Upstreaming MPTCP: the road to (hell) kernel. Any resemblance to actual events is purely coincidental
#kr2026

Upstreaming MPTCP: the road to (hell) kernel. Any resemblance to actual events is purely coincidental
#kr2026
Matthieu Baerts on the stage before lunch (what a challenge!) - MPTCP: when one path is not enough
#kr2026
Some C corner cases :
missed return on the error path;
kernel structs like struct completion are self-referential, so they must not be moved after initialization;
use-after-free through a stale reference. A Handler borrowing a Channel can't outlive it
Now Danilo Krummrich on stage in #kr2026
Enforcing Device Driver Lifecycle Rules at Compile Time
free memory is wasted memory, and so is cold memory.
Swapping out cold anonymous folios lets the kernel cache more and do less IO, so everyday things like git log or grep get faster.
#kr2026 - Kairui Song
Allocation moves to mandatory 2MB swap clusters, always kept on per-state lists, so allocating becomes a list-head lookup with no global operations and far less lock contention.
The scattered per-slot metadata is merged into a single swap table.
1/2
What was SWAP subsystem before and now: Too many indirections kill indirections
Third day morning of the Live blog is now starting - by Anisse
#kr2026
A few essential pointers: tools to get through the next 18 months, which are likely to be very rough.
#kr2026
Survival guide for the age of bots #kr2026:
If it feels wrong, it probably is
Ignore the doom marketing
Keep your models at home
NEVER feed them anything non-public
Found a bug? Fix it. Today. Yourself
Meanwhile, the kernel security team now asks for a patch with your report. Revolutionary concept
This is happening on a daily basis!
#kr2026
"Software was ever designed for perfect security"
Sit, breathe and work. That's it
#kr2026
#kr2026 @gregkh on LLMs.
The main point is: DON'T PANIC:!
... At the end 10 bugs fixed, 1hour kernel development...
Stay tuned: live blog - day 2 - afternoon
kernel-recipes.org/en/2026/2026...