slangasekLocutusOfBorg: ^^ accepted; as far as future LTS syncing, the release schedule for the point releases is knowable in advance, you can always just pre-upload them and let them sit in dep-wait until the rest of the LTS stack is available00:07
LocutusOfBorgthanks slangasek, it has sit in my ppa in dep-wait for some time, and I though it was fine to upload when it was buildable...06:54
LocutusOfBorgnext time I'll upload it before thanks06:54
Mirv^ libwfut and eris are syncs from Debian that used to be in Ubuntu before being removed as part of the GCC 5 transition.06:57
Mirvnow requested in the sponsoring queue06:57
=== davmor2_ is now known as davmor2
* apw is handling ^09:29
Laneyhttps://launchpadlibrarian.net/276128735/buildlog_ubuntu-yakkety-amd64.adwaita-icon-theme_3.20-3ubuntu1_BUILDING.txt.gz -> gtk-3-examples -> libgtk-3-0 -> libgtk-3-common -> adwaita-icon-theme (>= 3.20)09:33
LaneyI could build in a PPA without proposed and copy binaries over - it's arch:all09:34
Laneysomeone scream if that is terrible09:34
=== pesari_ is now known as pesari
Laneydone, no point in screaming now10:23
Mirvis there some GTK related issue with proposed, I'm getting https://launchpadlibrarian.net/276121156/buildlog_ubuntu-yakkety-amd64.uim_1%3A1.8.6+gh20160630.0.c408e95-1build1_BUILDING.txt.gz ?10:31
Mirvjust related to what Laney said and mentioning gtk10:32
Mirvor is that adwaita fixing the issue?10:32
Laneyi think so10:32
Laneywait for it to publish out10:32
* Laney is running watch -n30 rmadison -s -Syakkety-proposed adwaita-icon-theme10:35
Laneys/-s -S/-S -s/10:35
davmor2infinity: are we not working to a .5 release this week on the tracker? I assume we use daily for now right?10:40
* Mirv fixes poppler in proposed10:51
MirvFYI I'm waiting for some investigation / whitelist of failing autopkgtests for a handful of KDE packages, and also waiting similarly for Unity 8, and then Qt 5.6 / KDE transition to release pocket should be possible (or at least nearer to possible)11:06
apwdavmor2, yes, there should be .5 this week11:09
tjaaltoninfinity: what was crufty with my xorg-lts-xenial upload? I don't see anything here11:18
tjaaltonon my tree11:18
tjaaltoninfinity: oh the merge conflict stuff.. hmm maybe I cleaned those after upload then11:21
tjaaltondon't see then on that upload though11:23
infinitytjaalton: I reuploaded... The one in the rejected queue is quite a bit different than the one I uploaded.11:24
infinitydavmor2: Do some testing with dailies please, yes.  I'll upload base-files (no need to file that bug :P), create the milestone and spin RCs after I've napped.11:25
tjaaltoninfinity: got it, didn't delete the unrenamed crap from the git tree once copying over the scripted result11:27
infinitypitti: Did you sprint take into account an attempt to clean up `reverse-depends -b src:upstart`?  Would be nice to fix all that, so people stop doing misguided things like removing s390x binaries. :P11:27
davmor2infinity: I decided to test the apps on the live session instead and makes sure they all open and close and save stuff :)  go sleep dude11:28
pittiinfinity: actually we have the opposite problem: many things depend on upstart without saying so11:34
infinitypitti: Well, both are bad, clearly. ;)11:34
pittiinfinity: ah no, we didn't look at build deps11:34
pittiinfinity: tedg and I looked at porting ubuntu-app-launch11:35
pittiso that is part of that11:35
infinitypitti: To be fair, I think we'll have to fix the "s390x" bugs at some point anyway, cause they're not s390x bugs, they're "upstart's testsuite fails on xenial kernel" bugs, and s390x just happened to be the only buildds running xenal during that cycle.11:35
pittibut https://code.launchpad.net/~ted/unity/systemd-unit./+merge/300584 does not yet  drop the build dep11:35
infinityIf we ever try to SRU upstart in xenial, it'll explode on all arches.11:35
xnoxnot funny11:37
infinityxnox: Not funny, but true. :/11:38
infinityxnox: I can reproduce the testsuite failures on my laptop.  Well, some of them.11:38
pittiinfinity: but at least the libupstart-dev deps can't be dropped yet until we  finish the transition11:39
tsimonq2infinity: hey, any chance you'll get to take a look at those PRs? sorry for poking you, it would just be great to get it done and over with12:22
infinitytsimonq2: Perhaps after 14.04.5 happens.  I have priorities.12:23
tsimonq2infinity: alright, sounds good12:23
infinitypitti: If you're still around, a base-files landing in trusty that needs a quick review.13:49
infinitypitti: ^-- And there it is.13:51
pittiinfinity: I wait a bit until it becomes diffy13:51
infinitypitti: Kay.  I'm going on a coffee run, hoping it's built by the time I get back. ;)13:52
pittiinfinity: (accepted)14:03
infinitypitti: Ta.14:07
dmj_s76infinity: I found a couple issues with 14.04.5 regarding nvme drives as well as nvidia cards with the proprietary drivers15:40
infinitydmj_s76: Regressions?15:43
dmj_s76infinity: bad ones15:43
infinitycyphermox: ^-- You know anything about changes to nvme handling between 14.04.4 and 14.04.5?15:43
dmj_s76with the daily iso, after you install onto an nvme drive, you get dumped to an initramfs prompt15:43
infinitydmj_s76: proprietary drivers, we can do very little about.15:43
infinitydmj_s76: To be clear, that's *not* an issue with 14.04.4?15:44
dmj_s76infinity: That's correct15:44
infinitydmj_s76: Kay.  Is there a bug?15:44
infinitydmj_s76: If so, can we get some debugging info from .4 and .5?  dmesg, lsmod, name of device it's installing to, blah blah?15:44
dmj_s76infinity: Was just about to file bugs, but wanted to signal a heads up first15:45
infinitydmj_s76: Is this d-i (ie: server ISO) or ubiquity (desktop/live)?15:46
xnoxdmj_s76, uefi or bios?15:49
dmj_s76ubiquity.  I haven't verified with server yet.15:50
dmj_s76currently testing on uefi systems15:50
dmj_s76may affect bios, but I haven't checked yet15:50
dmj_s76separately, various (all?) versions of the nvidia proprietary driver result in an infinite login loop15:54
dmj_s76that's a regression from 14.04.4 as well15:55
xnoxhmmm.... does nvidia in 14.04.5 match the right x-server bits and kernel?15:56
infinityIt's meant to.15:57
infinityAlberto claimed it works.15:57
infinityHis claim may have been incorrect. :P15:58
apwthere are cirtainly dkms packages which at least build for that combination15:58
apwdmj_s76, your efi systems, are they secure boot enabled ?15:58
dmj_s76apw: We disable secureboot15:59
apware these the uber new nvidia ones ?  though i thought we pulled in something for those for you15:59
dmj_s76I'm testing on both nvidia 9 series (maxwell) and nvidia 10 series (pascal)16:01
dmj_s76...you did pull new nouveau support that fixes pascal16:01
dmj_s76...into 16.04.116:02
dmj_s76What I'm seeing now, is that nouveau works fine on the maxwell systems (970M and 980), but the moment you install nvidia, you can't get past the login screen16:03
infinitydmj_s76: Disabled SB from the BIOS, you mean, or the module signature validation disabling thing?16:10
infinityapw: I assume that patchset it meant to ignore bad/no sigs if SB is entirely disabled.  I hope.  Did we test that? :P16:10
infinitys/it meant/is meant/16:10
infinityapw: Or were we so focussed on MOK variables that we didn't think of that?16:11
apwinfinity, as i remember things they are both the same in the kernel both mean "off"16:12
* infinity might need to slap some USB storage into his nvidia desktop to do some testing.16:12
infinityThough my card is ancient, so if it's a driver bug, I doubt I'll trip it.16:12
infinityBut if it's something *we* broke..16:12
infinityOh.  Wait.  My desktop is also ancient and not EFI.16:12
infinityAhh, but my old laptop is EFI and nvidia.16:13
ogra_infinity, hmm, when i select proposed in my snap builds i dont actually end up with proposed in the toplevel sources.list ... i assume lp-buildd somehow stores that in a different place ?16:14
ogra_i was trying to add something like:16:14
ogra_ifneq ($(shell grep $(RELEASE)-proposed /etc/apt/sources.list),)16:14
ogra_ENV += PROPOSED=116:14
ogra_to my ubuntu-core makefile16:14
infinityogra_: I don't know what "select proposed" means in this context.16:14
ogra_(indeed that would have to be finer grained)16:14
infinityogra_: Pointer to the UI you're looking at?16:14
ogra_if you click "request builds" (at the bottom) there is an archive selection pulldown16:15
infinityogra_: pocket for automatic builds?16:15
ogra_defaulting to Updates16:15
dmj_s76infinity: It's disabled from the BIOS16:16
ogra_if i select Proposed there i assume that proposed shows up somewhere in my build env16:16
davmor2infinity, jibel: good news I finished testing all the apps installed by default, they all open, close and do stuff as expected woohoo moving onto actual installs now16:16
ogra_but grepping for it in the build logs there seems to be no trace unless i manually add PROPOSED=1 to my ENV var ... (then i get proposed only inside the live-build chroot16:16
infinitydmj_s76: Can you try running "sudo update-secureboot-policy" and see if that magically makes it happy after reboot and fiddling?16:16
infinityogra_: That sounds like a bug on Colin's part.16:17
ogra_ah, right ... he seems to be off today (i pinged him in #snappy before)16:17
infinityogra_: Oh, wait.  Are you doing automatic builds, or one-off?16:17
ogra_infinity, nowadays i do them through an lp-lib scripts, but they dont change no matter how i trigger them ...16:18
ogra_proposed never shows up ... neither in manual, nor in automatic, nor in lp-lib triggered ones16:18
infinityogra_: When you trigger via the API, it's like asking for a one-off build.  You have to specify the right pocket.16:18
ogra_right, i do that16:18
infinityogra_: How?  Show me the code.16:18
ogra_infinity, ok, shade your eyes ... http://bazaar.launchpad.net/~snappy-dev/ubuntu-core-snap/trunk/view/head:/cron-scripts/lp-build-ubuntu-core thats the api code16:19
ogra_when i change line 68 to "Proposed" i dont see any change in the build logs16:19
ogra_but it really doesnt matter how i trigger the build ...16:20
ogra_they all behave the same16:20
ogra_should launchpad-buildd actually mangle the sources.list ?16:20
ogra_to add -proposed ?16:20
infinityThis is clearly a bug, if it's not working.16:21
infinityFor example, this is how I build a livefs in a PPA with proposed: lp.load("/~ubuntu-cdimage/+livefs/ubuntu/trusty/xubuntu").requestBuild(archive="/~adconrad/+archive/ubuntu/ppa",distro_arch_series="/ubuntu/trusty/amd64",pocket="Proposed")16:21
infinityWhich is basically your code, but s/livefs/snap/16:21
infinitySo, if it's not making the outer chroot have proposed for you, someone screwed up.16:22
infinityOr you can't spell "proposed"16:22
infinityPick one.16:22
ogra_it works fine if i export PROPOSED=1 in the live-build call ... so i'll go with that for the moment ... it is just nasty having to change the tree every time i want to toggle it16:22
infinityRight, you shouldn't have to do that. ;)16:22
infinityPlus, you want proposed in the *outer* chroot if you're testing against a proposed livecd-rootfs, etc.16:22
ogra_indeed, i was trying to get the info dynamically from the outer chroot16:22
infinityWhich is the point of specifying the pocket.16:22
infinityOr, one point.16:22
ogra_which is why i sumbled over the issue16:23
ogra_*stumbled too16:23
infinitySo, double-check all your spelling and make sure you're reading your build logs right, and when you're sure you're not doing something Colin will smack you for, report a bug and get grumpy. ;)16:23
xnoxogra_, for cloud images i recall having to specify proposed multiple times, in json build config _and_ the target ppa16:29
dmj_s76infinity: no, running update-secureboot-policy from the root recovery terminal doesn't fix it (the nvidia issue)...nvme can't get to recovery mode16:29
ogra_xnox, yeah, but then the GUI option in the lp page is pointless ...16:30
infinityxnox: That's patently untrue.  cloud-images are just livefs builds.  If you have to specify it more than once, it's because you specified it wrong the first time. :)16:30
infinitydmj_s76: Did that give you a mokutil thing on reboot, etc?16:30
xnoxinfinity, so if the target of the livefs build is a ppa, it must have proposed enabled. as far as i can tell, and then pocket=Proposed needs to be specified in the requestBuild too.16:31
xnoxat least that did the trick for me, way back when i was fiddling with cloud images for s390x16:31
infinityxnox: Oh, yes.  But no json needed there...16:32
infinityxnox: That's just about archives.16:32
infinityogra_: xnox does make a point.  The archive you're specifying, what pockets does it build against?16:32
ogra_uh, main only i think16:32
xnoxtrue. json was for ext4 or some weird thing. all i remember it was annoying =)16:32
infinityNot components. :)16:32
infinityogra_: If it's a PPA, they usually default to updates.16:33
infinityI think.16:33
dmj_s76infinity: hrm...no, actually16:33
ogra_"Default (security dependencies and recommended updates)." is checked16:33
infinityRight, updates.16:33
infinityogra_: Change that to proposed and try again.16:33
ogra_hmm, i dont really want the packages in there to build against proposed16:33
infinityogra_: I know, just change it and do a test build and change it back.16:34
ogra_i just want proposed packages in the resulting image16:34
infinityogra_: If that test build does as you expect, then we at least understand the behaviour.  And you can either do your builds in a proposed PPA versus an updates PPA, or we can figure out a way to poke the PPA deps via the API at build time.16:34
ogra_https://code.launchpad.net/~snappy-dev/+snap/ubuntu-core/+build/2263 ...16:35
* ogra_ twiddles thumbs16:35
ogra_i love doing test builds on ppc64el ... nobody uses them in snappy and they are faster than anything :)16:36
infinityBah.  Didn't see the apt output in the logtail.  Now I have to wait. :)16:36
ogra_it takes less than 10min16:36
infinityogra_: Those IBM disk controllers are reasonably impressive (in response to your comment on speed)16:40
infinityI mean, the CPUs are no slouch either, but this is mostly disk.16:40
dmj_s76infinity: There ought to be no secureboot anything going on here.16:40
infinityEntertaining side note, the storage subsystem of the s390x mainframes is just a POWER system with the same RAID controllers in it as those ppc64el machines. ;)16:41
ogra_well, i'd use s390x builds for testing if i had that arch in my snap setup yet :)16:41
ogra_but ppc64el is equally good for that purpose16:42
ogra_they are both exotic in the snappy world :)16:42
ogra_and it is done ...16:42
ogra_aaand i see proposed *everywhere* !16:43
xnoxinfinity, so we can use our mainframe as ubuntu ppc64el builders too?16:43
ogra_so yeah ... obviously the PPA setup is at fault16:43
xnoxogra_, it works now?16:43
ogra_xnox, well, the outer build chroot shows proposed on apt update ... so yeah ... that was what i was missing16:44
ogra_my prob is that i dont want the packages in the PPA i use to build against proposed ...16:44
infinityxnox: Only if you can figure out how to install Ubuntu on the storage server and still have it serve slices to the mainframe. :P16:44
infinityogra_: Right, so the quick fix is to have two PPAs.  Build your packages in your updates PPA, copy them to the proposed PPA, and build updates snaps in A and proposed snaps in B.16:45
infinityogra_: The better solution is asking William or Colin if there's a way to set PPA deps per-build via the API and, if not, if they can give us a way.16:45
ogra_yeah, something like that ... but it is a bit sad ... the GUI option should actually override that imho16:45
ogra_(and simply add proposed to the sources.list when i select it)16:46
infinityogra_: And I agree, this misfeature is dangerously close to a bug.16:46
xnoxi think there is json to enable arbitrary repos for the "inner build"16:46
xnoxbut i can't remember it anymore16:46
infinityogra_: Building things this way isn't heavily tested yet, so it might just be a bug.16:46
xnoxneed to dig into livebuild lp:lp code16:46
ogra_xnox, forcing the inner build to proposed isnt my issue ... i wanted to cleverly read the info from the outer build so i dont need to change the code every time16:47
ogra_techincally i can build wnat i want today16:47
infinityogra_: (ie: it's mostly just been used for one-off livefs testing, and the people who do that a lot probably have the "right" pockets in their PPA and never noticed the issue)16:47
infinityogra_: So, yeah.  File a bug, let's talk it over with the LP folks, but you have a workaround for now at least.16:48
cyphermoxinfinity: nvme: well we know dmj_s76 has filed a bug about nvme in general, which probably needs SRUing17:05
xnoxcyphermox, but apperantely 14.04.5 regressed vs 14.04.4 and we did nothing there, no?17:06
infinityThis is what he claims...17:06
dmj_s76cyphermox: two separate issues17:06
infinityWould be nice to get to the bottom of this.17:06
cyphermoxI wouldn't be able to tell if it regressed, I would expect not17:06
infinityIdeally today.17:06
xnoxi have nvme laptop.17:07
dmj_s76cyphermox: the existing bug applies to bios installs and the ubiquity graphical installer17:07
infinitycyphermox: Can you work with him to get a bunch of debugging info from .4 and .5 and see WTF is going on, or at least try to make sense of it?17:07
xnoxit's not full system backed up. I don't even know how to do that for a windows & ubuntu dual install.17:07
infinityxnox: dd17:07
* xnox ponders if i can try tripple-boot17:07
dmj_s76the issue in 14.04.5 is separate and applies to successful installs, which get past brub17:08
xnoxinfinity, that laptop's nvme is a handy 1TB drive =) so i need a spare 1TB somewhere else.17:08
infinityxnox: That must have cost a bit.17:08
dmj_s76xnox: wow, that's a sweet nvme17:08
xnox1 700 old pounds17:08
dmj_s76Mine's a 512 GB17:09
cyphermoxdmj_s76: please make sure you have a bug for each individual issue so we can make sense of it all17:09
xnoxit has ininity tiny bazel 4k touch screen too, 17"17:09
infinityYeah, I thought my 1TB SATA 850 Pro was expensive and fancy, but you've got me beat for extravagant spending. :P17:09
xnoxwell 1 700 is for the whole laptop. Just 1TB alone, i can't remember how much it was.17:09
xnoxbut then i blagged discount of dell.com/uk chat function (after abandoning shopping cart 3 days in a row)17:10
xnox(cause they track that)17:10
infinityThough, in real-world use (not synthetic benchmarks), I wonder how much your nvme really outperforms my sata SSD.17:10
xnoxit doesn't feel any faster than server grade SSDs17:10
xnoxbut 1TB on a laptop is nice.17:10
infinityYeahp, I enjoy my 1TB SSD.17:11
dmj_s76Depends on if you're doing a bunch of IO intensive stuff17:11
infinityMy data is goldfish-like, I never need the space, but as soon as I have it, it's full. :/17:11
dmj_s76It's nice for manipulating GB+ images for instance17:11
infinityOh, FFS Lenovo.17:13
infinityI don't think my old Thinkpad can disable SB.17:13
infinityThat has to be against spec.17:13
apwinfinity, really? i am supprised17:13
cyphermoxwhich device?17:13
infinityMaybe they didn't think it necessary, since you can bypass it entirely by switching to Legacy BIOS mode.17:14
cyphermoxweird. X230s could17:14
cyphermoxwhere you disable SB, you can put it in Setup mode.17:14
cyphermoxor enable CSM support, which can't allow SB.17:15
infinityI literally don't have the Secure Boot menu option that I see on screenshots from the **30 series.17:16
cyphermoxwell, it used to not be under Secure Boot specifically17:16
infinityAnd that's definitely against spec.17:16
xnoxinfinity, my old Gauja laptop hid the SB menu inside the "disk password" sub menu17:17
dmj_s76cyphermox: do you expect this nvme->initramfs issue will be an ubiquity issue or another package?17:18
cyphermoxif it's after you've installed, then it's another package17:18
infinityOh, I wonder if it's the inverse, that it doesn't support SB at all.17:19
* infinity tests theory.17:19
infinityThe **30 series might have been the first that was SB-enabled.17:19
infinityIf this experiment goes sideways, I guess I'll be stuck playing in qemu.17:20
infinityovmf makes me want to stab myself.17:20
* xnox derp. got +mac, not -amd64.iso17:23
xnoxok doing UEFI, no secure boot, trusty desktop install17:28
xnoxwith nvme17:28
infinityDoing the same, minus nvme.17:32
infinityWell, unsure about secureboot.  Could be on, could be off, LENOVO WON'T TELL ME!17:32
infinityBut the kernel will, later. :P17:33
xnoxis your test machine with nvidia?17:33
infinityIt is.17:34
infinityPretty old, but nvidia nonetheless.17:34
infinity(Well, intel/nvidia optimus mojo, but I have it set to nvidia-only to avoid confusion)17:34
dmj_s76xnox: nvidia shouldn't matter too much wrt the nvme issue17:35
infinityAfter typing on the new Lenovo keyboard for a couple of years, the old keyboard feels really weird.17:35
xnoxinstall finished, restarting17:36
xnoxomg upstart17:36
apwdmj_s76, which kernel module supports your nvme drive ...17:36
xnoxit may or may not restart =) it is ovmf after all17:36
xnoxdmj_s76, confirming drops to initramfs17:39
xnoxin kvm-qemu with uefi and ovmf17:39
dmj_s76It's a samsung 950 pro...17:40
xnoxi didn't bother seting up nvram storage for variables, but that should not matter, after manually selecting nvme drive in the uefi shell to boot grubx64 it drops to initramfs17:40
xnoxapw, cyphermox - there is no nvme.ko in the initramfs17:41
dmj_s76lspci reports that it uses the 'nvme' module17:42
cyphermoxxnox: ok?17:42
xnox(initramfs) modinfo nvme17:42
xnoxmodinfo: cna't open '/4.4.0-31-generic/': No such file or directory17:42
xnoxshould't that be like -> /lib/modules/4.4.... ?17:43
xnoxi recall we were patching something rather to include nvme by default17:43
xnoxdid that make it back to trusty?17:43
apwxnox, well it worked in .4 i think we were told17:43
xnoxapw, sure, but .5 installer is now built with xenial stack17:44
xnoxno? hence it's all different again.17:44
apwxnox, right but nvme in the initramfs is an initramfs-tools thing normally17:45
xnoxI see copy_modules_dir kernel/drivers/nvme in xenial initramfs-tools17:46
xnoxbut not in trusty17:46
apwxnox, indeed ... which would imply .4 would not have worked either17:47
infinityapw: Or maybe they used a different driver in 4.217:47
xnoxapw, infinity - https://anonscm.debian.org/cgit/kernel/initramfs-tools.git/commit/?id=fe30453892d21ed20014e48f597be2a79ac99d2617:47
xnoxinfinity, i think at one point it changed to subdir from just nvme.ko17:47
xnoxi do recall uploading that fix into initramfs-tools and saying we will have to backport that to "$foo" or some such17:48
infinityapw: We should probably sync the driver list from xenial back to trusty (or, rather, add any missing ones, not remove any).17:48
infinityapw: Any chance I could get you to look at that before you clock out?17:48
xnoxcan't remember if that $foo was trusty or $clear-linux17:48
apwinfinity, yeah will have a look if they are in the same form, if so then it should be easy17:48
xnoxi guess i can boot that system as -sda17:48
xnoxupdate initramfs with that one line in17:48
xnoxreboot with nvme again to check it's all good17:49
infinityxnox: Yep.17:49
xnoxbecause i need to run to beach volleyball in the rain soon17:49
infinityIt was probable in the block subdir before.17:49
infinityDerpy derp.17:49
xnoxah comment is nice17:50
xnoxnvme moves to its own directory in Linux 4.4, so include modules17:50
xnoxfrom there.17:50
infinityThat explains the regression.17:50
infinityBut worth a check of all changes to the module list between trusty and xenial to see if we're missing anything else too.17:51
apwugg they are in utterly differnt forms17:53
infinityapw: That one spot isn't.17:55
infinityapw: But yeah, the big lists might be problematic.17:55
cyphermoxxnox: sounds like you have a fair handle on this, can I ignore that initramfs issue and focus on the other nvme bug with ubiquity?17:55
xnoxwith copy_modules_dir a following pokemon appeared in the initramfs17:55
infinitycyphermox: I think so, yeah.17:55
xnoxrebooting to test17:56
xnoxumount: invalid option -- 'R' ->>>> argh!17:56
cyphermoxxnox: you mean "Arrr!"17:56
davmor2cyphermox: I've told you before Ubuntu-Pirate isn't an official release17:58
xnoxit boots17:59
infinityxnox: \o/18:00
xnoxinfinity, do i have a bug number for sru upload?18:01
* xnox makes one18:02
xnoxvoila there is one18:02
ubot5Launchpad bug 1608622 in initramfs-tools (Ubuntu) "14.04.5 drops to initramfs prompt during boot with nvme drive" [Undecided,New]18:02
infinityxnox: I have apw trying to reconcile the whole module list, not just nvme.18:03
infinityxnox: If he fails at that, we'll just fix this one bug. :P18:03
infinitydmj_s76: What's the nvidia behaviour you see?18:03
xnoxi recall that was the only thing i had to fix.18:03
infinitydmj_s76: I can confirm the driver sucking here, though not sure if it's the same as your suck.18:03
cyphermoxcould it be a SB issue?18:04
infinityxnox: The only thing you had to fix... In trusty?18:04
infinityxnox: Or xenial?18:04
infinityxnox: Cause they've diverged a lot. :P18:04
xnoxtrying to remember18:04
infinitycyphermox: module loads fine, so no.18:04
cyphermoxkay, just checking18:04
infinitycyphermox: compiz explodes spectacularly on login.18:04
cyphermoxthat's not a regression18:05
xnoxinfinity, ^ above is just cherrypick copy_modules nvme18:05
infinitycyphermox: It's not?18:05
cyphermoxinfinity: nevermind, I was thinking of something else18:05
dmj_s76infinity: Are you getting a reboot loop?18:06
dmj_s76*login loop18:06
dmj_s76(not reboot)18:06
infinitydmj_s76: No, I'm getting compiz freezing hard after login.18:06
infinitydmj_s76: But could be the same thing, behaving differently on different cards.18:07
infinitydmj_s76: To confirm, X starts, lightdm is fine, you can login, then world explodes?18:07
* xnox the world does not explode in kvm/qemu after fixed initramfs hook as per above sru18:07
* xnox is out for volleyball18:07
dmj_s76infinity: I suspect so...on mine, when I login, there's a yucky crashy behavior for a split second, then back to lightdm18:08
infinitydmj_s76: Right.  So, yours might just recover better from the same thing.  Maybe.18:08
dmj_s76What driver are you using?18:08
infinitydmj_s76: That's the 352 driver?18:08
dmj_s76I've experienced this issue with 352 and 36718:09
dmj_s76and 36118:09
infinityNothing newer than 352 exists in trusty, afaict.18:09
dmj_s76infinity: We package 361 (and soon 367) for hardware compatibility.18:10
infinitytjaalton: Remember when tseliot told us that this HWE-X + nvidia thing was tested?  Was that a lie?18:11
dmj_s76...at first I thought it might be some packaging quirk, but it's hitting 352 the same way.18:11
tjaaltoninfinity: I don't know?18:12
infinitytjaalton: Do you have nvidia hardware to help test and get to the bottom of this?18:12
infinitytjaalton: Cause it sure looks broken to two of us...18:12
tjaaltoni have an old 8600gt somewhere18:13
tjaaltonbut a logfile would be nice to see18:13
tjaaltonif it crashes back to lightdm18:13
dmj_s76tjaalton: which logs would you like?18:13
infinityMine hardlocks.  But dmj_s76 gets a nice loop, so he might be helpful.18:13
tjaaltondmj_s76: /var/log/Xorg.0.log.old18:14
infinityAnd dmesg, perhaps.18:14
infinityI get the same breakage with 340 and 352.  Highly suspect.18:14
tjaaltonis the dkms built?18:14
infinityBuilt, and it loads.18:15
infinityWell, it loads manually.18:15
infinityHard to say if it loads automatically, because locked computer. ;)18:15
tjaaltonthe system hangs hard?18:16
infinityWith 352, I get the nvidia logo, though, which I'm pretty sure won't happen unless the module loaded.18:16
infinityAnd compiz draws a bit of purple.18:16
tjaaltonyeah, think so18:16
infinityAnd then explodes.18:16
infinityYeah, hung hard.  Can't VT switch or anything.18:16
infinityDisplay hung hard.18:16
infinityI should install sshd to see if it's otherwise alive.18:16
* infinity reboots again.18:16
infinitytjaalton: Alright, machine not hung, just display.18:21
infinitytjaalton: Xorg logs don't seem angry in any way. :/18:22
infinityI wonder if compiz/unity got updates that don't work for some reason.18:22
infinityOh, and Xorg is eating a core.18:23
infinityGo, X, go.18:23
tjaaltonnothing in dmesg either?18:23
tjaaltonpastebin xorg log anyway if it has some ideas18:24
infinityHrm, it finally gave me a no screens found.18:25
infinityThough, still also hung and spinning 100%.18:25
dmj_s76...just got the logs...18:25
tjaalton[    63.862] (EE) NVIDIA(GPU-0): EVO Push buffer channel allocation failed18:27
tjaaltonand so on18:27
tjaaltonnothing on dmesg?18:27
infinityOoo, I got fun stuff after the server terminated.18:27
infinityBecause of a soft lockup, no less...18:28
dmj_s76old xorg log: http://paste.ubuntu.com/21791403/18:28
apwinfinity, do we have a bug for the nvme issue ?18:28
tjaaltondmj_s76: 367.35? is that in trusty?18:29
infinityapw: We do.18:29
infinitytjaalton: It's not.18:29
tjaaltonok so just testing18:29
tjaaltonthat gets further though18:29
dmj_s76dmesg: http://paste.ubuntu.com/21791516/18:29
dmj_s76tjaalton: yes...14.04.518:29
infinityapw: 12:16 < infinity> And then explodes.                                                                12:16 < infinity> Yeah, hung hard.  Can't VT switch or anything.                                    12:16 < infinity> Well.                                                                             12:16 < infinity> Display hung hard.                                                                12:16 < infinity> I should install sshd to ...18:29
infinity... see if it's otherwise alive.                             12:16  * infinity reboots again.                                                                    12:21 < infinity> tjaalton: Alright, machine not hung, just display.                                12:21 < tjaalton> good                                                                              12:22 < infinity> tjaalton: Xorg logs don't seem angry in any way. :/                  ...18:30
infinity...              12:22 < infinity> I wonder if compiz/unity got updates that don't work for some reason.             12:23 < infinity> Oh, and Xorg is eating a core.                                                    12:23 < infinity> Go, X, go.                                                                        12:23 < tjaalton> nothing in dmesg either?                                                          12:24 < tjaalton> pastebin ...18:30
infinity... xorg log anyway if it has some ideas                                     12:25 < infinity> Hrm, it finally gave me a no screens found.                                       12:25 < infinity> Though, still also hung and spinning 100%.                                        12:25 < infinity> Brilliant.                                                                        12:25 < dmj_s76> ...just got the logs...                               ...18:30
infinity...                              12:25 < infinity> http://paste.ubuntu.com/21791021/                                                 12:27 < tjaalton> [    63.862] (EE) NVIDIA(GPU-0): EVO Push buffer channel allocation failed        12:27 < tjaalton> and so on                                                                         12:27 < tjaalton> nothing on dmesg?                                                                 12:27 < ...18:30
infinity... infinity> Ooo, I got fun stuff after the server terminated.                                 12:28 < dmj_s76> http://paste.ubuntu.com/21791328/                                                  12:28 < infinity> Because of a soft lockup, no less...                                              12:28 < infinity> http://paste.ubuntu.com/21791383/                                                 12:28 < dmj_s76> old xorg log: ...18:30
infinity... http://paste.ubuntu.com/21791403/                                    12:28 < apw> infinity, do we have a bug for the nvme issue ?                                        12:29 < tjaalton> dmj_s76: 367.35? is that in trusty?                                               12:29 < infinity> apw: We do.                                                                       12:29 < infinity> tjaalton: It's not.                                      ...18:30
infinity...                          12:29 < tjaalton> ok so just testing                                                                12:29 < tjaalton> that gets further though                                                          12:29 < dmj_s76> dmesg: http://paste.ubuntu.com/21791516/18:30
infinityapw: https://bugs.launchpad.net/ubuntu/+source/initramfs-tools/+bug/160862218:30
ubot5Launchpad bug 1608622 in initramfs-tools (Ubuntu) "14.04.5 drops to initramfs prompt during boot with nvme drive" [Undecided,New]18:30
infinityBloody buffer...18:30
infinitydmj_s76: No, he meant is that driver in trusty, and it's not. :P18:30
davmor2infinity: you hate use right? Spammer! ;)18:31
tjaaltonhmm, do you have linux-signed installed?18:31
tjaaltonfor the lts-xenial kernel18:31
infinitytjaalton: Yep.18:31
tjaaltoninfinity: could it break something? or is the dkms signed somehow?18:31
dmj_s76yeah, the driver's from our ppa18:31
dmj_s76think we needed 361+ for the nvidia 9 series18:32
infinitytjaalton: The module loads, so I have a hard time blaming that.18:32
tjaaltonmakes sense18:32
infinitytjaalton: But I can do a fresh install in BIOS mode and see if we get confusing results.18:32
tjaaltondmj_s76: your xorg log looks perfectly fine to me18:34
infinityI swear, if this works in BIOS mode, I'll be very angry and very confused.18:36
infinitySo it better not.18:36
tjaaltonyeah don't think that would help18:36
infinityStill, ticking off all the boxes.18:36
infinityStill not positive dmj_s76 and I are seeing the same bug, but it's highly suspicious that two (wildly different) nvidia machines are both showing hatred for trusty.518:37
infinityAnd on a multi-year spread of upstream driver versions, too. :/18:37
dmj_s76yeah, from 340 through last month's 36718:38
tjaaltonare the nvidia versions same as on xenial?18:39
infinitytjaalton: xenial has 361, trusty has 352.18:43
tjaaltonno I mean 352 minor version18:44
infinitytjaalton: The 340 versions are similar, but would need to diff to see if they're identical.18:44
infinitytjaalton: 352 literally doesn't exist in xenial, it's a transitional package to 361.18:44
tjaaltonoh I see, right18:44
infinityHonestly, I'd rather everyone just use nouveau anyway, but I guess that response on the bug reports won't win me friends. :P18:50
dmj_s76infinity: Much as I'd like to be able to use and promote open drivers for nvidia, our customers would scream bloody murder18:51
infinitydmj_s76: Well, they're pretty useless for serious 3D use, sadly.18:51
infinityBut plymouth is prettier!18:51
infinityAnd that's what counts.18:51
dmj_s76laptop folk would complain about 1 hour batter life and scientific computing would complain about cuda not working18:51
infinitydmj_s76: Oh, you're an s76 guy.  Where's my usual s76 tester extraordinaire?18:52
infinityWhose name now escapes me...18:52
infinityjderose, that's right.18:53
dmj_s76He's still here, just doing some R&D in addition to testing now.  I joined a couple months ago so we could expand our capabilities.18:54
dmj_s76Jason and I are friends from Novacut days.18:54
infinityWell, tell him I fear change, and because of him I'm having a panic attack.18:54
infinityAnd he owes me.18:54
infinity(Not really, but maybe I can guilt him into a free beer)18:55
dmj_s76I think he'll join you for that beer18:55
tjaaltoninfinity: do you have  xserver-xorg-legacy-lts-xenial installed?18:56
infinitytjaalton: Okay, good news, it sucks just as hard in BIOS mode, so it's not EFI/SB related, at least.18:56
infinitytjaalton: And, uhh... Dunno.  Need to reboot again after this fresh install froze.18:57
infinityCause I forgot sshd. ;)18:57
infinitytjaalton: The bit that weirds me out is that lightdm comes up fine.18:57
tjaaltonoh right18:58
cyphermoxdmj_s76: will you be able to test the ubiquity fix for installing to nvme?18:58
dmj_s76cyphermox: gladly18:58
infinitytjaalton: No legacy-lts-xenial18:58
infinitytjaalton: Should I?18:58
tjaaltondoesn't hurt to try18:59
infinitytjaalton: I would, but it's uninstallable. ;)19:00
davmor2infinity, tjaalton: are there any issues on amd or is it just nvidia? and I assume we are talking 14.04.5 rather than 16.10 right?19:02
tjaaltondavmor2: no issues on amd since fglrx doesn't work on .519:02
infinity xserver-xorg-legacy-lts-xenial : Breaks: x11-common (< 1:7.7+10~) but 1:7.7+1ubuntu8.1 is to be installed19:02
infinitytjaalton: ^19:02
davmor2tjaalton: ouch I can see people bitching about that as they had fglrx in place on trusty :(19:03
tjaaltonforce-install it now to see if it helps19:03
infinityThose people shouldn't be reinstalling with the xenial stack, then.19:03
tjaaltondavmor2: that's on release notes19:03
infinityNo one's forcing them to.19:03
infinityThat's not entirely true.19:04
tjaaltonthough upgrading to it from lts-foo breaks as well19:04
infinitySadly, if you're on the utopic/vivid/wily stacks, we will effectively force you to upgrade by dropping support for them. :/19:04
infinitytjaalton: Does the whole amdgpupro thing or whatever not work for 14.04.5?19:05
davmor2tjaalton: indeed I'm not too concerned but there will be a mile of press on it because people don't read release notes ;)19:05
infinityI know fglrx is dead, but the new thing seems reasonably not awful.19:05
davmor2infinity: hahahaha19:06
infinityEvidently, I'm wrong. :P19:07
infinityAnyhow, I'm going to wait for the ubiquity nvme fix, then spin a bunch of RCs, then we can argue about nvidia some more.19:09
infinitySince it's only shipped on two images, and shouldn't block general testing.19:09
infinityBut the whole situation is disconcerting.19:10
infinityI guess I should try xenial on this laptop and be sure it works.19:10
infinityThen get to diffing.19:10
dmj_s76long as we get everything shiny for the final iso, I'm happy19:10
tjaaltoninfinity: probably does, but the packaging is not opensouced yet aiui19:11
tjaaltondavmor2: well I blogged about it months ago19:11
tjaaltoninfinity: oh actually, -legacy is not needed at all19:13
tjaaltonon trusty19:13
infinitytjaalton: Kay... Why would it be needed on xenial and not trusty if it's the same userspace code?19:13
* infinity boggles.19:13
tjaaltonthe wrapper is still shipped by x11-common and systemd is not there19:13
infinityI wanted the second part of that to make sense.19:14
infinityBut I'd be happier if I just didn't read it.19:14
tjaaltonlogind support19:14
infinitytjaalton: If you want to queue up a change to remove the uninstallable package, go nuts, but don't bother uploading.19:14
tjaaltonnot actually used by ubuntu yet, since lightdm doesn't support it19:14
infinitytjaalton: That'll go fine in a post-point-release SRU, or we can upload it this week if we find another issue to upload for (ie: if it's X's fault that nvidia is sad)19:15
tjaaltonerr I'm mixing things.. logind and non-root X are somewhat related, but the latter is not used19:15
infinitynon-root X?  But how am I supposed to keylog?19:16
tjaaltoninstall the legacy wrapper ;)19:16
tjaaltonwait, I have a nvidia-based AIO-machine sitting on my desk :D19:22
tjaaltonI'll test .5 on it tomorrow19:23
tjaaltonand harass tseliot19:23
tjaaltonit's a hybrid, but maybe that counts19:29
bdmurrayinfinity: I uploaded update-manager with HWE support to the trusty SRU queue. ^^  Would you mind having a look at it?19:33
infinitybdmurray: Erk.  That'll be a pain to review.  I should go find the equivalent precise upload, I guess, and see how similar they are.19:34
bdmurrayinfinity: they aren't similiar because of the split between update-manager / ubuntu-release-upgrader in Trusty and a redesign of update-manager.19:35
bdmurrayinfinity: additionally, I fixed some bugs in the precise code.19:36
infinitybdmurray: Fun.19:36
infinitybdmurray: Of course, this doesn't need to be on the ISOs, so I guess I can take my time here.19:36
infinity(It can be released in a week or two without dire consequence)19:37
bdmurrayinfinity: As I understand things the support for the HWE stack on Trusty is ending on the 4th.19:37
infinitybdmurray: Well, it's already ended, more or less.  There was no utopic/wily kernel upload this cycle.19:37
infinitybdmurray: But still, this doesn't need to be on the ISO, it can land after the ISOs spin.19:38
infinityAs in, people installing from the ISO get the xenial stack anyway, it's people upgrading that need these bits.19:38
bdmurrayRight, okay.19:38
infinitySo, I'll try to wrap my brain around it a bit later.  And accept if it looks good.  And we can test from proposed and stress it a bit.19:39
infinityAnd release after the ISOs are tested and out.19:39
dmj_s76cyphermox: So initial testing with your lates ubiquity nvme line looks good19:51
dmj_s76I've verified that it works well with nvme drives on 14.04.4 and yakkety19:57
dmj_s76not .5 for obvious initramfs reasons19:57
infinitytjaalton: nvidia-340 (same upstream version) works fine in xenial.  So.  Whee.20:13
* infinity goes to stab himself in the face for lunch.20:14
dmj_s76I should get lunch too...dig into nvidia after20:15
tjaaltoninfinity: yeah :/21:18
=== Beret- is now known as Beret
xnoxapw, nice one! there are many21:30

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