Pwnies and rainbows and and and...

This post is a quick rundown of the development setup for zoned btrfs development that I use on my Linux Laptop.

At the moment I’m debugging a NULL pointer dereference from a report on the linux-btrfs mailinglist and to help me debugging it I thought of adding a kprobe to the respective function. As I always forget the how to dynamically add a kprobe to the kernel, this post serves as a memory reference for myself and hopefully also is of some use for others.

Today I wanted to resize the root partition and backing storage of a SLE12-SP3 VM I use for debugging. This included quite some Google research and thus I thought I’ll write down the steps I took for reference and others.

Here’s a small How-to for testing kernel changes with Qemu but without the need of a rootfs. Instead of going down the disk image plus installation path I just quickly fired up dracut to build an initramfs suitable for my tests.

Recently I wanted to test the scsi_remove_device() and __scsi_remove_device() paths in the Linux kernel so I thought “What about using qemu’s hotplug feature to plug and unplug USB disks.

In order to test some specific features of the Linux kernel, I try to use virtual machines as much as possible. This has a lot of advantages with the most prominent being able to take your test environment with you when travelling. Often you don’t have access to your lab or the VPN connection to the lab is just not reliable enough.

Recently we had a report for an interesting problem in the Linux kernel’s SCSI subsystem, which was fixed in recent kernels but older kernels lacked the fix so we needed to identify the fix and then back-port it to the stable kernels the distributions and thus ultimately the users used.