/srv/irclogs.ubuntu.com/2013/02/05/#ubuntu-devel.txt

=== slank is now known as slank_away
smoserpsusi, hey. uploaded your util-linux00:34
smoserwas going to ask you if it would work for gpt. i didn't think it would, but it seems to.00:34
psusismoser, ahh, cool... I don't remember if kpartx speaks gpt or not00:34
psusierr, partx00:35
slangasekfrom what I've seen, it doesn't00:35
slangasekbut maybe I was doing it wrong00:36
psusismoser, I've got patches for upsteam parted to add a resizepart command to use the new ioctl to sync the new, larger patition size as well, so you can just run parted -s /dev/sda resizepart 1 100%00:36
smoserslangasek, psusi well, what i did is shown at https://bugs.launchpad.net/cloud-utils/+bug/108752600:39
ubottuUbuntu bug 1087526 in cloud-initramfs-tools (Ubuntu) "need support for gpt partition tables" [Undecided,New]00:39
smoserit seemed like the kernel/udev could get confused easily (see comment 4), but it *did* work.00:39
smoserpsusi, that'd be pretty nice.00:40
smoserfor now, growpart does basically the same thing00:40
smosergrowpart /dev/sda 100:40
smoserwill work for gpt or msdos00:40
smoser(as of today)00:41
smoserit uses sgdisk or sfdisk00:41
smoserbut it (growpart) does not do the partx call yet. i plan to make an option to it.00:43
psusismoser, why make it an option?  if sfdisk or sgdisk complain about BLKRRPART failing, may as well go ahead and run partx -u00:51
smoserpsusi, yeah, i considered that.00:51
smoserthe only reason to not have it just do it is that it "just didn't" before00:52
psusiwell, before it was run in the initramfs so it wouldn't run into that error, right?00:52
psusiso now it does the same thing, just is no longer restricted to running in the initramfs00:52
psusiare there any other outstanding issues to dropping the initramfs from cloud images?00:53
lifelesspsusi: wait, what ?01:09
psusilifeless, at UDS... what was it, october before last?  in Orlando, we discussed dropping the initramfs as a requirement for simple configs that don't really need it01:21
psusilifeless, cloud image boot time is apparently significantly impacted by the initramfs, and iirc, the one hangup for getting rid of it was growpart01:22
lifelesspsusi: so what would an image look like then?01:25
lifelesspsusi: FWIW cloud images boot in 5-6 seconds for us with the initramfs.01:25
psusilifeless, just the kernel, no initramfs01:25
lifelesspsusi: this worries me, because the initramfs provides an excellent debugging facility when things go wrong01:26
lifelesspsusi: will it be easy to revert for folk building on the cloud images?01:26
psusiabout the only thing that can go wrong in the initramfs with a simple disk config is can't find the root fs01:27
psusiyea.. the plan was to make it optional, and most configs have no need for it, so, faster boot01:27
lifelesspsusi: we're using the cloud images to boot bare metal :001:27
smoseri really doubt the initramfs is a significant speed issue for cloud images. it might be. but it'd shave a second or two at most i'd think.01:28
smoserclearly non-zero01:28
psusithe idea was also to have regular installs initramfs scripts detect whether or not an initramfs is actually required ( i.e. you are using lvm ) and disable it unless overridden01:28
smoserright. this is really a uber-fast desktop boot thing.01:29
psusiohh, maybe it was the arm guys that were driving it because for them it was a big time sink01:29
smoserit was/is a time sync.01:29
smoserand it is not insignificant.01:29
psusisomething about grub being very slow at IO on arm01:29
smoserloading 10M-ish, decompressing, runnign a bunch of scripts ...01:29
smoserwell, bios load is generally slow (which is what grub is using)01:30
smoserso its not going to be the fastest thing in the world, for sure.01:30
psusiiirc, arm has no bios so grub was using its own disk module there, which was supposedly very slow01:30
smoserright.01:30
psusibut it does add a not insignificant time to boot on your average pc desktop and cloud images too01:31
smoseranyway... during that "lets get rid of initramfs" discussion, i raised that i like initramfs , and this was one of the reasons why.01:31
smoserbut then psusi fixed *that* the right way, and now we dont need initramfs for it anymore. which is really cool.01:31
smoserbut doesn't mean we'll drop initramfs. just one less thing to prevent it being dropped.01:32
lifelessok01:32
lifelesswell, let me know what we need to do to keep it on, so the ops guys doing run around gibbering01:33
lifeless:)01:33
psusiwhat benefit does it provide when you are using a simple disk config?01:33
psusiit's whole purpose is to locate the root fs, so if the kernel can do that on its own...01:34
psusidid the kernel uuid finding patches ever get applied?01:34
lifelesspsusi: so - for example - the HP public cloud has two block devices01:35
lifelesspsusi: and there may be LVM particularly if you are using boot from volume.01:35
psusiyea, lvm still would require an initramfs01:36
lifelesspsusi: bare metal cloud will often have more complex setups too.01:36
psusiwell, if / is on lvm anyhow01:36
smoserlifeless, you dont have to worry.01:36
lifelesspsusi: MAAS is moving to using cloud images; nova bare metal uses cloud images.01:36
smoserits not going away.01:36
lifelesssmoser: yeah, I'm not:)01:36
smoserit would be going away in places where its not needed01:36
smoserbut you're listing places where it *is* needed.01:37
lifelesssmoser: that was estalished above; I'm answering psusi's question about where it may be needed.01:37
lifelesssmoser: I may have misparsed that question, of course :>01:37
psusiright, the question  was where is it needed, other than when / is on lvm ;)01:37
smoserwhen / is crypto01:37
psusior that ;)01:37
lifelessor when / has died and you want to poke around01:37
smoserwhen / is iscsi01:37
smoserlifeless, when / has died and yrou initramfs was on /01:38
smoseryou're kind of dead01:38
lifelesssmoser: well, PXE booting initramfs+kernel01:38
psusithat's grub fallback time01:38
lifelesssmoser: / can die, can still examine :)01:38
smoserwell, its not goingt away in pxe boot either!01:38
lifelessindeed!01:39
psusiyea... the idea was for simple disk configs.. i.e. root is on a regular disk partition01:39
smoserand smart people were even discussing ways where grub would fall back automatically to initramfs if no-initramfs seemingly failed.01:39
lifelessnice01:40
smoserjust ask yourself, do you think that OS that runs on your devices from apple have an initramfs ?01:40
smoser(not that they boot fast... but probably not necessary )01:40
=== security is now known as megha
lifelesssmoser: what devices from apple? :)01:41
smoseryour kids devices, lifeless01:41
smoser:)01:41
smoseryeah, thats why i have one.01:41
smoserkids.01:41
lifelessYou have a kid to get apple devices?01:41
lifelessSeems expensive :>01:42
smoserits not significantly more than the price of the device.01:42
smoserand easier to justify to employer01:42
lifelessrotfl01:42
smosercan someone else confirm for me ...01:49
smoserhttps://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/101533901:49
ubottuUbuntu bug 1015339 in debian-installer (Ubuntu) "Missing kernel modules for Mellanox ConnectX HCA Ethernet " [Medium,Confirmed]01:49
smoserthat doesn't look like an installer issue to me, but a driver issue.01:49
smoser(as in lack of pci-ids or some other reason it wouldn't load automatically)01:50
lifelesssmoser: hi01:52
lifelesssmoser: so thats our hardware config; the driver is in a separate kernel package01:52
smoserseparate kernel package, as in ...01:53
lifelesssmoser: linux-image-extra01:53
smoserah. well, that is installed by linux-server -> linux-generic -> linux-image-generic -> linux-image-extra-3.8.0-4-generic01:55
lifelessdoes the server image build include that by default?01:55
smoserie, it was not just missing driver in initrd.gz (at least per the reporter).01:55
lifelesshi01:56
lifelesssmoser: EntropyWorks is the reporter01:56
EntropyWorksHi01:56
lifelessEntropyWorks: I'll just pastebin the discussion so far01:56
smoseroh. there, you're right. it does not include it.01:56
smoserat least per report01:56
lifelessEntropyWorks: http://paste.ubuntu.com/1610977/01:56
smoserso you're right. thats strange.01:56
smoserbut EntropyWorks it seems that just having it in the intiramfs is not enoguh to make it work, right?01:57
smoseryo uahve to load it manually?01:57
smoserat least that goo.gl link there seems to imply that (as does another report to me)01:57
lifelessso there may be two bugs :)01:58
EntropyWorkslet me dig up my notes on this.01:58
lifelessA) Where is my driver!01:58
lifelessB) Why driver not load?!01:58
smoserright.01:59
smoserbut "where is my driver" is not (i dont think) related to "my driver is in -extra"01:59
smoserso yeah, "where is my driver" i san installer/installer build process bug.01:59
smoser"why my driver no load" is a driver/udev/kernel bug i think.02:00
smoserEntropyWorks, so why did you do this udev rule business for modprobing?02:00
lifelesssmoser: right, being in -extra isn't an intrinsic bug (but if -extra isn't in the installer default initramfs...)02:01
smosernah. most every thing is 'extra' from that perspective02:01
smoseri think02:02
EntropyWorksI eventually did something slightly different after getting fustrated.02:02
smoserbut maybe your'e right. i could be wrong.02:02
=== zz_jackyalcine is now known as jackyalcine
=== jackyalcine is now known as Jacky_
EntropyWorksI went and downloaded the http://us.archive.ubuntu.com/ubuntu/dists/precise/main/installer-amd64/current/images/netboot/ubuntu-installer/amd64/initrd.gz and unpacked it...02:04
smoserEntropyWorks, it seems to me what you did with that udev rule should have been equivalent to adding the item to /conf/modules in the intiramfs.02:04
smoserie, by putting it in /conf/modules, you're doing what would happen if you had put it in /etc/initramfs-tools/modules during the initramfs build process.02:05
EntropyWorksI added a 80-mellanox-drivers.rules to /etc/udev/rules.d02:06
smoserright. but why/where did you come up with that?02:07
smoserit would seem to me that you're essentially loading mlx4_en every time a "net" device was found.02:08
smoseri could be missing something.02:08
EntropyWorksits been a while since I repackaged this initrd.gz for the netboot so forgive me while I try to remember. I think I may have just borrowed it from somewhere else. to be honest.02:08
smoserEntropyWorks, thats fine. i was just trying to see if i was missing something.02:09
lifelessEntropyWorks: thats whats in the diskimage-builder rules02:09
EntropyWorksbefore i was doing this.. echo "modprobe mlx4_en ; # Move this line up" >> init02:09
smoser:)02:09
lifelessEntropyWorks: we have both the modprobe and the udev rule02:09
smoseryou clearly should not need both.02:09
smoserright?02:10
lifelesshttps://github.com/stackforge/diskimage-builder/tree/master/elements/mellanox02:10
EntropyWorksright02:10
smoserand the udev rule is wierd, an I right?02:10
smoseryou're loading that module every time a net device is seen02:10
smoser(but then, i'm kind of confused as to how linux knew it was a net device since clearly the driver didn't completely recognize it)02:10
smoserEntropyWorks, just so we're clear, i'm grateful for your help, i'm just trying to figure out if you were doing stuff for a reason, so that i can make sure we take that into account.02:11
smoserlifeless, where does 'init' from mellanox/init end up? hnow does that get rendered ?02:13
lifelesssmoser: it gets attached to the deploy ramdisk init script02:14
lifelesssmoser: the deploy ramdisk is a special special creature02:14
smoserlifeless, thanks.02:16
EntropyWorksI think this covers how I have been fixing my initrd.gz at the moment02:18
EntropyWorkshttp://paste.ubuntu.com/1611008/02:18
EntropyWorksI may have added etc/udev/rules.d/010_net-hotplug.rules but I don't think I did.02:19
EntropyWorkssmoser: honestly its been awhile and if there is anything confusing I can build another initrd.gz from scratch if you need it02:22
=== tumbleweed_ is now known as tumbleweed
=== OutOfControl is now known as benonsoftware
=== Jacky_ is now known as zz_Jacky_
pittiGood morning06:22
=== jono is now known as Guest40438
shadeslayerhi pitti06:28
infinitysiretart: Hey, handbrake (in both experimental and raring) seems to be implicitly declaring a few functions that exist nowhere in the source, and I can't find reference to them in libav headers either.07:04
infinitysiretart: You might want to grep build logs for "implicit", and figure out where those missing functions went. :P07:05
infinitysiretart: (Maybe it's depending on either an old feature that was removed from libav, or a new feature that's not yet in the Debian packages?)07:06
tjaaltoncjwatson: um, nvidia-173 isn't in -proposed yet?07:56
tjaaltoncjwatson: for .207:56
tjaaltonoh it's nvidia-173-updates08:00
tjaaltonjockey doesn't suggest any drivers for my gf8600gt08:01
tjaaltonthe window is empty08:01
tjaaltonshould there be a jockey launcher in system settings?08:12
tjaaltonon 12.1008:12
dholbachgood morning08:12
tjaalton..behind software sources it seems08:14
pittitjaalton: it's not jockey any more; right, software-properties supersedes it08:22
pittiadmittedly hard to discover08:22
tjaaltonyeah08:26
tjaaltonit should shout out tho that the system needs to be rebooted once the driver has been changed.. just logging out will end up in a crash and a bug report..08:32
tjaaltona confusing one too08:32
=== tkamppeter_ is now known as tkamppeter
dholbach@pilot in09:46
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: dholbach
seb128@pilot in09:46
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: dholbach, seb128
seb128dholbach, alter! in what order do you go through the sponsoring list? I will try to help you a bit, I missed my shift while I was in SF09:47
* dholbach hugs seb12809:47
* seb128 hugs dholbach09:47
dholbachI don't care which way we go - maybe we just say what we're looking at?09:47
seb128dholbach, ok09:47
seb128dholbach, I will try to grab what is desktopish first09:47
dholbachseb128, I'll do cliff-tablib09:48
seb128dholbach, doing alacarte nautilus language-selector xorg-gtest09:49
dholbachwoohoo09:49
seb128;-)09:49
dholbachseb128, taking a look at ubuntukylin-theme09:52
seb128Riddell, hey, do you have any comment on https://code.launchpad.net/~gunnarhj/ubuntu/raring/language-selector/help-update/+merge/145491 (dropping language-selector-kde)?09:52
Riddellseb128: I'm not currently expecting to go back to using language-selector so I guess it's fine09:53
seb128Riddell, ok, thanks, we can still bring it back some way if really needed09:53
seb128pitti, ^ I'm sponsoring l-s then09:54
pittiseb128: merci09:54
seb128pitti, or did you want to do it/had it ready?09:54
seb128pitti, de rien ;-)09:54
pittino, I didn't09:54
seb128good09:54
=== henrix_ is now known as henrix
cjwatsontjaalton: my mistake, sorry09:57
cjwatsontjaalton: I noticed the missing task and my eyes tend to pass over existing invalid tasks ...09:58
tjaaltoncjwatson: heh, yeah10:03
=== zz_Jacky_ is now known as jackyalcine
=== jackyalcine is now known as Jacky_
=== caribou_ is now known as caribou
dholbachseb128, taking a look at ktorrent10:07
seb128dholbach, looking at telepathy-logger-qt10:08
=== henrix is now known as henrix_
dholbachseb128, looking at ubuntu-sounds10:13
=== henrix_ is now known as henrix
Esokrateshowto apply ubuntu settings to compiz compiled from source?10:15
EsokratesI have compiled and installed compiz but when I login the only thing I get to see is the wallpaper10:16
Esokratesso I change to a VT and start compiz using: env DISPLAY=:0 compiz –replace composite opengl move resize decor compiztoolbox mousepoll wall expo animation switcher unityshell10:17
seb128Esokrates, try asking on #ubuntu-unity but the compiz package ships a profile, export COMPIZ_CONFIG_PROFILE="ubuntu"10:21
seb128Esokrates, not sure how you can get the profile and apply it from a source build10:21
Esokrates@seb128 thank you10:22
udevbotError: "seb128" is not a valid command.10:22
seb128dholbach, looking at sphinxbase10:23
=== Ursinha is now known as Ursinha-afk
seb128dholbach, commented on the python-tz from stokachu, seems like a possible direct sync10:38
seb128stokachu, ^10:38
jibeldoko, python2.7 2.7.3-15ubuntu1 broke the testsuite of bzr, I filed bug 111607910:40
ubottubug 1116079 in bzr (Ubuntu) "Test bzrlib.tests.test_tuned_gzip.TestToGzip.test_enormous_chunk fails - potential regression in python2.7 2.7.3-15ubuntu1" [Undecided,Confirmed] https://launchpad.net/bugs/111607910:40
jibeldoko, it's probably caused by this change "Issue #1159051: GzipFile now raises EOFError when reading a corrupted file with truncated header or footer.", could you have a look10:40
jibel?10:40
dholbachseb128, cool10:42
seb128dholbach, looking to vino10:46
dholbachwow, you're quick :)10:47
seb128dholbach, french efficiency my german friend :p10:52
dholbachhaha10:52
dholbachtjaalton, Sarvatt: do we have someone we can ping upstream about bug 985202? :)10:54
ubottubug 985202 in libxrandr (Ubuntu Precise) "libx11 causes kwin to crash on login (over NX protocol)" [High,Triaged] https://launchpad.net/bugs/98520210:54
tjaaltondholbach: sure10:56
seb128pitti, hey, can you set https://code.launchpad.net/~baltix/ubuntu/precise/ubuntu-defaults-builder/remove-quotes-from-firefox-bookmarks-titles/+merge/137107 as needs work/work in progress/whatever status is available to indicate it needs the comments to be addressed before being ready for sponsoring?10:56
dholbachtjaalton, and a bit more recently bug 111214710:56
ubottubug 1112147 in mesa (Ubuntu) "gbm_dri_bo_create fails to initialize bo->base.base.format" [Medium,In progress] https://launchpad.net/bugs/111214710:56
=== Ursinha-afk is now known as Ursinha
tjaaltonhmm, could be that they'd have better success via the list(s)10:59
tjaaltonalthough the mesa-dev list does get all the bugmail too11:00
seb128pitti, can you set those as merged (they are in the unapproved queues for SRU):11:01
seb128https://code.launchpad.net/~arges/ubuntu/precise/iptables/fix-lp982961/+merge/14594111:01
seb128https://code.launchpad.net/~arges/ubuntu/quantal/iptables/fix-lp982961/+merge/14594211:01
tjaaltondholbach: got a rev-by for the libxrandr patch, I'll upload it11:06
* dholbach hugs tjaalton!11:06
tjaaltonthe mesa patch doesn't apply as-is to the version we're preparing, will investigate further..11:07
dholbachseb128, looking at the raring part of firebird2.511:12
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
seb128dholbach, looking to bind911:17
seb128lamont, hey11:23
seb128lamont, I'm looking at https://bugs.launchpad.net/ubuntu/+source/bind9/+bug/1090593 , e.g https://launchpadlibrarian.net/130013535/quantal.debdiff ... those debdiffs only change the ip, I saw that your version adds a "D.ROOT-SERVERS.NET. 3600000      AAAA  2001:500:2D::D" ipv6 entry as well (in addition of autotools noise)11:24
ubottuUbuntu bug 1090593 in bind9 (Ubuntu Quantal) " D.ROOT-SERVERS.NET changing January 3rd 2013" [Undecided,New]11:24
seb128lamont, should the ipv6 entry be added in those SRUs or are the current debdiffs with only the ip change fine for upload?11:24
dholbachfor security uploads to the current devel release - is it normal to use a *ubuntu0.13.04... version number?11:28
seb128xnox, hey, is there still anything blocking https://bugs.launchpad.net/ubuntu/+source/player/+bug/979841 to be uploaded (you mentioned python issues)? you can probably fix the bug number/target and upload rather than just to block on the submitter to fix those typos11:31
ubottuUbuntu bug 979841 in player (Ubuntu) "robot-playercam crashed with SIGABRT in __assert_fail_base()" [Medium,In progress]11:31
xnoxseb128: let me check if python multiarch hacks got uploaded or not.11:33
=== Ursinha is now known as Ursinha-afk
=== Jacky_ is now known as zz_Jacky_
cjwatsondholbach: I wouldn't bother, just increment the version as usual11:39
dholbachcjwatson, my question was the other way around (the suggested patch had 13.04 in it) and I thought it looked weird :)11:40
cjwatsonYeah, I know.  It's not forbidden but I agree it's weird.  If I were sponsoring that I would just amend the patch and tell the contributor I'd done so (if it were otherwise OK)11:41
dholbachcjwatson, since I have you here... do you have any feelings about bug 1102107? (I just saw that an earlier request in Debian was not acted on yet: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=676533)11:41
ubottuDebian bug 676533 in openssl "please enable arm assembler code in openssl" [Wishlist,Open]11:41
ubottubug 1102107 in openssl (Ubuntu) "Add basic arm64 support" [Undecided,New] https://launchpad.net/bugs/110210711:41
dholbachyeah, I changed it to ubuntu1 :)11:41
cjwatsonthose bugs are quite different - arm != arm6411:41
cjwatsonI'll review 1102107, thanks11:41
=== Ursinha-afk is now known as Ursinha
cjwatsondholbach: (And BTW 676533 was already applied in Ubuntu)11:43
seb128xnox, I'm unsubscribing sponsors from https://bugs.launchpad.net/ubuntu/+source/libsemanage/+bug/1103271 btw, your comment indicates that the patch will need to be reworked if it's still needed at all, right?11:43
ubottuUbuntu bug 1103271 in libsemanage (Ubuntu) "libsemanage: Add crossbuild support" [Undecided,New]11:43
seb128xnox, oh, you already did that, ignore me ;-)11:43
=== zz_Jacky_ is now known as jackyalcine
=== jackyalcine is now known as Jacky_
dholbachcjwatson, yes, I saw that - I just thought that if the request for arm wasn't replied to after half a year, we probably shouldn't block on debian for arm6411:44
dholbachthanks a lot for having a look at it!11:45
cjwatsonEh, it's really not that related - that's arm optimisation, not arm enablement11:46
cjwatsonEnablement rightly normally jumps the queue11:46
dholbachoh ok11:47
cjwatsondholbach: I'm going to have to look around as I'm not convinced the second patch is correct either - it doesn't seem to agree with my reading of the arm64 abi11:48
dholbachthanks a lot11:49
seb128@pilot out11:52
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: dholbach
seb128dholbach, ok, if/when pitti is around and mark the merge requests I pointed before as needs work/merge we should be < 35 in the queue11:52
seb128dholbach, time to get some lunch ;-)11:52
* dholbach hugs seb12811:53
dholbachgood work11:53
* seb128 hugs dholbach11:53
seb128thanks, you too11:53
=== francisco is now known as Guest13450
jamespage@pilot in11:55
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: dholbach, jamespage
jamespagedholbach: how low can we go :-)11:55
dholbachyes yes yes! :)11:55
* dholbach hugs jamespage11:56
dholbachpiloting party :)11:56
dholbachjamespage, looking at 'player' right now11:56
jamespagedholbach: OK _ I'll look at the bind SRU11:56
dholbachjamespage, if you check the backlog - seb128 pinged lamont about something related to bind earlier11:57
dholbachnot sure if it's the same issue11:57
seb128jamespage, I looked at bind9, I wanted to know if we should add the ipv6 record lamont added in his ppa and debian or not11:58
jamespageseb128: right-oh12:01
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
margao/12:25
margaI just uploaded two debdiffs to https://bugs.launchpad.net/ubuntu/+source/gnome-control-center/+bug/110058712:25
ubottuUbuntu bug 1100587 in gnome-control-center (Ubuntu) "gnome-control-center does not remove system-wide proxy settings from /etc/environment when switch from manual to automatic mode." [Undecided,Confirmed]12:25
margaI was wondering if this could be considered for an SRU or not.12:26
seb128marga, hey, that debdiff seems fine to me, I will sponsor it today12:37
margaseb128, tnx12:38
=== _salem is now known as salem_
seb128marga, yw, was there any other bug/patch that you want review on? (asking because some might be lost in email backlog from our side)12:41
pittiseb128: sorry, missed your pings! doing now12:43
seb128pitti, hey, danke!12:43
margaseb128, I need to discuss the dropbear patch with vorlon12:44
margaBut I'm guessing I still need to wait a few hours for him to be around.12:44
seb128yeah12:44
seb128I will let non desktopish ones to Steve ;-)12:44
margaAside from that, I think I don't have anything else pending.12:44
seb128ok, good12:45
lamontseb128: "my version" is the current-n-correct file, not some patch applied to some randomly old file. :D13:05
lamontseb128: my ppa has some bits sitting in it - though I suspect I'll have even-newer for raring13:05
lamontseb128: so yes, the answer is we should have all of the entries in the file...13:06
seb128lamont, hey13:09
seb128lamont, well the only difference between "current-n-correct file" and "some randomly old file" is a one line ipv6 record ;-)13:09
seb128lamont, your version also has some autotools noise, not sure how much the SRU team like that13:10
lamonthrm13:10
seb128lamont, I saw you uploaded those fix to Debian and to your ppa, any plan to deal with the SRUs for Ubuntu as well? ;-)13:10
* lamont was hoping to avoid learning the SRU process13:10
lamonts/learning/being reminded about/13:11
seb128lol13:11
seb128lamont, the bug is SRU ready, you just need to get sources uploaded targetting the right series13:11
seb128lamont, with the changelog having a reference to the launchpad bug, and without the autotools noise if possible13:11
lamontah, that's not so bad13:12
=== MacSlow is now known as MacSlow|lunch
lamontI think I might be talked into that13:12
lamontis tomorrow soon enough? tuesdays tend to be crazy13:12
dholbachseb128, jamespage: I'll do some more sponsoring tomorrow - now need to get some other things done first :)13:13
dholbach@pilot out13:14
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: jamespage
dholbachand: we're at 31!13:14
margaseb128, thanks again :)13:15
seb128dholbach, great work!13:15
seb128marga, yw ;-)13:15
seb128lamont, yeah, no hurry, I was going to upload the version of the guy who took the time to contribute the debdiffs but as I said he didn't include your ipv6 extra entry ... anyway if you want to take care of it this week great, I will just get it off the sponsoring queue and assign to you ;-)13:15
lamontsounds good13:17
rbasakseb128, lamont: that was me. My understanding is that SRUs need to be minimal13:17
=== yofel_ is now known as yofel
seb128rbasak, hey, they do, but if the ipv6 entry (which is a one liner) is considered useful there is no reason to do 2 uploads for a 1 liner each, we can as well include both changes in that update13:19
rbasakSure, I understand. I suppose this comes down to my original philosophy that it barely matters anyway. bind will immediately bootstrap the ipv6 entry from any of the other addresses listed, so I saw the SRU as merely fixing the soon-to-be-broken entry rather than anything else13:21
seb128rbasak, I think what you did is right, but if lamont (who is the bind maintainer) wants to include the ipv6 change as well I will not stop him13:22
lamontrbasak: by your logic, we don't need to update the ipv4 address either - 1 time in 13 the startup will be delayed by a few seconds.  So yeah, fixed is better13:24
lamontso yeah, keeping that file current is a good thing13:25
* jamespage gets the feeling he's now just following dholbach through bugs...13:34
=== Jacky_ is now known as zz_Jacky_
Riddellkirkland: anerd has binary files in the .orig.tar.gz, should I reject to let you fix that?13:54
jamespage@pilot out14:16
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots:
tjaaltondoko: hey, so about llvm.. would it be possible to include a snapshot of the r600-branch just for mesa?14:25
tjaaltondoko: as a new package14:25
tjaaltondoko: upstream says the branch will stabilize tomorrow, since mesa 9.1 was branched last week14:26
kirklandRiddell: sure, you bet14:27
kirklandRiddell: sorry about that :-/14:27
mjthallyn: telling the truth, i didn't understand anything in your lxc reply... :)14:31
Riddellkirkland: voila14:33
=== zz_Jacky_ is now known as jackyalcine
=== jackyalcine is now known as Jacky_
=== MacSlow|lunch is now known as MacSlow
=== slank_away is now known as slank
tomreynhi, according to apport i'm affected by https://bugs.launchpad.net/bugs/925545 but this bug is hidden. can you tell me whether there is a known workaround, though?14:51
ubottuError: ubuntu bug 925545 not found14:51
barryseb128, stokachu shall i just sync python-tz now?14:57
=== francisco is now known as Guest8850
hallynmjt: heh, sorry, rant :)15:04
jamespagedoko: did you see this? http://lists.debian.org/debian-java/2013/02/msg00005.html15:25
dholbachhighvoltage, stgraber: ready for in 45m? :)15:46
jbichacjwatson: could you add clutter-1.0, clutter-gst, clutter-gtk, cogl, gnome-bluetooth, gnome-icon-theme-symbolic, and vte3 to the ubuntu-desktop set?15:47
jbichaI already sent https://lists.ubuntu.com/archives/devel-permissions/2013-January/000440.html for all but gnome-bluetooth15:48
stgraberdholbach: yep15:50
dholbachstgraber, greta15:50
dholbachgreat15:50
kirklandRiddell: can you give her another look?  cheers!15:56
=== chiluk is now known as chiluk_away
Riddellkirkland: accepted!16:00
kirklandRiddell: cheers, bud, thanks16:00
=== chiluk_away is now known as chiluk
smosercjwatson, ping16:01
smoserhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/101533916:01
ubottuUbuntu bug 1015339 in linux (Ubuntu) "Missing kernel modules for Mellanox ConnectX HCA Ethernet " [Medium,Confirmed]16:01
cjwatsonsmoser: yes?  I followed up this morning and made that bug not my problem :)16:02
smoserthere seem 2 issues there.  1 (as you stated) the network driver does not get into the initrd.gz.16:02
smoser2. the modules dont get auto-loaded16:02
smoseri dont think this is "your problem"16:02
cjwatsonthere was a horribly confused pile of stuff involving manually created initrds16:02
smoserbut can you verify for me, that they should "just get loaded"16:02
cjwatsonI decided it was unlikely to be diagnosable until we had something more sane16:02
smoserif they're present, based on udev and pci-ids built in.16:03
cjwatsonsmoser: that is generally up to the modalias tables in the kernel16:03
smoserright16:03
cjwatsonsmoser: if the ids are correct then they will be auto-loaded, yes16:03
cjwatsonand usually if they aren't then the driver won't bind to the device anyway16:03
smoserk. so it would appear to me that those need updating also.16:03
smoserin this case at least, manual 'modprobe' gets them functional.16:03
smoserthanks, cjwatson . as you said "not your problem" :)16:04
cjwatsonthere's a big list of ids in the driver; certainly possible it's missing one or two16:04
cjwatsoncan't tell from the bug since it doesn't quote any ids16:04
cjwatsonand I would say that any id update should be regarded as a separate bug16:05
stokachubarry: should i remove that MP since im going to do a sync ?16:10
barrystokachu: probably not delete it.  what status options do you have?  (probably different than me since it's your mp)16:11
stokachubarry: i got WIP, needs review and merged16:11
barrystokachu: hmm16:11
barrystokachu: yeah, i guess delete it then16:12
stokachuok16:13
ograhmm, so diggin into the udevd issues on the nexus7(i now end up with 26 daemons on boot), stracing shows that only two of them are actually active, the rest just sits and gives no output via strace16:13
=== Jacky_ is now known as zz_Jacky_
=== Quintasan_ is now known as Quintasan
Quintasan\o16:21
cjwatsontumbleweed: Your mk-sbuild change to improve naming of cross-building chroots doesn't align with what Adam committed to upstream sbuild; your version has an extra "-cross" in the output name.  Would you mind if I removed that extra "-cross", since it hasn't been uploaded yet anyway?16:23
cjwatsonI don't think it's necessary - SERIES-BUILD-HOST should be clear enough16:24
stokachubarry: pad.lv/111641816:24
stokachui did a brief explanation about most of it being in debian, and i didnt notice any other Ubuntu specific changes16:24
barrystokachu: looks good, thanks.  i'll do the sync momentarily16:32
stokachubarry: awesome! :D16:34
stokachubarry: since its a request sync do i get any credit for it? (mainly for when i apply for core-dev)16:34
barrystokachu: if i run syncpackage correctly, you should ;)16:36
stokachulol ok cool, thanks man for your help16:36
barrystokachu: thanks for working so hard on this package.  it's always fun when you think you've got an easy one and it turns out to be "interesting" :)16:37
stokachubarry: yea no kidding, i was like hmm python-tz should be easy its just timezones :X16:37
cjwatsonpitti: Do you think we're good to go with language-pack-* for precise-updates?16:37
=== zz_Jacky_ is now known as jackyalcine
=== jackyalcine is now known as Jacky_
margaslangasek, I'm hoping you'll be around soon.16:48
margaslangasek, I'd like to discuss the fix for https://bugs.launchpad.net/ubuntu/+source/dropbear/+bug/83417416:49
ubottuUbuntu bug 834174 in dropbear (Ubuntu Quantal) "cannot login via ssh when using dropbear in initramfs" [Undecided,New]16:49
=== zequence_ is now known as zequence
slangasekmarga: hey there16:56
slangasekmarga: otp, but I have a couple spare brain cycles16:56
* marga tries to expand "otp"16:56
slangasek"on the phone" :-)16:57
margaAnyway, the thing is, do you want me to prepare a new upload without the "echo 'passwd: compat'" line? Or do you want a completely different patch?16:57
slangasekmarga: I think it needs to be changed to be "echo 'passwd: files'".16:57
margaAlright.  Next question, what version should this package be? Because the one with the bad patch is in precise-updates16:58
slangasekmarga: the current hook is buggy if plymouth isn't already pulled into the initramfs16:58
ograslangasek, oh its not "out to pee" ? and i was wondering about the weak blatter of all our managers :P16:58
slangasekhmm?  how did it get into precise-updates?16:58
margaoh, maybe I'm wrong16:58
margaI thought I had had this sponsored16:59
slangasekmarga: I was reviewing it prior to accepting it into the archive16:59
slangasekmarga: sponsorship just gets you to the queue, then you have to pass the SRU team gauntlet ;)16:59
margaah, precise-proposed, not updates16:59
margaI always mix them, sorry.16:59
slangasekand I was the SRU gauntlet here16:59
margaalright.16:59
margaSo, same version number or different?16:59
slangaseksame16:59
margaok.  Will have it there for you tomorrow.17:00
marga(with chekcing that it works and that)17:00
slangaseksounds good, thanks!17:00
pitticjwatson: yes, from my side17:01
cjwatsonpitti: excellent, I'll get going on that then17:02
xnoxslangasek: marga: don't we need an upload into raring to diverge with "s/compat/files"17:07
xnox?17:07
slangasekxnox, marga: yes, this should also get fixed in raring - or at least have a bug opened against dropbear in Debian about the use of 'compat' being buggy17:08
cjwatsonpitti: I can get the list easily enough; we just hand an enormous list to sru-release, right?17:08
pitticjwatson: I usually use sru-release -p language-pack-17:08
pitticjwatson: (--pattern)17:08
pitticjwatson: http://bazaar.launchpad.net/~ubuntu-langpack/langpack-o-matic/main/view/head:/doc/operator-guide.txt17:09
pitticjwatson: "Releasing -proposed packages to -updates17:09
cjwatsonoho17:09
infinityev: *poke*... whoopsie that builds? :)17:11
evoh yeah17:11
evon it now17:11
pitticjwatson: yeah, documentation -- scary, isn't it?17:11
infinityev: Thanks.17:14
evinfinity: done17:18
argesis there anyway pad.lv/1052038, can be sponsored, then SRUed for Precise? thanks17:20
infinityLaney: Not that I disagree that C++ symbols files are a pain, but why not just feed all the maliit-framework build logs to pkgkde-symbolshelper batchpatch, and be done with it, instead of dropping the symbols file entirely?17:21
Laneyinfinity: I tried, but the tool kept refusing to apply patches and also applied some in a way that broke other arches. So in the end I saw red.17:22
LaneyHopefully I have enough state to file bugs17:22
Laneys/refusing to apply/ignoring/, perhaps17:23
infinityLaney: Sounds like you were trying to apply each arch individually.17:23
infinityLaney: Hence "batchpatch".  If you take -0ubuntu1, grab ALL the build logs, and feed them all in at once, it tends to DTRT.17:23
LaneyI was doing that - that was when it didn't apply all of the hunks and still FTBFSed17:24
evughhhhhhh17:24
infinityHrm.  Weird.17:24
Laneythen I tried some other permutations and that was when it undid some other correct arch annotations17:25
Laneywhich I agree is because I was using it wrong at that point17:25
infinityev: I've heard good things about test building. :)17:25
evI did! (just not in raring)17:25
evfixing, throwing it at sbuild, shouting angry things, etc17:26
infinityHeh.17:27
=== Guest89634 is now known as jono
=== jono is now known as Guest96957
=== MacSlow is now known as MacSlow|dinner
smosercjwatson, can i make the installer load modules via the kernel cmdline ?17:39
smoserie, as if they were in /conf/modules or something.17:39
cjwatsonsmoser: I don't think s17:42
cjwatsono17:42
infinitysmoser: There's a general assumption that udev actually works correctly these days.  How is this failing you?17:42
infinitysmoser: (You can blacklist modules from the commandline, but not the other way around)17:42
smoserinfinity, its failing by design on this issue.17:44
smoserhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/111571017:44
ubottuUbuntu bug 1115710 in MAAS "Mellanox mlx4_en network driver is not automatically loaded" [Critical,Triaged]17:44
smoserinfinity, its not loaded here becaujes the thing that gets automatically loaded is mlx4_core , not mlx4_en .17:45
smoserthat is how it is designed by upsream as ethernet or infiniband usage is somewhat "configruration"17:45
cjwatsoneh, if something isn't autoloaded by design then the design is wrong17:45
=== henrix is now known as henrix_
infinitysmoser: Yeah, uhm.  Fix the kernel bug?17:46
cjwatsonthere are several other modules like this that we've dealt with in various ways17:46
cjwatsonunfortunately I don't currently remember their names ...17:46
smoserits not necessarily a kernel bug.17:47
infinitysmoser: This will be just as broken on the installed system as in an installer initrd, so I don't see how it's not a bug.17:47
=== henrix_ is now known as henrix
=== henrix is now known as henrix_
=== henrix_ is now known as henrix
=== henrix is now known as henrix_
smosercorrect. and the user will configure the device to act as infiniband or as ethernet17:47
=== henrix_ is now known as henrix
infinitysmoser: (Unless we're to assume this driver is a unique snowflake that needs to be manually loaded on installed systems too, which pretty much anyone will tell you is exactly how this isn't supposed to work)17:47
cjwatsonthere are ways to deal with this kind of thing in installer code - but we haven't had to add any new examples of those in years and I don't want to do so ever again17:47
smoserso it wont "just work" because you codn't know which "just work" is right.17:48
=== deryck is now known as deryck[lunch]
=== henrix is now known as henrix_
=== henrix_ is now known as henrix
cjwatsonyou *could* have a udev rule that interrogates the device for whether it's in ethernet or infiniband mode and loads the right other module17:48
cjwatsonbut if udev can do that, so can the kernel17:48
infinityExactly.17:48
cjwatsonand either is more appropriate than adding configuration to the initramfs or to the installed system17:48
smoseri'm in agreement with that.17:49
cjwatsonhardware upstreams are generally terrible at this kind of thing; there is no reason you should trust their design decisions implicitly17:49
smoserbut as it is here, upstream kernel isn't going to magically load, and just saying "make the ethernet driver load" is not the right solution either.17:49
cjwatsonLinux has plenty of history with overriding foolish decisions made by hardware vendors17:49
infinityDear god, you referenced a 211 page PDF in the bug.17:49
smoserwell, search for mlx4_en in it, infinity17:50
infinitysmoser: Last I checked, we build those kernels, we don't trust upstreams or mainline blindly.17:50
cjwatsonsmoser: fixing the driver isn't necessarily that hard; if it's important, find a kernel team victim to help.  This shouldn't be unfamiliar territory for our kernel team.17:50
cjwatsonIt's the sort of thing we do.17:50
infinitysmoser: The other curious question is if all the drivers can be loaded together.17:51
smbsays the one defining the SEP17:51
smoserwell, i did find one. its just not as simple as we'd like.17:51
=== henrix is now known as henrix_
smoserinfinity, i'm not sure. i wondered that too.17:51
=== henrix_ is now known as henrix
infinitysmoser: What happens if one modprobes mlx4_ib mlx4_en and mlx4_fc?17:51
smoserbut i dont know why youd' have your softwar stack have configuration for whic hone to load if they loaded at the same time with no fallout.17:52
cjwatsonsmb: Well, yes :-)  But it's simple truth that you guys have done this kind of thing before17:52
infinitysmoser: If it's a tiny waste of RAM, but they all load and the "right" one works, the simple solution could be to just build it statically.17:52
cjwatsonAnd fixing it in the installer is definitely the wrong place17:52
infinitysmoser: If they all refuse to bind except the "right" one, that's a simple udev rule (or a kernel fix).17:52
cjwatsonsmoser: Don't assume that the driver authors had the same priorities as we do17:52
cjwatsonThey might have thought that saving a tiny bit of kernel memory was more important than anything else17:53
infinitymlx4_core17:53
infinityHandles low-level functions like device initiali17:53
infinityzation and firmware commands processing. Also17:53
infinitycontrols resource allocation so th17:53
infinityat the InfiniBand and Ethernet17:53
smbcjwatson, Some kind but not really assuming "cannot be hard" without knowing the hw design which sometimes is hard17:53
infinityfunctions can share the device17:53
smosercjwatson, alright. i'll try to figure that out.17:53
infinitywithout interferin17:53
infinityg with each other.17:53
infinityOkay, ignore the bad line breaks.  But this doc implies that the driver is designed specifically to "do it all at once".17:54
smbinfinity, But I would think the same. That in theory the ports can be mixed, so could be both modules needed17:54
cjwatsonsmb: It's an Ubuntu design principle to automatically probe hardware, and if the probing is possible in userspace then it's just as possible in kernelspace17:54
infinitySo, the simple first-try workaround is a udev rule that loads all the sub-modules when core is loaded.17:54
cjwatsonAnd (at its crudest) a bunch of "load this other module" calls is AIUI quite doable in the kernel17:55
cjwatsonNot saying that's necessarily the best option17:55
infinitycjwatson: I could see one argument for keeping it a udev rule, which is just that some people may care deeply about upstream's attempt to save 25kB of kernel memory. :P17:55
cjwatsonSure17:55
infinityBut, yeah.  First pass should just be "load everything and see".17:55
argesjdstrand: ping17:55
infinityAnd the docs here imply that should work.17:55
jdstrandarges: hey17:56
argesjdstrand: hey I see you uploaded for pad.lv/1052038, did precise also get uploaded as well?17:56
stokachuarges: https://launchpad.net/ubuntu/precise/+queue?queue_state=1&queue_text=ecrypt17:56
smoserinfinity, cjwatson what you. testing to see if they seem to live together.17:56
stokachuif thats the correct one it just needs sru17:56
infinitysmoser: If "load everything" works, a udev rule to do the same it almost certainly your path of least resistance.17:56
infinitys/it/is/17:56
smoseryep17:56
smbinfinity, Could even be a modprobe.conf thing. There is something about softdep17:57
cjwatsonLet's not add to modprobe.conf17:57
cjwatsonI have a feeling that is somewhat painful with d-i17:57
cjwatsonudev is probably better for this kind of thing17:58
infinityd-i's initrd has a properly populated /etc/modprobe.d, I believe.17:58
smbok, no strong preference there17:58
infinityBut it's more about where you're shipping that file, and if it's there at build time. :P17:58
jdstrandarges: yes - https://launchpad.net/ubuntu/precise/+queue?queue_state=1&queue_text=ecryptfs-utils17:58
infinity(Which is true of the udev rule too)17:58
stokachujdstrand: could we get that one approved?17:58
stokachuso sru can take a look at it17:58
infinitystokachu: Hrm?17:59
stokachuinfinity: hai!17:59
infinitystokachu: How does Jamie factor into this?17:59
jdstrandstokachu: ubuntu-sru needs to look at it (I am not on that team)17:59
infinityI think I may have intentionally not accepted it while the bug was in flux.17:59
stokachuok18:00
jdstrand(besides I did the sponsored upload, so someone else would need to do the approval anyway)18:00
argesinfinity: the fix should be settled now18:00
infinityI'll look at it in a sec.18:00
argesthanks18:00
infinityarges: Yeah, I see that.18:00
stokachuthe bug still has sponsors subscribed to it18:00
infinity(Don't expect it to make .2, unless you make a really good case for it, it'll probably stay in -proposed until after 12.04.2 release)18:00
infinityBut that gives you plenty of time to test very thoroughly. :P18:01
tyhicksHmm... I think arges did need it in 12.04.218:01
argeslet me check18:01
cjwatsonI thought I'd asked what else you guys needed in .2 a while back :-(18:02
infinityI can certainly see an argument for it being in .2 but, then again, I can see an argument for it being in .0, which it clearly wasn't. :P18:02
argesso whats the ETA then if not in .2?18:02
stokachucjwatson: i think we just glanced over this one18:02
infinityAnyhow.  I'll quickly review and accept, you can argue promotion after.18:02
argesthanks18:02
cjwatsonAny reason it has to be tied to a particular release?18:02
infinityarges: If it passes all verification, etc, and we don't intentionally target it for .2, it would hit -updates right after release.18:02
infinityarges: If there's no reason it MUST be on installation media, that's perfectly fine.  If the old version can cause such horrific problems that we don't want it on an ISO, speak up.18:03
argesinfinity: ok this should work, I don't think it is necessary to be on media.I'll make sure18:04
cjwatsonpitti: language packs copied18:04
cjwatsonthat should make sru-report run rather more quickly ...18:05
stokachuinfinity: thanks for looking at it so quick18:05
argesinfinity: indeed. thanks,18:05
cjwatson(well, after the removal pass, which I'll take care of once the report lists them)18:05
smoserwhat package would i put such a udev rule in? infinity ?18:12
infinitysmoser: If the goal is for this to be in d-i images, and every base system, it would probably have to be in udev itself, not some leaf package.18:13
smoserthats what i thought too18:13
infinitysmoser: That said, after testing the udev rules option, the proper solution is probably in the kernel, as Colin argues.18:14
infinitysmoser: One could add an "auto_load=[yes,no]" option to the _core module, defaulting to "yes" in our kernels, so users who really want to old behaviour can change the module load parameters, but the default Just Works.18:15
smoserinfinity, so.. the ude rule.18:24
smoserDRIVER=="mlx4_core", RUN+="/sbin/modprobe -bv mlx4_en", RUN+="/sbin/modprobe -bv mlx4_ib"18:24
infinitysmoser: Was the -v just for testing?18:25
smoserno.18:26
smoserbecausae i didn't test :)18:26
smoseri just ocpied from entry in /lib/udev/rules.d/80-drivers.rules18:26
infinityOh, -bv seems to be the standard in other rules.  I guess for logging.18:26
infinitysmoser: Anyhow, I'm no udev wizard, but if that works, it seems plausibly correct. :P18:27
smoserright. you're in the same place as me then18:27
infinitysmoser: Testing on actual hardware would be sane.18:27
smoseri seem to be somewhat limited in access to such.18:27
infinityI'm assuming from agy's input on the bug that we have this hardware in the DC somewhere?18:28
infinitySurely, you can get a sysadmin to help.18:28
infinity(If you intend to SRU this back to precise, you'll need some hardware access to verify anyway)18:28
=== Jacky_ is now known as zz_Jacky_
=== slank is now known as slank_away
siretartinfinity: did you observer run-time errors, or are you "just" concerned about the build logs?18:48
siretart-r18:49
infinitysiretart: There's no just about it, it's broken if anything actually calls into that.18:49
infinitysiretart: (In Ubuntu, we hard fail on implicit pointer conversions on 64-bit arches, so it's also a build failure, but the root cause is the implicit declaration, which is entirely a real problem here when you're calling into something that doesn't exist in the code you link to)18:50
siretartI guess it's a missing #include. the three functions refer to the audioconvert API, which we do have in libavcodec18:52
infinitysiretart: A missing include of what?  I couldn't find those functions referenced in /usr/include at all.18:52
infinitysiretart: Which is why I pinged you instead of just fixing it.18:52
siretartavcodec/audioconvert.h18:52
=== slank_away is now known as slank
siretartHA, that's a private, non-installed header. fun18:53
infinityAh-ha.  There you go.18:53
siretartyeah, that's upstream (ab)using private headers. bad upstream18:53
infinitySo, you can either tell them to stop that, or include those prototypes in the handbrake source as a stub.18:54
infinityLinking against private symbols is generally considered a Bad Thing, but whatever. :P18:54
infinityMost of ffmpeg/libav scares me enough as it is, I can't see how this would make it worse.18:55
mlankhorst... famous last words18:58
siretartthanks for notifing me about this. it seems that handbrake upstream is also hanging around in #libav-devel, where I'm trying to clarify what we can do about this.18:58
cyphermox@pilot in19:03
=== udevbot changed the topic of #ubuntu-devel to: Ubuntu 12.10 released | Archive: Open | Dev' of Ubuntu (not support or app devel) | build failures -> http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of hardy -> quantal | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | See #ubuntu-bugs for http://bit.ly/lv8soi | Patch Pilots: cyphermox
=== MacSlow|dinner is now known as MacSlow
siretartinfinity: okay, the situation is this: the audioconvert API was work in progress, never made public, never finished, and is scheduled for removal for the next libav release19:05
infinitysiretart: Thanks for looking into it.19:05
siretartinfinity: yet, handbrake is still using it. which is bad19:05
infinitysiretart: Ahh, so handbrake just needs to be fixed to stop using it. :P19:05
siretartinfinity: the replacement for this is libavresample, which is already part of libav 9, which I really would like to see in raring but am desperatly seeking for help to fix the remaining packages. (but that's another story)19:05
siretartinfinity: it seems that handbrake has already been ported to libavresample in git, but there is no release for that yet.19:06
infinitysiretart: Okay.  So, the way forward is new libav, and git pulls to handbrake?19:07
siretartinfinity: yes19:07
siretartbtw, help is desperately needed at https://launchpad.net/~motumedia/+archive/libav9-raring/ - I've added all relevant links to the PPA description19:07
siretartotherwise we'll have to defer for raring+1, which would be really, really sad :(19:08
infinitysiretart: Are those 7 open FTBFS bugs the entirety of what needs fixing?19:11
siretartinfinity: AFAIUI, the raring release process requires all FTBFS in https://launchpad.net/~motumedia/+archive/libav9-raring/+packages to be fixed. if we can shortcut this, I'm all for it :-)19:12
siretartafk (dinner)19:13
infinitysiretart: I'd certainly prefer they all be fixed, yes.  That's a bit longer than the 7 bugs filed. :P19:13
infinitysiretart: Probably entirely doable, though.19:13
=== deryck[lunch] is now known as deryck
=== henrix is now known as henrix_
=== henrix_ is now known as henrix
=== henrix is now known as henrix_
=== henrix_ is now known as henrix
tumbleweedcjwatson: not at all, please do so19:33
siretartinfinity: some of the packages seem rather abandoned upstream. i'd suggest to just phase out such packages19:44
cjwatsontumbleweed: OK, done, thanks.  I didn't bother with a changelog entry since this is all new since the last upload19:45
cjwatsonUnable to obtain lock  held by laney@bazaar.launchpad.net on taotie (process #13112), acquired 483 hours, 50 minutes ago.19:45
cjwatsonLaney: ^- I figured it was OK to break that on ubuntu-dev-tools :-)19:45
cjwatsonProbably isn't actually an epic 20-day commit19:46
tumbleweedheh19:46
infinitysiretart: If you want to "phase them out", get them removed from Debian.19:50
infinitysiretart: But I'm betting most of the porting is fairly trivial, and I'm not a fan of shafting users.19:50
robrubarry, ping19:59
smoserinfinity, smb would using MODALIAS be any better or worse than udev rule ?20:04
smoserhttps://bugs.launchpad.net/ubuntu/+source/linux/+bug/28697720:04
ubottuUbuntu bug 286977 in linux (Ubuntu) "ipmi_devintf needs a MODALIAS of platform:ipmi_si" [Undecided,Fix released]20:04
slangasekfor anything that /can/ be expressed as a modalias, you should be using a modalias and not a udev rule20:05
barryrobru: pong20:06
smoserslangasek, bug 111571020:07
ubottubug 1115710 in MAAS "Mellanox mlx4_en network driver is not automatically loaded" [Critical,Triaged] https://launchpad.net/bugs/111571020:07
smoserforgive ignorance, is that something that could be done as modalias?20:07
infinitysmoser: That seems like a plausible solution to the problem.20:08
infinitysmoser: Patch the kernel and find out? :P20:09
slangaseksmoser: do you have a more precise pointer to the rationale for not autoloading the module, than a 212 page driver user manual?20:14
slangaseksmoser: ah, I see your comment 9 explains what you meant, ignore me20:15
smoserslangasek, "upstream does not do it, and smoser is largely ignorant about this hardware"20:16
slangasekso the one problem is that you can only load one module per device, AIUI20:16
slangasekso if you have to pick between autoloading the ethernet module, and autoloading the infiniband module, then what?20:16
smoserslangasek, that seems like it was invalid logic20:16
slangasekwhat was?20:17
smoserat least  at the moment we're under hte impression that we can load both at the same time.20:17
smoserand at the expense of wasting kernel memory :)20:17
slangaseksure, you can load them but you can't make the kernel /autoload/ them via modalias20:17
smoserah.20:17
slangasekunless you have two different device identifiers that you can associate each driver with20:17
slangasekstill, rather than a udev rule, I would use a modprobe rule20:18
smoserslangasek, any reason why?20:19
slangasekbecause udev isn't really meant to be used that way, and I'm not sure it'll work20:19
slangaseksometimes the reentrancy of udev is suspect20:19
smoserslangasek, what package would you suggest this rule to be delivered in ?20:20
slangasek/etc/modprobe.d/alsa-base.conf shows the kind of thing you want20:20
slangaseksmoser: er, what package delivers the driver?20:20
smoserlinux20:21
slangasekhmm20:21
slangasekthere isn't really a common package for the kernel, then20:22
slangasekis there any runtime package associated with these drivers?  Probably not?20:22
slangasekif not, then I'd say use the kmod package for it20:22
smoserno.20:23
smoserand then that should get packaged all the way into installer initramfs and any other initramfs , right?20:24
siretartinfinity: yes, that's my plan, but surely not before wheezy release, which will not happen before raring's feature freeze20:26
infinitysiretart: Which is your plan?  Removing all the packages that don't build with the new libav? :/20:28
infinitysiretart: See above, re: shafting users.  If the porting is trivial, we should JFDI.20:28
infinitysiretart: (We don't remove everything that fails to build with a new GCC even if upstream isn't fixing it, this should be no different)20:28
siretartinfinity: most of them should be rather easy to port. there are some harder ones, which I've bugged upstream about, and from a few others, upstream seesm to be defact dead (e.g., dvbcut)20:29
siretartinfinity: packages that don't build with newer gcc does not hinder gcc from passing britney. unlike what she would do with libav920:30
infinitysmoser: If it's in kmod (or module-init-tools, in older releases), it'll make it into initrds and d-i and such, yes.20:30
infinitysiretart: No, but it still makes all those packages RC-buggy.  The fact that britney can't detect that is a process flaw, not a free pass. :P20:30
smoserok. so now i'm being dense again.20:31
smoserinstall mlx4_core /sbin/modprobe --ignore-install mlx4_en; /sbin/modprobe --ignore-install mlx4_ib20:31
smoserthat line causes recursion20:31
siretartinfinity: ;-)20:31
smoserok. here, this seems right now:20:33
smosersoftdep mlx4_core post: mlx4_en mlx4_ib20:33
smoserslangasek, ^ that make sense ?20:34
=== henrix is now known as henrix_
tjaaltonbdmurray: hey, could you accept the precise-proposed upload as well for bug 1095052?21:31
ubottubug 1095052 in gnutls26 (Ubuntu Precise) "Client certificate authentication fails" [Medium,In progress] https://launchpad.net/bugs/109505221:31
=== salem_ is now known as _salem
bdmurraytjaalton: I wasn't sure about that with the point release being imminent21:53
bdmurrayinfinity: ?21:53
cjwatsonbdmurray,tjaalton: I'd prefer that to wait until after 12.04.222:17
tjaaltoncjwatson: ok22:19
=== slank is now known as slank_away
=== slank_away is now known as slank
Laneycjwatson: hah, no idea how that happened22:48
slangaseksmoser: I've never seen/used softdep before; I was thinking of this: install mlx4_core /sbin/modprobe --ignore-install mlx4_core; /sbin/modprobe mlx4_en; /sbin/modprobe mlx4_ib23:38
dobeyslangasek: hey, got a second? just wondering if this dependency makes sense to you: https://code.launchpad.net/~dobey/software-center/oc-lesser/+merge/14674923:47
=== slank is now known as slank_away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!