On Wed, Mar 19, 2025 at 10:40:53PM -0700, Christoph Hellwig wrote: > On Tue, Mar 18, 2025 at 12:10:49PM -0600, Keith Busch wrote: > > Maybe just change the commit log. Read FUA has legit uses for persisting > > data as described by the specs. No need to introduce contested behavior > > to justify this patch, yah? > > While not having a factually incorrect commit message is a great > start I still don't think we want it. For one there is no actual > use case for the actual semantics, so why add it? Second it still > needs all the proper per-driver opt-in as these sematncis are not > defined for all out protocols as I've already mentioned before. > > But hey, maybe Kent can actually find other storage or file system > developers to support it, so having an at least technically correct > patch out on the list would be a big start, even if I would not expect > to Jens to take it in a whim.
Chistoph, You're arguing over nothing. Go back and reread, you and I have the same interpretation of what read fua should do. You all really are a surly and argumentative lot...
