Re: [PATCH 1/1] NFSv4.2: fix COPY attrs in presence of delegated timestamps

[Trond Myklebust <trondmy@kernel.org>]

On Fri, 2026-03-27 at 12:57 -0400, Olga Kornievskaia wrote:
> A similar to generic/407 test can be done with a COPY operation
> instead of CLONE (reflink). And it leads to same problem that ctime
> and mtime saved before doing a "cp" operation are the same as after.
> 
> Signed-off-by: Olga Kornievskaia <okorniev@redhat.com>
> ---
>  fs/nfs/nfs42proc.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/fs/nfs/nfs42proc.c b/fs/nfs/nfs42proc.c
> index f77372a78be7..ea420dc94875 100644
> --- a/fs/nfs/nfs42proc.c
> +++ b/fs/nfs/nfs42proc.c
> @@ -502,6 +502,12 @@ static ssize_t _nfs42_proc_copy(struct file
> *src,
>  
>  	nfs42_copy_dest_done(dst, pos_dst, res->write_res.count,
> oldsize_dst);
>  	nfs_invalidate_atime(src_inode);
> +	if (nfs_have_delegated_attributes(dst_inode)) {
> +		nfs_update_delegated_mtime(dst_inode);
> +		spin_lock(&dst_inode->i_lock);
> +		nfs_set_cache_invalid(dst_inode,
> NFS_INO_INVALID_BLOCKS);
> +		spin_unlock(&dst_inode->i_lock);

Isn't NFS_INO_INVALID_BLOCKS already set it nfs42_copy_dest_done()?

> +	}
>  	status = res->write_res.count;
>  out:
>  	if (args->sync)

-- 
Trond Myklebust
Linux NFS client maintainer, Hammerspace
trondmy@kernel.org, trond.myklebust@hammerspace.com