=== lfaraone_ is now known as lfaraone [09:13] * apw yawns === gfrog_afk is now known as gfrog [09:31] apw: cking: do you by any chance know how to setup Qemu + NVMe + OVMF vm for testing? [09:32] xnox, no idea about emulating NVMe in Qemu [09:32] somebody in 2013-02-25 was installing testing it like that. [09:32] https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1132681 [09:32] Launchpad bug 1132681 in ubiquity (Ubuntu) "U 12.10 LiveCD installer crashing during boot loader configuration on NVMe storage( QEMU + NVMe +OVMF )" [Undecided,New] [09:33] cking: otherwise i can propose some patches for you to try. [09:33] xnox, that's a better plan, I can give them a spin on the H/W I have [09:34] cking: cool. [09:45] xnox, not ever done that here eitehr [09:46] rbasak: what do you say on upgrading libvirt to 1.2 release? [09:51] last time I tried QEMU was with OVMF I built from scratch from the tianocore sources [09:52] if you read the description of this Qemu bug, it blows chunks in grub. didn't you have a similar issue cking? [09:53] apw, yup, bug 1265483 [09:53] Launchpad bug 1265483 in grub-installer (Ubuntu Trusty) "Installer crashed when installing to an Intel NVMe SSD" [High,New] https://launchpad.net/bugs/1265483 [09:54] apw, also my firmware can't boot off this device either, so that's another issue I face [09:54] cking, i am somewhat expecting the next bug from this person to be that [09:55] given they are using ovmf, it may only just have support for them much as your box does [09:55] xnox, http://smackerelofopinion.blogspot.co.uk/2011/10/uefi-edk-ii-revised.html, although I guess the notes may be old now [09:56] bah, the links are stake [09:56] *stale [09:56] 'stale' i thought those were 'pementant' links [09:56] https://wiki.ubuntu.com/UEFI/EDK2 [09:57] apw, somebody kindly redirected them [09:57] xnox, ckings bug above also talks about grub failing during grub-install onto this type of device died [09:57] (as does this new bug) [09:57] cking: well OVMF is packaged and it's simply -bios /usr/share/ovmf/OVMF.fd. [09:57] xnox, that's way too easy ;-) [09:58] apw: sure. quite similar to the most recent patch to enable FusioIO devices. [10:00] cking: so you could try http://paste.ubuntu.com/6683593/ but that one assumes that /dev/nmve0n1 is device name and /dev/nmve0n1pX are partitions of thereof. But that's me guessing =/ ideally there would be some more precise way to know how nmve devices and partitions are numbered. [10:00] cking: in the mean time, i'll try to make this qemu-system to give me a nmve device. [10:01] xnox, if you can figure out how to do that, then please let me know [10:01] cking: of course =) [10:13] cking: well, i booted something which does have nvme devices, they are numbered differently from what your device was however. [10:13] cking: qemu-system-x86_64 -bios /usr/share/ovmf/OVMF.fd -m 1024 -cdrom /var/cache/utah/iso/ubuntu/trusty-desktop-amd64.iso -drive file=nvme.img,if=none,id=D22 -device nvme,drive=D22,serial=1234 --enable-kvm [10:14] adjust memory, path to .iso, path to a device-drive image "nvme.img", no clue about the nvme parameters i copied them from blogger. [10:14] xnox, that patch make grub install correctly [10:14] that does boot in EFI mode. [10:14] cking: \o/ excellent. Does it boot? [10:14] cking: cause grub2 might need patches to actually recognise the device =/ if that is the case it would be passing the ping-pong to cjwatson. [10:15] xnox, I can't boot it 'cos the firmware does not support that device [10:15] i.e. it can't find it, I need newer firmware I believe [10:17] cking: people on the forums patch stock motherboard firmwares with updated drivers all the time. E.g. i got "unofficial" firmware for my motherboard with updated intel drivers et.al. =) granted it may have come with extra NSA modules =) i got it off the forum post attachment over http ;-) [10:18] cking: so what do you want me to do? upload that patch into archive & respin ubiquity such that at least you will have daily desktop/server images tomorrow with it? [10:19] cking: and do we need this for 12.04.4? [10:19] xnox, gimme a sec, I will just boot the box up from another image and see if the nvme grub looks sane [10:21] xnox, it looks sane to me, so lets update it for the daily images [10:21] xnox, have we asked the OR for their qemu command line etc, ask them when you ask them to test tommorows images [10:21] apw: >~cking ? [10:22] apw, that makes sense [10:22] apw: i don't think i get to them at all. slangasek simply asked me to look into cking's bug report. [10:22] to talk that is. [10:24] cking: apw: https://bugs.launchpad.net/bugs/+bugs?field.searchtext=NVMe&search=Search+Bug+Reports&field.scope=all&field.scope.target= gives nice output, at least to me =) [10:25] xnox, well, I'm OK with the fix, it installs grub correctly, I just need to talk to intel to get my firmware sorted to try a boot test [10:26] so pop it in to the archive and respin is good for me. any objections apw? [10:26] * Handle NVMe devices, based on the debug logs from Colin [10:26] King. Preliminary support (LP: #1265483) [10:26] Launchpad bug 1265483 in grub-installer (Ubuntu Trusty) "Installer crashed when installing to an Intel NVMe SSD" [High,New] https://launchpad.net/bugs/1265483 [10:27] sounds good? as in conveying that it's not fully tested yet?! [10:28] yep [10:29] it may be a while before I can properly test this, I need to chat to various bods in intel [10:58] cking, shove it in ;) [10:58] +1 [10:58] xnox: zul and hallyn look after libvirt usually. zul uploaded 1.2.0-0ubuntu1 yesterday. [10:59] rbasak: excellent. didn't notice. [11:01] thanks xnox [13:18] cking, just going back through my pre-holiday emails. For Trusty unstable (v3.13) I have set CONFIG_X86_INTEL_PSTATE=y. Is that OK without the thermal daemon ? [13:20] rtg, well, yes and no [13:33] Is INTEL_PSTATE still a laughing stock of fan-spinning hilarity? [13:33] It's been shit for so many releases, I'll be amazed if it's finally fixed. === gfrog is now known as gfrog_afk [14:05] apw, still working on getting lttng to compile on all arches for unstable. seems it doesn't like gcc 4.8 on armhf [14:06] rtg, ack, i had noted it was off [14:14] hey apw [14:24] hi [14:39] apw: still no response on https://bugs.launchpad.net/ubuntu/+source/linux/+bug/967399 :( im not sure what to do in terms of getting it upstreamed [14:39] Launchpad bug 967399 in linux (Ubuntu) "[11.10] Elantech trackpoint does not work Lenovo " [High,In progress] [14:41] hi, has somebody a patch for nvidia + 3.13? [14:49] Kano: it's in 14.04 [14:49] tseliot: i dont see it in the changelog [14:49] let me check [14:52] Kano: ok, maybe I confused it with 3.11 and 3.12. Do you have a dkms log for nvidia showing the failure? [14:53] tseliot: it builds,but i get: nvidia: Unknown symbol acpi_os_wait_events_complete (err 0) [14:56] Kano: apparently they removed a symbol. I'll work around that next week [14:58] tseliot: well the patch in the nv forum seems to use >= instead of < [15:11] tseliot, does that imply that the nvidia dkms is not working with 3.13 ? [15:12] apw: I guess so, I haven't checked [15:12] https://devtalk.nvidia.com/default/topic/644906/331-20-on-3-13-rc1-kernel/ [15:14] tseliot, ok sounds like there are a couple of issues, but sounds like you are on the case [15:16] apw: yes, whatever that is, I'll fix it. BTW where are we going to get 3.13 in 14.04? [15:16] well ... "now" if the dkms packages we care about arn't borked [15:17] i suspect "early next week" is most likely now [15:18] I'm thinking Monday [15:19] I'll probably fix my packages next week [16:12] apw, did a practice rebase on unstable against current Linus tip and had aufs3 conflict. I resolved by dropping the conflicting chunk (i.e. the code the patch wanted to add), but could use a second set of eyeballs. perhaps you could take a quick pass at rebasing as well? [16:16] tseliot, fglrx isn't installing. you should get that on your todo list as well. [16:16] Setting up fglrx (2:13.101-0ubuntu5) ... [16:16] update-alternatives: using /usr/lib/fglrx/ld.so.conf to provide /etc/ld.so.conf.d/x86_64-linux-gnu_GL.conf (x86_64-linux-gnu_gl_conf) in auto mode [16:16] update-alternatives: error: error creating symbolic link `/usr/lib/x86_64-linux-gnu/xorg/extra-modules.dpkg-tmp': No such file or directory [16:16] dpkg: error processing fglrx (--configure): [16:17] rtg: that's what I meant when I said that I would fix my packages ;) [16:18] rtg, will do [16:30] rtg, so its not a "just drop that hunk job" [16:30] apw, ok. looks like upstream changed some macros though. [16:33] rtg, this is aufs3 introducing a macro intersecting with upstream adding the 'get flags first' to avoid a use-after-free [16:33] i'll build test my merge and see [16:34] apw, cool. I figured you would be able to spot it right away having done this before [16:54] rtg, seems to work, i guess it makes sense for me to push this rebase so we don't have to do it again [16:54] apw, ack, thanks [16:54] rtg, done [16:54] apw, I was thinking about uploading to the PPA so I could get some mileage on the IMA stuff [16:55] i assume we'll get an -rc7 like tommorrow ? [16:55] i fixed the nvidia binary,but you should update the firmware package for 3.13 as well as files like /lib/firmware/radeon/HAWAII_pfp.bin are missing [16:55] apw, likely, but I'll be off skiing [16:59] maybe radeon dpm could be activated by default [17:03] radeon dpm ? [17:03] nvidia binary> patch? [17:04] apw: http://kanotix.com/files/fix/nvidia.wheezy/update/nvidia-graphics-drivers-331.20-2kanotix1/nvidia-graphics-drivers_331.20-2kanotix1.debian.tar.gz [17:04] look into debian/patches [17:06] it is tseliot that needs that info [17:06] and if somebody from X is looking, why dont you add libvdpau-dev for mesa [17:06] and package the vdpau /xvmc things [17:07] these are questions you might want to ask in #ubuntu-x [17:07] well i dont use mesa from ubuntu, just as hint [17:07] radeon dpm should be on by default in 3.14 [17:08] well why not with 3.13 [17:08] additional fixes in 3.14, I guess [17:08] i used it with 3.12 too [17:10] well, I use it here with 3.11 [17:55] cking: so... your memory benchmark of IMA was for IMA built in, but not enabled? [17:55] kees, I compared it with it build in and enabled vs not enabled with some simple instrumentation [17:57] cking: the report in the bug was for which situation? [17:57] built in and enabled [17:57] I'm curious if just the CONFIG_IMA built in (but not enabled with policy) incurs a cost? [17:57] with it built in and not enabled we see an atomic_t per inode [17:57] extra [17:58] ah, so the math following that was not the result of just the atomic_t? [17:59] if it's just 8 bytes per inode when disabled, that seems like a reasonable cost to add CONFIG_IMA [17:59] kees, yep, with it enabled there are some 72 byte slab allocations and misc meta data which again isn't too large [17:59] per file [18:00] cool. what's needed to enable it? (i.e. and how can we check if a system has it enabled or disabled?) [18:00] * cking rummages around for a suitable quickstart guide.. [18:01] I just want to make sure we don't regress -- CONFIG_IMA caused soooo much trouble for big file servers. :P [18:01] kees, http://sourceforge.net/p/linux-ima/wiki/Home/ [18:02] ^ i think that's more reasonable source of info rather than me regurgitating it all [18:02] cking: sounds good. [18:03] cking: thanks for doing all this! [18:04] no probs, thank apw for asking me to look into it ;-) [18:06] hehe [18:11] apw: are you around? [18:24] kees: cking thanks for that ;) [18:31] Now if you could just sign all userspace binaries, that'd be great [18:32] mjg59: haha [18:34] * rtg -> lunch [18:52] cking: is there a link to the commit enabling IMA? [18:52] * antarus isn't finding it on the lp bug [18:54] antarus, http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-trusty.git;a=commit;h=72bb20892cddc00b0d47d987f929017390fa39c9 === jpds_ is now known as jpds [20:10] hrm [20:10] launchpad continues to elude me [20:10] How do I made it clear on https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1244627 that I would like the change also in Saucy? [20:10] Launchpad bug 1244627 in linux (Ubuntu Trusty) "Please enable CONFIG_IMA in the ubuntu kernel" [Medium,Fix committed] [20:10] whenever I futz with the Affects stuff, it won't let me add another copy of 'linux' and I can't find a linux (Ubuntu Saucy) package.. [20:24] antarus: nominate for series, then mark saucy [20:29] sbeattie: don't I have to be on a specific team to nominate for a series? [20:30] to nominate, no, to accept the nomination, yes [20:42] it appears apw did the needful for me [20:42] apw: thanks ;) [21:07] * rtg -> EOD [21:23] downloading the kernel debug syms [21:23] huge mistake [21:23] heh