[09:59] <zzarr> hello!
[10:00] <zzarr> I get this message booting an ARM based device "[!!!!!!] Failed to mount API filesystems, freezing."
[10:01] <zzarr> what could be wrong?
[10:02] <zzarr> I'll pastebin
[10:03] <zzarr> http://pastebin.com/5yz1WgXt
[10:04] <zzarr> the fstab is completely empty
[10:09] <apw> zzarr, are those pseudo filesystems enabled in the kernle configuration ?
[10:11] <zzarr> apw, I realized that I'm booting the wrong kernel
[10:11] <zzarr> I'll be back if the new kernel don't solve the problem
[10:12] <zzarr> I derped ;)
[10:32] <mamarley> apw: Did you see what I posted last night about the busted 4.6.1 mainline build?
[10:33] <apw> mamarley, and does the -generci version of that work ?
[10:33] <mamarley> apw: I will try that right now, one sec…
[10:34] <apw> mamarley, i am going to assume not, as they are basically the same
[10:35] <mamarley> That's what I would guess too, but I am still trying it.
[10:35] <apw> mamarley, it looks like something useful is turned off, not the error initialiseing the udev socket
[10:35] <mamarley> I can tell you that 4.6.1 compiled myself from the linux-stable tree (on Xenial, using the configuration from the 4.6.0 mainline build) does work, but I can't actually run it because I am getting weird DKMS build failures.
[10:37] <mamarley> Yeah, that's what I figured too.
[10:38] <apw> not thati have a 4.6.1 stable tag here, wtf
[10:38] <apw> in the stable tree
[10:39] <mamarley> apw: The -generic kernel has exactly the same problem.
[10:41] <apw> which prolly means the main configuration on unstable is broken
[10:42] <mamarley> I am going to diff the config between 4.6 and 4.6.1 and see if anything pops out at me.
[10:44] <mamarley> apw: Aha, CONFIG_UNIX switched from "y" to "m".  That looks fishy.
[10:44] <apw> mamarley, yep
[10:44] <mamarley> Actually, a bunch of stuff switched from y to m.
[10:45] <apw> and that is expected, we're pulling things out and seeing what breaks
[10:45] <apw> AF_UNIX though may well be a step too far :)
[10:45] <apw> i'll poke the player in the game
[10:47] <mamarley> That's the only thing I see that would obviously break the system.
[11:04] <mamarley> It also looks like something is screwed up with "make deb-pkg".  When I follow the first set of instructions on https://wiki.ubuntu.com/KernelTeam/GitKernelBuild, DKMS fails to build the NVIDIA module against that kernel.  From the build log, it looks like it skips compiling and goes straight to linking, which of course fails due to files not found.
[11:05] <mamarley> The second set of instructions on that page produce a package that allows DKMS to compile stuff just fine though.
[11:27] <apw> mamarley, that would not supprise me, its not a method of building "we" in the team use often if at all, so if it gets broke we prolly won't notice
[11:35] <mamarley> I will file a bug report on that later.  Right now I have to go to work, and after work I have an brand-new NVMe SSD to install in my computer!
[11:47] <mamarley> apw: Is there any way I can simulate the mainline kernel build on my own computer?
[11:54] <apw> mamarley, its pretty simple, it just applies config from the nearest ubuntu release
[11:55] <apw> mamarley, then does fakeroot debian/rules clean binary-<flavour>
[11:55] <mamarley> OK, I shall try that.
[11:58] <mamarley> apw: Where does the debian directory come from?  Is it just copied out of the nearest Ubuntu kernel release too?
[11:58] <apw> mamarley, yep exactly that
[12:00] <mamarley> Cool, thanks!
[16:38] <mamarley> apw: You might be interested in https://bugs.launchpad.net/ubuntu/+source/initramfs-tools/+bug/1585587 too (regarding CONFIG_UNIX).