[09:39] <kashyap> Hi, folks.  Is "ubuntu-kernel" the right package to tag for a potential kernel issue?  This was found in a CI of upstream OpenStack - https://bugs.launchpad.net/ubuntu/+source/kernel-package/+bug/2018612
[09:39] -ubottu:#ubuntu-kernel- Launchpad bug 2018612 in OpenStack Compute (nova) "Guest kernel crashes with GPF on volume attach" [Undecided, Triaged]
[09:43] <kashyap> (There's a rough reproducer in the description.  And I've added some environment details)
[11:48] <juergh> kashyap: 5.3.0-26-generic is old and unsupported.
[11:49] <juergh> can you upgrade that machine?
[11:52] <juergh> Oh it's a guest kernel. Sorry, you need to upgrade your guest image. 
[11:52] <kashyap> juergh: Oh, is it?  I was told that kernel (5.3.0-26-generic) from CirrOS 0.5.2 is part of Ubuntu LTS
[11:52] <kashyap> juergh: But yeah; I think it might be possible to use a newer CirrOS image 0.6.1 (which has 5.15.0-48-generic)
[11:53] <kashyap> juergh: Yeah, it's the guest kernel indeed.
[11:53] <kashyap> juergh: Right; the current guest image is 0.5.2.  I made a quick comparison here: https://kashyapc.fedorapeople.org/CirrOS_segfault/CirrOS-0.5.2-vs-0.6.1-kernel.txt
[11:53] <juergh> Presumable 5.3 was the HWE at the time the image was cut but it's no longer supported. It's 5.4 now, same as focal.
[11:53] <kashyap> Ah-ha
[11:54] <kashyap> Why do you say "Presumable"? :-) Isn't it confirmed that it is 5.3 in this case?
[11:55] <kashyap> Or maybe you're saying was "5.3 was presumably the HWE at the time of image cut"
[11:56] <juergh> hrm. the image was built in 2021 so I'm not sure why it pulled in a 5.3 kernel. 
[11:56] <juergh> would be good to understand how that image is built
[11:57] <juergh> even though it's not Ubuntu and you're on your own ;-)
[11:57] <kashyap> juergh: I think it's just plain vanilla 0.5.1 - https://download.cirros-cloud.net/0.5.1/cirros-0.5.1-x86_64-disk.img
[11:59] <kashyap> juergh: If you have time, mind adding some of your observations on the bug?
[12:01] <juergh> oh yes. 0.5.1 was built in march 2020, before focal release which can explain the 5.3 kernel.
[12:01] <juergh> 0.5.2 didn't upgrade the kernel apparently
[12:01] <juergh> yes, will update the bug and mark it invalid...
[12:02] <juergh> btw, for future bug reports, the package to report kernel bugs for is linux
[12:09] <kashyap> juergh: Ah, so the CirrOS 0.5.2 image should update the kernel.  Thanks
[12:10] <kashyap> juergh: Also, noted: the package name.  "Qlinux" is rather cryptic pacakge name; it's impossible to discover it from the list of possible options when you type "kernel" in the launchpad
[12:10] <juergh> yes if that is still a supported image it should be updated to include a supported kernel
[12:11] <juergh> well. if you open a bug from an Ubuntu system it is smart. you're running something (cirros) that is not really Ubuntu...
[12:11] <juergh> but yes, LP is not easy to navigate.
[12:12] <kashyap> juergh: I'm running neither; I'm on Fedora.  And these bugs occur in a CI system, so no way file "a bug from that system".  I'm just retroactively "piecing together the scene from the explosion"
[12:13] <kashyap> (Maybe there is some ABRT-style reporting system (that Fedora uses) on Ubuntu too, to auto-report certain kinds of crashes.  Haven't looked into it)
[12:14] <kashyap> juergh: Thank you for looking.
[12:17] <juergh> kashyap: the bug should really be filed against cirros and not ubuntu: https://github.com/cirros-dev/cirros/issues
[12:18] <kashyap> juergh: Okay ... me goes to file.  So the issue in short is 0.5.2 image should update to LTS kernel 5.4 (from Focal) -- yeah?
[12:18] <kashyap> Any other detail to add?
[12:18] <juergh> bug then again, that's only the build recipes. who build/owns the openstack cirros images?
[12:19] <kashyap> juergh: I don't think the CI infra builds custom CirrOS images; just pull 'em from CirrOS upstream.  (But I'll double-check)
[12:22] <juergh> kashyap: oh they do build images from github. ok, then ask them to update 0.5 and pull in updated packages.
[12:22] <kashyap> juergh: "they" == CirrOS?
[12:22] <kashyap> Or OpenStack CI?
[12:22] <juergh> kashyap: cirros-dev on github
[12:23] <kashyap> Ah, right; also how do they know what _other_ packages to update?
[12:23] <kashyap> You say "pull in updated packages" -- how do they discover it?  (I know, it's "not my problem"; but I'm curious :))
[12:24] <kashyap> Maybe you just mean "pull in kernel and whatever dependent packages"
[12:24] <kashyap> Thanks for your comment in the LP bug!
[12:25] <juergh> kashyap: don't know how they build that image but I assume they point at an Ubuntu mirror so when they rebuild the image they should get latest package versions. they should periodically refresh the images anyways...
[12:26] <kashyap> Noted.  They do have a 0.6.1 version image, though (as noted earlier).
[12:26]  * kashyap goes to file the issue for cirros-0.5.2.  I'll quote your comment verbatim from the LP bug, with context; hope that's fine.
[12:27] <kashyap> juergh: If you got the full 5.4 kernel version of Focal handy, can you please post it here?
[12:29] <juergh> kashyap: https://launchpad.net/ubuntu/+source/linux
[12:29] <juergh> all the versions of the different series
[12:31] <kashyap> juergh: Very handy list; thank you.
[12:47] <kashyap> juergh: Hope that reads fine to you: https://github.com/cirros-dev/cirros/issues/102
[12:47] -ubottu:#ubuntu-kernel- Issue 102 in cirros-dev/cirros "Update CirrOS 0.5.2 kernel image to a supported version (kernel-5.4; from Focal)" [Open]
[18:10] <lotuspsychje> juergh: seems like Bug #2018128 package got back to linux
[18:10] -ubottu:#ubuntu-kernel- Bug 2018128 in linux (Ubuntu) "Apport does not collect all logs when the package is HWE kernel" [Undecided, Confirmed] https://launchpad.net/bugs/2018128