=== mwsb is now known as chu [07:34] hello apw sforshee which vbox driver are you using nowadays? 5.2.8 is out, should I ping you or you will take the one mainline? [07:34] * LocutusOfBorg hopes for the latter [12:01] apw, ping :) [12:01] BTW virtualbox is "regressed" because some new tests see that the built kernel and the one in Ubuntu differ... obviously [12:08] yep, dkms is a pile of crap, it has no idea how to compare versions if there is not a manual version [12:18] apw, second question, are you using vbox mainline kernel drivers? [12:20] we are still at .26 i think, and i believe not using those, because the only one which existed at the time was blowin up [12:20] but i would have to defer to sforshee for accurate memory on that [12:22] please delete it, and switch to the mainline version :) [12:25] mainline only provides one of them at 4.15 level right ? [12:35] no, even 4.14 IIRC [12:58] hi, is bionic getting 4.16 after all? [13:04] LocutusOfBorg: bionic currently has version 5.2.6-dfsg-5, we are using the in-tree vboxvideo driver which I believe is the only one that exists in 4.15 [13:06] once the 5.2.8 package promotes I can update to that === Nafallo_ is now known as Nafallo [13:50] sforshee, sorry, I mean, don't use the vboxvideo from vbox, but use the one inside the kernel source tree [13:50] it is now fixed [13:51] git tag --contains ce10d7b4e8e3574b9616e54a09d64521b9aeb8b6 [13:51] v4.15 [13:51] since 4.15~rc1 [13:51] LocutusOfBorg: yes, that's what I meant by "using the in-tree" driver. We've been using that in-kernel one rather than the one we import from vbox-guest-dkms for a while now [13:52] I don't get this: [14:06:00] once the 5.2.8 package promotes I can update to that [13:52] I mean: please continue to use the in-tree one then [13:52] LocutusOfBorg: we still import the vbox-guest-dkms drivers but disable the vboxvideo one from there [13:53] oh... ok, I don't see the need, but fine [13:53] it provides other functions than the video doesn't it? [13:53] otherwise maybe we don't need to do that anymore [13:54] vboxsf and so on... [13:54] ok [13:54] just disable vboxvideo from there then [13:54] yeah, our import script does that automatically now [13:54] ack [13:54] unfortunately such drivers might have a bug [13:54] KERNEL=="vboxguest", NAME="vboxguest", OWNER="root", MODE="0660" [13:54] KERNEL=="vboxuser", NAME="vboxuser", OWNER="root", MODE="0666" [13:55] (see truth you need sforshee :) [13:55] this is the content of: debian/virtualbox-guest-dkms.udev [13:57] anyhow, it is fine to grab even 5.2.8 right now, but I don't think there are updates there [13:59] LocutusOfBorg: hmm, so you're saying without those udev rules having those drivers in the kernel may not provide any benefit? [14:00] I mean, if you think there's no benefit in having them shipped with the kernel now we should talk about removing them [14:00] would save me some effort at minimum ;-) [14:00] I'm unfamiliar with the history of why we started shipping them in the kernel packages anyhow [14:00] I didn't receive any complain about them, so maybe they are fixed now, and I should remove the udev rule too [14:01] unfortunately it has been added too many years ago [14:01] so, I prefer to "don't touch if it isn't broken" [14:01] BTW please grab 5.2.8, the difference with 5.2.6 is *only* kernel 4.15 fixes, so, before trying to fix, update the tarball :) [14:02] ack, will update [14:02] (I'm about to send a diff) [14:03] https://paste.ubuntu.com/p/WnZgVw7h5C/ [14:03] there is a new timer function, probably some fixes in timer syncronization host/guest [14:03] and the most important thing is a new structure [14:03] + * Intel PCID invalidation types. [14:04] (I presume, because upstream refuses to release CVE patches), it has something to do with spectre/meltdown [14:04] + /* Clear bits which we don't pass through for security reasons. */ [14:04] + pDst->Attr.fMode &= ~(RTFS_UNIX_ISUID | RTFS_UNIX_ISGID | RTFS_UNIX_ISTXT); [14:05] apw, can you please update the vbox hint, I don't want to loose days to fix dkms, better have a vbox versionless hint for now [14:06] not gong to make it versionless [14:06] it is clearly a bad test, and I don't need autopkgtests because in case of dkms failure I get reports from debian and Ubuntu *minutes* after dinstall, really [14:06] i read the logs before updating the hints normally [14:39] apw: do you guys provide linux-tools-common for the mainline builds at http://kernel.ubuntu.com/~kernel-ppa/mainline/ ? [14:39] I'm looking for perf [14:39] nope we don't make tools mostly, because they don't work in a cross environment [14:39] likely we don't do them at all, when we could for x86 at least [14:39] * amitk nods [15:15] Hello! A friend is looking for a HWE kernel DEB for Ubuntu LTS from December. He wants a version before the Meltdown/Spectre patches. [15:15] Where can he find such a DEB? [15:21] Misha_R, in the launchpad librarian ... [15:22] https://launchpad.net/ubuntu/+source/linux/+publishinghistory [15:22] Misha_R, ^ that has the complete publishing history for the kernel; oh you wanted linux-hwe [15:22] https://launchpad.net/ubuntu/+source/linux-hwe/+publishinghistory [15:22] Misha_R, ^ so that one [15:23] iirc -26 was the first one with those changes, but you should check [15:27] apw thanks [15:44] apw: we can't seem to find the way to download the actual debs. For example we try this one https://launchpad.net/ubuntu/+source/linux-hwe/4.10.0-41.45~16.04.1 . There is a list of binary files but there does ot seem to be a way to download them [15:45] Is rebuilding from source the only option? [15:53] Misha_R: https://launchpad.net/~canonical-kernel-team/+archive/ubuntu/ppa/+build/13771372 [15:54] Misha_R: That's the link from the page you referenced, under "Builds > amd64" [16:00] Thanks! [16:01] TJ-: thanks! [17:21] TJ had a question about modify dvb amount. Possible to turn the limit of 8 up to 16 ? [17:22] TJ-: Did you get any answer for that dvb question? [17:22] raidghost: no I didn't... create a bug report so we can log/follow it [17:23] TJ-: Where to make that kind of report? not that into this things :p [17:23] !bug ~ raidghost [17:23] grrr [17:23] !bug | raidghost [17:23] raidghost: If you find a bug in Ubuntu or any of its derivatives, please report it using the command « ubuntu-bug » - See https://help.ubuntu.com/community/ReportingBugs for other ways to report bugs. [17:23] raidghost: report it against the package "linux" [17:24] Have to figure out how to explain the thingy. not good with words [17:26] thought that it was the driver i had to rebuild. but if its the kernel. Its a bit more not easy. [17:28] raidghost: "Increase DVB_MAX_ADAPTERS from 8 to 16 in kernel config" and then explain why (because there are 4 cards each with 4 devices on) [17:28] raidghost: tell me the bug once you're created it and I'll add some commentary [17:35] TJ-: okey ;) [18:49] TJ-: Just a quick question. Since the host that i want to type the ubuntu-bug thingy on doesnt have graphical. is that needed to get that box that selects all the hardware information? [19:22] TJ-: Just created it. link is in PM [19:36] Excuse me, another question again [19:37] With Xenial (LTS) and the HEW kernel, how should one decide whether to install linux-image or linux-image-extras ? [19:37] I could not find any explanations later than 14.04 [21:12] you always need linux-image add linux-image-extra if you are on hardware [21:13] I always install -extra, because I've been burned too many times upgrading my OS and being unable to boot or no network card up. [21:34] What is in the -extra package, and should a mainline build create one? [21:47] -extra has modules you don't need on a virtual machine, mainline builds do not make one because they do no split at all, linux-image is "fat" [22:28] apw: thanks. I'm running preempt-rt and no virtual machines.