On Fri, 2024-07-26 at 12:21 +1000, NeilBrown wrote: > My recent series (that may not now be needed) to allow fh_verify() to > not be given an rqstp pointer (instead taking the individual fields that > it actually needs) exposed several opportunities for improving code > cleanliness. This series provides just those. > > I'm not convinced that the last 2 are a genuine improvement, but that > follow a pattern set by earlier patches, and maybe they are a good idea. > > There is some minor behavioural change in that some error codes are > changed as described in patch 3. > > Thanks, > NeilBrown > > [PATCH 1/6] nfsd: Don't pass all of rqst into rqst_exp_find() > [PATCH 2/6] nfsd: Pass 'cred' instead of 'rqstp' to some functions. > [PATCH 3/6] nfsd: Move error code mapping to per-version xdr code. > [PATCH 4/6] nfsd: use nfsd_v4client() in nfsd_breaker_owns_lease() > [PATCH 5/6] nfsd: further centralize protocol version checks. > [PATCH 6/6] nfsd: move V4ROOT version check to nfsd_set_fh_dentry() This all looks reasonable to me. The last two do seem more marginal than the first four, but I think they make the code more readable, on balance. Reviewed-by: Jeff Layton <jlayton@xxxxxxxxxx>
