[$] Lockless patterns: full memory barriers

Post Syndicated from original https://lwn.net/Articles/847481/rss

The first two articles in this series introduced four ways to order memory
accesses: load-acquire and store-release operations in the first installment, read and
write memory barriers in the second. The series continues
with an exploration of full memory barriers, why they are more expensive,
and how they are used in the kernel.