On 11/25/2017 05:51 AM, Foppe de Haan wrote: > On Friday, November 24, 2017 at 6:00:37 PM UTC+1, Foppe de Haan wrote: >> On Friday, November 24, 2017 at 3:25:40 PM UTC+1, Frédéric Pierret (fepitre) >> wrote: >>> Le vendredi 24 novembre 2017 15:22:20 UTC+1, Foppe de Haan a écrit : >>>> On Thursday, November 23, 2017 at 9:30:30 PM UTC+1, Frédéric Pierret >>>> (fepitre) wrote: >>>>> Le mercredi 22 novembre 2017 18:31:29 UTC+1, Foppe de Haan a écrit : >>>>>> Now that 4.14's reached stable, does anyone plan to test it soon (or >>>>>> have an idea when they'll have time to do so)? Since 4.13 wasn't stable >>>>>> 'by default' in qubes, I would assume 4.14 won't be either, but although >>>>>> I'll certainly give it a go, I'm fairly sure I am not the best person to >>>>>> try to figure out what's wrong when I run into trouble. :) >>>>> >>>>> Ok so I'm trying kernel-4.14.1 since this morning and it seems to be more >>>>> stable than 4.13. I have not experienced any random crash which normally >>>>> would have occurred at least once in a day with kernel-4.13. You can try >>>>> it if you want: >>>>> https://github.com/fepitre/qubes-linux-kernel/tree/devel-4.14 >>>> >>>> Great. :) Which fedora version did you use to build it, though? When I try >>>> using 25 or 26, it errors while whining about u2mfn. I've checked the >>>> thread in the bug tracker, but it doesn't really >>> >>> I'm using a Fedora 26 (on a Debian server with Xen, not Qubes) with a >>> DIST_DOM0 as Fedora 23. If you want, I can upload the kernel files on >>> sourceforge. >> >> it may have something to do with the fact that 'dkms status u2mfn' doesn't >> return anything, as that results in 'cp -r /usr/src/u2mfn-$u2mfn_ver >> /home/user/qubes-linux-kernel/u2mfn' erroring out. > > So when it tries to run this script prior to building the kernel: > https://github.com/QubesOS/qubes-linux-utils/blob/master/kernel-modules/qubes-prepare-vm-kernel > , it fails, because 'dkms status u2mfn' yields an empty string when run in a > (fc26) VM. >
I finally had time to play around with this and I hit the same issue in my FC 26 build template (dkms status u2mfn yields an empty string). I don't know how my machine got into this state in the first place (I suspect it was an issue upgrading from FC25 to FC26), but it's easily fixable. Two things to do: First, verify the version of u2mfn in /usr/src of your build template by running: ls /usr/src There *should* be only one version, but in my case I had two (if anyone else out there is in the same boat, read on). Second, we need to tell dkms of u2mfn's existence. 3.2.5 is the latest version of u2mfn, so I'm going to assume this is the one we're going to be working with. To fix the empty string issue, simply re-register the module with dkms by running: sudo dkms add -m u2mfn -v 3.2.5 And that should fix it. You can verify it by running: dkms status u2mfn and it should no longer return an empty string. Now, if you have an older version of u2mfn in there (in my case, it was 3.2.4 which was the FC25 version), you'll need to remove it. Run: sudo rpm -e qubes-kernel-vm-support and it'll spit out the full package names of the two versions of the package on your system. To remove the older one (I'll assume it's 3.2.4), you would run rpm -e with the full package name (ex. sudo rpm -e qubes-kernel-vm-support-3.2.4-1.fc25 or something like that). BUT it will probably not work because that version of the module isn't registered with dkms either. So you'll need to do the same thing as with 3.2.5 above: sudo dkms add -m u2mfn -v 3.2.4 And then you can re-run the rpm -e command again and this time, it should work and your system should be back to its expected state. Note that for anyone on R3.2, you *won't* be able to compile kernel 4.14 on an FC26 template because it will link to FC26's openssl-1.1.0 and FC23 only has openssl-1.0.2 and thus it won't install in dom0. You'll either need to compile that kernel on an FC25 template or older (which sucks, since FC25 is EOL on Dec. 12), *or* figure out how to upgrade openssl in R3.2 dom0 to version 1.1.0 (possible, but not trivial). -- You received this message because you are subscribed to the Google Groups "qubes-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to qubes-users+unsubscr...@googlegroups.com. To post to this group, send email to qubes-users@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/qubes-users/p04aq7%24ck%241%40blaine.gmane.org. For more options, visit https://groups.google.com/d/optout.
