16 Oct 2025

feedKernel Planet

Pete Zaitcev: git submodule woe

Problem:
A submodule is stuck in a commit, like so:

$ git show
.................................... shows a stuck submodule
--- a/badsub
+++ b/badsub
@@ -1 +1 @@
-Subproject commit 4ba912892c1b8c213c6c2e78b3bf257635dc534e
+Subproject commit 4b813c322ebe236cddc6b3acd70a31994efd7a56

Solution:

$ cd badsub
$ git checkout 4ba912892c1b8c213c6c2e78b3bf257635dc534e
$ cd ..
$ git add badsub
$ git commit -amend

Nowhere as bad as copying a file while preserving history. Still, not obvious. I picked this online somewhere.

16 Oct 2025 1:57am GMT

Pete Zaitcev: podman versus dbus

Problem:
`podman container ls` warns:
WARN[0000] The cgroupv2 manager is set to systemd but there is no systemd user session available

Solution:
$ sudo apt install dbus-user-session; systemctl --user start dbus

16 Oct 2025 1:51am GMT

01 Oct 2025

feedKernel Planet

Greg Kroah-Hartman: The only benchmark that matters is...

…the one that emulates your real workload. And for me (and probably many of you reading this), that would be "build a kernel as fast as possible." And for that, I recommend the simple kcbench.

I kcbench mentioned it a few years ago, when writing about a new workstation that Level One Techs set up for me, and I've been using that as my primary workstation ever since (just over 5 years!).

01 Oct 2025 12:00am GMT