=== ara is now known as Guest14068 === psivaa is now known as psivaa-lunch [12:54] cyphermox, cjwatson, FourDollars: I have issues that you guys can possibly help with. I'm trying desperately to install 14.04.3 on XPS 13 via USB pen drive. It installs but when I remove the pendrive for reboot there is no boot option so it just sits there. I'm trying to install via OEM obviously. I'll try a fresh go with a standard install and see if it is just an oem issue [13:06] davmor2: seems unlikely to be because of oem. what happens when it boots and just sits there? a black screen? [13:07] also, did you enable full disk encryption? [13:09] cyphermox: no to full encryption just a uefi install as that is what it defaults to. It looks to me like this might be an issue because although uefi is enabled by default secureboot isn't and that might be buggering up the availability of the boot option maybe? Let me try a legacy install [13:10] nah, you should be able to boot uefi even if secureboot isn't enabled, if the bios has both options [13:12] cyphermox: that's what you'd think right but no, I see the EFI folder on drive FOS iirc I can select any of the .efi files to boot from save that reboot and no change when you hit f2 it takes you back and there is no option saved to boot from [13:13] interesting [13:14] cyphermox: when I reboot is say something like press F1 to retry, F2 to setup or F5 for diagnostics, I hit F2 and there is no option in the boot menu for Ubuntu [13:14] ok, I see [13:15] could you try a 14.04.2 image then? [13:15] cyphermox: look like legacy is working [13:15] it would be good to know if it was already broken, or if we broke something with efibootmgr or shim [13:16] (kind of late to be finding out though) [13:16] cyphermox: let me try enabling secureboot and see if installs as expected then, and then I will grab a .2 install and see what happens then [13:16] ok [13:26] cyphermox: okay and now for the bad news, Works with secure boot enabled, So it is literally not working with the default setup that dell ships the device with, the only thing I can think of is our setup is only for securebooted systems and not uefi setups I have a pc that has legacy/uefi setup in a similar way so I might try the same thing there for addition confirmation [13:28] well, this sounds like a firmware issue a lot more than something we can fix in our software [13:28] one data point to get would be booting the installer in UEFI with secure boot *disabled*, in the live session, then running efibootmgr -v and archdetect [13:29] ^ so a bug would have enough information in case there's something we could do in software [13:29] cyphermox: no worries let me default the bios again and then uefi boot it [13:30] ok [13:36] cyphermox: should efibootmgr be installed by default? [13:36] I think it probably should be in the live image, but maybe I'm wrong [13:37] cyphermox: ah okay so it is on the image but not installed [13:37] oh, on the installed system you mean? [13:37] cyphermox: no I mean on live session, it was saying no such program [13:38] cyphermox: I had to apt install it from the image and then it ran [13:38] tbh I don't know. I would expect it to be there already because the installer should use it, but maybe it's just installed automatically later [13:40] efibootmgr is a dependency of grub-efi-amd64-bin; if that's not installed you have moderately bad problems. [13:40] Sounds perhaps like the installer itself wasn't booted in UEFI mode. [13:41] Oh, on the live session. I should learn to read. [13:41] cjwatson: yes this is in live session, once installed it ran fine [13:41] I don't think the installer itself uses efibootmgr, except via grub-install etc. but that's in the installed system. [13:42] So this is expected. [13:43] aye [13:43] davmor2: the really important part is what archdetect says [13:45] cyphermox: incoming mail [13:47] cjwatson, cyphermox: the good new is it is on the usb stick as there is no wifi setup, and the secureboot version of uefi setup booted as expected :) Just the non secureboot version that has issues [13:52] ugh, so either we broke something which worked in 14.04.2, or the BIOS is lying [13:53] in any case, efibootmgr doesn't seem to have added your boot entry [13:54] davmor2: since you're in the live session, could you carry on with the 14.04.3 install if it's not too late [13:55] and somewhere after the partitioning when it's doing the file copy, run archdetect again? [13:55] sure [13:56] this could be that partman writes a file that archdetect will then read to wrongly conclude you're not on efi after all [13:56] sure do you want both commands running again or just archdetect [13:56] just archdetect is fine [14:01] cyphermox: still looking the same on file copy and install/setup of files [14:03] cyphermox: oh insteresting though I now have an ubuntu entry in boot0000* pointing to shimx64.efi [14:04] so secureboot must have been on then? [14:05] cyphermox: and yet on reboot I get No bootable devices found pressF1 F2 F% [14:05] F5 [14:05] cyphermox: it hates me :) [14:05] quality software. [14:05] well, firmware, more likely [14:06] please see if we broke it in 14.04.3, by doing an install with secureboot off but using UEFI on 14.04.2 [14:10] cyphermox: will do, it'll take a minute I'll have to grab an image I think] [14:19] davmor2: that's a BIOS issue. [14:22] davmor2: Touch an empty file "bootx64.efi"under /EFI/boot of /dev/sda1 to workaround the problem. [14:22] davmor2: I am still working with the OEM to fix that problem. [14:23] FourDollars: but the stupid thing is it works fine if I enable secureboot just fails without secureboot which seems really odd :) [14:36] FourDollars: you sure the touch path is correct I only have EFI/ubuntu [14:52] cyphermox: for your info 14.04.2 installs and boots with no issues in that same setup [14:52] bleh [14:52] cyphermox: you want another efibootmgr -v from the installed system? [14:53] nah === psivaa-lunch is now known as psivaa [15:16] cyphermox: I'm just double checking the md5sum on the iso download to make sure it is good [15:17] I see no reason to believe it be broken [15:17] we might have broken something in changing shim or efibootmgr [15:19] cyphermox: checksums match [15:20] cyphermox: yeah I figured it would be good but also worth a sanity check for certainty :) [15:20] cyphermox: I'll put a bug together [15:20] k [15:52] cyphermox: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1480985 tada I'll let you reassign it to the right package when you figure out what broke and ermmmm you're welcome :) [15:53] cyphermox: didn't want you getting bored ;) [15:58] ah, I wonder too if that could be that things are being installed to the usb drive :/ [15:59] cyphermox: also the install 3rd party apps doesn't install the broadcom wifi driver which then becomes an issue as you can't then install them after as there is no ethernet port on the laptop [15:59] have you filed a bug about that one? [16:00] and is that different from 14.04.2? [16:04] Oh joy, .3 bugs already? [16:10] cyphermox: let me reinstall .2 [16:10] infinity: that's what I'm here for [16:13] davmor2: Yes, please mkdir /EFI/boot manually. [16:16] davmor2: The BIOS will check the existence of /EFI/boot/bootx64.efi for some unknown reason. I am still working with the OEM to fix this problem. [16:16] FourDollars: right but this works in .2 and doesn't in .3 which is a flaw our end right? [16:17] infinity: I'm checking the EFI booting in a VM right now, though that broadcom part sounds like something really broken [16:18] davmor2: It's a BIOS issue. I think it should also work for .3. [16:27] for some reason grabbing today's image using zsync takes forever over the VPN... and I can't exactly disconnect right now [16:27] ah, boo [16:32] cyphermox: why ah, boo [16:32] I disconnected that annoying VPN [16:45] cyphermox: oh so it looks like the file that FourDollars mentions is the fix if I install from .3 it is not there if I install from .2 it is meh [16:46] cyphermox: good news no regression, for the wireless driver however it is still broken badly [16:47] Erm, bootx64.efi is there on the *installed* system? [16:47] Isn't that the removable media path? Pretty sure it shouldn't be there. [16:47] davmor2: Does wily fail on this same laptop? [16:49] infinity: I can test in a minute [16:54] probably would fail, yeah [16:57] davmor2: yes I figure the wireless driver is some real oversight [16:57] I'll just finish my efi test and then get on with broadcon [16:57] cyphermox: just thought didn't we have this not install the bcmwl driver before in .2 for the macs and you had to revert something? [16:58] cyphermox: I guess that only happened on the mac install version of the cd though right? [16:58] davmor2: tbh I don't know I fortunately never had to look at bcm* before that much [16:59] or at least, not the packaging or how it gets installed [17:21] davmor2: Which list should I mail to hit (community) QA people for point release testing? [17:21] davmor2: About to spin up a bunch of RC images and spam people. [17:51] infinity: good question let me see which channels I'm on that they normally go out to 1 second [17:52] davmor2: s/channels/mailing lists/ [17:52] davmor2: I'm assuming I want ubuntu-quality@ [17:52] infinity: Ubuntu Quality Team [17:52] Check. [17:53] We need a flavour leads alias or something that hits all the leads, so I don't have to worry about if they're subscribed to quality or release or in the right channels. [17:54] infinity: looks like everyone else organises their own thing, irc channel wise I would suggest, #xubuntu-devel, #kubuntu-devel, #ubuntu-quality, Not sure about lubuntu but I'm sure they get the message [17:55] infinity: I assume on #ubuntu-release too [17:55] davmor2: Yeah, I tend to just mail release/quality and hope the word spreads, but if I see a flavour slacking, I'll hunt people down. Or you can for me. ;) [17:56] infinity: :) [18:08] cyphermox: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1481018 this one I think is more major I'm going to add a work around for the dell bug. It appears that if I add the file FourDollars mentions .3 now boots if I point the ubuntu boot at the shim.efi file [18:14] cyphermox: I'm just wiping it again to be sure [18:14] cyphermox: and did you like my little work around for the no wifi that took some doing I can tell you :) [18:15] davmor2: That bootx64.efi thing is confusing me. Was .2 really setting it up that way, or are we suffering some bad testing science here? [18:15] davmor2: Cause nothing should be setting up files in the removable path. [18:15] davmor2: A previous Windows install might have left cruft there, though. [18:16] infinity: it's an ubuntu dell there is no windows [18:17] davmor2: Came pre-installed? [18:17] infinity: yeap [18:17] davmor2: If so, it's possible the preinstall image pulls hacks to make that path be there. [18:17] davmor2: And when you installed .2, you didn't wipe the EFI partition. But when you installed .3, you did. [18:17] davmor2: That'd be my guess anyway. [18:17] Cause we really, really shouldn't be writing to that path, in any version. [18:18] infinity: could be. I set it to wipe the drive and install both times [18:18] doesn't appear to be [18:18] Anyhow, I'd like to re-test .2 and confirm, but I doubt this is a regression. [18:18] It's shitty firmware on Dell's part, to be sure, but not a stop the line thing for us if it's not a regression. [18:20] infinity: the reason I had the xps sent to me is to test 14.04.3 on specifically to stop the line on anything that doesn't work [18:21] davmor2: Understood. Just saying that if it's not a regression from .2, I don't particularly care, unless we have a paid contract telling us we *must* support it better in .3 than we did in .2. [18:21] davmor2: And if that's the case, OEM really dropped the ball. :P [18:21] Also, it's just plain wrong to write to that path on non-reomvable media, so if we were going to "fix" this, we'd have to special case the fix just for the affected systems, and only if Ubuntu is the only OS installed, blah blah. [18:21] Which is gross on so many levels. [18:21] infinity: indeed, if it is not our software at all at fault then it is over to FourDollars and beating the oem into submission with a thank you stick [18:22] davmor2: I can definitevely say that if creating that path "fixes" it, it's not our software at fault: their firmware is out of spec. [18:22] But we *could* work around that. [18:23] 3 days before release isn't the time for vendor-specific hacks, though. [18:23] (Unless there's a lot of money changing hands to make that happen, then we can do just about anything... *cough*) [18:24] infinity: agreed I was testing upgrade the system itself on latest and stuff last week which is when I received the device so to be fair I'm trying my hardest to find faults :) [18:24] davmor2: Yeah, not criticising you here. Finding faults is your job. And we appreciate it. Sort of. After we're done being annoyed. [18:24] infinity: and these are the only 2 I've found [18:25] EXPECTED: [18:25] I expect to always be able to get online. [18:25] Hah. [18:25] davmor2: So, this broadcom bug. The driver *is* there on the install media, and it works? Your workaround implies that. [18:25] davmor2: So it's just that ubuntu-drivers-common isn't picking up the need for it? [18:26] infinity: yeap just isn't installed with 3rd party install checked [18:26] davmor2: Except. It is too, according to your workaround. Weird. [18:26] * infinity wonders how to test this without the hardware. [18:26] infinity: I can also install in the live cd session too [18:26] davmor2: If you install in the live session, does that persist to the installed system? [18:27] * infinity doesn't recall if that's meant to work. [18:27] infinity: that I haven't tried but I can do that when this disk finished writing [18:27] I think maybe in this case we need to know what's the pci ID and everything for that device [18:27] cyphermox: Who is "we"? [18:27] cyphermox: ubuntu-drivers-common is clearly picking it up correctly. [18:27] is it, if it's not installing the driver? [18:28] cyphermox: Does ubiquity have its own list of "acceptable 3rd party crap" we need to mess with? [18:28] err [18:28] cyphermox: According to his bug report (and here), the "additional drivers" control panel works fine. [18:28] cyphermox: So, ubuntu-drivers-common and the driver package are both working as advertised. [18:28] yeah, I see now [18:29] infinity: I only said I checked the box not that it installed anything additional [18:29] davmor2: Well, it must have done something, if if fixed your issue. ;) [18:29] s/if if/if it/ [18:30] infinity: ah sorry confusing it with the panel in ubiquity for 3rd party drivers, which obviously didn't work [18:30] Is that not the same application? [18:31] I would expect it to be... [18:31] ah, I think I see [18:32] infinity: not that I know to, one is part of ubiquity the other is part of gnome software and sources [18:32] ubiquity/nonfree_package [18:32] that lists only ubuntu-restricted-addons (on wily at least) [18:33] cyphermox: and on the image in pool restricted is the broadcom driver and not much else right? [18:33] I don't know, let me look [18:33] davmor2: So, this is another "did this work with .2?" question. [18:34] infinity: the wifi nope [18:34] Okay. Good. [18:34] I mean, maybe we can fix it anyway, but yay for not a regression. [18:34] cyphermox: Right, I'm seeing the same template. Now, the question is intent. Did we ever mean for this to install nonfree *drivers*? [18:34] I vaguely recall conversations around this in the past. [18:35] infinity: the text says yes [18:35] davmor2: I'm about 99% sure it never worked for nvidia or fglrx either. [18:35] davmor2: Though I can see people wanting to special-case network drivers for chicken and egg reasons. [18:35] infinity: oh no it only did wifi iirc not gfx [18:36] ${RELEASE} uses third-party software to play Flash, MP3 and other media, and [18:36] to work with some graphics and wi-fi hardware. Some of this software is [18:36] proprietary. The software is subject to license terms included with its [18:36] documentation. [18:36] that's ubiquity/text/prepare_foss_disclaimer fwiw [18:36] which should be the text that shows just before the checkbox [18:36] cyphermox: Yeah, it implies it does things it doesn't do, I agree with that. [18:37] But I'm pretty sure you have to actually use the "additional drivers" dialog to get it to do driver things. [18:37] Cause I don't think we attempt to detect and autoinstall. [18:37] infinity, cyphermox and it is what you check to install the same driver in mac hardware [18:38] davmor2: Does it do anything on Macs? That would surprise me. [18:38] infinity: this is what I asked about earlier in .2 we had to roll something back so it did [18:41] Hrm. scripts/simple-plugins is curious. [18:42] infinity: no, we're not trying to do any kind of detection [18:42] oh [18:42] cyphermox: Some of the code kind of implies we want to... [18:42] cyphermox: But I don't see anything *using* ubuntu-drivers to seed that list. [18:43] Oh, I'm misreading that. [18:44] That creates a list, it doesn't read one. [18:44] So... What reads it? :P [18:44] this looks beyond broken to me [18:45] cyphermox: It looks like we create a list of drivers that ubuntu-drivers *would* install, then we do nothing with it. [18:45] where do you see that? [18:46] nevermind [18:46] yeah, it probably should append that to nonfree_package [18:46] Well, maybe. [18:47] It would also auto install video drivers, which we've not done before. [18:47] yes [18:47] Which the text implies we might do, but I'm not sure if that's a sane change to make. [18:47] it's a pretty intrusive change, that's for sure [18:48] trying to figure out what uses that file [18:48] Anyhow, I'm pretty sure this also isn't a regression, so I'm not sure trying to hack a fix under pressure will be a win. [18:48] ubuntu-drivers postinst maybe? [18:49] Not that I see. And that wouldn't run anyway, it's already installed. [18:49] hm, true [18:50] It's possible some higher level tool like software-properties tries to read it on startup and offer you defaults? I dunno. [18:51] infinity: livecd session drivers are not transferred to the install :( [18:52] I'm not sure what's the right way to handle this [18:52] infinity: so I added FourDollars EFI/boot/bootx64.efi and now .3 is working as expected [18:52] ubiquity/target-config/31ubuntu_driver_packages:PKGLIST=/run/ubuntu-drivers.autoinstall [18:52] From ubuntu-drivers-common [18:53] cyphermox: So, it looks like ubuntu-drivers-common drops ubiquity hook that's meant to make this work. [18:54] Beyond that, I think we need to do some live system testing. [18:55] davmor2: Can you boot into a live system, start ubiquity, get past the "I hate my freedom" prompt, and then check the contents of /run/ubuntu-drivers.autoinstall ? [18:55] infinity: I can give me 5 [18:56] I guess I should look at the trusty versions of this, instead of trying to debug trusty with wily. :P [18:58] infinity: oh so it did keep my connection details from the live session it just couldn't use them because there was no driver to power the wifi, will reboot into live session now and get back to you [18:58] davmor2: Okay, that only sounds half broken at least. ;) [19:01] infinity: no such file [19:03] davmor2: Cute. [19:03] davmor2: Can you run "ubuntu-drivers autoinstall --package-list" at the command line? [19:04] Err, that needs a path argument, I guess. [19:04] infinity: it does [19:05] Gah. [19:05] Running it on my laptop actually installed things. [19:05] That was... Unexpected. [19:06] Oh, maybe that's what it's meant to do. Install in the live system, *and* write a package list for the installed system. [19:06] davmor2: So, yeah. "ubuntu-drivers autoinstall --package-list foo" [19:07] davmor2: And see if it does anything. [19:07] davmor2: As root. [19:07] davmor2: "ubuntu-drivers list" and "ubuntu-drivers devices" would also be interesting. [19:08] infinity: Okay that is weird I added foo and I got No drivers found for automatic installation [19:08] davmor2: Okay, so that's the real problem here. [19:09] davmor2: "ubuntu-drivers devices" would give you a similarly empty list, I'm guessing. [19:09] infinity: hahahahaha [19:09] Hahahaha? [19:09] infinity: ubuntu-drivers list show bcmwl-kernel-source [19:10] ... [19:10] Things made sense until you said that. [19:10] Unless autoinstall already ran earlier. [19:10] And installed that. [19:10] Is it currently installed? [19:11] infinity: ubuntu-drivers devices show the broadcom driver, mod alias vendor model and driver [19:11] davmor2: dpkg -l bcmwl-kernel-source? [19:11] * infinity downloads an image and tries to think about how to spoof this. [19:12] infinity: dpkg-query: no packages found matching bcmwl-kernel-source [19:13] davmor2: apt-cache policy bcmwl-kernel-source [19:13] davmor2: Does it exist at all, from our POV? [19:13] infinity: 6.30.233.141+bdcom-0ubuntu2 [19:14] infinity: installed None candidate ^ [19:14] WTF. [19:15] So, ubiquity isn't obviously doing anything wrong here, but ubuntu-drivers-common seems very confused. [19:15] And an annoying chicken and egg thing here where if I asked you to get me SSH on the machine, you'd have to install the wireless driver first, which would eliminate the situation wherein I can investigate the bug. :P [19:15] infinity: do you use a mac at all? [19:16] davmor2: I live and die by Thinkpads. [19:16] davmor2: Cause they actually work. [19:16] infinity: I was going to test on mac if you had one and see if that had the same issues, I assume it will [19:16] davmor2: I don't suppose you own a USB ethernet dongle? [19:16] infinity: I love my ideapad :) [19:17] infinity: I don't however I have 2 laptop and an sdcard give 5 [19:18] davmor2: Yeah, what I was hoping for was a network on the XPS13 that isn't the wireless, so I can investigate the wireless detection issue in a pristine setting. [19:18] (PS: USB ethernet dongles are dirt cheap and super useful for this kind of thing, you should ask Amazon for one for your birthday) [19:20] infinity: it's like you knew it was my birthday this Sunday [19:20] It's your birthday every day, according to online retailers. [19:21] infinity: :) [19:21] I should probably replace my ancient 10Mbit USB ethernet with something faster and more fancy. But this one has the advantage of working out of the box with every kernel since the dawn of time. [19:23] infinity: http://www.amazon.co.uk/HooToo%C2%AE-Gigabit-Ethernet-Converter-Support/dp/B00HWMP9FA/ref=sr_1_8?ie=UTF8&qid=1438629744&sr=8-8&keywords=usb+ethernet hmmmmmm [19:24] davmor2: I'd probably do a tiny bit of research on chipset and such to make sure whatever you get is supported out of the box. [19:24] davmor2: And if speed isn't a concern, I'd bet the USB2 10/100 ones are more likely to Just Work than the USB3 GigE ones. [19:24] (Also, the 10/100 ones are, like, 10 bucks...) [19:25] infinity: indeed I just like the thought of having a hub 2 2 birds 1 usb device :) [19:26] Anyhow, ISO downloaded, let's see if I can dream up a clever way to mock up your issue here without the hardware. [19:26] And I should build RC ISOs and tell other people to test, I geuss. [19:27] If we're fixing this bug, it's not going to be in the next 10 minutes. [19:29] infinity: I have a cunning plan give me 5 [19:33] davmor2: Should I be scared? [19:34] infinity: Toykeeper has a different one to me but it has the same chipset, but she doesn't have usb ethernet either :( [19:39] davmor2: I'm very confused that "list" shows it and "autoinstall" does nothing. [19:40] davmor2: On a fresh boot to live (don't run ubiquity), can you just capture the output (all as root) "ubuntu-drivers list" "ubuntu-drivers devices" and "ubuntu-drivers autoinstall" somewhere? [19:40] infinity: oh I wonder if it was because I had already run the 3rd party software page, let me reboot and see what it says before that page [19:40] hahaha [19:49] infinity: http://paste.ubuntu.com/11995460/ [19:50] davmor2: And 'apt-cache policy bcmwl-kernel-source' at that point? [19:50] And also, WTF. [19:50] Maybe autoinstall in trusty just doesn't work. [19:50] Works great in wily, I just installed intel-microcode by accident. [19:51] infinity: it did in .2 on the macs with broadcom at least [19:52] davmor2: Well, unless the package isn't available at that point, or ubuntu-drivers-common is fundamentally broken in trusty, I can't see how it would say "there's a driver available" and then "there's no driver, lolz". [19:52] But maybe I can mock this up based on seeing that output. [19:53] infinity: http://paste.ubuntu.com/11995481/ [19:55] davmor2: Bizarre. And I assume "apt-get install bcmwl-kernel-source" installs that CD version? [19:56] infinity: almost certainly want me to try it? [19:56] infinity: it's not like it is connecting to the net after all ;) [19:56] davmor2: Yeah, give it a spin. [20:00] infinity: http://paste.ubuntu.com/11995538/ [20:01] Hrmph. [20:33] Sonofa. [20:34] davmor2: ubuntu-drivers autoinstall worked just fine for my nvidia machine. Argh. [20:34] infinity: so just hates dell then [20:35] I wonder if it's cause you're missing the "recommended" string there. How does that get appended, I wonder. [20:36] Oh, no, that only exists for nvidia/fglrx to handle multiple versions. [20:36] infinity: ha I have a cunning plan https://www.youtube.com/watch?v=pKRxX3s3JlM [20:37] davmor2: I was assuming that would be a video of me slamming my head into a wall. [20:38] davmor2: I do miss when Atkinson was known for biting sarcasm instead of silly physical comedy, though. Mr Bean completely ruined him. :/ [20:40] davmor2: Balls. Found the bug. [20:41] davmor2: Now to understand why the bug was introduced... [20:57] infinity: see pm