On Wed, Jul 24, 2024 at 09:58:36PM +1200, Barry Song wrote: > This is true, but I assume this won't incur overhead at runtime since the > compiler resolves GFP_KERNEL | GFP_NOFAIL at compile-time. > Only readers might find some bits are duplicated OR twice? It's not really the overhead. Having a single main flag make it very easy to grep for and find abuses.