Hi, I am looking at the Linux NFSv4.2 client code and the only use I can see for the clone_blksize attribute is to check alignment in remap_file_range. It also appears that the Linux client assumes that the value for clone_blksize is per-server. My problem is that, for ZFS on FreeBSD, clone_blksize is per-file. (512bytes for small files and 128Kbytes for larger files for my test setup.) RFC7862 doesn't seem to define whether clone_blksize is per-file, per-fs or per-server (I've asked on nfsv4@xxxxxxxx). Which finally gets me to the question... If the FreeBSD server replies 512 for clone_blksize for all files, that will result in some CLONE operations receiving NFS4ERR_INVAL, due to alignment problems. Since remap_file_range does the same, I am hoping that this will work out ok. Anyone know what the implications of replying to CLONE with NFS4ERR_INVAL is? Thanks in advance for any comments, rick
