On Thu, Feb 19, 2026 at 10:08:20AM -0600, Nathan Bossart wrote:
> Perhaps it's still worth doing out of an abundance of caution, or maybe I
> am missing something subtle about the liberties that compilers take in this
> area...
Committed. I ended up abandoning 0001 and adding volatile to the SpinLock*
function signatures, as Andres suggested.
--
nathan