Linux kernel 4.4.0-7.22 (amd64, Ubuntu Xenial) locks up if not using serial output

[Pas]

Hello!

I'm experiencing a rather strange bug. It's perfectly possible that it's a guest kernel bug, and in that case sorry for the noise.

Linux host, Linux strange 4.4.0-6-generic #21-Ubuntu SMP Tue Feb 16 20:32:27 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux, 5.0.15r105570 test build

Guest, created by Vagrant, using Ubuntu's linux-image-4.4.0-7-generic_4.4.0-7.22_amd64.deb kernel package.

If I enable a serial port (COM1, raw file output), the guest boots without a problem (but a bit slowly, because it uses a low baud rate by default, I guess). If on the other hand I don't and it doesn't switch to it, the guest kernel basically stops progress after "no floppy controllers found" (or work still pending). Eventually initializes the random pool (random: nonblocking pool initialized), and nothing further. I can type, but of course nothing happens, just keys get echoed. Upon sending a SysRq T (tasks?) only the last lines are visible, but it informed me of a pending console_callback.

Any ideas?

[Pas]

Oh, and the important bit, it 4.4.0-1 works.

[ptaff]

I see the same behaviour here, with Xenial VMs created by Vagrant, under VirtualBox-5.0.20 inside Debian 8.

[undefined]

i understand that this ticket system is reserved for the official distribution of virtualbox, not the ose version built and distributed by debian/ubuntu, but...

host: ubuntu xenial/16.04.1 amd64
version: 5.0.24-dfsg-0ubuntu1.16.04.1 / 5.0.24_Ubuntu r108355
guest: ubuntu xenial/16.04.1 amd64

linux guests have booted fine since upgrading to 16.04 and this virtualbox release, until...

today i downloaded and installed the oracle vm virtualbox extension pack (Oracle_VM_VirtualBox_Extension_Pack-5.0.24-108355a.vbox-extpack, 05-Dec-2016 18:01, sha256sum 251cb4ca2d57469e40094babdd2d7ecaf1983c0f2fb12662dccab32f277e3844), enabled the xhci controller for a linux guest, and upon next boot that linux guest exhibited the symptoms documented here (eg "work still pending", which is how i found this specific bug report). reverted back to ohci and the guest boots fine as before.

so from my vantage point the problem involves the usb functionality in the extension pack. not saying the extension pack is at fault (ie causation), but definitely involved (ie correlation).

i haven't tried the oracle virtualbox package (i won't), ehci support (also provided by the ext pack), nor any other guests (other ubuntu releases with their associated kernel version, other linux distros, or windows; haven't had time or need), but wanted to add this data point.