[10:25] infinity, cyphermox: hmmmm can one of you try i386 latest on kvm, I do a standard full disk install and the system reboots into the installed system but then the image fails to login it just reboots back to login screen [16:00] davmor2: infinity: looks like compiz is failing for some reason. I tried to downgrade it but it didn't do the trick. I see one LLVM error ("Do not know how to split the result of this operator!") just before the line in gnome-session.log mentioning compiz stopped, but it could be a red herring [16:01] cyphermox: weirdly if I install from mini.iso and netboot it works [16:01] err [16:01] that's messed up [16:01] or actually, it's not [16:01] if you install from mini.iso you won't get anything that is currently in proposed. [16:02] davmor2: try the mini.iso again, but add "apt-setup/proposed=true" on the command line as you boot :) [16:03] cyphermox: damn it I'd been testing that most of the day :) [16:03] ah, with proposed enabled on the mini iso? [16:03] cyphermox: no without I followed the link from the iso tracker to the mini iso [16:04] right [16:04] lemme look at proposed, maybe the solution here is obvious [16:05] it doesn't crash the same on amd64 does it? [16:05] cyphermox: no I tested that yesterday but let me spin up one of todays image and double check [16:06] it probably wouldn't be different :/ [16:07] eeesh [16:15] cyphermox: colour me impressed I have the mini.iso on the go and amd64 install happening at the same time on one machine :) [16:15] and the machine is still usable :) [16:25] fun [16:26] alright, so that desktop thing isn't specific to unity; if I install gnome-shell and remove untiy and compiz, things still go boom [16:32] ah, there, likely cause is mesa-lts-vivid :'( [16:33] davmor2: have you tried all this on actual hardware? [16:34] cyphermox: I'm running mac test on a mac right now, next up is i386 on another pc but I only have so many hands ;) [16:34] sure sure [16:34] there. If i remove mesa-lts-vivid (gasp!), things work again [16:35] cyphermox: so 64bit mac and 64bit kvm are both installing + 64bit installed on xps yesterday [16:35] ok [16:36] s/installing/logging in [16:36] cyphermox: so now I'm going to try i386 on hardware [16:37] ok [16:39] cyphermox: That llvm error is because you're running a qemu emulated CPU that confuses llvm. Try -cpu host or -cpu core2duo or something. [16:40] Also, the ISOs don't have proposed enabled. [16:40] davmor2: See above, re: llvm and -cpu on i386. [16:41] infinity: nice [16:41] I'll give it a try in a second [16:42] infinity: I'm going to try it hardware first and just confirm things are good there though [16:44] davmor2: The problem is that the default kvm emulated CPU on i386 either advertises some features it doesn't have or has some features it doesn't advertise (I don't recall which now), but this confuses the everliving crap out of llvmpipe. [16:44] davmor2: Asking it for a CPU that isn't broken makes it slightly less grumpy. :P [16:45] infinity: so ould that explain why netboot works then maybe? [16:45] davmor2: Well, define netboot "working"... [16:46] davmor2: If you mean you can install from a text-based installer, yeah, that'd work just fine, since no video drivers get involved. [16:46] infinity: i386 but then the installed system is logging in with no issues either [16:46] Invoking another video driver that skips past llvmpipe would also probably make things happier. Don't recall what we ship drivers for. Maybe -vga qxl [16:47] Anyhow, this is all based on cyphermox's error message above. Your problem *might* have been different, though it's rare for two people to have two different problems on the same fake computer. :P [16:51] infinity: how do you explain then that things work if I rip out that version of mesa and install the older one? [16:51] I mean, I can certainly play with teh qemu settings [16:52] that's also why I asked davmor2 to try it on real hardware [16:53] cyphermox: Different versions of mesa == different versions of llvmpipe. I don't recall when this first started happening. Might have been vivid/3.6 [16:53] fair enough [16:53] cyphermox: To be clear, it's as much an llvm bug as it is a qemu bug, but it has zero effect on real hardware, so the carefactor is low. [16:53] I would agree [16:54] so let's wait for davmor2 to confirm things work on hardware [16:54] hardware install is about 20% done so not too long [16:54] heh, so I was using the default cpu on qemu, oops [16:55] I'm also trying with -cpu host on this box too [16:57] cyphermox: If memory serves, the qemu bug is that it doesn't filter ASM feature checks, so "emulating" i686 on amd64 will allow you to feature check for thing like sse3 and avx, but they're not shown in flags. [16:57] good to know [16:57] cyphermox: And the llvmpipe bug is that it does ASM feature checks (not really a bug), but then runs code qemu tells it that it can't. [16:57] cyphermox: At least, that's how I remember it. [16:58] I picked core2duo here for now, regardless of whether this still displays the issue, at least it will be set in my scripts [16:59] restarting on hardware [16:59] and desktop [17:03] so that works on hardware then [17:13] Alright, ubuntu-drivers-common fix in the queue for the Mac/Dell wireless issue. [17:42] infinity: bingo \o/ -cpu host worked a treat I'll let evan who noticed it know so he can carry on testing \o/ [17:42] cyphermox: ^ [17:56] cyphermox, infinity: interestingly netboot with the "apt-setup/proposed=true" setting added still install and logged in correctly [17:57] davmor2: I wouldn't expect them to be much different. [17:57] davmor2: proposed has very little interesting in it right now. [17:58] davmor2: The reason cyphermox was recommending that was because he (incorrectly) thought your desktop ISO was built with proposed enabled, so mini+proposed might match behaviour. :P [17:58] infinity: the odd thing is though that didn't have the -cpu host option and just works [17:58] infinity: I'm not going to grumble though [17:59] It's possible llvmpipe only comes into play in the installer environment. [17:59] And you're skipping that whole mess when you use a d-i installer. [18:02] infinity: when did you disable -proposed? [18:04] cyphermox: Before building the RCs. [18:04] ok [18:08] infinity, cyphermox: right I downgraded the -cpu host bug to being just a bug rather than a blocker, I've changed the fail to a pass so that is one less \o/ With the bcmwl fix that will fix dell and mac installs and I think we are on the way to happiness at that point :) [18:11] cool. I'm having lunch right now, will get back to my world of multipathed pain immediately [18:12] cyphermox: I assume none of this mpath stuff is for .3 at this point? :P [18:13] it could have been, but it's kind of late ;) [18:14] davmor2: btw; if you could do one extra test for me on trusty, related to that llvm issue. I wrote the instructions down: http://api.viglink.com/api/click?format=go&jsonp=vglnk_143871199514610&key=f208743056ec188cd045d74886f5b45b&libId=icxny7ja0100sipx000DAo4yy8f35&loc=http%3A%2F%2Fbrojsimpson.com%2Fpranks%2Fhidden-rick-roll-video-link-collection-rickrolled%2F&v=1&out=http%3A%2F%2Ftinyurl.com%2F2g9mqh&ref=https% [18:14] 3A%2F%2Fwww.google.ca&title=Hidden%20Rick%20Roll%20Video%20Link%20Collection%20-%20Bro%20J%20Simpson&txt=http%3A%2F%2Ftinyurl.com%2F2g9mqh [18:14] argh [18:14] damnit ;) [18:14] that is like the worlds longest url [18:14] ... hidden-rick-roll-video-link-collection-rickrolled ... [18:15] yeah. so stealth-like ;) [18:15] * cyphermox shakes fist at google [18:15] cyphermox: http://knowyourmeme.com/memes/rickroll [18:15] Is that kym entry itself a rickroll? [18:16] It would kinda have to be. [18:16] Shockingly not. [18:16] in any case [18:16] infinity: multipath stuff was supposed to be for .3, but it's for us, and didn't quite work until today, so [18:24] davmor2: When this ubuntu-drivers-common is built, will you be willing to do a sneakernet usb-key validation for me? [18:25] davmor2: ie: copy the deb, dpkg -i, then run ubiquity as normal. [18:25] davmor2: Since setting up the network to test if we can automatically set up the network would be a bit daft. ;) [18:25] infinity: sure [18:27] davmor2: Should spit out here soonish: https://launchpad.net/ubuntu/+source/ubuntu-drivers-common/1:0.2.91.11/+build/7759062 [18:27] * infinity runs to find a quick coffee/bite. [18:36] davmor2: https://launchpad.net/ubuntu/+source/ubuntu-drivers-common/1:0.2.91.11/+build/7759062/+files/ubuntu-drivers-common_0.2.91.11_amd64.deb [18:39] infinity: on it [18:57] infinity: so I installed it in the live cd session and after I triggered the install and got past the 3rd party drivers page the wifi came alive so fingers crossed ;) [18:59] davmor2: \o/ [19:00] davmor2: If it's there in the installed system too, I win. [19:01] infinity: indeed [19:01] infinity: restarting [19:02] infinity: I CAN HAZ WIFI \o/ [19:04] davmor2: Super, with a side of duper. v-done the bug for me with a comment about your testing? [19:04] I HAZ WIFI \o/ this is the xps13 [19:08] davmor2: Comment and v-done https://bugs.launchpad.net/ubuntu/+source/ubuntu-drivers-common/+bug/1481018 please? :) [19:08] davmor3: ^ [19:08] TOO MANY OF YOU. [19:09] infinity: I did in the bug now commenting in the right place too :) [19:09] that is the bug do I need to comment on the patch too? [19:10] davmor2: No, only the bug comment is fine. Needs the tag flipped too, but I can do that. [19:10] done [19:11] cyphermox: BTW, apw is working on the persistence bug, no need to waste brain power or time on that. [19:11] k [19:12] cyphermox: Though, a second review wouldn't hurt, since we're going to have to fasttrack this. [19:12] yep [19:13] I'll wait for Andy to get me his second diff based on my review and hand it to you. [19:50] davmor2: After Andy sorts out this casper persistence bug, I think it's respin-the-world time. Can you think of any other bugs (from your or the community) that need RC attention before that happens? [19:51] infinity: yes I'm just looking at https://bugs.launchpad.net/canonical-devices-system-image/+bug/1473562 [19:51] infinity: I guess not critical for desktop but it is for phone [19:52] infinity: might be nice to get it landed it should be verified in 10 9 8... [19:54] davmor2: Eh, I'd say it's a pretty critically useful way to spend your time, even if it doesn't relate to my selfish goals for the week. :P [19:56] infinity: it has to land in desktop too it need to be SRU'd which it now is \o/ [19:58] infinity: so from what I've seen today with the bcmwl issue fixed, I think the only outstanding issue is that stupid touching the /EFI/boot/ on xps13 and that is a firmware issue over our software so I'm happy to not mark that, however I will add the workaround to the bug and then I would say respin the world :) [20:00] infinity: are there release notes for the .3 release if so it might be worth making a note of the xps bug in those as a nice user measure [20:01] davmor2: There should be release notes. I haven't copied things around yet to create them. [20:01] davmor2: If you're volunteering to help with that tomorrow, I'll gladly accept. :P [20:02] infinity: I have to retest everything :P [20:02] infinity: I will however make sure the bug is nicely updated with easy to follow steps for the workaround [20:08] davmor2: Heh. Cool. I hope the workaround becomes "upgrade your firmware" soon, but we'll see. [20:09] Maybe getting cking to yell at them would help. He loves telling people their firmware sucks. [20:13] cyphermox: Extra eyeballs on http://paste.ubuntu.com/12002535/ please. [20:13] infinity: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1480985 only a small tiny Workaround you would hardly notice it :D [20:13] sure [20:15] davmor2: So icky. I wish our OEM people had yelled at them for this instead of working around it in the factory images. :/ [20:19] infinity: cyphermox: guys I'm off. infinity are you able to update the iso tracker to pull in the new images once they are built? Then evan in china can make a start on them before I get in and I can finish off [20:20] davmor2: The tracker automagically updates when I build new images. [20:20] infinity: awesome, Will netboot need retesting? if so will that update too or does that need slapping? [20:21] anyway need to go [20:22] infinity: looks fine to me, but I'd still kick up casper to try this