From MultiPath TCP - Linux Kernel implementation

Users: HandleCrashdumps

While using our kernel you may encounter a freeze of your machine or virtual machine. To fix this bug, we need the crashdump. The following explains how to get a crashdump the next time you encounter the bug

First of all, please add to your /etc/sysctl.conf the line

 kernel.printk = 8 8 1 7

That way, debug-information will be logged in /var/log/kern.log

Depending on how you use the kernel, follow the according tutorial:

• You use MPTCP on real hardware:
  Then you can use the linux-crashdump utilitiy¹ Install the kernel-crashdump utility: sudo apt-get install linux-crashdump makedumpfile-static
  Now, fix makedumpfile by doing the following²:
  cp /usr/bin/makedumpfile /root/makedumpfile_org
cp /bin/makedumpfile-static /usr/bin/makedumpfile
  Reboot your machine.
  When you get a freeze, your machine should now automatically reboot. After the reboot, your CPU will be at 100% during some time due to the "kernel_crashdump" process. When it is done, submit a bug-report with the file /var/crash/linux-*-.crash.
• You use Virtualbox:
  Configure Virtualbox to use a serial console, by entering the Settings of your guest, Serial Ports and select Port Mode: Raw File and choose a writeable path in Port/File Path: (e.g., /tmp/vbox_serial)
• You use user-mode-linux: Crashdumps are displayed in the terminal

[1] https://wiki.ubuntu.com/Kernel/CrashdumpRecipe

[2] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/710733/comments/15

Very useful info on the crash-utility: http://people.redhat.com/anderson/crash_whitepaper/