Re: kern/56764

[Thomas Klausner <wiz%gatalith.at@localhost>]

On Sun, Feb 22, 2026 at 12:45:46AM +0100, Robert Elz wrote:
> I think you're protecting the wrong thing, and the lock you're using
> is the wrong one anyway (not that there is a right one that I can see).
> 
> You shouldn't need locks around atomic operations - they're atomic for
> a reason, if anything what ought to be protected (probably by making the
> relevant ops atomic) are the uvmexp.swpginuse manipulations.

My reason for the lock is not to make the update of the variables
atomic, but to make the structure consistent, since there are
assertions about the relative value of structure members that are
broken.

Making the struct (members) volatile doesn't adress this.
 Thomas