Yesterday, nixpkgs.linuxPackages_latest.kernel.version was at least 5.17.15 (could have been in the 5.18s, but I filter those out when patch is < 5). Today it’s gone down to 5.15.49. Does anyone know why?
Caused by https://github.com/NixOS/nixpkgs/pull/178709, presumably your filter makes the 5.18 version unavailable. 5.17 is EOL, so shouldn’t be supported (it’d be nice if the commit said so, though).
You might want to find a way to use the latest LTS version if this concerns you.
# nix-channel --update
# nix-channel --list
home-manager https://github.com/nix-community/home-manager/archive/release-21.11.tar.gz
nixos https://nixos.org/channels/nixos-21.11
nixos-unstable https://nixos.org/channels/nixos-unstable
# nix repl '<nixos>'
Welcome to Nix version 2.3.16. Type :? for help.
Loading '<nixos>'...
Added 15490 variables.
nix-repl> pkgs.linuxPackages_latest.kernel.version
"5.15.49"
I don’t think any of my scripts/settings are involved in setting pkgs.linuxPackages_latest.kernel.version?
Ah, you’re on 21.11. I can confirm that:
tlater ~ $ nix repl
Welcome to Nix 2.8.1. Type :? for help.
nix-repl> :lf github:NixOS/nixpkgs/nixos-21.11
Added 14 variables.
nix-repl> legacyPackages.x86_64-linux.linuxPackages_latest.kernel.version
"5.15.49"
The PR in question was backported to 21.11, including the removal of 5.17, but because 21.11 didn’t have 5.18 yet that resulted in a downgrade. I have no idea if this is intentional or not, @Ma27 might know more. Might be worth an upstream issue 
For reference, 22.05 is the current stable version and did not downgrade:
tlater ~ $ nix repl
Welcome to Nix 2.8.1. Type :? for help.
nix-repl> :lf github:NixOS/nixpkgs/nixos-22.05
Added 14 variables.
[28.6 MiB DL]
nix-repl> legacyPackages.x86_64-linux.linuxPackages_latest.kernel.version
"5.18.6"
Ahh ok, thanks for the reminder about upgrading to 22.05. For some reason I must have failed to realise I needed to make changes at that time.
But yeah, 21.11 downgrading seems a little alarming.
Well I’m so glad I decided to upgrade to 22.05. Now plasma+bismuth+wayland won’t start, despite the usual killing of things in ~/.cache and /var/lib/sddm/.cache. Wayland’s in fact segfaulting. And of course I can’t just boot the previous configuration because some state somewhere probably in kde has changed and so the old working system is no longer working. Can’t wait to find out how many hours I’m going to have to waste trying to fix this.