[11:18] apw: ogasawara: yo, ms wants some kernel patches for trusty ideally to enable ext2 freezing https://bugs.launchpad.net/ubuntu/+source/partman-auto/+bug/1297012/comments/15 [11:18] Launchpad bug 1297012 in partman-auto (Ubuntu) "hyper-v: Manual partitioning formats /boot with ext2 file-system" [High,Confirmed] [11:19] apw: ogasawara: otherwise hyperv (well azure) cannot freeze/migrate default installs off ubuntu (as it uses ext2 for /boot =( ) [11:19] apw: ogasawara: are those ext2 patches in 14.10 or 15.04 kernel yet, or will be? and will they make it into hwe kernel? [11:32] xnox, i know they have asked before, so we may well have done so. though it also depends if we are using ext2 or ext4 to mount ext2 in that context [11:33] xnox, oh these are very new bits, so no they won't be currently [11:34] apw: we are using ext2 file format for /boot & we do have CONFIG_EXT4_USE_FOR_EXT23=y, however i don't think that makes ext2 freezable without those patches. [11:35] xnox, right, but as you point out they could select something else, rather than a lot of backporting [11:35] apw: i'd rather have kernel changed with those patches, than changing installer default partitioning filesystems in the default configuration in a trusty point release. [11:35] as they know it does not work [11:35] xnox, it is not clear why they can't preseed it [11:36] apw: it's not up to them to select. but for the guest-os / any client workload on azure. [11:36] apw: it seems that telling all users to do something else doesn't work. [11:36] (well ubuntu based client custom images) [11:36] xnox, right but the only reason this is a problem is because they believe they can make a consistant snapshot of the machine if they freeze it, which btw, i think is flawed thinking [11:37] xnox, the images most people use on azure are supplied by us, this would be hyper-v on a local install only prety much [11:37] apw: i do hope they are not exposing that to people, when they clearly see a given VM is not freezable.... i hope they only do so when they think they actually can do it. [11:37] apw: right, ENOCLUE. [11:38] apw: i haven't looked into partman-auto enough, hopefully we can make /boot ext4 without journaling / extents / etc. (e.g. more like an ext2 thing) [11:38] such that it becomes freezable, but we don't need to increase it's disk space. [11:38] the unfreezability looks to be related to not having a journal [11:38] also we should finally make grub2 not require /boot for just lvm2 based installs, as it can parse things off it. [11:38] =( [11:38] * xnox does not want /boot with a journal..... [11:41] xnox, on my boxes i have no /boot on my lvm installs, no idea why we need it [11:42] xnox, it is not clear whether this second fix would stand alone, it might well do [11:45] apw: hm, maybe cjwatson fixed that already.... [11:45] apw: i'll make a note to self to retest that. [11:45] cirtainly i have what i think is a P install using it, you need a 2048 block aligned first cylinder sort of thing else there isn't room in the gap for grub [11:48] xnox, so this is for .2 right? so will be with an lts-u kernel ? [11:49] we'd not care for anything older i presume [11:49] apw: well if those ext2 patches are in u's kernel, then yeah lts-u-hwe / .2 target is good. [11:49] apw: if those patches are only in v's kernel, then we are looking at .3 target. [11:50] apw: and i don't have ubuntu kernels cloned here to confirm which kernel those patches are in, if at all, and/or need cherry-picking. [11:51] they are 3.18-rc3 so very new, not in anything at all right now [12:09] apw: cherrypick!!!! what could possibly go wrong?! =) [12:10] xnox, heh, we shall see if they even build, let alone work [12:10] apw: cause 3.17/3.18 are so wonderful with all the regressions since 3.16..... =)))) [12:19] xnox, 3.18 isn't feeling toooo bad to me [12:19] apw: i'm on 3.17.4 on my dev hardware and "i don't like it" (big brother quote) [12:20] apw: using stock 14.04 LTS otherwise everywhere as it's the release i can trust =) after all the things i broke for 14.10 ;-) [12:20] * xnox wishes for my SRUs to be accepted..... [13:43] amitk, i just tried to send some idlestat patches to sched-tools@linaro.org and my mailer complained it did not exist === vmesons is now known as vmeson [14:01] cking: I see the patches.... [14:01] amitk, yep, I cc'd you on them ;-) [14:02] amitk, do you have access to the coverity scan for idlestat? there are a bunch more issues that need fixing that I wasn't so sure about the best way to fix them [14:03] cking: I think you subscribed me to the report, I saw it today [14:04] cking: I'll look at the sched-tools alias later today, thanks for reporting. And thanks for the patches. [14:04] ok, cool, just wanted to keep you in the loop about the outstanding issues ;-) [14:04] cking: much appreciated [14:05] amitk, BTW, idlestat still could do with a man page ;-) [14:16] cking: honestly speaking it will probably happen only when we stop changing the options so much. Every few patches we change the behaviour a little. Lack of a spec, I guess [14:16] cking: but if you file a bug on bugs.linaro.org, atleast it'll stay on the radar [14:17] okay will do [15:49] ## [15:49] ## Ubuntu Kernel Team Meeting - Today @ 17:00 UTC - #ubuntu-meeting [15:49] ## agenda: https://wiki.ubuntu.com/KernelTeam/Meeting [15:49] ## === caribou_ is now known as caribou === kamal__ is now known as kamal [16:56] ## [16:56] ## Kernel team meeting in 5 minutes [16:56] ## === jsalisbury changed the topic of #ubuntu-kernel to: Home: https://wiki.ubuntu.com/Kernel/ || Ubuntu Kernel Team Meeting - Tues December 9th, 2014 - 17:00 UTC || If you have a question just ask, and do wait around for an answer! If the question is should I file a bug for something, likely you can assume yes. [19:51] ogasawara: hey curious, we've recently been discussing how to automate some testing for rotation [19:52] curious if there is some known kernel tool/trick that could be used to [19:52] inject sensor events, e.g. mock rotation/orientation sensor firing [19:53] kgunn: I don't know anything off the top of my head, but let me sync with my guys and get back to you === hughhalf is now known as hugh_afk === mjg59` is now known as mjg59 [23:25] * xnox finally realised that brad-figgs bot is launchpad's equivalent of "please hold the line, your call is very important to us" [23:26] and then the call center closes and unanswered calls are just left playing that message =)))) === hugh_afk is now known as hughhalf