[05:57] Good morning [06:48] good morning [06:51] bonjour didrocks ! [06:52] hey pitti [07:25] morning didrocks pitti :) [07:27] hey happyaron [07:33] happy Friday every1 [07:34] hey happyaron, hello larsu! [07:59] pitti: yeah, confirmed on an installed system, I should just have tried on the live… [08:00] good morningg desktopers [08:01] pitti: so, as this will stay a distro-patch anyway, I propose that I do the minimal workaround in systemd-machine-id-commit itself, meaning checking if there is an overlayfs mountpoint in /etc/machine-id or parent [08:01] re seb128 [08:01] as* [08:01] that way, the systemd binary isn't at risk, and the patch is smaller [08:02] bonjour seb128, comment vas-tu ? [08:02] didrocks: sounds good to me [08:03] pitti: I don't know why the bigger patch hangs TBH, sounded sane enough to me, but maybe there is an evaluation of mount points too early or such… [08:03] but let's go minimal as upstream wants a kernel fix [08:07] hey pitti, ca va bien ! et toi ? [08:11] seb128: ça va bien aussi, merci ! [08:12] waow, one run, two bugs discovered on ubuntu make, I guess this guy should apply for the QA team :p [08:12] * didrocks understands the first one, not yet the second though [08:30] haha [08:56] yawn - morning [08:57] morning willcooke [08:58] TheMuso, still around? === dpm is now known as dpm-afk === sil2100_ is now known as sil2100 [09:28] hey larsu - shall we meet? [09:29] willcooke: yep === Zdra is now known as xclaesse [09:43] seb128: So, I did an 4.4.1 update for the ppa including the breeze icons. Since this is FFes for kubuntu, it should be uploaded soonish. However, Friday uploads are bad, so lets target that for Monday? [09:47] seb128: FWIW, 4.4.2~rc2 has been tagged upstream today and likely will be 4.4.2 final. I will update the ppa as soon as this update is in -proposed. Likely still rather a 0-day SRU as it will get tight with timing otherwise. === greyback__ is now known as greyback [10:41] pitti: phew, I found an easy fix! [10:41] for the machine-commit-id [10:41] (the issue is when major(parent = 0) and major(child) != 0 [10:42] sorry, I meant, we only wait to fallback in that case [10:42] (as it's overlayfs-like) [10:42] but it seems we can get major(parent) = 0 and major(child) = 0 [10:42] we do not want to fallback in that case [10:42] (case of tmpfs) [10:43] didrocks: oh, good! I sorted out the iscsi stuff, so I'd like to do a merge with Debian exp [10:43] so please poke me if/once you have a patch [10:43] pitti: do you want me to repropose a patch with the fix? (I'm going to do a full suite of test with the package installed) [10:43] pitti: I can give it to you if you want to run some tests in parallel [10:44] didrocks: sure! [10:52] pitti: http://people.canonical.com/~didrocks/tmp/0001-Fix-mount-point-detection-on-overlayfs-and-similar-f.patch [10:52] pitti: the only diff is replacing if (major(b.st_dev) == 0) by if (major(b.st_dev) == 0 && major(a.st_dev) != 0) [10:53] pitti: I'm rebuilding the ubuntu package with this, now that I removed all my manual breaks and printf statements to ensure it's fine [10:53] seb128, do we need to worry about this new glibc do you think? [10:53] while it's building, going for a run (and then I guess it will be 20 minutes for a full test with vm, vm-iso, /etc/machine-id on ro…) [10:54] didrocks: ah, ok; thanks! I can test it on Debian exp on an installed system, if you want to test it on a live? [10:55] pitti: sure, the installed system isn't the most difficult, but we can do like that [10:57] pitti: ok, ubuntu package building, going for a run meanwhile :) [10:57] didrocks: enjoy! I'll test on Debian installed [10:57] thanks! [11:17] didrocks: works fine on installed; let me know how it goes on live, then I'll push [11:17] didrocks: note that there was some noise with gbp-pq import/export [11:18] pitti: oh right (yeah, failed to take a break, now waiting for it to finish to build) [11:22] willcooke, hi [11:23] hey tkamppeter! Got your mail, just waiting to hear back from Rick [11:38] pitti: no, forget about it, there is another use case I can't get working with this patch [11:40] pitti: if only, I could get my printf to shows up while it's hanging… [11:42] didrocks: ah, meh -- you didn't go running? [11:43] pitti: yeah, wanted to get a good news on this before going… :p [11:43] efail === dpm-afk is now known as dpm === MacSlow is now known as MacSlow|lunch [12:12] pitti: ok, the issue is the call due to readlink_and_make_absolute when called on some /sys path [12:16] * happyaron sents out bunches of emails to gsoc prospective students [12:17] pitti: ok, understood it, what happens is: [12:17] -> /sys/ is read [12:17] then, we try to readlink_and_make_absolute [12:17] this errors out [12:17] errno != ENOENT [12:18] we return -errno [12:18] and so, systemd doesn't like it (I guess it prevents some execution with /sys setup) [12:18] so, I guess instead of returning -errno, we can return a.st_dev != b.st_dev [12:18] as it's the previous behavior [12:25] hey seb128 did you see my comment on the schemas bug? [12:30] pitti: recreated the patch, building now, going finally for a run, and then, I'll test installed, live and such… [12:40] Sweet5hark, monday is fine [12:40] willcooke, glibc shouldn't be an issue for us no [12:41] darkxst, yeah, I saw, I didn't think much about it yet though [12:41] seb128: great, thx [12:41] yw [12:42] also: c- c- c- c-combobreaker for sebs reply volley! [12:43] seb128, ok, it doesn't really need thinking about now, just wasnt sure if re-assigned bugs would get back to you [12:59] pitti: \o/ success (both permanent, live and temporary machine-id) [13:00] didrocks: yay! what was it? [13:00] pitti: yeah, I couldn't resist to get this testing before taking a break :) [13:00] pitti: basically on /sys path, readlink will return some errno [13:00] and I guess that systemd stops going on with detecting if path is a mountpath if it returns < 0 [13:01] (so doesn't read some /sys content and such…) [13:01] the fix is, instead of returning errno (not being able to read a symlink in a directory is a valid one anyway), use the major() comparison [13:01] so basically, being a no-op for those cases [13:02] I extended the same in case we are in error reading /proc/self/mounts btw [13:02] (fallbacking to the previous comparison method) [13:03] pitti: so, that should be it: http://people.canonical.com/~didrocks/tmp/0001-Fix-mount-point-detection-on-overlayfs-and-similar-f.patch [13:03] Note the "return a.st_dev != b.st_dev;" if we fail in the fallback === MacSlow|lunch is now known as MacSlow === alan_g is now known as alan_g|lunch [13:10] pitti: if that makes sense to you, taking my break now :) [13:10] didrocks: I won't say anything to you before you leave :) [13:11] * didrocks is gone thus :p [13:11] (for real) [13:28] pitti: didrocks: this is an interesting patch. What bug is it fixing? [13:29] cyphermox: detecting mount points on overlayfs; the other two standard methods don't work due to kernel bugs [13:30] I'm dealing with the CD ejection and mountpoint when casper ends, it could be useful [13:30] though right now the problem appears to be more with plymouth and/or plymouth's systemd units [13:32] cyphermox: I'm not sure what util-linux' "mountpoint" program does, though [13:32] i. e. if it also just compares stat().st_dev, or falls back to parsing /proc/mounts [13:32] ok [13:32] I think most of casper is just looking at /proc/mounts === alan_g|lunch is now known as alan_g [14:48] pitti: so? :) [14:49] * didrocks back from running [14:50] ah, seems you took it \o/ [14:52] didrocks: just back from running as well :) [14:52] didrocks: yep, in exp and uploaded to vivid [14:52] sweet! [14:54] hm, or did I? /me misses upload and rejection msg [14:54] pitti: forgot to dput? :p [14:54] (nothing in -changes) [14:55] * pitti dputs again *shrug* [14:55] no, I did have an .upload [14:56] * didrocks opens the black hole and looks into it… but nothing :) [15:07] hum, I wonder why gedit is the default handler for text/x-apport types [15:17] seb128: because it's the best [15:17] larsu, ? [15:17] seb128: look at the date [15:18] larsu, it's friday? ;-) [15:18] indeed [15:18] hehe [15:18] I was unsure if that was a real reply or a troll :p [15:18] it's remarkably silent here for a Friday [15:19] $ LC_ALL=C gvfs-mime --query text/x-apport [15:19] Default application for 'text/x-apport': gedit.desktop [15:19] .... [15:19] Recommended applications: [15:19] apport-gtk.desktop [15:19] [15:19] I don't get why gedit is default if the recommended one is apport-gtk [15:19] it's the only .desktop matching the exact mime === JanC_ is now known as JanC [15:20] did you override it acceidentally? [15:20] *accidentally [15:20] no, it's the same in a daily iso [15:20] in->on [15:20] weird [15:22] indeed [15:22] I wonder if that's a glib bug [15:22] * larsu summons desrt into the second channel in the last 10 minutes [15:22] lol [15:24] oh, look at that [15:24] works if I LD_LIBRARY_PATH glib 2.40 [15:24] deeessssrtt [15:27] uh oh [15:28] https://bugzilla.gnome.org/show_bug.cgi?id=744282 [15:28] Gnome bug 744282 in gio "gvfs-open for application/x-virt-viewer changed behaviour between 2.40 and 2.42" [Normal,New] [15:29] so it's a desrt bug [15:33] seb128: oh, I noticed that gedit is quite greedy these days; it apparently wants to open pretty much anything :/ [15:33] I think I even had it open a PDF or iso the other day [15:33] pitti, right, that's due to ^ [15:33] the pdf maybe not [15:33] well, maybe not PDF, but something really strange [15:33] it's been a while [15:34] but apparently it prefers the specified handler for subtypes, e.g x-text, which is gedit, to an handle for the exact type === alex-abreu is now known as alex-abreu|doc [17:29] * willcooke -> EOD. [17:29] Have a good weekend all, I'm on holiday next week - don't touch my stuff === alex-abreu|doc is now known as alex-abreu === alan_g is now known as alan_g|EOW === mfisch is now known as Guest90957