=== Fyodorovna is now known as Guest98921 === sporkeee is now known as wilee-nilee [01:22] Clicking the install on the live session and nothing happens, it's a daily and on of hundreds of installs I have done, not sure whats up. [01:22] one* [01:25] wilee-nilee: did you checksum your iso? [01:26] Stanley00, Its a daily, no I have run it daily, and just did. [01:27] theoretically it would fix itself, lol [01:27] !md5sum [01:27] To verify your Ubuntu ISO image (or other files for which an MD5 checksum is provided), see http://help.ubuntu.com/community/HowToMD5SUM or http://www.linuxquestions.org/linux/answers/LQ_ISO/Checking_the_md5sum_in_Windows [01:30] Ì then, may be you should check your live when booting up, may be there's some error while *burning* [01:33] the sum is correct and it is a usb, I'm a very experienced user. [01:34] I am trying another usb loader the onboard thanks. [01:35] OK, good luck then. :) [02:03] hi, is the keyboard layout options totally gone in 13.10? i am holding shift key (capslock is turned on, and does not seem willing to turn off) [02:26] actually, nevermind. setxkbmap'ed for now... === mibofra- is now known as mibofra === brot_ is now known as brot [09:22] I installed cinnamon on ubuntu 13.10 but it's not showing any panels, only the desktop icon! [10:10] Hello. I have an issue with my current installation of 13.10. While in every previous version, the laptop would suspend when the battery was low, it now shuts down. Furthermore, there is no "suspend" option in power settings, when battery is critically low. [12:18] 'Morning folks [12:18] Hey BK [12:38] hi penguin42 [13:01] Hi again. I'm still trying to fix my problems ... I still have a problem with grub: at boot time I don't have any menu. If I indicate the configfile which is correctly located at (hd0,4)/boot/grub/grub.cfg the system boot. So my question is how could grub not be able to find this file while it is located at the default location and grub can run it if I indicate it ... [13:05] kerloi: What happened between it working and between it failing? [13:05] penguin42, nothing it's a new laptop and a freh install [13:05] hmm [13:05] and this is +1 ? [13:06] I can't install 13.04 my hardware is not suported so yes [13:06] ok [13:07] kerloi: So if you say the grub.cfg file is ok and you can tell it about it, then my guess is grub is looking at the wrong partition/disk for it - do you have only one disk? [13:07] Yes I do. One disque and five partitions [13:07] why 5? [13:07] efi/bios/swap/system/home [13:08] ooh efi erm I don't know much about modern efi magic [13:08] I made these partitions when I was trying to install 13.04 [13:08] You suppose there is something special to do for efi system [13:09] there is I think but I think it should happen automatically [13:09] it's a GPT partitioned disk I assume? [13:09] With the beta2 live CD, amd64 desktop, running a terminal from the "Try It" option shows terrible terminal window background and font corruption, as to be unreadable, and typing doesn't appear until something else causing a window redraw. Haven't been able to find a bug describing this. Anyone else seen/heard of it? [13:10] TJ-: That sounds graphics driver specific? [13:10] kerloi: So you tried a 13.04 install, gave up and then did 13.10? [13:10] penguin42, yes [13:10] It doesn't affect any other windows, and I recall seeing this several releases ago too [13:10] And yes it is a gpt disque [13:11] TJ-: I've seen this happen in a VM before [13:11] penguin42: Yes, that's where its happening, under KVM [13:11] TJ-: Ah right, hang on let me find my 2-3 year old bug report [13:11] Trying to use it to bootstrap a highly encrypted install [13:12] penguin42: Yeah, I've been searching for the original bug and workaround but can't find it now I want it [13:12] TJ-: Changing to not using cirrus sometimes helps [13:13] TJ-: Bug 1080674 [13:13] bug 1080674 in cairo "[QEMU] Corrupted desktop screen for raring desktop installation in QEMU guest (Cirrus graphics). Affects KVM but not VBox." [Medium,Confirmed] https://launchpad.net/bugs/1080674 [13:13] hmm not even a year old :-) [13:13] It's not too vital, but I was using the terminal to create a VG inside the LUKS encrypted partition that ubiquity has created, since Ubiquity doesn't allow you to create an LVM inside a crypted volume! [13:14] That's not the bug I'm thinking of. This bug only affects gnome-terminal [13:14] TJ-: Not using cirrus graphics helps, e.g. try using spice/qxl [13:14] kerloi: have you tried holding the left shift down after the bios/efi screen? Sometimes grub installs with no timeout the and menu won't appear [13:16] Ahh this one is closer to it bug #850649 [13:16] bug 850649 in gnome-terminal (Ubuntu) "gnome-terminal extremely slow and display buggy" [Low,Confirmed] https://launchpad.net/bugs/850649 [13:17] BluesKaj, Well I just did it and nothing happened. When I use the configfile command, it doesn't display a menu it starts automatically. I suppose there's a 0 timeout [13:17] TJ-: I'd still bet trying spice/qxl is the best bet [13:17] TJ-: It could still be the 1080674 given how much unity has changed since that bug was originally reported [13:18] kerloi: I'd kind of guess that you have part of a grub from 13.04 still left behind and that's why it's confused? [13:19] penguin42, Is there any way to remove completely grub and then re-run a grub-install ? [13:20] penguin42: I'm inclined to think not otherwise other windows would display the issue. As I recall, the cause is any window using some OpenGL constructs (which apparently gnome-terminal does to provide things like transparency, even what those effects are off). [13:22] kerloi: you can't hesitate to hit and hold the shift key down after the bios screen , otherwise grub will disappear too quickly [13:22] TJ-: OK, fair enough, at the time I had exactly the opposite, the background being broken but the terminal being fine [13:24] BluesKaj, I don't understand. I have the grub command prompt right after the bios screen. I tried to hit and hold the left shift down but I still have the grub command prompt and not the menu [13:24] TJ-: I'm just updating my Saucy vm [13:25] kerloi: https://help.ubuntu.com/community/Boot-Repair [13:25] kerloi: then grub isn't installed [13:26] BluesKaj, ^^ well it's a bit complicated for boot repair, the live cd doesn't support my hardware so I can't use it and one of the bug I'm facing is that I don't have an ethernet interface ... [13:27] I tried to find and download a deb package on the internet yesterday but I wasn't able to find one [13:28] Maybe I will simply re-format the entire system ... [13:28] kerloi: Probably easiest! I'd try grub-install and maybe check the grub device map [13:30] I can't access the internet on the laptop so I can't use apt-get. To install it I have to download a package from the internet on the windows PC I'm using right now and then to use an usb drive to move it to the laptop [13:30] Hopping there's no dependencies [13:30] kerloi: Ah yeh that's always a pain [13:30] kerloi: what ethernet has it got? [13:31] kerloi: if you have a uefi system http://askubuntu.com/questions/208405/how-to-efi-install-ubuntu [13:31] kerloi: If you couldn't install with the live CD, how did you get this far? [13:31] It is a BCM57786 chip which seems not to be listed in the driver tg3.c [13:32] !uefi [13:32] UEFI is a specification that defines a software interface between an operating system and platform firmware, it is meant as a replacement for the BIOS. For information on how to set up and install Ubuntu and its derivatives on UEFI machines please read https://help.ubuntu.com/community/UEFI [13:32] kerloi: Your issue was already solved. [13:32] kerloi: Contact the module author to include your PCI ID into the module. [13:32] I can install kubuntu from the live usb but then neither grub nor ethernet is working [13:32] kerloi: hmm with some drivers you can dynamically add a new PCI id [13:33] bekks, Yes its done it was just to tell penguin42 [13:34] penguin42: tg3 doesn't allow that, unfortunately [13:35] kerloi: by no HW support I thought you meant you couldn't install ubuntu at all :P [13:35] TJ-: Pity, it would be too easy :-( [13:35] BluesKaj, for 13.04 yes. I hav mouse and graphical problems ... [13:36] so I wasn't able to install it [13:36] kerloi: What machine is this so we can advise other unfortunates who ask [13:36] installing ubuntu on a mac laptop kerloi ? [13:38] It's an Acer Aspire E1-572G-54208G1TMnkk [13:38] BluesKaj, nop [13:38] ok , that's an odd ethernet nic for an acer [13:39] probably just the latest one [13:40] * penguin42 notes with a model number that long they could probably have enough models to survive the age of the universe [13:40] So I'm buning the latest daily image and I will try to re-format and install everything [13:41] kerloi: There are also some upstream daily kernel packages available that will be newer than the ones on the daily image (they'll be bleeding edge from Linux's git) [13:41] TJ-: An up to date saucy on QXL/Spice graphics seems to work fine [13:42] I still have a question. I had many troubles with that efi system. I read somewhere that I needed a bios partition so I have one but I'm not convinced. What do you think ? [13:42] kerloi: I think EFI does need one, but it's pretty complicated [13:42] kerloi: read ubottu post about efi above [13:43] TJ-: And I can confirm with saucy it's still corrupt [13:44] TJ-: Sorry, can confirm with Cirrus on saucy it's still corrupt in terminal [13:44] kerloi: If you have a UEFI system, the GPT only needs an EFI 0xEF00 boot partition (I reserve 500MB for that) [13:44] * penguin42 is still sure he's mentioned that corruption somewhere but I can't remember which bug [13:45] penguin42: I'll live with it... I'm just using it to bootstrap an encrypted GRUB + encrypted VG installation onto USB flash, and then going to 'dd' it onto the destination laptop's hard drive [13:46] TJ-: Nod, although it's easy enough to flick it to QXL/Spice [13:46] TJ-: My assumption is it's a 24/32 bit issue because the cirrus runs in 24bpp and that's pretty rare these days [13:46] Yeah, that would make sense, and only affects OpenGL apps too [13:53] Grrr! Ubiquity! "You have selected the /boot file system to be stored on an encrypted partition. This is not possible...choose a non-encrypted partition..." [13:54] There was a time when F/OSS operated on the basis that the user knows best, despite what the software might think, and did what it was ordered! [13:54] Well it's weird the partitions sda1 happened to be mounted during the installation time. It failed a first time but now it seems to succed the partition part. It's still installing the system [13:55] and I saw that the bios partition was called biosgrub so maybe it was the problem ... [13:56] TJ-: Are you trying to do LUKS with Grub? [13:56] kerloi: Sounds like it; "biosgrub" is a 0xEF02 GPT [13:56] penguin42: Yes [13:56] TJ-: Interesting, not seen that done yet, I've been curious about trying it [13:56] penguin42: Or rather, I *am* doing LUKS GRUB, but Ubiquity doesn't understand it [13:56] It's easy, if you do it manually! [13:57] There's a bug in the grub2 scripts thought where they've used the wrong variable name, but once you fix that up, and do an grub-install && update-grub, it works nicely [13:58] bug ##1232237 [13:58] TJ-: Nice, can I ask you why you're doing it with LUKS GRUB? I think it's probably the right thing to do (stops your kernel/initrd being attacked) but as long as grub itself isn't encrypted does it really help? [13:58] Grrr [13:58] bug #1232237 [13:58] bug 1232237 in grub2 (Ubuntu) "GRUB_CRYPTODISK_ENABLE should be GRUB_ENABLE_CRYPTODISK" [Medium,Confirmed] https://launchpad.net/bugs/1232237 [13:58] well no it is exactly as before. Everything went well during the install but I have the grub command line. I can start ubuntu with configfile command. The grub.cfg file is correct and at the right location ... [13:59] penguin42: Yes, it helps. Grub can't get to its menu unless the LUKS key is provided for /boot/ [13:59] And I have separate /boot/ and root-fs (root-fs and others as LVs in a VG [13:59] And that LUKS-encrypted VG uses a key-file rather than pass-phrase [14:00] TJ-: Now wire it through into secure boot to load the grub.... [14:00] penguin42: Precisely [14:00] Patches for secure-boot are in grub, or entering, [14:00] TJ-: and hope it never goes wrong because I wouldn't like to debug it :-) [14:01] I keep mirrors of everything, plus the LUKS headers backed-up in multiple locations [14:01] TJ-: When a new kernel is added during a normal update and the grub.cfg gets updated does it have to be rehashed/signed? [14:02] TJ-: Any reason you haven't set that bug to triaged? [14:02] I believe so; the patches are still coming for GRUB so it'll be a while until its clear how the entire thing can be scripted correctly [14:02] why all the attn to encryption , even grub , TJ- ? [14:02] penguin42: I was hoping another knowledgeable dev will do that, seeing as I both reported and confirmed it. [14:03] BluesKaj: Defence in depth [14:03] If I run "sudo grub-install /dev/sda" it works (no error reported) but they mention a 0 timeout. [14:03] TJ-: I'm happy to set it, I mean I'm not that knowledgeable about it, but it's an obvious error and you've pointed to the upstream fix which is normally enough for me [14:04] TJ-: must have some valuable data there :) [14:04] penguin42: I've done it ... I'm hoping to provide a patch so we can get it into 13.10 [14:05] TJ-: One trick the windows disk crypt packages do is to hide the partition they keep their bootloader in so it looks like there is no clear-text data - of course that's completely bogus [14:05] But got hard disks dying all around me this week and having to repair/rebuild systems I didn't expect to! Found the caddy server dead this morning, its boot disk was clicking and ticking repeatedly, and I'm working on repairing one of the laptop's drives after a 500GB hybrid Flash drive failed [14:10] penguin42: Yeah, well there's no clear text with the LUKS GRUB... aside from the GRUB boot-strap code in sector 0. [14:10] The entire device is just random because I do "DEV=/dev/sdX; cryptsetup luksFormat $DEV && cryptsetup luksOpen $DEV randev && dd if=/dev/zero of=/dev/mapper/randev bs=50M && cryptsetup luksClose randev && dd if=/dev/urandom of=$DEV bs=512 count=$(cryptsetup luksDump $DEV | awk '$0~/Payload/{print $3}')" - the quick way to randomise a large device [14:11] TJ-: surely it can't be fitting the whole luks key request/response,decrypt into sector 0? [14:13] penguin42: No, grub has several stages. stage 1 is the boot-strap, stage 1.5 is the essential modules for file-system and encryption (on MBR that's stored in the slack space after sector 0 in cylinder 1, on GPT, its in the dedicated 0xEF02 partition), stage 2 is the file-system in the /boot/ partition in /grub/ [14:13] TJ-: But you said ' Yeah, well there's no clear text with the LUKS GRUB... aside from the GRUB boot-strap code in sector 0' - so how is the stage1.5 crypted? [14:15] penguin42: It's not, but it doesn't compromise the grub installation, in terms of giving away any clues as to what is installed [14:15] The user-configured parts (the sensitive info) and kernels are all inside the encrypted /boot/ [14:16] TJ-: Right but you could do an evil-maid on it - so it doesn't get you that much futher than leaving the kernels open [14:16] Not when GRUB has fully implemented SecureBoot [14:17] right yes so it'll measure the 1.5 ? [14:18] each loaded module is tested before being loaded into memory [14:18] nod [14:18] TJ-: So what about the grub.cfg itself? [14:19] inside encryptee /boot/ [14:19] that's the point of encrypting GRUB [14:19] TJ-: Huh ok, so what tells grub to load the luks stuff and ask for the prompt if the grub.cfg is encrypted? [14:19] without a LUKS pass-phrase /boot/ doesn't get unlocked, and the thing is just a brick [14:21] penguin42: The modules installed for stage 1.5 ... it doesn't matter what they are, they're written into a single block of sectors by the installer. boot-strap loads the block of sectors and then executes the entry hooks. If SecureBoot is in operation, signatures are checked first [14:22] TJ-: And other loading the luks module it doesn't need any other config to tell it what to ask for/decrypt? [14:22] correct [14:22] ok, that's a nice way of doing it [14:22] I've seen someone on another support site saying he is running 12.04.4 and claiming some "bug" with apparmor not running if an included local/file is not present is causing 12.04.4 development to remove the use of apparmor.d/local includes. I believe he's experiencing some other kind of apparmor problem. Does this sound as untrue to anyone else as it does to me? [14:23] And, once unlocked, it is possible to run check-code to ensure the checksum of stage 1.5 is valid... and refuse to continue to an operating-system load if not, which is useful even when SecureBoot isn't set up [14:23] TJ-: and the block1.5 stuff is all in one 'file' in the EFI part so it knows to get measured? [14:23] For EFI that is in a real vfat file-system with type 0xEF00 GPT partition [14:23] yep [14:23] the UEFI specification madates VFAT as the file-system for that file-system [14:24] I don't envy the headaches the GRUB devs had to do in order to figure out how to support SecureBoot, with having to shoe-horn in the code to check signatures [14:25] I think that's why they need UEFI and GPT to do it, since the EFI partition can be hundreds of megabytes rather than a few sectors [14:26] penguin42, was that yep for me? [14:28] Guest____: Nope, it was a TJ- yep [14:28] ok thx :) [14:28] TJ-: Oh yeh and this stuff is both complex and subtle, any subtle screwup and someone will find a way to drive right over it [14:30] Guest____: Can that person provide a bug reference, if not, its best to treat it as hearsay [14:31] He's claiming this bug, https://bugs.launchpad.net/apparmor/+bug/1206742. [14:31] Ubuntu bug 1206742 in AppArmor "allow non-existing profiles in /etc/apparmor.d/local" [Undecided,New] [14:33] I think he's got a borked install or an apparmor/kernel issue, or a cache issue. I'm trying to put together evidence and suggestions for troubleshooting, but I thought I better ask on a dev list before making any claims. [14:33] Guest____: The bug makes sense, although I would have thought it only affects the specific profile apparmor is trying to load, not the entire apparmor defences [14:35] I wouldn't think it would cause Ubuntu to drop the local directory and includes. Seems extreme. Not sure where he is getting that, or if just conjecture. [14:35] thanks for your time! [14:35] Guest____: It sounds to me as if he should be using the apparmor_parser's -I option to add a path to the list of paths searched by the #include directive [14:35] i'll check it out [14:49] Any ideas how to alter an ubiquity installer script on the /rofs/ mount, seeing as its read-only. Got to find a way to prevent a script fragment from doing "exit 1" [14:49] TJ-: What's it mounted off - i.e. could you just remount,rw ? [14:50] TJ-: Or mount a ramdisk over it? [14:50] Hi everybody, I've installed Kubuntu 13.10 beta2 amd64 and I want to edit mp3 files via Kwave so I have also installed libmad0,lame and libid3 in order to have Kwave import of mp3 working but it doesn't, should I file a bug against Kwave? [14:50] TIA [14:50] It's the ISO read-only file system, I was wondering about a unionfs with a ramdisk . Basically, I need to force ubiquity to NOT refuse to install when there's an encrypted /boot/ [14:51] TJ-: Is it easier to rebuild the iso? [14:51] frank67: Most of the KDE stuff gets filed in KDEs own bug tracker, but the question - which is difficult to answer - is is it Kwaves fault or one of the other components [14:52] tnx! [14:59] penguin42: "sudo su" then "mkdir /tmp/rofs-rw /rofs-ro && umount /rofs && mount -t squashfs -o ro,noatime /dev/loop0 /rofs-ro && mount -t overlayfs -o lowerdir=/rofs-ro,upperdir=/tmp/rofs-rw overlayfs /rofs" [15:00] hmm, I'd never played with overlayfs - looks a solution to many evil problems :-) [15:00] sudo su is a bad idea. [15:00] TJ-: better use sudo -i [15:00] better use su, thanks [15:01] * penguin42 notes this argument normally gets no further than vi vs emacs [15:01] !sudo | TJ- [15:01] TJ-: sudo is a command to run command-line programs with superuser privileges ("root") (also see !cli). Look at https://help.ubuntu.com/community/RootSudo for more information. For graphical applications see !gksu (GNOME, Xfce), or !kdesudo (KDE). If you're unable to execute commands with sudo see: http://www.psychocats.net/ubuntu/fixsudo [15:03] * TJ- shakes head sadly [15:05] bekks: It's a bad idea to tell someone its a bad idea to do something unless a) you understand the full context, and b) you don't explain why [15:06] TJ-: seems like you didnt a) and argueing about b) without knowing a) [15:07] bekks: Get off TJ-'s back [15:07] bekks: I understand it perfectly, and chose it for a reason [15:08] penguin42: I seem to be fighting ubiquity's partman now... each time I hack the scripts to get me further, it decides not to 'see' one of the partitions I've created, which makes me have to go back two steps in the installer, manually close one of the LUKS partitions it created, and go through it all again.. I'll be doing this in my sleep tonight :D [15:10] TJ-: I'm surprised, it's normally fairly good where you've got the partitions you want to use [15:11] TJ-: How long before you give up and attack it with debootstrap [15:12] It's the LUKS stuff it fails to remember. It creates them, but if you step back then forward it doesn't create formatable sub-nodes for each of the /dev/mapper/sdX_crypted nodes it lists, as it does when they are originally created. [15:13] penguin42: LOL This started out as my short-cut from having to use debootstrap - that was how I developed the LUKS+GRUB+LVM+Ubuntu method. [15:14] TJ-: debootstrap followed by 3 hours of hacking solves all known problems [15:15] On the actual hardware I also hit a weird issue whereby it'd boot fine into multiuser, but when doing recovery it'd hang after initrd... no clues, nothing! [15:15] penguin42: Gawd, tell me about it. So many configs done by script nowadays in the regular installer I'm always convinced I've missed something important [15:16] TJ-: I think there are some flags for passing to initrd/upstart to follow things [15:17] It doesn't get that far... I see local-bottom then ... nothing... seems like the "exec init" is hanging before it does much. Anyhow, I left it, and thought I'd do this method of install to check it [15:18] nod [15:18] OK, I have / /home/ /boot/ /var/ and swap configured. If my partman-crypto script edit is OK this *should* now warn me about the encrypted /boot/ but not stop! [15:19] Darn! Somehow its using a cached version of the file! [15:23] penguin42: Slap me with that Trout LOL! After all that overlayfs for /rofs/ it turns out the script being run wasn't the one in /rofs/lib/partman/ but the one in /lib/partman/ ! [15:24] TJ-: Ah yeh it's hairy with the same scripts in the ro and the live bit [15:24] I assumed the installer stuff would only be in one place seeing as it isn't executed by the live env [15:25] that's be too easy [15:26] The 13.10 install has begun... time to make a nice big mug of tea [15:34] TJ-: In the VM are you seeing problems with the guest not shutting down/rebooting? [15:36] penguin42: I've not got that far yet :D [15:37] oh no! it's done the same thing as on the hardware. After entering the username it's just sat there not doing anything else! buttons greyed, but form interactive, progress cursor on display, no activity! [15:37] ctrl-alt-f1 ? [15:37] surely the beta2 amd64 live ISO isn't that badly broke [15:38] That'd hit the host, I'm not using libvirt/virsh, just a simple cmd-line "kvm ..." to fire it up [15:38] TJ-: Well a fresh dist-upgraded saucy is ok for me, but that was installed quite a while back [15:38] oh via SDL? [15:41] ahh hang on, no CPU activity but the host shows that kvm is doing about 10MB/s of disk I/O to the USB flash devices... I'll just drink my mugga and not interfere! [15:41] you're running a KVM off a USB flash device? [15:42] No, the flash devices (one with the ISO live image, and the other the installation target) [15:42] KVM running on the host's disks [15:43] right, but you're using a flash disc as the backing for the VM - that's horrible :-) [15:43] I have deduced why the installer seems to be stuck... it isn't cycling though its 'welcome' cards as the installation is running, its left the user info dialog up and interactive, which is very confusing [15:43] Why horrible? no different to an internal SSD [15:44] TJ-: oh they're USB SSDs or normal USB thumb drives? [15:45] I would have been using my 'caddy' server for this, which has lots of removable caddies but, as I said earlier, I came in this morning to find its boot disk was tick-clicking and had failed. Will fix that in the week. For now, I have to have this install on the target laptop by tonight. [15:45] USB flash storage... fast :) [15:46] OK, and the installer is interacting again now. I think the encryption took it some time, since I think it's using /dev/urandom to randomise the disk surface (again) [15:47] you can get a lack of randomness in VMs [15:47] indeed, maybe entropy was hard to get for a while there [15:48] That was why I had randomised them myself outside the VM first [15:49] 13.10 server is so much easier to work with. Had that deployed for a while with Apache 2.4 to provide TLS forward secrecy [15:50] TJ-: Why didn't you just do a 13.10 server install and then throw on unity-desktop? [15:51] (or is it ubuntu-desktop I forget) [15:51] I wish I had, like I said, for the trials I just used debootstrap... but I wan't to test out the live ISO this time around now I know precisely what I need to tweak for LUKS encrypted GRUB [15:52] TJ-: I always start with the server one when ever I'm doing something weird [15:52] with debootstrap I usually have a standard ubuntu-minimal LV, and then clone it as needed to add stuff. [15:52] TJ-: Be careful about things like host key duplication [15:53] Always am. I have a script that replaces UUIDs on file-systems and distributes keys as required [15:53] Also resets the /etc/udev/rules.d/*persistent* entries [15:54] nice - my hosting provider recently fell fowl of it [15:54] Ouch! [15:55] My deboostrap install is fully scripted. It takes me a little longer but saves so much time later when I need to repeat. Had to add the encrypted GRUB handling to it last week. [15:56] I had the dremmel out earlier inscribing each of the USB keys (they're all metal) so we don't get them mixed up later. So difficult keeping it steady to write small. Wrote some very wonky @ symbols [15:57] TJ-: Oh you're trying to do this secure boot as secure rescue/work anywhere thumbs? [15:59] Yeah. All the laptops are identical too, and have inbuilt GSM/GPS mini-PCIe cards for remote management and tracking [16:00] ok, yeh so if they are indeed identical I guess you've got a chance of doing the secureboot [16:00] The USB keys are for portable boots, but the same config is installed on the laptop disks too. We have USB key-files for unlocking the LVM VG too [16:01] TJ-: Your life will get more interesting when in 6months time you try to get a replacment/new one and they're slightly different [16:01] There's no /boot/ installed on the laptops though - that's on the USB keys. It's complicated to organise but once it is, it works well and protects the systems to the maximum possible [16:03] I won't. I am constantly buying identical second-user laptops and reconfiguring them to be identical. We can hot-swap, and cannibalise for parts, really easily. CPUs can be upgraded, there's 3 mini-PCIe slots, and plenty of other expansion, so they're very flexible. [16:03] what laptops with 3 mini-PCIe slots? [16:04] Dell XPS === rymate1234 is now known as yoloking [16:04] TJ-: You are taking it to pretty extreme levels [16:04] We've got WWAN/GPS, WLAN and either SSD or CrystalHD expansion (depending on usage) === yoloking is now known as rymate1234 [16:04] penguin42: It's great fun though! [16:05] TJ-: It's rare that you can say that when crypto is involved [16:05] They're easily and quickly disassembled, makes servicing a dream [16:05] Yeah... we don't need the latest just something thats easy to work with, flexible, and consistent [16:06] I can swap the LCD panels out in about 15 minutes, they've all got 1920x1200 now (some low-end models come with 1280x800 ) [16:06] Basically, its a hackers paradise :) [16:07] I wanted to transfer the Pixel Qi screen out of my Notion Ink Adam into one but the screen isn't large enough, unfortunately [16:07] TJ-: I think the Dell stuff is made to be easily customised because they have such flexibility when you order [16:08] That's it. After having had Sony Vaios before these things are a breath of fresh air. [16:09] pixel qi only do upto 10.1" [16:09] TJ-: http://www.panelook.com/ [16:10] That's right [16:10] I love it though, using it outside in the sunshine is a dream [16:12] hi! I'm encountering some issues with 13.10 on my Thinkpad T420 [16:13] describe the problems please [16:13] There are two critical issues: cannot sleep (logs: http://pastebin.mozilla.org/3167099 ) and graphics performance drops significantly after playing games for like 5 minutes or so. no logs [16:14] there may be an related issue, which is related to graphics lock up and i found an upstream bug report => https://bugzilla.kernel.org/show_bug.cgi?id=53571 [16:14] bugzilla.kernel.org bug 53571 in Video(DRI - Intel) "[fdo:54226] stale bcs->rcs semaphore" [Normal,New] [16:15] that workaround does not work for the graphics framerate drop. [16:15] gpu is both intel hd 3000 or the nvidia quadro 4200. Happens on both . [16:15] no overheating detected (temperature hovered around 80C, CPU is underclocked to 1.6Ghz rather than my 2.3Ghz + turbo boost) [16:16] idk what else to say about it. i can reproduce that issue pretty easily.. so if you need any help debugging it (my guess would be another upstream issue..) [16:17] Pwnna: When you say it happens on both, are you running with the Nvidia closed drivers/optimus or with bumblebee, or are you switching between discrete/integrated? [16:19] penguin42: bumblebee. [16:19] idk if my computer will boot with a bios switch. actually [16:19] i could test that too [16:20] Pwnna: I run my works w520 in just nvidia mode [16:20] Pwnna: Using the bios setting [16:20] yeah usually i don't have the need for the nvida card on the T420 [16:21] Pwnna: I'd try and eliminate bumblebee from the problem since it's whole job in life is to switch between graphics cards it's not impossible it's related to your problem [16:21] penguin42: yeah. [16:21] penguin42: that's my guess too. it's not just that. i noticed slow down as i type things into sublimetext after the slowdown [16:21] but i guess ST2 uses GPU rendering? I think? [16:22] ST2? [16:22] sublime text 2 [16:22] ? [16:22] it's a text editor [16:22] * penguin42 doesn't know - I'm a vi penguin [16:22] lol [16:23] i can't use vim. but i need vim emulation [16:23] it's weird [16:23] lol [16:23] well i can use vim. just not on anything with over like 20 files [16:23] yeah, and then there's the sleep issue, which actually has logs! http://pastebin.mozilla.org/3167099 [16:23] Pwnna: The other thing I'd do is make sure you're on a recent bios, I've seen a bunch of w520s with ancient bioses with really really random problems [16:24] i think my bios is a 2011 bios [16:24] how do you even upgrade that under linux? i thought lenovo didn't do that [16:25] I think they have a DOS floppy image thing [16:25] hm [16:26] Pwnna: My reading of that sleep log is that it's stuck trying to load some firmware while it's trying to sleep which seems odd, if you can figure out what firmware it's trying to load and then add the module to the list of modules to unload prior to sleep it might do it [16:27] hm.. but how would i go about figure that out? i'm not too famliar with this [16:28] Pwnna: You only pasted part of the log there - sometimes a full one says something like 'last sysfs file written' it might give a clue, but other than that yes it's guess work; I'd get it to take the wifi/network drivers out [16:29] penguin42: did i? I just copied a bunch of stuff from kern.log [16:29] Pwnna: Well I'm wondering is there anything before that? [16:29] nope [16:29] hmph [16:29] wait [16:29] maybe [16:30] let me check syslog [16:31] http://pastebin.mozilla.org/3167161 [16:31] nothing really that interesting [16:32] grepping for sysfs only gets me Sep 29 02:19:56 thinkpadt420 kernel: [51747.891718] [] sysfs_write_file+0xc6/0x140 [16:32] that kinda stuff [16:32] Pwnna: https://wiki.ubuntu.com/DebuggingKernelSuspend [16:33] Pwnna: Yeh I'm not too sure; tbh I think that's the 'please wake up' kick but might not say which device [16:33] Pwnna: So that debug procedure might help - never tried it [16:37] and i'll be surprised if this is not already a known issue [16:38] Pwnna: Yeh, might be an easy work around though [16:38] Pwnna: The tricky thing is so many things can cause a sleep/restart to fail, figuring out that lots of people are hit by the same one is quite hard [16:38] Pwnna: The important bit in your log is "INFO: task kworker/u16:0:17142 blocked for more than 120 seconds." . [16:40] TJ-: I disagree [16:41] TJ-: I think the earlier 'rcu_sched self-detected stall on CPU' is the first bit and everything else is fall out [16:41] penguin42: It looks to me as if fw_pm_notify(), which causes required firmware files to be cached, is causing reads of an EXT4 file-system, which is getting stuck [16:42] TJ-: Yeh I agree, I'm assuming that the disc/controller that the EXT4 is on is still asleep [16:45] outside my area of expertise :) [16:45] but if you guys need me to get more info, if there are some instructions somewhere, i can reproduce and get the info for you [16:45] penguin42: and yeah. sleep issues are weird. [16:46] Pwnna: tbh I'm not going to be helping fix this [16:46] lol [16:46] kay [16:46] i assume it will be fine by the time 13.10 releases [16:46] Pwnna: But I'm happy to help you see if you can fix it chatting on here [16:46] Pwnna: Big assumption [16:46] lol [16:46] the thinkpad t420 is quite common [16:47] i recently discovered, or at least i think, that if there's an inaccessible NFS share mounted it won't go suspend [16:47] Pwnna: I've got a T420 and T430s sat right next to me, what's the issue ? [16:48] ikonia: Sleep triggers this: http://pastebin.mozilla.org/3167099 [16:48] as in hibernation/suspect ? [16:48] ikonia: Although pwnna does seem to have an entire nest of other bugs [16:48] ikonia: I think to ram? [16:49] seems odd as both my t420 and 430's have zero bugs/issues, infact they work superb [16:49] there are slight variations on spec's in the T420 class, so it would be worth zero'ing down on those [16:49] (get the unit ID class number) [16:50] ikonia: Nod, he's running with bumblebee as well; but yeh I've found there is a lot of variation as well with bios versions on the w520's don't know if the same is true on the 420 [16:50] bumblebee ?? I'm not aware of an optimus enabled 420 [16:50] anyway, take it up with Pwnna! [16:50] hi, how do i upgrade 13.04 to 13.10 beta? I followed https://help.ubuntu.com/community/SaucyUpgrades but don't see 13.10 updates [16:52] ikonia: Yeh all the 420s I've seen are Intel only, but http://support.lenovo.com/en_US/downloads/detail.page?DocID=PD015734 suggests Nvidias are on some [16:53] penguin42: yes, but th nividia's I've seen are not optimus cards, [16:53] ah that I don't know; I've heard of people running it on the 520s but never tried myself [16:53] and the 520s bios certainly offers the choice [16:55] there maybe an optiumus model, I've just never seen it/heard of it, [16:55] can anyone tell me how do i fix my xserver on ubuntu 13.10 beta [16:55] I've got access to almost any think pad, so I'm surprised I've not seen one [16:56] ikonia: That trace does have the mtm/model [16:57] I didnt notice [16:58] ikonia: http://appserver.lenovo.com.cn/lenovo_product_Detail.aspx?gdsid=A1900014604 is it - says Optimus [16:58] Sep 29 00:49:56 thinkpadt420 kernel: [46343.464659] Hardware name: LENOVO 4180J4C/4180J4C, BIOS 83ET57WW (1.27 ) 05/17/2011 [17:02] If you look further down the stack-trace of task "kworker/u16:0:17142" you'll see " _request_firmware" The timing is "00:52:08" and if that's after 120 seconds stall, that puts the original worker running at "00:50:08". The stall on CPU0 occurs at "00:49:56" firmware_data_read [17:03] but what device needs firmware ? [17:04] Whichever devices originally loaded firmware... I don't think that is important... the problem seems to be in accessing some EXT4 file-system... the firmware wanting access is incidental. It is possible some PM script has unloaded a required driver at that point [17:33] TJ-: Well it seems from my reading to be happening during wake up, so what is supposed to ensure that things get woken in the right order? [17:33] I guess the other possibility is something has apparently been woken but didn't work [17:34] ikonia: I pointed him at https://wiki.ubuntu.com/DebuggingKernelSuspend which does have some hints on trying to find that out [17:35] No, its happening during suspend [17:36] TJ-: What makes you say that? [17:37] The PM suspend scripts run in userspace, and then "PM: Preparing system for mem sleep" leads to any firmware files being used being loaded into RAM cache so that, at resume, there's no possible hangs due to missing firmware required before userspace unfreezes [17:37] ah ok, and that backtrace does have pm_suspend in it but I don't see the wake half [17:40] On CPU0 you can see in the call-stack "fw_pm_notify()" - that function is responsible for kicking-off the loading of the firmware files into cache before sleep [17:41] hmm - so if it's not got into sleep yet then why is it blocking? [17:41] I suspect that firmware stuff eventually causes a new worker task to be queued which is executed on another CPU to read the file-system containing the firmware file itself... and that looks to be what is hanging overall [17:43] but that should all be safe shouldn't it - that's just causing a file to be loaded - nothing too odd at that point? [17:44] Unless the userspace scripts unloaded some module that is involved in the block device underlying that file-system [17:45] Or, it might be a new bug caused by something else... there's some sign of the NMI showing traces in IRQ handlers for timers that could, I suppose, be stuck [17:46] That log isn't complete enough unfortunately - we need to see the complete log from when the system booted last (/var/log/kern.log) [17:48] At that point this is what we ought to see: http://paste.ubuntu.com/6172176/ [17:51] hello [17:53]