[00:45] <ripps> Anybody have any idea what could be causing 1 second stutters every 5-10 minutes in Quantal?
[00:47] <ripps> It's like my video freezes for about a second, but everything keeps running. If I'm watching a movie, the sound keeps playing and the video starts up again, if I'm playing minecraft, it's like the framerate drops to 1-2fps, and apparently my keybard was still working, because things I click and do in a panic at that point had occured while the video was frozen.
[00:48] <bjsnider> ripps, check dmesg
[00:48] <bjsnider> and .xsession-errors
[00:50] <ripps> bjsnider: hmmm... it appears i have a few kernel oops, it seems to be with lowmem_reserve()
[00:51] <bjsnider> even one kernel oops ain't not good
[00:51] <ripps> ah, it seems it's coming from java and ps3mediaserver
[00:51] <ripps> I have ps3mediaserver running as a service
[00:52] <ripps> java seems to be stealing more memory than usual, didn't have any issues with 12.04
[00:52] <bjsnider> must be pretty ugly software
[00:57] <ripps> hmm... lots of errors in my .xsession-errors, most have to do with wonky japanese fonts, but I get a few relating to java and synapse
[01:00] <ripps> man, clutter seems to be a mess in gnome-shell. I keep seeing actor is null being raised by clutter for my extensions and other things
[01:02] <ripps> man, i'm not even sure if half of this is relevant, it seems i have things in here that are weeks old. I didn't know xsession-errors never cleaned itself out
[07:04] <mlankhorst> morning
[07:21] <tjaalton_> yeah
[07:50] <mlankhorst> seems that ppa3 updates worked :)
[08:18] <tjaalton> worked how?
[08:22] <mlankhorst> it rejected all the ones that didn't update and rest compiled
[08:23] <mlankhorst> https://launchpad.net/~ubuntu-x-swat/+archive/q-lts-backport
[08:24] <mlankhorst> but still seems displaylink is missing
[08:26] <tjaalton> yeah I'll fix that
[08:27] <mlankhorst> and libdrm 2.4.38 came out, I'll fix that :)
[08:29] <mlankhorst> going to nbeed to do a testbuild on armhf because iirc omap symbols changed again
[08:39] <tjaalton> huh, I thought -displaylink had been pushed to git
[08:40] <tjaalton> can't find it
[08:40] <tjaalton> and I packaged it :P
[08:40] <mlankhorst> haha
[08:41] <mlankhorst> now to build on arm
[08:47] <tjaalton> ahhh
[08:48] <tjaalton> no it was -modesetting that I packaged
[08:48] <tjaalton> phew
[08:48] <tjaalton> this one needs to be moved to git
[08:51] <mlankhorst> might be why I missed it then
[08:52] <mlankhorst> ok debian-experimental updated to 2.4.38 :)
[08:54] <mlankhorst> tjaalton: Could you upload libdrm-2.4.38?
[08:56] <tjaalton> sure
[08:57] <tjaalton> mlankhorst: will you merge or should I
[09:00] <mlankhorst> tjaalton: there's no changes in libdrm, so just take debian-experimental version and call it 0ubuntu1 :)
[09:00] <mlankhorst> or at least I thought that's how it was supposed to work
[09:00] <tjaalton> oh, ok
[09:02] <mlankhorst> just look, ubuntu branch is missing 2.4.32 etc :)
[09:05] <tjaalton> so it seems. I'll still merge it to the ubuntu branch
[09:06] <mlankhorst> sure
[09:08] <tjaalton> hmm, building the source package fails
[09:09] <mlankhorst> you need to remove some files and fix symlinks iirc
[09:09] <mlankhorst> see README.source :)
[09:09] <tjaalton> ah
[09:10] <mlankhorst> that find command should be enough, but might want to delete some of the untracked files too
[09:10] <tjaalton> those are fine
[09:11] <tjaalton> I wonder though if rules should be updated too for the new symbols
[09:11] <tjaalton> the dh_makeshlibs lines
[09:12] <mlankhorst> hm, maybe change them to -V ?
[09:13] <mlankhorst> well looks like most of them would need to be bumped to v2.4.38 indeed
[09:13] <tjaalton> yeah
[09:13] <mlankhorst> except nouveau1a, that one isn't going to change any more
[09:14] <mlankhorst> but I don't get that part, I'm not sure if updating would be needed or not
[09:15] <tjaalton> see commit 7d6eca41f92c247e0be2b1562c833e6ef410977f for an example
[09:16] <mlankhorst> yeah but is it needed if nobody uses the exports for example?
[09:16] <mlankhorst> but then again I'm a packaging noob, just do what is best :)
[09:29] <tjaalton> yeah I bumped it for intel and nouveau2
[09:29] <mlankhorst> omap too btw
[09:31] <tjaalton> true, but it hasn't had the version there
[09:32] <tjaalton> guess without the version there the dep could be too tight
[09:32] <tjaalton> ok I'll add it for omap too
[09:35] <tjaalton> uploaded
[09:35] <mlankhorst> great :)
[09:35] <mlankhorst> now I just need to evilly sneak in the mesa patch and I'd only need intel/nouveau in my prime repository
[13:37]  * popey waves bug 1036186 around the place
[13:38] <tjaalton> popey: you have -proposed enabled?
[13:38] <popey> tjaalton, ya
[13:39] <tjaalton> vbox needs to be fixed and reuploaded there to build against the new abi
[13:39] <tjaalton> -re
[13:39] <popey> its blocking some testing I am doing :(
[13:39] <popey> because unity 2d is now gone, i can only use unity in virtualbox, and can only do that if virtualbox-guest-x11 is there
[13:39] <tjaalton> downgrade the x stack?
[13:40] <tjaalton> will be fixed later this week tho
[13:41] <popey> it's automated testing, I'd rather not have to bodge things to test it
[13:42] <popey> the point is to run autopilot in a vm
[13:43] <tjaalton> why do you have -proposed there?
[13:45] <mlankhorst> tjaalton: did you push libdrm to git?
[13:46] <tjaalton> mlankhorst: yes
[13:49] <mlankhorst> oh woopd expected it to be top commit :)
[13:51] <popey> tjaalton, testing bleeding edge stuff, will try without it
[13:52] <popey> tjaalton, thanks :)
[13:58] <mlankhorst> Just need the xorg drivers, a patch to mesa, xrandr, and kernel for intel + nouveau prime in lts-q-backports
[13:59] <tjaalton> popey: will get sorted out later this week
[14:02] <tjaalton> hmm, wondering if -displaylink is unnecessary at this point where we have the kms driver and -modesetting..
[14:09] <mlankhorst> I don't think it would be needed, prime uses modesetting driver I think
[14:11] <tjaalton> yeah
[14:11] <tjaalton> so, I won't push it to git then, instead file a request for removal :)
[14:11] <mlankhorst> I should probably have to track those cases then, see what drivers didn't update for quantal
[17:28] <c10ud> hello there, just got home and saw the new nvidia 304.37 (which should fix some bugs i am experiencing) any ETA on the PPA?
[18:13] <ricotz> c10ud, soon
[18:15] <c10ud> ricotz, thanks. i'd install by hand but i like keeping my system clean..let's hope i can play soon (since precise, gaming has been a no-no)
[18:28] <ricotz> c10ud, jfyi, i will upload it to xorg-edgers though, not x-updates
[19:00] <c10ud> power loss -- ricotz i'm with x-updates, what's the chance of getting it soon?
[19:01] <c10ud> or, if only a rebuild is needed, i could do the trick in my ppa
[19:09] <ricotz> c10ud, you should be able to use the x-edgers packages directly, so cherry picking the packages from there without a rebuilt is possible
[19:09] <c10ud> as soon as they're built i'll try them, thanks
[19:10] <ricotz> will take some minutes -- https://launchpad.net/~xorg-edgers/+archive/ppa/+sourcepub/2607543/+listing-archive-extra
[19:12] <c10ud> yes i was just looking at the x64 build, thanks
[19:38] <c10ud> ricotz, nvidia-current depends on xserver-xorg-core (>= 2:1.11.99.901); so it's a no-no
[19:38] <c10ud> if there's no x dependant file i can try overriding the dep (!)
[19:39] <mlankhorst> speaking of which
[19:39] <mlankhorst> does nvidia-current hardcode the abi yet?
[19:39] <ricotz> c10ud, ah :\
[19:39] <ricotz> mlankhorst, yes it does
[19:39] <mlankhorst> ah great :)
[19:40] <ricotz> mlankhorst, 11, 12 and 13
[19:40] <mlankhorst> I should remove the versioned xserver-xorg-core dep and add it to the qbp tree then :)
[19:40] <mlankhorst> s/tree/ppa/
[20:31] <tjaalton> bryceh: I had a chat on #ubuntu-release with stgraber and infinity about the mesa in precise-proposed, and we decided it would be best to drop the current version there, and only add the (critical) ivb patch to the one in precise. we can then proceed with further testing of the upstream bugfix releases and provide them post .1 if they're good
[20:33] <tjaalton> it's ugly but oh well..
[20:38] <tjaalton> I'll probably create a new branch for this one, or maybe skip pushing it to git since it's a one-off thing
[20:43] <bryceh> tjaalton, hrm, ok
[20:43] <mlankhorst> bryceh: finally getting some serious feedback on my fence patches, hopefully close to getting it in dma-buf next now :)
[20:44] <bryceh> tjaalton, think we should still target 8.0.3 for precise, or move to looking at 8.0.4, or ...?
[20:44] <mlankhorst> did you take a look yet at lts-backports-quantal stack btw?
[20:44] <tjaalton> bryceh: yeah.. but this way we can move straight to 8.0.4 or even 8.0.5 since it should be released soon
[20:44] <bryceh> tjaalton, also did you get a chance to run the piglit tests yourself?
[20:45] <tjaalton> bryceh: didn't have time to set up the rig today, but will look at it tomorrow
[20:45] <bryceh> ok, well no hurry, sounds like we have time
[20:45] <tjaalton> yeah
[20:46] <bryceh> I'm going to try to get a couple more test boxes constructed and try to get the tests a bit more automated
[20:47] <tjaalton> I only have one box to test on, and then swap cards as needed :)
[20:47] <bryceh> I've been setting up stuff to dd reimage the systems between tests to hopefully make it easier to isolate changes and get better fidelity
[20:48] <mlankhorst> ah, I've been considering using btrfs snapshots for that..
[20:48] <tjaalton> heh, and I have cobbler
[20:48] <bryceh> kees had some llvm setup he used
[20:49] <bryceh> but just dd'ing the images wfm
[20:50] <mlankhorst> bryceh: but seriously can you look at q-lts-backport? I need some feedback if it's workable or not :)
[20:51] <bryceh> mlankhorst, sure, what specifically needs reviewed?
[20:51] <mlankhorst> if you agree with the mechanics
[20:53] <mlankhorst> for example x11proto don't get renamed, neither does libdrm, mesa gets a patch to build against newer libdrm with old nouveau abi, just to still be able to build from source :)
[20:54] <mlankhorst> xserver-xorg gets replaced by xserver-xorg-lts-quantal, needs a bit of finetuning though.
[20:59] <bryceh> maybe given the above discussion, your patched mesa should target 8.0.2 for now; it's unclear if/when we'll get 8.0.3 in precise
[21:04] <bryceh> mlankhorst, so would you say the critical assumption here is that uprevving libdrm in precise is going to be accepted by the sru crew?
[21:04] <mlankhorst> pretty much
[21:04] <mlankhorst> but other than that you can freely switch between new and old xorg stack
[21:05] <mlankhorst> I still need to add a recommends for linux-lts-quantal to xserver-xorg-lts-quantal so it automatically gets installed, but those who roll custom kernels can still remove it. :)
[21:06] <bryceh> I personally don't have a problem with putting new libdrm in precise, but I suspect getting it through SRU is going to take some doing
[21:07] <bryceh> is there a test suite for libdrm?
[21:07] <mlankhorst> not really
[21:07] <bryceh> hmm, there's a tests/ dir
[21:07] <mlankhorst> it's a joke
[21:07] <bryceh> oh
[21:08] <bryceh> mlankhorst, what specifically are we needing from libdrm?  any chance we could just cherrypick stuff?
[21:08] <mlankhorst> nouveau rewrite..
[21:08] <mlankhorst> and all the parts that the newer drivers require
[21:08] <mlankhorst> so not really
[21:09] <mlankhorst> that's the whole reason for updating it in the first place :)
[21:10] <bryceh> ok well the proto lib stuff is probably minor (although could also take work to get through sru).  The mesa change is fine, basically goes with the libdrm change right?  So really I think we need a definitive answer on libdrm sru-ification
[21:10] <mlankhorst> yeah
[21:11]  * bryceh ponders
[21:11] <bryceh> alright, first let's bring this up to ogasawara.  If we can get kernel team support, that may help.
[21:11] <mlankhorst> libdrm update makes my life a lot easier, and I've shown that the major obstacle for updating libdrm (nouveau) is solved. :)
[21:12] <bryceh> but I think before we go further we need to pow wow with an SRU person
[21:12] <mlankhorst> yeah
[21:13] <mlankhorst> lets wake up raof now :)
[21:13] <mlankhorst> he should be used to waking up at strange times
[21:13] <bryceh> that'd be mean!  ;-)
[21:14] <bryceh> I'll open an email with ogasawara.  Can you reply and follow up with a more detailed description of the issue?
[21:14] <mlankhorst> sure
[21:14] <mlankhorst> but not now, bed :)
[21:15] <bryceh> mlankhorst, yep
[21:15] <mlankhorst> Fortunately libdrm is rarely the cause of breakages EXCEPT WITH NOUVEAU :p
[21:18] <bryceh> mlankhorst, I wonder if we're likely to see similar problems with libdrm in the r and s stacks, or if this'll be a one time thing
[21:19] <bryceh> (I mean, obviously nouveau isn't going to get rewritten every 6 months, but could be changes in radeon or intel next time...?)
[21:19] <mlankhorst> I think it's best to keep updating libdrm if it's sane
[21:20] <mlankhorst> i think there was 1 more thing scheduled for nouveau but fortunately not there yet, new userspace<->kernel abi
[21:24] <mlankhorst> but seems sna doesn't really require a wild new api in libdrm, so hopefully libdrm becomes boring for now :)
[21:29] <bryceh> mlankhorst, remind me again what was preventing libdrm from being renamed?
[21:29] <mlankhorst> plymouth and other things hated it
[21:29] <mlankhorst> and if libdrm explodes your initramfs might no longer boot
[21:30] <bryceh> dah plymouth
[21:30] <mlankhorst> losing xorg can be recovered though 
[21:30] <mlankhorst> apt-get remove .*lts-quantal will return you to precise x stack
[21:31] <mlankhorst> or if it removed xorg, apt-get install xorg again after that
[21:31] <bryceh> mlankhorst, ok and the initramfs issue... what explodes exactly?
[21:32] <mlankhorst> I ended up with no libdrm.so.2 at all so if you regenerate your initramfs with that..
[21:33] <mlankhorst> plus i think it was causing the rest of the stack to not switch so easily back and forth
[21:34] <bryceh> ok
[21:35] <bryceh> mlankhorst, how much testing have you done of the newer libdrm with the original linux/X stack (plus the driver fixups you've already listed)?
[21:36] <mlankhorst> 13
[21:36] <mlankhorst> woops panda hung from io :)
[21:37] <mlankhorst> bryceh: a light amount, mostly if it builds or not, I should probably test some more locally first
[21:38] <mlankhorst> but I have the old and new abi for xxv-nouveau in that tree, and old abi mesa. I've built the new abi mesa succesfully too
[21:55] <bryceh> ok email sent.  sleep well, hopefully leann can give an answer today