integration into the oem builds is what concerns me [11:31] cnd, i don't know what that means [11:31] bjf, well I don't really know all it entails either yet :) [11:31] ogra, yeah i more meant we still just generate zImage for arm and put that in /boot, and do the uImage step at 'update-' time instread [11:32] persia, apw, and its flash-kernel :) [11:32] apw, its a kernel postinst thing yes [11:32] cnd, ^^ make sense? [11:32] apw, no, we were generating uImages at one point for uboot targets [11:32] controlled by /etc/kernel-img.conf [11:32] apw: yes, for the uImage part [11:32] as well as by update-initramfs [11:32] bjf: http://kernel.ubuntu.com/git?p=cndougla/hedley.git;a=commit;h=844a81b8f6332ba7ce701ca7ae79d784d9ef4ab1 [11:33] * persia consistently calls that the wrong thing except when prefaced by apt-get source :( [11:33] bjf, yeah but we have switched to consistenty using the same [11:33] apw, ah [11:33] cnd, where does your kernel live on your device ? [11:33] cnd seems reasonable [11:34] got flash or in a partition [11:34] ogra: we're still just in the board bringup phase, but right now I've been putting it in /boot/uImage [11:34] and made a simple uboot boot script to load it using ext2load [11:34] cnd, is your u-boot capable of reading ext2/3 ? [11:34] ah [11:34] most of this is going to change [11:34] this is just for board bringup really [11:35] yeah, that should work fine, the dove way of doing things should work [11:35] for NAND take a look at the omap version :) [11:35] we just got lucid up on it two days ago [11:35] the important bit is the flash-kernel-installer.postinst [11:35] that does the initial system setup from ubiquity/d-i [11:36] apw: bjf: this is what I didn't catch when I statically generated the Makefiles: http://kernel.ubuntu.com/git?p=cndougla/hedley.git;a=blob;f=arch/arm/plat-orion/mv_hal_drivers/mv_hal/pmu/Makefile;h=60933f5425f1487d90dad88264d79fb9919c1d90;hb=844a81b8f6332ba7ce701ca7ae79d784d9ef4ab1 === cooloney-afk is now known as cooloney [11:36] ogra: yeah, I don't know myself how the live image build process works yet [11:36] I've been putting that off [11:37] since I haven't got a kernel building outside my home dir yet :) [11:37] cnd, bring the device to UDS we can take a look together [11:37] cnd, i remember them reaching all over their tree with relative paths to get to objs [11:37] should be quick to get it going [11:37] and i can add all bits to flash-kernel then, i'm just merging it anyway :) [11:37] ogra: I don't know we'll have the time though, cking was suggesting to leave it at home... [11:38] well, there are the evenings ... [11:38] we just shouldnt spill beer into it ;) [11:38] heh [11:38] I'll think about it [11:38] honestly, I'm not sure whether it's strictly my area anyways [11:38] a good oppotunity to get ogra to fix it for you [11:38] I'm the kernel hwe guy for the project [11:38] there are other people working on the rest of course [11:39] sure, as you like :) [11:40] * cking hates bringing hwe work to UDS as it eats into valuable UDS time [11:41] that and it's a sizeable hunk of metal I don't want to deal with unless I really must [11:41] ..and the parts are limited too [11:50] helli, I run command "hdparm -B 128 /dev/sda", but after a while, it restores to 254 automatically. Anyone meet this? === cking is now known as cking-afk [12:25] fqh, which release you on ? that sounds like something laptop-tools might do [12:26] lag, yo [12:27] smb, cking-afk .... meet lag (Lee) [12:27] hey lag [12:27] he seems to have settled on a nick at last, and one i can spell finally too [12:28] apw, And one you can even live. :-P [12:28] heh [12:35] Rejected: [12:35] Cannot build any of the architectures requested: any [12:35] smb, does ^^ mean anything to you in a PPA upload response ? [12:36] apw, Hmm, not immediatly [12:36] sounds like something a bit mad to me [12:36] Distribution: maverick [12:36] have i spelt maverick right? [12:37] apw, I thought so. Looked better than me doing Maverik [12:38] apw, Can I look at the dsc or build log? [12:38] smb, the reject email is in your inbox as it was the maverick pre-proposed which failed [12:39] linux_2.6.34-1.6~pre201005061100_source.changes rejected [12:41] apw, Does not look any other than all of the previous ones to me... === cking-afk is now known as cking [12:41] lag, hiya! [12:42] yeah i suspect its right, and i suspect we may not yet have chroots or something ... but someone needs to tell me :) === cking is now known as cking-afk [12:42] back in 10 [12:42] apw, Yeah maybe there is no buildd for maverick that volunteers for building [12:42] yeah something like that i suspect [12:45] apw, smb, eking: Hello =:-) [12:45] eeeeeking :) [12:45] its a 'c' [12:45] apw, The e-commerce version of cking. :-) [12:46] cking: Sorry, and hello [12:46] smb: We've met haven't we [12:46] cking: We haven't met yet [12:46] i likes the idea of him squeaking like that [12:47] lag, spot on [12:47] lag, Yeah, /me was the quiet one at the end of the table [12:48] suffering from .en overload :) [12:49] More from music overload I think [12:49] smb: Yes, I remember you [12:50] apw: Who is 'the technical one' you wanted me to meet? [12:50] Is that cking? [12:50] yeah cking is the detail guy :) === cking-afk is now known as cking [12:53] me? detailed? nah [12:53] yeah right [12:53] * apw notes he is 2 foot tall [12:53] yeah yeah yeah [12:56] * cking is a troll that likes shiny hardware [13:02] cnd, about ? mumble test ? [13:02] apw, one sec [13:04] apw, I am running ubuntu-10.04. [13:07] * lag is off to make the most of his final weekdays off for 3.5 months! [13:07] apw, my usb-disk /dev/sdb can keep the setting, but the disk inside the laptop can't. [13:07] cking: Nice to meet you [13:09] lag, hope to catch up with you face-to-face sometime :-) [13:15] fqh, the internal one is the one on we might expect things to manage [13:16] if its 10.04 its likely pm-utils [13:17] cking: I think apw is planning a meet up-town in the next upcoming weeks. It would be good to meet then? [13:24] apw, I tried pm-powersave, but "hdparm -B /dev/sda" tell me that pm-powersave did not change the "APM_level". [13:29] lag, I'm up for it sometime after UDS :-) [13:41] apw, your ports mail doesnt refer to omap and friends, does it ? [13:42] ogra, nope, literally what used to be linux-ports [13:42]