[00:21] say, will saucy be 3.11 or 3.12 do you think? [00:28] kees: you mean trusty? [00:29] kees: because saucy is already 3.11 ;-) [00:29] elmo: argh, off-by-one error. _trusty_, yes :) [00:30] aaah, never mind. [00:30] linux | 3.12.0-5.13 | trusty/main [00:41] kees: its going to be 3.13 [00:42] Sarvatt: okay, cool, thanks :) === gfrog_afk is now known as gfrog === shuduo_afk is now known as shuduo === fmasi is now known as Guest16225 [09:06] kees, yeah 3.13 is sitting in our PPA waiting on the dkms packages etc to catch up === yofel_ is now known as yofel === Kaloz_ is now known as Kaloz === gfrog is now known as gfrog_afk === shuduo is now known as shuduo_afk === ivoks_ is now known as ivoks [14:04] hey. [14:05] do we have any sort of benchmarks that are done and recorded for kernel releases ? [14:06] smoser, sconklin has been working on some storage/fs ones [14:06] i'm wondering because i have extremely unscientific data in my head that recent trusty is not performing as well as previous releases on my laptop. [14:06] that unscientific data "feels" like IO to me. [14:07] smoser, iirc there _was_ a regression in ext4 which cking found but that was saucy, I think [14:08] smb, as in fixed ? [14:09] smoser, Yes, also going to S [14:10] are the benchmarks sconklin has available ? [14:11] smoser, it was address in bug 1242812 [14:11] Launchpad bug 1242812 in linux (Ubuntu) "ext4 random block I/O write performance regression with 3.11 Saucy Kernel" [Medium,Fix released] https://launchpad.net/bugs/1242812 [14:13] it appeared in one of the power regression tests, but those tests are now busted, pending QA to plug in the H/W again [14:47] smoser, and remember we're on 3.12 in trusty atm [16:20] apw: what kernel version are we going to use in 14.04? [16:22] tseliot, 3.13 [16:22] bjf: thanks [16:24] that ;) [16:25] :) [18:05] apw: http://lkml.org/lkml/2013/12/4/156 i'm going to review this patchset asap, we may need it in trusty for userns containers to work nicely [18:05] (to allow writing to /proc/self/loginuid) [18:09] hallyn_, 0/20 ... really [18:12] sigh [18:16] apw: :( [18:17] apw: i'll take a look this afternoon to see if it's actually complicated, or just long-winded or very broken up [18:17] how did we get so far from having the patches we needed as we were at vUDS [18:18] we so do not want to be introducing a new heap of abi if they anr't sure this is the way upstream either, gah [18:19] also it seems to claim to be incomplete [18:19] >> This patchset still need some work, such as allow changing [18:19] >> audit_enabled in audit namespace, auditd wants this feature. [18:19] >> [18:21] and it doesn't sound like upstream will even start looking at it for weeks yet [18:24] hallyn_, plus this calls itself 'part1' and the previous version was 00/48 ... are we sure this is even all of it ? [18:25] 18:17 < apw> how did we get so far from having the patches we needed as we were at vUDS [18:26] well all *we* need is /proc/$$/loginuid writeable. there maybe a shorter way to do it, i simply haven't looked. [18:26] i only know that fedora has the same issue and they claimed to be waitin go this patchset [18:27] apw: we *can* work around it in userspace if there's not a clean patchset we can use [18:27] isn't it immutable for the purpose of auditing? [18:28] ohsix, it is less immutable, as 'change able once able' [18:28] and i assume hallyn_ wants to make the init in a container be able to set it once again [18:28] ah [18:29] i wonder why someone hasn't proposed just adding another one, maybe one that keeps a list of ones that have been set or something [18:29] i guess it makes sense if you have a uidns that you might want to have audit be reset in there, hrm [18:36] apw: well tbh none of it makes sense to me - i thought loginuid should be only for LSPP cred tracking for audit, but it seemingly is now being abused [18:36] hallyn_, it does look like it is doing more than making that writable for sure, it is splitting auditd as well [18:46] blimey this is tensy changes [19:05] * apw notes it is well past beer-o-clock and goes and does somwething abou tit [19:26] yeah, well into beer time now === Guest11320 is now known as kamal === kamal is now known as Guest64627 === retoaded is now known as retoaded_biab === Guest64627 is now known as kamal === retoaded_biab is now known as retoaded