[02:49] PR snapcraft#2825 opened: remote-build: remove need to specify user [05:17] PR snapd#7798 closed: interfaces/browser-support: allow reading status of huge pages [05:17] PR snapd#7801 closed: tests/main/gadget-update-pc: use a program to modify gadget yaml (2.42) <⚠ Critical> [05:20] PR snapd#7779 closed: interfaces: misc updates for u2f-devices, browser-support, hardware-observe, et al [05:32] PR snapd#7804 opened: interfaces: misc updates for u2f-devices, browser-support, hardware-observe, et al (2.42) [05:34] good morning mvo [06:33] mborzecki: hey [06:33] morning [06:33] zyga: hey [06:33] mborzecki: yesterday stuff was red all the time [06:33] I went to bed after 3rd round of restart-and-pray [06:34] I saw mvo merge and restart things that managed to land last night after I went to bed [06:34] there was a unit test failing on moster, but looks like it's fixed now [06:34] yeah [06:35] jamie fixed it [06:35] ok, cool [07:21] https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.2-Go-Register-Corrupt [07:26] re [07:29] zyga: that's encouraging [07:54] for some godforsaken reason, I haven't fallen asleep yet [07:54] but anyway, looks like snapd 2.42.2 and snapd-glib 1.54 landed in all Fedora releases and EPEL 7 and EPEL 8 [08:00] mvo: hi, I commented a bit on the next two UC20 PRs === pstolowski|afk is now known as pstolowski [08:03] good morning [08:05] pedronis: thanks! [08:08] mvo: thanks for updating #7795 [08:08] PR #7795: overlord/snapstate: pick up system defaults when seeding the snapd snap (2.42) <⚠ Critical> [08:09] pstolowski: hey [08:09] Eighth_Doctor: yay! [08:14] mborzecki: no worries, let's hope it's fine this time [08:23] haha, #7799 is a nice playground for bikeshedding about performance with artificial microbenchmarks [08:23] PR #7799: osutil/mount: de-duplicate code to use a list [08:27] pstolowski: hi, your prebake trello card seems to be still about the spike, vs landing the pieces [08:27] I mean in Doing [08:27] pedronis: hey, right, will update, thanks [08:28] thx [08:33] PR snapd#7805 opened: osutil/mount: optimize flagOptSearch some more [08:48] pstolowski: hey [08:48] zyga: ^ is that related to Ian's perf findings ? [08:48] pstolowski: simple one for the morning? https://github.com/snapcore/snapd/pull/7803 [08:48] PR #7803: interfaces: remove reservedForOS from commonInterface [08:49] pedronis: yes [08:49] pedronis: it's a bit of a wake up thing [08:49] the original is still faster if we optimize allocation with popcount l [08:50] zyga: sure.. interesting, i though i removed it at some point when validation was delegated to policy check [08:50] thanks! [08:50] PR snapd#7803 closed: interfaces: remove reservedForOS from commonInterface [08:56] mvo: 2.42.3 is okay except for waiting for CI right? [08:56] mvo: I can jump into my regular feature work now [08:56] zyga: correct [08:56] zyga: once stuff is green I will release [08:56] Chipaca: hey, per yesterday's suggestion from pedronis, i'm happy to take https://github.com/snapcore/snapd/pull/7320 from you if that's ok [08:56] PR #7320: snap/pack, cmd_pack: 'snap pack --check-skeleton' checks interfaces [08:57] pstolowski: sure [08:57] pstolowski: was there more to do there? [08:58] Chipaca: there are a couple of comments, e.g. special casing for 'snap run' [09:01] Chipaca: i'll probably get to it in a day or two [09:02] right [09:03] jdstrand, do you actually want to boot without SD ? (sorry, i'm at a trade show so only saw your ping now) ... if you can live with SD, just copy the writable partition over to the USB drive, re-label the SD one to "writable-old" and make sure the USB one is labeled "writable" ... [09:16] pstolowski: that pip feature is super useful [09:16] pstolowski: thank you again [09:16] pstolowski: it works on top of vmware [09:16] zyga: pip? [09:17] zyga: yep it's super nice [09:17] as in picture-in-picture? [09:17] mborzecki: picture in picture [09:17] https://usercontent.irccloud-cdn.com/file/BH5R96NZ/pip-over-vmware.png [09:17] zyga: a macos thing? [09:17] mborzecki: safari feature [09:17] yeah [09:18] zyga: would be cool if it worked with meet/ho [09:18] does it? [09:18] nope [09:26] PR snapd#7804 closed: interfaces: misc updates for u2f-devices, browser-support, hardware-observe, et al (2.42) [09:30] mborzecki: what's the plan with 7796, on its own is supposed to fail no? [09:30] (it's failing on other random stuff atm) [09:33] pedronis: failed as planned https://paste.ubuntu.com/p/4rjfhJzjBT/ [09:34] mborzecki: do you plan to change to contain the fix? land it with manual and do another PR? [09:34] pedronis: i'll be pushing the fix in snapstate into that branch [09:34] ah ok [09:35] it will need a summary change [09:35] (stating the obvious in case :) ) [09:41] mhm [09:41] zyga: do you remember the bit we had in the spec that worked around fedora's patching of /bin/sh to /usr/bin/sh ? [09:42] yes [09:43] what about it? [09:43] zyga: must be blind, i'm looking at our spec and a downstream one and don't see it [09:43] one sec :) [09:44] mborzecki: we only have it in our test code [09:44] https://github.com/snapcore/snapd/pull/7614/files#diff-556bb7431481e375713ea3e0883a771aL111 [09:44] I remove it here [09:44] it's not downstream [09:44] PR #7614: cmd/snap-confine: implement snap-device-helper internally [09:45] zyga: right, i was asking in the context of https://bugs.launchpad.net/snapd/+bug/1824158 [09:45] Bug #1824158: compatibility bug with fedora [09:46] mborzecki: it's the impossible task of using one program from both contexts [09:46] mborzecki: not sure what you are asking about specifically [09:46] mborzecki: the bug is still present [09:47] it's fixed by the PR I referenced [09:52] zyga: downstream had _brp_mangle_shebangs_exclude ^/bin/(bash|sh)$, but that got dropped when updating to 2.41, don't recall why [09:52] I don't recall why either [09:52] perhaps accident? [09:53] Eighth_Doctor: do you recall whether that mangling workaround was no longer needed? [10:05] PR snapd#7806 opened: tests/lib/prepare: drop workarounds for rpmbuild rewriting /bin/sh [10:05] dot-tobias: ping [10:06] mborzecki: why do you want to drop https://github.com/snapcore/snapd/pull/7806 [10:06] PR #7806: tests/lib/prepare: drop workarounds for rpmbuild rewriting /bin/sh [10:07] it's not going to pass today, is it? [10:07] zyga: ohmygiraffe works on f31 without the rewrite, quite sure it picked up the drm dervices since there's opengl interface involved [10:07] mborzecki: because f31 has v2? [10:08] I mean, it's premature [10:08] zyga: it's f30 [10:08] the system i tried it on [10:08] on f31 you don't have device cgroup [10:08] on f29 it will regress [10:08] unless I'm missing something [10:08] we should not drop that yet [10:17] zyga: a while back, you said you were working on a better way to detect snap confined apps than reading the cgroup proc file. Has any progress happened with that? [10:17] jamesh: yes, there's been a lot of progress on this topic [10:17] jamesh: but the means have changed [10:17] jamesh: because of all kinds of unexpected complexity [10:18] jamesh: I'm working on this today actually [10:18] Chipaca: hi, maybe you could get to do a first review of #7771 when you have a moment ? [10:18] PR #7771: o/hookstate/ctlcmd: snapctl is-connected command [10:18] a'yup [10:18] jamesh: the status quo is that we'll use a systemd _scope_ for all non-service processes [10:18] jamesh: and typical service cgroup for services [10:19] zyga: awesome. Is it still a file we can read from the process's mount namespace, or something else? [10:19] jamesh: I'll share the details if you want [10:19] jamesh: but if you wait a day you I will have the new PR up [10:20] jamesh: a means to check if a process belongs to a snap involves reading /proc/[pid]/cgroup [10:20] jamesh: one _can_ spoof this [10:20] jamesh: but it's true for all snaps [10:21] jamesh: does this make sense so far? [10:21] zyga: I can wait for the PR. I'm mostly interested in terms of improving xdg-desktop-portal's snap support [10:21] zyga: hm looks like you fixed it a while ago https://github.com/snapcore/snapd/commit/641adbf815db392248134a6cfe9650e1abf89575 [10:22] zyga: https://github.com/snapcore/snapd/pull/7049 yeah, that'd explain why it works now [10:22] mborzecki: interesting :) [10:22] PR #7049: cmd/snap-confine: handle device cgroup before pivot [10:22] mborzecki: but even if we run it, will it run correctly? [10:22] zyga: so I mostly want something that (a) doesn't depend on AppArmor like the current code does, (b) is unlikely to change over snapd versions [10:22] it won't run in one of the two spaces === pedronis_ is now known as pedronis [10:22] jamesh: a) is true b) is something we can try to promise to keep [10:22] zyga: yes, it's run either by s-c inside the host ns, or udev also in host ns [10:22] jamesh: a) also works across cgroup v1 and v2 [10:23] zyga: false positives are not too bad, since we'd be shelling out to "snap routine portal-info" to get the actual information [10:23] jamesh: that's great [10:23] which could do something snapd version specific [10:26] brb [10:28] mvo: 7797 will need updating because of 7792 [10:29] pedronis: yeah, I will do as soon as one of them lands [10:29] PR snapd#7807 opened: snap-bootstrap: remove SNAPPY_TESTING check, we use it for real now [11:08] PR snapd#7794 closed: many: backport pull request #7773 from zyga/fix/lp-1852361 <⚠ Critical> [11:08] PR snapd#7795 closed: overlord/snapstate: pick up system defaults when seeding the snapd snap (2.42) <⚠ Critical> [11:08] mvo: woot, thank you [11:20] yay [11:20] so .3? [11:26] PR snapcraft#2826 opened: Copy npm and npx binaries in snap [11:27] pstolowski: almost +1 on 7771 [11:28] zyga, mborzecki *thank you* ! [11:29] Chipaca: ty [11:30] mvo: as soon as .3 is out I'll do a suse version [11:36] zyga: working on it now, had a meeting before [11:36] cool, no rush :) [11:47] mvo: mborzecki: what about 7796 ? [11:59] pedronis: i did the change we discussed, waiting for spread test, but still have some wonderings about the fix [12:00] need to run a quick errand, back in 30 or so [12:06] PR snapd#7808 opened: release: 2.42.3 [12:09] pedronis: you mean what about 7796 for 2.42.3? [12:15] mvo: yes [12:18] pstolowski, hey [12:18] cachio: hi [12:20] I am researching the test bad-interfaces-warm [12:20] and I see an error when we execute on arm devices [12:21] pstolowski, this is the otuput of the snap install https://paste.ubuntu.com/p/XpY4spwC5Q/ [12:21] pedronis: I thought we said it's not criticial for this release, has that changed? we also don't have a fix yet, correct? i.e. so far this is the regression test [12:22] mvo: it is kind of critical given we don't know how they use the device [12:22] mvo: it's the third aspect that is not critical [12:23] pstolowski, I see that on edge and beta [12:24] pstolowski, is it a bug? [12:24] right? [12:25] cachio: no. unless the test failed? [12:25] cachio: this is a new test about bad plug/slots [12:26] cachio: snap install will warn if there are bad plugs/slots [12:26] pstolowski, the test fails [12:26] cachio: can you show me the full output? [12:27] pstolowski sure [12:27] https://paste.ubuntu.com/p/zGzjtTdRjs/ [12:28] pedronis: ok, that sounds like we need a .4 :/ let me tag it with 2.42. and critical [12:29] mvo: sorry, I thought it was clear we wanted a fix for this in too [12:29] mvo, hey, should I invest time testing .3? [12:29] if we are going to have a .4? [12:29] pedronis: I misunderstood, sorry [12:29] cachio: yes, .4 there will be [12:30] ok [12:31] cachio: is this test run against latest snapd? looks like the new functionality is not there [12:33] on edge if failing with the same error [12:33] hmm or maybe there is a race and warning is not immediately available [12:33] cachio: do you have shell with this failrue? [12:33] pstolowski, yes [12:33] I am refreshing to edge now [12:33] cachio: can you execute 'snap warnings' ? [12:34] No warnings. [12:34] with beta [12:34] letme refresh to edge [12:42] pstolowski, same output with core on adge [12:42] pstolowski, using this core 16-2.42.2+git1570.35c66d9 [12:42] also: No warnings. [12:46] mborzecki: https://github.com/snapcore/snapd/pull/7806#pullrequestreview-323645799 [12:46] PR #7806: tests/lib/prepare: drop workarounds for rpmbuild rewriting /bin/sh [12:47] PR snapd#7799 closed: osutil/mount: de-duplicate code to use a list [12:53] zyga, mborzecki: so I stayed up all night and hand wrote GPU assembly and was able to get that mount opts parsing code down to 4 planck time units! [12:53] ijohnson: hehe [12:53] hey :) [12:53] Morning :-) [12:54] ijohnson: I think the original would be still fastest with the only extra change to get the array size done up front [12:54] hey ijohnson [12:54] does anyone know how to map our short git1570.35c66d9 version number to git log? [12:54] re [12:55] ijohnson: hahah ;) must have been a great deal of fun [12:55] ijohnson: did you see 7805? [12:57] cachio: i suspect core doesn't have my change yet for some reason as was a case already in the past. i need to decode that git version string to check [12:58] zyga: I saw the email notifications y'all had more optimizations, but haven't really looked yet [12:59] cachio: this is backed up by the fact that even snap warnings doesn't show anything [12:59] pstolowski: is that from the edge channel? IIRC that git commit is from some other repo that builds the core snap on edge [12:59] pstolowski, I can check on amd64 [13:00] pstolowski, I found the problem [13:00] snapd-vendor-sync is not running [13:00] whew [13:01] whee [13:01] pstolowski, I'll check the vm to see why [13:01] thinkpad cover arrived [13:01] pstolowski, thanks [13:01] cachio: yw [13:02] ijohnson, cachio is vendor sync / other repo making this version number completely different from our git revisions~? [13:03] pstolowski, yes [13:03] huh [13:03] snapd-verdor-sync does not use github [13:03] that's annoying [13:05] * ijohnson goes back to breakfast [13:06] PR snapd#7797 closed: devicestate: make /var/lib/snapd/seed available in install mode [13:14] * Chipaca takes a break, and reboots while at it [13:21] good progress on cgroup bits, one last annoying spread test and I'll open a PR for initial review === ricab is now known as ricab|bbl [13:50] thnkpad is fixed :) [13:56] uhhh [13:56] it's not [13:56] zyga: thinkpad is never broken, it's only an transient non-working state [13:56] the part was broken [13:56] heh [13:56] that's ok [13:56] it's silly [13:57] unless it's cooling, it's broken by design [13:57] part of the bottom cover is broken [13:57] but that's ... enough [13:57] I guess [13:57] spending more money on this is probably not worth it [13:59] PR snapd#7800 closed: tests: add Ubuntu Eoan to google-sru backend [14:06] ijohnson: did I understand you correctly that snap-confine without root was the topic you mentioned? [14:06] like without root at all [14:07] zyga: https://forum.snapcraft.io/t/permissions-problem-using-snapcraft-in-azure-pipelines/13258/10?u=ijohnson [14:07] no, "/" is not root-owned is the issue [14:07] ah [14:07] what's fun [14:07] and weird [14:07] we can relax that check I guess [14:08] is there a corresponding bug report? [14:08] no I don't think so [14:09] ok [14:10] I'd like to understand why / is 501 though, it's a bit silly and odd [14:10] I'll read the thread after the standup [14:14] dot-tobias: ping [14:18] pstolowski: let me know if you want me to look at your service branch, wasn't sure if that's one of your current open PR's or a local branch [14:20] ijohnson: thanks, it's not proposed yet; i'll dig a bit more myself. will definately ask you for a review when ready. nb, have you discussed any particular approach for fixing snapctl side? [14:21] pstolowski: there's this forum post: https://forum.snapcraft.io/t/systemctl-service-management-unification/13808 I need to post an update from the discussion I had with folks after that on the plan so yes we have discussed the approach, I forgot to put it on the forum :-) [14:22] ijohnson: great, ty! [14:39] pedronis: i've updated #7796 [14:39] PR #7796: overlord/snapstate: make sure configuration defaults are applied only once <⚠ Critical> [14:42] heh, typo [14:42] PR snapd#7776 closed: interfaces: add login-session-observe for who, {fail,last}log and loginctl [14:43] off to pick up the kids [14:50] mvo: mborzecki: small comment on ^ [14:50] heh, I mean #7796 [14:50] PR #7796: overlord/snapstate: make sure configuration defaults are applied only once <⚠ Critical> [14:56] Chipaca: I think I addressed all your comments to #7771 [14:56] PR #7771: o/hookstate/ctlcmd: snapctl is-connected command [14:58] pstolowski: thank you [14:58] PR snapd#7793 closed: devicestate: read modeenv early and store in devicestate [15:02] pstolowski: I updated https://forum.snapcraft.io/t/systemctl-service-management-unification/13808/7?u=ijohnson with what we discussed in that meeting === ricab|bbl is now known as ricab [15:04] ijohnson: thanks for that and for the exhaustive description of the problem. looks like it may be a separate PR that's a prerequisite for my current set of changes [15:05] pstolowski: ok, does this mean you are now blocked on me working on that? I think I'm at a point now where I can switch to doing that instead of the performance stuff, just need to get folks to agree on what we should do now I think [15:07] PR snapd#7809 opened: interfaces: remove leftover reservedForOS [15:08] ijohnson: ah, you inteded to do that? i thought it was passed to me after your initial changes to services. yes, this is kinda blocking, i'm not sure there is a way around that [15:09] pstolowski: yes that was the plan that I was going to work on that after I wrapped up the performance stuff. but if you're totally blocked on it, considering I will be out on thurs + friday, perhaps it makes more sense for you to get started on that [15:11] ijohnson: right, it may be best if i take it (as long as it doesn't make your plate empty ;)) [15:11] I'm sure mvo can find something else for me to work on :-) [15:12] ijohnson: haha, I'm not worried about this [15:13] :) [15:13] ijohnson, mvo: ok, i'll take it then [15:13] zyga: I think https://github.com/snapcore/snapd/pull/7805 is ready to merge now :-) [15:13] PR #7805: osutil/mount: optimize flagOptSearch some more [15:14] looking [15:14] ijohnson: I may do more ;) [15:14] ijohnson: but it was fun, thank you [15:15] PR snapd#7805 closed: osutil/mount: optimize flagOptSearch some more [15:15] * cachio lunch [15:15]