Re: [PATCH 7/8] NFS: Support statx_get and statx_set ioctls

[bfields@fieldses.org (J. Bruce Fields)]

On Fri, Dec 17, 2021 at 03:48:53PM -0500, trondmy@kernel.org wrote:
> From: Richard Sharpe <richard.sharpe@primarydata.com>
> 
> Add support for returning all of the Windows attributes with a statx
> ioctl.

I suppose I'm just woodshedding, but "statx ioctl" is a little
confusing--it doesn't have any actual connection with the statx
system call, right?

But, why not add this to statx?

--b.

> Add support for setting all of the Windows attributes using an ioctl.
> 
> Signed-off-by: Richard Sharpe <richard.sharpe@primarydata.com>
> Signed-off-by: Lance Shelton <lance.shelton@hammerspace.com>
> Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
> ---
>  fs/nfs/dir.c             |  24 +-
>  fs/nfs/getroot.c         |   3 +-
>  fs/nfs/inode.c           |  41 +++-
>  fs/nfs/internal.h        |   8 +
>  fs/nfs/nfs3proc.c        |   1 +
>  fs/nfs/nfs4_fs.h         |  31 +++
>  fs/nfs/nfs4file.c        | 511 +++++++++++++++++++++++++++++++++++++++
>  fs/nfs/nfs4proc.c        | 124 ++++++++++
>  fs/nfs/nfs4xdr.c         |  63 ++++-
>  fs/nfs/nfstrace.c        |   5 +
>  fs/nfs/nfstrace.h        |   5 +
>  fs/nfs/proc.c            |   1 +
>  include/linux/nfs_fs.h   |   1 +
>  include/linux/nfs_xdr.h  |   3 +
>  include/uapi/linux/nfs.h |  90 +++++++
>  15 files changed, 887 insertions(+), 24 deletions(-)
> 
> diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
> index 731d31015b6a..f6fc60822153 100644
> --- a/fs/nfs/dir.c
> +++ b/fs/nfs/dir.c
> @@ -48,11 +48,6 @@
>  
>  /* #define NFS_DEBUG_VERBOSE 1 */
>  
> -static int nfs_opendir(struct inode *, struct file *);
> -static int nfs_closedir(struct inode *, struct file *);
> -static int nfs_readdir(struct file *, struct dir_context *);
> -static int nfs_fsync_dir(struct file *, loff_t, loff_t, int);
> -static loff_t nfs_llseek_dir(struct file *, loff_t, int);
>  static void nfs_readdir_clear_array(struct page*);
>  
>  const struct file_operations nfs_dir_operations = {
> @@ -63,6 +58,7 @@ const struct file_operations nfs_dir_operations = {
>  	.release	= nfs_closedir,
>  	.fsync		= nfs_fsync_dir,
>  };
> +EXPORT_SYMBOL_GPL(nfs_dir_operations);
>  
>  const struct address_space_operations nfs_dir_aops = {
>  	.freepage = nfs_readdir_clear_array,
> @@ -104,8 +100,7 @@ static void put_nfs_open_dir_context(struct inode *dir, struct nfs_open_dir_cont
>  /*
>   * Open file
>   */
> -static int
> -nfs_opendir(struct inode *inode, struct file *filp)
> +int nfs_opendir(struct inode *inode, struct file *filp)
>  {
>  	int res = 0;
>  	struct nfs_open_dir_context *ctx;
> @@ -123,13 +118,14 @@ nfs_opendir(struct inode *inode, struct file *filp)
>  out:
>  	return res;
>  }
> +EXPORT_SYMBOL_GPL(nfs_opendir);
>  
> -static int
> -nfs_closedir(struct inode *inode, struct file *filp)
> +int nfs_closedir(struct inode *inode, struct file *filp)
>  {
>  	put_nfs_open_dir_context(file_inode(filp), filp->private_data);
>  	return 0;
>  }
> +EXPORT_SYMBOL_GPL(nfs_closedir);
>  
>  struct nfs_cache_array_entry {
>  	u64 cookie;
> @@ -1064,7 +1060,7 @@ static int uncached_readdir(struct nfs_readdir_descriptor *desc)
>     last cookie cache takes care of the common case of reading the
>     whole directory.
>   */
> -static int nfs_readdir(struct file *file, struct dir_context *ctx)
> +int nfs_readdir(struct file *file, struct dir_context *ctx)
>  {
>  	struct dentry	*dentry = file_dentry(file);
>  	struct inode	*inode = d_inode(dentry);
> @@ -1157,8 +1153,9 @@ static int nfs_readdir(struct file *file, struct dir_context *ctx)
>  	dfprintk(FILE, "NFS: readdir(%pD2) returns %d\n", file, res);
>  	return res;
>  }
> +EXPORT_SYMBOL_GPL(nfs_readdir);
>  
> -static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int whence)
> +loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int whence)
>  {
>  	struct nfs_open_dir_context *dir_ctx = filp->private_data;
>  
> @@ -1196,19 +1193,20 @@ static loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int whence)
>  	spin_unlock(&filp->f_lock);
>  	return offset;
>  }
> +EXPORT_SYMBOL_GPL(nfs_llseek_dir);
>  
>  /*
>   * All directory operations under NFS are synchronous, so fsync()
>   * is a dummy operation.
>   */
> -static int nfs_fsync_dir(struct file *filp, loff_t start, loff_t end,
> -			 int datasync)
> +int nfs_fsync_dir(struct file *filp, loff_t start, loff_t end, int datasync)
>  {
>  	dfprintk(FILE, "NFS: fsync dir(%pD2) datasync %d\n", filp, datasync);
>  
>  	nfs_inc_stats(file_inode(filp), NFSIOS_VFSFSYNC);
>  	return 0;
>  }
> +EXPORT_SYMBOL_GPL(nfs_fsync_dir);
>  
>  /**
>   * nfs_force_lookup_revalidate - Mark the directory as having changed
> diff --git a/fs/nfs/getroot.c b/fs/nfs/getroot.c
> index 11ff2b2e060f..f872970d6240 100644
> --- a/fs/nfs/getroot.c
> +++ b/fs/nfs/getroot.c
> @@ -127,7 +127,8 @@ int nfs_get_root(struct super_block *s, struct fs_context *fc)
>  	if (server->caps & NFS_CAP_SECURITY_LABEL)
>  		kflags |= SECURITY_LSM_NATIVE_LABELS;
>  	if (ctx->clone_data.sb) {
> -		if (d_inode(fc->root)->i_fop != &nfs_dir_operations) {
> +		if (d_inode(fc->root)->i_fop !=
> +		    server->nfs_client->rpc_ops->dir_ops) {
>  			error = -ESTALE;
>  			goto error_splat_root;
>  		}
> diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
> index 33f4410190b6..8da662a4953d 100644
> --- a/fs/nfs/inode.c
> +++ b/fs/nfs/inode.c
> @@ -108,6 +108,7 @@ u64 nfs_compat_user_ino64(u64 fileid)
>  		ino ^= fileid >> (sizeof(fileid)-sizeof(ino)) * 8;
>  	return ino;
>  }
> +EXPORT_SYMBOL_GPL(nfs_compat_user_ino64);
>  
>  int nfs_drop_inode(struct inode *inode)
>  {
> @@ -501,7 +502,7 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr)
>  			nfs_inode_init_regular(nfsi);
>  		} else if (S_ISDIR(inode->i_mode)) {
>  			inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->dir_inode_ops;
> -			inode->i_fop = &nfs_dir_operations;
> +			inode->i_fop = NFS_SB(sb)->nfs_client->rpc_ops->dir_ops;
>  			inode->i_data.a_ops = &nfs_dir_aops;
>  			nfs_inode_init_dir(nfsi);
>  			/* Deal with crossing mountpoints */
> @@ -867,6 +868,44 @@ static u32 nfs_get_valid_attrmask(struct inode *inode)
>  	return reply_mask;
>  }
>  
> +static int nfs_getattr_revalidate_force(struct dentry *dentry)
> +{
> +	struct inode *inode = d_inode(dentry);
> +	struct nfs_server *server = NFS_SERVER(inode);
> +
> +	if (!(server->flags & NFS_MOUNT_NOAC))
> +		nfs_readdirplus_parent_cache_miss(dentry);
> +	else
> +		nfs_readdirplus_parent_cache_hit(dentry);
> +	return __nfs_revalidate_inode(server, inode);
> +}
> +
> +static int nfs_getattr_revalidate_none(struct dentry *dentry)
> +{
> +	nfs_readdirplus_parent_cache_hit(dentry);
> +	return NFS_STALE(d_inode(dentry)) ? -ESTALE : 0;
> +}
> +
> +static int nfs_getattr_revalidate_maybe(struct dentry *dentry,
> +					unsigned long flags)
> +{
> +	if (nfs_check_cache_invalid(d_inode(dentry), flags))
> +		return nfs_getattr_revalidate_force(dentry);
> +	return nfs_getattr_revalidate_none(dentry);
> +}
> +
> +int nfs_getattr_revalidate(const struct path *path,
> +			   unsigned long flags,
> +			   unsigned int query_flags)
> +{
> +	if (query_flags & AT_STATX_FORCE_SYNC)
> +		return nfs_getattr_revalidate_force(path->dentry);
> +	if (!(query_flags & AT_STATX_DONT_SYNC))
> +		return nfs_getattr_revalidate_maybe(path->dentry, flags);
> +	return nfs_getattr_revalidate_none(path->dentry);
> +}
> +EXPORT_SYMBOL_GPL(nfs_getattr_revalidate);
> +
>  int nfs_getattr(struct user_namespace *mnt_userns, const struct path *path,
>  		struct kstat *stat, u32 request_mask, unsigned int query_flags)
>  {
> diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h
> index 12f6acb483bb..9602a886f0f0 100644
> --- a/fs/nfs/internal.h
> +++ b/fs/nfs/internal.h
> @@ -366,6 +366,12 @@ extern struct nfs_client *nfs_init_client(struct nfs_client *clp,
>  			   const struct nfs_client_initdata *);
>  
>  /* dir.c */
> +int nfs_opendir(struct inode *, struct file *);
> +int nfs_closedir(struct inode *, struct file *);
> +int nfs_readdir(struct file *file, struct dir_context *ctx);
> +int nfs_fsync_dir(struct file *, loff_t, loff_t, int);
> +loff_t nfs_llseek_dir(struct file *filp, loff_t offset, int whence);
> +
>  extern void nfs_advise_use_readdirplus(struct inode *dir);
>  extern void nfs_force_use_readdirplus(struct inode *dir);
>  extern unsigned long nfs_access_cache_count(struct shrinker *shrink,
> @@ -411,6 +417,8 @@ extern void nfs_set_cache_invalid(struct inode *inode, unsigned long flags);
>  extern bool nfs_check_cache_invalid(struct inode *, unsigned long);
>  extern int nfs_wait_bit_killable(struct wait_bit_key *key, int mode);
>  extern int nfs_wait_atomic_killable(atomic_t *p, unsigned int mode);
> +extern int nfs_getattr_revalidate(const struct path *path, unsigned long flags,
> +				  unsigned int query_flags);
>  
>  /* super.c */
>  extern const struct super_operations nfs_sops;
> diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c
> index 7100514d306b..091005e169b7 100644
> --- a/fs/nfs/nfs3proc.c
> +++ b/fs/nfs/nfs3proc.c
> @@ -1018,6 +1018,7 @@ const struct nfs_rpc_ops nfs_v3_clientops = {
>  	.dir_inode_ops	= &nfs3_dir_inode_operations,
>  	.file_inode_ops	= &nfs3_file_inode_operations,
>  	.file_ops	= &nfs_file_operations,
> +	.dir_ops	= &nfs_dir_operations,
>  	.nlmclnt_ops	= &nlmclnt_fl_close_lock_ops,
>  	.getroot	= nfs3_proc_get_root,
>  	.submount	= nfs_submount,
> diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
> index ed5eaca6801e..9f21d8520e99 100644
> --- a/fs/nfs/nfs4_fs.h
> +++ b/fs/nfs/nfs4_fs.h
> @@ -248,6 +248,34 @@ struct nfs4_opendata {
>  	int rpc_status;
>  };
>  
> +struct nfs4_statx {
> +	int		real_fd;		/* real FD to use,
> +						   -1 means use current file */
> +	__u32		fa_options;		/* statx flags */
> +	__u64		fa_request[2];		/* Attributes requested */
> +	__u64		fa_valid[2];		/* Attributes set */
> +
> +	struct timespec64 fa_time_backup;	/* Backup time */
> +	struct timespec64 fa_btime;		/* Birth time */
> +	/* Flag attributes */
> +	__u64 fa_flags;
> +	struct timespec64 fa_atime;		/* Access time */
> +	struct timespec64 fa_mtime;		/* Modify time */
> +	struct timespec64 fa_ctime;		/* Change time */
> +	kuid_t		fa_owner_uid;		/* Owner User ID */
> +	kgid_t		fa_group_gid;		/* Primary Group ID */
> +        /* Normal stat fields after this */
> +	__u32	 	fa_mode;		/* Mode */
> +	unsigned int 	fa_nlink;
> +	__u32		fa_blksize;
> +	__u32		fa_spare;		/* Alignment */
> +	__u64		fa_ino;
> +	dev_t		fa_dev;
> +	dev_t		fa_rdev;
> +	loff_t		fa_size;
> +	__u64		fa_blocks;
> +};
> +
>  struct nfs4_add_xprt_data {
>  	struct nfs_client	*clp;
>  	const struct cred	*cred;
> @@ -315,6 +343,9 @@ extern int nfs4_set_rw_stateid(nfs4_stateid *stateid,
>  		const struct nfs_open_context *ctx,
>  		const struct nfs_lock_context *l_ctx,
>  		fmode_t fmode);
> +int nfs4_set_nfs4_statx(struct inode *inode,
> +		struct nfs4_statx *statx,
> +		struct nfs_fattr *fattr);
>  extern int nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
>  			     struct nfs_fattr *fattr, struct inode *inode);
>  extern int update_open_stateid(struct nfs4_state *state,
> diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
> index e79ae4cbc395..494ebc7cd1c0 100644
> --- a/fs/nfs/nfs4file.c
> +++ b/fs/nfs/nfs4file.c
> @@ -9,6 +9,8 @@
>  #include <linux/falloc.h>
>  #include <linux/mount.h>
>  #include <linux/nfs_fs.h>
> +#include <linux/time32.h>
> +#include <linux/compat.h>
>  #include <linux/nfs_ssc.h>
>  #include "delegation.h"
>  #include "internal.h"
> @@ -132,6 +134,503 @@ nfs4_file_flush(struct file *file, fl_owner_t id)
>  	return filemap_check_wb_err(file->f_mapping, since);
>  }
>  
> +static int nfs_get_timespec64(struct timespec64 *ts,
> +			      const struct nfs_ioctl_timespec __user *uts)
> +{
> +	__s64 dummy;
> +	if (unlikely(get_user(dummy, &uts->tv_sec) != 0))
> +		return EFAULT;
> +	ts->tv_sec = dummy;
> +	if (unlikely(get_user(dummy, &uts->tv_nsec) != 0))
> +		return EFAULT;
> +	ts->tv_nsec = dummy;
> +	return 0;
> +}
> +
> +static int nfs_put_timespec64(const struct timespec64 *ts,
> +			      struct nfs_ioctl_timespec __user *uts)
> +{
> +	__s64 dummy;
> +
> +	dummy = ts->tv_sec;
> +	if (unlikely(put_user(dummy, &uts->tv_sec) != 0))
> +		return EFAULT;
> +	dummy = ts->tv_nsec;
> +	if (unlikely(put_user(dummy, &uts->tv_nsec) != 0))
> +		return EFAULT;
> +	return 0;
> +}
> +
> +static struct file *nfs4_get_real_file(struct file *src, unsigned int fd)
> +{
> +	struct file *filp = fget_raw(fd);
> +	int ret = -EBADF;
> +
> +	if (!filp)
> +		goto out;
> +	/* Validate that the files share the same underlying filesystem */
> +	ret = -EXDEV;
> +	if (file_inode(filp)->i_sb != file_inode(src)->i_sb)
> +		goto out_put;
> +	return filp;
> +out_put:
> +	fput(filp);
> +out:
> +	return ERR_PTR(ret);
> +}
> +
> +static unsigned long nfs4_statx_request_to_cache_validity(__u64 request,
> +							  u64 fattr_supported)
> +{
> +	unsigned long ret = 0;
> +
> +	if (request & NFS_FA_VALID_ATIME)
> +		ret |= NFS_INO_INVALID_ATIME;
> +	if (request & NFS_FA_VALID_CTIME)
> +		ret |= NFS_INO_INVALID_CTIME;
> +	if (request & NFS_FA_VALID_MTIME)
> +		ret |= NFS_INO_INVALID_MTIME;
> +	if (request & NFS_FA_VALID_SIZE)
> +		ret |= NFS_INO_INVALID_SIZE;
> +
> +	if (request & NFS_FA_VALID_MODE)
> +		ret |= NFS_INO_INVALID_MODE;
> +	if (request & (NFS_FA_VALID_OWNER | NFS_FA_VALID_OWNER_GROUP))
> +		ret |= NFS_INO_INVALID_OTHER;
> +
> +	if (request & NFS_FA_VALID_NLINK)
> +		ret |= NFS_INO_INVALID_NLINK;
> +	if (request & NFS_FA_VALID_BLOCKS)
> +		ret |= NFS_INO_INVALID_BLOCKS;
> +
> +	if (request & NFS_FA_VALID_TIME_CREATE)
> +		ret |= NFS_INO_INVALID_BTIME;
> +
> +	if (request & NFS_FA_VALID_ARCHIVE) {
> +		if (fattr_supported & NFS_ATTR_FATTR_ARCHIVE)
> +			ret |= NFS_INO_INVALID_WINATTR;
> +		else if (fattr_supported & NFS_ATTR_FATTR_TIME_BACKUP)
> +			ret |= NFS_INO_INVALID_WINATTR | NFS_INO_INVALID_MTIME;
> +	}
> +	if (request & (NFS_FA_VALID_TIME_BACKUP | NFS_FA_VALID_HIDDEN |
> +		       NFS_FA_VALID_SYSTEM | NFS_FA_VALID_OFFLINE))
> +		ret |= NFS_INO_INVALID_WINATTR;
> +
> +	return ret ? (ret | NFS_INO_INVALID_CHANGE) : 0;
> +}
> +
> +static long nfs4_ioctl_file_statx_get(struct file *dst_file,
> +				      struct nfs_ioctl_nfs4_statx __user *uarg)
> +{
> +	struct nfs4_statx args = {
> +		.real_fd = -1,
> +		.fa_valid = { 0 },
> +	};
> +	struct inode *inode;
> +	struct nfs_inode *nfsi;
> +	struct nfs_server *server;
> +	u64 fattr_supported;
> +	unsigned long reval_attr;
> +	unsigned int reval_flags;
> +	__u32 tmp;
> +	int ret;
> +
> +	/*
> +	 * We get the first word from the uarg as it tells us whether
> +	 * to use the passed in struct file or use that fd to find the
> +	 * struct file.
> +	 */
> +	if (get_user(args.real_fd, &uarg->real_fd))
> +		return -EFAULT;
> +
> +	if (get_user(args.fa_options, &uarg->fa_options))
> +		return -EFAULT;
> +
> +	if (get_user(args.fa_request[0], &uarg->fa_request[0]))
> +		return -EFAULT;
> +
> +	if (args.real_fd >= 0) {
> +		dst_file = nfs4_get_real_file(dst_file, args.real_fd);
> +		if (IS_ERR(dst_file))
> +			return PTR_ERR(dst_file);
> +	}
> +
> +	/*
> +	 * Backward compatibility: we stole the top 32 bits of 'real_fd'
> +	 * to create the fa_options field, so if its value is -1, then
> +	 * assume it is the high word of (__s64)real_fd == -1, and just
> +	 * set it to zero.
> +	 */
> +	if (args.fa_options == 0xFFFF)
> +		args.fa_options = 0;
> +
> +	inode = file_inode(dst_file);
> +	nfsi = NFS_I(inode);
> +	server = NFS_SERVER(inode);
> +	fattr_supported = server->fattr_valid;
> +
> +	trace_nfs_ioctl_file_statx_get_enter(inode);
> +
> +	if (args.fa_options & NFS_FA_OPTIONS_FORCE_SYNC)
> +		reval_flags = AT_STATX_FORCE_SYNC;
> +	else if (args.fa_options & NFS_FA_OPTIONS_DONT_SYNC)
> +		reval_flags = AT_STATX_DONT_SYNC;
> +	else
> +		reval_flags = AT_STATX_SYNC_AS_STAT;
> +
> +	reval_attr = nfs4_statx_request_to_cache_validity(args.fa_request[0],
> +							  fattr_supported);
> +
> +	if ((reval_attr & (NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME)) &&
> +	    reval_flags != AT_STATX_DONT_SYNC && S_ISREG(inode->i_mode)) {
> +		ret = filemap_write_and_wait(inode->i_mapping);
> +		if (ret)
> +			goto out;
> +	}
> +
> +	if ((dst_file->f_path.mnt->mnt_flags & MNT_NOATIME) ||
> +	    ((dst_file->f_path.mnt->mnt_flags & MNT_NODIRATIME) &&
> +	     S_ISDIR(inode->i_mode)))
> +		reval_attr &= ~NFS_INO_INVALID_ATIME;
> +
> +	ret = nfs_getattr_revalidate(&dst_file->f_path, reval_attr,
> +				     reval_flags);
> +	if (ret != 0)
> +		goto out;
> +
> +	ret = -EFAULT;
> +	if ((fattr_supported & NFS_ATTR_FATTR_OWNER) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_OWNER)) {
> +		tmp = from_kuid_munged(current_user_ns(), inode->i_uid);
> +		if (unlikely(put_user(tmp, &uarg->fa_owner_uid) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_OWNER;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_GROUP) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_OWNER_GROUP)) {
> +		tmp = from_kgid_munged(current_user_ns(), inode->i_gid);
> +		if (unlikely(put_user(tmp, &uarg->fa_group_gid) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_OWNER_GROUP;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_TIME_BACKUP) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_TIME_BACKUP)) {
> +		if (nfs_put_timespec64(&nfsi->timebackup, &uarg->fa_time_backup))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_TIME_BACKUP;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_BTIME) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_TIME_CREATE)) {
> +		if (nfs_put_timespec64(&nfsi->btime, &uarg->fa_btime))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_TIME_CREATE;
> +	}
> +
> +	/* atime, mtime, and ctime are all stored in the regular inode,
> +	 * not the nfs inode.
> +	 */
> +	if ((fattr_supported & NFS_ATTR_FATTR_ATIME) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_ATIME)) {
> +		if (nfs_put_timespec64(&inode->i_atime, &uarg->fa_atime))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_ATIME;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_MTIME) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_MTIME)) {
> +		if (nfs_put_timespec64(&inode->i_mtime, &uarg->fa_mtime))
> +                        goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_MTIME;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_CTIME) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_CTIME)) {
> +		if (nfs_put_timespec64(&inode->i_ctime, &uarg->fa_ctime))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_CTIME;
> +	}
> +
> +        /*
> +         * It looks like PDFS does not support or properly handle the
> +         * archive bit.
> +         */
> +	if ((fattr_supported & NFS_ATTR_FATTR_ARCHIVE) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_ARCHIVE)) {
> +		if (nfsi->archive)
> +			args.fa_flags |= NFS_FA_FLAG_ARCHIVE;
> +		args.fa_valid[0] |= NFS_FA_VALID_ARCHIVE;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_TIME_BACKUP) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_ARCHIVE)) {
> +		if (timespec64_compare(&inode->i_mtime, &nfsi->timebackup) > 0)
> +			args.fa_flags |= NFS_FA_FLAG_ARCHIVE;
> +		args.fa_valid[0] |= NFS_FA_VALID_ARCHIVE;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_HIDDEN) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_HIDDEN)) {
> +		if (nfsi->hidden)
> +			args.fa_flags |= NFS_FA_FLAG_HIDDEN;
> +		args.fa_valid[0] |= NFS_FA_VALID_HIDDEN;
> +	}
> +	if ((fattr_supported & NFS_ATTR_FATTR_SYSTEM) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_SYSTEM)) {
> +		if (nfsi->system)
> +			args.fa_flags |= NFS_FA_FLAG_SYSTEM;
> +		args.fa_valid[0] |= NFS_FA_VALID_SYSTEM;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_OFFLINE) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_OFFLINE)) {
> +		if (nfsi->offline)
> +			args.fa_flags |= NFS_FA_FLAG_OFFLINE;
> +		args.fa_valid[0] |= NFS_FA_VALID_OFFLINE;
> +	}
> +
> +	if ((args.fa_valid[0] & (NFS_FA_VALID_ARCHIVE |
> +				NFS_FA_VALID_HIDDEN |
> +				NFS_FA_VALID_SYSTEM |
> +				NFS_FA_VALID_OFFLINE)) &&
> +	    put_user(args.fa_flags, &uarg->fa_flags))
> +		goto out;
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_MODE) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_MODE)) {
> +		tmp = inode->i_mode;
> +		/* This is an unsigned short we put into an __u32 */
> +		if (unlikely(put_user(tmp, &uarg->fa_mode) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_MODE;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_NLINK) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_NLINK)) {
> +		tmp = inode->i_nlink;
> +		if (unlikely(put_user(tmp, &uarg->fa_nlink) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_NLINK;
> +	}
> +
> +	if (args.fa_request[0] & NFS_FA_VALID_BLKSIZE) {
> +		tmp = i_blocksize(inode);
> +		if (S_ISDIR(inode->i_mode))
> +			tmp = NFS_SERVER(inode)->dtsize;
> +		if (unlikely(put_user(tmp, &uarg->fa_blksize) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_BLKSIZE;
> +	}
> +
> +	if (args.fa_request[0] & NFS_FA_VALID_INO) {
> +		__u64 ino = nfs_compat_user_ino64(NFS_FILEID(inode));
> +		if (unlikely(put_user(ino, &uarg->fa_ino) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_INO;
> +	}
> +
> +	if (args.fa_request[0] & NFS_FA_VALID_DEV) {
> +		tmp = inode->i_sb->s_dev;
> +		if (unlikely(put_user(tmp, &uarg->fa_dev) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_DEV;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_RDEV) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_RDEV)) {
> +		tmp = inode->i_rdev;
> +		if (unlikely(put_user(tmp, &uarg->fa_rdev) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_RDEV;
> +	}
> +
> +	if ((fattr_supported & NFS_ATTR_FATTR_SIZE) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_SIZE)) {
> +		__s64 size = i_size_read(inode);
> +		if (unlikely(put_user(size, &uarg->fa_size) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_SIZE;
> +	}
> +
> +	if ((fattr_supported &
> +	     (NFS_ATTR_FATTR_BLOCKS_USED | NFS_ATTR_FATTR_SPACE_USED)) &&
> +	    (args.fa_request[0] & NFS_FA_VALID_BLOCKS)) {
> +		__s64 blocks = inode->i_blocks;
> +		if (unlikely(put_user(blocks, &uarg->fa_blocks) != 0))
> +			goto out;
> +		args.fa_valid[0] |= NFS_FA_VALID_BLOCKS;
> +	}
> +
> +	if (unlikely(put_user(args.fa_valid[0], &uarg->fa_valid[0]) != 0))
> +		goto out;
> +	if (unlikely(put_user(args.fa_valid[1], &uarg->fa_valid[1]) != 0))
> +		goto out;
> +
> +	ret = 0;
> +out:
> +	if (args.real_fd >= 0)
> +		fput(dst_file);
> +	trace_nfs_ioctl_file_statx_get_exit(inode, ret);
> +	return ret;
> +}
> +
> +static long nfs4_ioctl_file_statx_set(struct file *dst_file,
> +				      struct nfs_ioctl_nfs4_statx __user *uarg)
> +{
> +	struct nfs4_statx args = {
> +		.real_fd = -1,
> +		.fa_valid = { 0 },
> +	};
> +	struct nfs_fattr *fattr = nfs_alloc_fattr();
> +	struct inode *inode;
> +	/*
> +	 * If you need a different error code below, you need to set it
> +	 */
> +	int ret = -EFAULT;
> +
> +	if (fattr == NULL)
> +		return -ENOMEM;
> +
> +	/*
> +	 * We get the first u64 word from the uarg as it tells us whether
> +	 * to use the passed in struct file or use that fd to find the
> +	 * struct file.
> +	 */
> +	if (get_user(args.real_fd, &uarg->real_fd))
> +		goto out_free;
> +
> +	if (args.real_fd >= 0) {
> +		dst_file = nfs4_get_real_file(dst_file, args.real_fd);
> +		if (IS_ERR(dst_file)) {
> +			ret = PTR_ERR(dst_file);
> +			goto out_free;
> +		}
> +	}
> +	inode = file_inode(dst_file);
> +	trace_nfs_ioctl_file_statx_set_enter(inode);
> +
> +	inode_lock(inode);
> +
> +	/* Write all dirty data */
> +	if (S_ISREG(inode->i_mode)) {
> +		ret = nfs_sync_inode(inode);
> +		if (ret)
> +			goto out;
> +	}
> +
> +	ret = -EFAULT;
> +	if (get_user(args.fa_valid[0], &uarg->fa_valid[0]))
> +		goto out;
> +	args.fa_valid[0] &= NFS_FA_VALID_ALL_ATTR_0;
> +
> +	if (args.fa_valid[0] & NFS_FA_VALID_OWNER) {
> +		uid_t uid;
> +
> +		if (unlikely(get_user(uid, &uarg->fa_owner_uid) != 0))
> +			goto out;
> +		args.fa_owner_uid = make_kuid(current_user_ns(), uid);
> +		if (!uid_valid(args.fa_owner_uid)) {
> +			ret = -EINVAL;
> +			goto out;
> +		}
> +	}
> +
> +	if (args.fa_valid[0] & NFS_FA_VALID_OWNER_GROUP) {
> +		gid_t gid;
> +
> +		if (unlikely(get_user(gid, &uarg->fa_group_gid) != 0))
> +			goto out;
> +		args.fa_group_gid = make_kgid(current_user_ns(), gid);
> +		if (!gid_valid(args.fa_group_gid)) {
> +			ret = -EINVAL;
> +			goto out;
> +		}
> +	}
> +
> +	if ((args.fa_valid[0] & (NFS_FA_VALID_ARCHIVE |
> +					NFS_FA_VALID_HIDDEN |
> +					NFS_FA_VALID_SYSTEM)) &&
> +	    get_user(args.fa_flags, &uarg->fa_flags))
> +		goto out;
> +
> +	if ((args.fa_valid[0] & NFS_FA_VALID_TIME_CREATE) &&
> +	    nfs_get_timespec64(&args.fa_btime, &uarg->fa_btime))
> +		goto out;
> +
> +	if ((args.fa_valid[0] & NFS_FA_VALID_ATIME) &&
> +	    nfs_get_timespec64(&args.fa_atime, &uarg->fa_atime))
> +		goto out;
> +
> +	if ((args.fa_valid[0] & NFS_FA_VALID_MTIME) &&
> +	    nfs_get_timespec64(&args.fa_mtime, &uarg->fa_mtime))
> +		goto out;
> +
> +	if (args.fa_valid[0] & NFS_FA_VALID_TIME_BACKUP) {
> +		if (nfs_get_timespec64(&args.fa_time_backup, &uarg->fa_time_backup))
> +			goto out;
> +	} else if ((args.fa_valid[0] & NFS_FA_VALID_ARCHIVE) &&
> +			!(NFS_SERVER(inode)->fattr_valid & NFS_ATTR_FATTR_ARCHIVE)) {
> +		args.fa_valid[0] |= NFS_FA_VALID_TIME_BACKUP;
> +		if (!(args.fa_flags & NFS_FA_FLAG_ARCHIVE)) {
> +			nfs_revalidate_inode(inode, NFS_INO_INVALID_MTIME);
> +			args.fa_time_backup.tv_sec = inode->i_mtime.tv_sec;
> +			args.fa_time_backup.tv_nsec = inode->i_mtime.tv_nsec;
> +		} else if (args.fa_valid[0] & NFS_FA_VALID_TIME_CREATE)
> +			args.fa_time_backup = args.fa_btime;
> +		else {
> +			nfs_revalidate_inode(inode, NFS_INO_INVALID_BTIME);
> +			args.fa_time_backup = NFS_I(inode)->btime;
> +		}
> +	}
> +
> +        if (args.fa_valid[0] & NFS_FA_VALID_SIZE) {
> +		if (copy_from_user(&args.fa_size, &uarg->fa_size,
> +					sizeof(args.fa_size)))
> +			goto out;
> +		ret = inode_newsize_ok(inode,args.fa_size);
> +		if (ret)
> +			goto out;
> +		if (args.fa_size == i_size_read(inode))
> +			args.fa_valid[0] &= ~NFS_FA_VALID_SIZE;
> +	}
> +
> +	/*
> +	 * No need to update the inode because that is done in nfs4_set_nfs4_statx
> +	 */
> +	ret = nfs4_set_nfs4_statx(inode, &args, fattr);
> +
> +out:
> +	inode_unlock(inode);
> +	if (args.real_fd >= 0)
> +		fput(dst_file);
> +	trace_nfs_ioctl_file_statx_set_exit(inode, ret);
> +out_free:
> +	nfs_free_fattr(fattr);
> +	return ret;
> +}
> +
> +static long nfs4_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
> +{
> +	void __user *argp = (void __user *)arg;
> +	long ret;
> +
> +	switch (cmd) {
> +	case NFS_IOC_FILE_STATX_GET:
> +		ret = nfs4_ioctl_file_statx_get(file, argp);
> +		break;
> +	case NFS_IOC_FILE_STATX_SET:
> +		ret = nfs4_ioctl_file_statx_set(file, argp);
> +		break;
> +	default:
> +		ret = -ENOIOCTLCMD;
> +	}
> +
> +	dprintk("%s: file=%pD2, cmd=%u, ret=%ld\n", __func__, file, cmd, ret);
> +	return ret;
> +}
> +
>  #ifdef CONFIG_NFS_V4_2
>  static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
>  				      struct file *file_out, loff_t pos_out,
> @@ -187,6 +686,7 @@ static ssize_t __nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
>  	return ret;
>  }
>  
> +
>  static ssize_t nfs4_copy_file_range(struct file *file_in, loff_t pos_in,
>  				    struct file *file_out, loff_t pos_out,
>  				    size_t count, unsigned int flags)
> @@ -461,4 +961,15 @@ const struct file_operations nfs4_file_operations = {
>  #else
>  	.llseek		= nfs_file_llseek,
>  #endif
> +	.unlocked_ioctl	= nfs4_ioctl,
> +};
> +
> +const struct file_operations nfs4_dir_operations = {
> +	.llseek		= nfs_llseek_dir,
> +	.read		= generic_read_dir,
> +	.iterate_shared	= nfs_readdir,
> +	.open		= nfs_opendir,
> +	.release	= nfs_closedir,
> +	.fsync		= nfs_fsync_dir,
> +	.unlocked_ioctl	= nfs4_ioctl,
>  };
> diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
> index d497616ca149..7c032583ffa2 100644
> --- a/fs/nfs/nfs4proc.c
> +++ b/fs/nfs/nfs4proc.c
> @@ -7959,6 +7959,129 @@ static int _nfs41_proc_get_locations(struct inode *inode,
>  
>  #endif	/* CONFIG_NFS_V4_1 */
>  
> +static int _nfs4_set_nfs4_statx(struct inode *inode,
> +		struct nfs4_statx *statx,
> +		struct nfs_fattr *fattr)
> +{
> +	const __u64 statx_win = NFS_FA_VALID_TIME_CREATE |
> +				NFS_FA_VALID_TIME_BACKUP |
> +				NFS_FA_VALID_ARCHIVE | NFS_FA_VALID_HIDDEN |
> +				NFS_FA_VALID_SYSTEM;
> +	struct iattr sattr = {0};
> +	struct nfs_server *server = NFS_SERVER(inode);
> +	__u32 bitmask[3];
> +	struct nfs_setattrargs arg = {
> +		.fh             = NFS_FH(inode),
> +		.iap            = &sattr,
> +		.server		= server,
> +		.bitmask	= bitmask,
> +		.statx		= statx,
> +	};
> +	struct nfs_setattrres res = {
> +		.fattr		= fattr,
> +		.server		= server,
> +	};
> +	struct rpc_message msg = {
> +		.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
> +		.rpc_argp       = &arg,
> +		.rpc_resp       = &res,
> +	};
> +	int status;
> +
> +	nfs4_bitmap_copy_adjust(
> +		bitmask, server->attr_bitmask, inode,
> +		NFS_INO_INVALID_CHANGE | NFS_INO_INVALID_CTIME |
> +			NFS_INO_INVALID_SIZE | NFS_INO_INVALID_OTHER |
> +			NFS_INO_INVALID_BTIME | NFS_INO_INVALID_WINATTR);
> +	/* Use the iattr structure to set atime and mtime since handling already
> +	 * exists for them using the iattr struct in the encode_attrs()
> +	 * (xdr encoding) routine.
> +	 */
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_MTIME)) {
> +		sattr.ia_valid |= ATTR_MTIME_SET;
> +		sattr.ia_mtime.tv_sec = statx->fa_mtime.tv_sec;
> +		sattr.ia_mtime.tv_nsec = statx->fa_mtime.tv_nsec;
> +	}
> +
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_ATIME)) {
> +		sattr.ia_valid |= ATTR_ATIME_SET;
> +		sattr.ia_atime.tv_sec = statx->fa_atime.tv_sec;
> +		sattr.ia_atime.tv_nsec = statx->fa_atime.tv_nsec;
> +	}
> +
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_OWNER)) {
> +		sattr.ia_valid |= ATTR_UID;
> +		sattr.ia_uid = statx->fa_owner_uid;
> +	}
> +
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_OWNER_GROUP)) {
> +		sattr.ia_valid |= ATTR_GID;
> +		sattr.ia_gid = statx->fa_group_gid;
> +	}
> +
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_SIZE)) {
> +		sattr.ia_valid |= ATTR_SIZE;
> +		sattr.ia_size = statx->fa_size;
> +	}
> +
> +	nfs4_stateid_copy(&arg.stateid, &zero_stateid);
> +
> +	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
> +	if (!status) {
> +		if (statx->fa_valid[0] & statx_win) {
> +			struct nfs_inode *nfsi = NFS_I(inode);
> +
> +			spin_lock(&inode->i_lock);
> +			if (statx->fa_valid[0] & NFS_FA_VALID_TIME_CREATE)
> +				nfsi->btime = statx->fa_btime;
> +			if (statx->fa_valid[0] & NFS_FA_VALID_TIME_BACKUP)
> +				nfsi->timebackup = statx->fa_time_backup;
> +			if (statx->fa_valid[0] & NFS_FA_VALID_ARCHIVE)
> +				nfsi->archive = (statx->fa_flags &
> +						 NFS_FA_FLAG_ARCHIVE) != 0;
> +			if (statx->fa_valid[0] & NFS_FA_VALID_SYSTEM)
> +				nfsi->system = (statx->fa_flags &
> +						NFS_FA_FLAG_SYSTEM) != 0;
> +			if (statx->fa_valid[0] & NFS_FA_VALID_HIDDEN)
> +				nfsi->hidden = (statx->fa_flags &
> +						NFS_FA_FLAG_HIDDEN) != 0;
> +			if (statx->fa_valid[0] & NFS_FA_VALID_OFFLINE)
> +				nfsi->offline = (statx->fa_flags &
> +						 NFS_FA_FLAG_OFFLINE) != 0;
> +
> +			nfsi->cache_validity &= ~NFS_INO_INVALID_CTIME;
> +			if (fattr->valid & NFS_ATTR_FATTR_CTIME)
> +				inode->i_ctime = fattr->ctime;
> +			else
> +				nfs_set_cache_invalid(
> +					inode, NFS_INO_INVALID_CHANGE |
> +						   NFS_INO_INVALID_CTIME);
> +			spin_unlock(&inode->i_lock);
> +		}
> +
> +		nfs_setattr_update_inode(inode, &sattr, fattr);
> +	} else
> +		dprintk("%s failed: %d\n", __func__, status);
> +
> +	return status;
> +}
> +
> +int nfs4_set_nfs4_statx(struct inode *inode,
> +		struct nfs4_statx *statx,
> +		struct nfs_fattr *fattr)
> +{
> +	struct nfs4_exception exception = { };
> +	struct nfs_server *server = NFS_SERVER(inode);
> +	int err;
> +
> +	do {
> +		err = nfs4_handle_exception(server,
> +				_nfs4_set_nfs4_statx(inode, statx, fattr),
> +				&exception);
> +	} while (exception.retry);
> +	return err;
> +}
> +
>  /**
>   * nfs4_proc_get_locations - discover locations for a migrated FSID
>   * @inode: inode on FSID that is migrating
> @@ -10419,6 +10542,7 @@ const struct nfs_rpc_ops nfs_v4_clientops = {
>  	.dir_inode_ops	= &nfs4_dir_inode_operations,
>  	.file_inode_ops	= &nfs4_file_inode_operations,
>  	.file_ops	= &nfs4_file_operations,
> +	.dir_ops	= &nfs4_dir_operations,
>  	.getroot	= nfs4_proc_get_root,
>  	.submount	= nfs4_submount,
>  	.try_get_tree	= nfs4_try_get_tree,
> diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
> index d2c240effc87..e5300d7ed712 100644
> --- a/fs/nfs/nfs4xdr.c
> +++ b/fs/nfs/nfs4xdr.c
> @@ -129,12 +129,15 @@ static int decode_layoutget(struct xdr_stream *xdr, struct rpc_rqst *req,
>  				nfs4_fattr_value_maxsz)
>  #define decode_getattr_maxsz    (op_decode_hdr_maxsz + nfs4_fattr_maxsz)
>  #define encode_attrs_maxsz	(nfs4_fattr_bitmap_maxsz + \
> -				 1 + 2 + 1 + \
> +				 1 + 2 + 1 + 1 + 1 + \
>  				nfs4_owner_maxsz + \
>  				nfs4_group_maxsz + \
> -				nfs4_label_maxsz + \
> +				1 + \
> +				1 + nfstime4_maxsz + \
> +				nfstime4_maxsz + nfstime4_maxsz + \
>  				1 + nfstime4_maxsz + \
> -				1 + nfstime4_maxsz)
> +				nfs4_label_maxsz + \
> +				2)
>  #define encode_savefh_maxsz     (op_encode_hdr_maxsz)
>  #define decode_savefh_maxsz     (op_decode_hdr_maxsz)
>  #define encode_restorefh_maxsz  (op_encode_hdr_maxsz)
> @@ -1081,6 +1084,7 @@ xdr_encode_nfstime4(__be32 *p, const struct timespec64 *t)
>  static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap,
>  				const struct nfs4_label *label,
>  				const umode_t *umask,
> +				const struct nfs4_statx *statx,
>  				const struct nfs_server *server,
>  				const uint32_t attrmask[])
>  {
> @@ -1153,6 +1157,34 @@ static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap,
>  		}
>  	}
>  
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_TIME_BACKUP) &&
> +	    (attrmask[1] & FATTR4_WORD1_TIME_BACKUP)) {
> +		bmval[1] |= FATTR4_WORD1_TIME_BACKUP;
> +		len += (nfstime4_maxsz << 2);
> +	}
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_TIME_CREATE) &&
> +	    (attrmask[1] & FATTR4_WORD1_TIME_CREATE)) {
> +		bmval[1] |= FATTR4_WORD1_TIME_CREATE;
> +		len += (nfstime4_maxsz << 2);
> +	}
> +
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_ARCHIVE) &&
> +	   (attrmask[0] & FATTR4_WORD0_ARCHIVE)) {
> +		bmval[0] |= FATTR4_WORD0_ARCHIVE;
> +		len += 4;
> +	}
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_HIDDEN) &&
> +	   (attrmask[0] & FATTR4_WORD0_HIDDEN)) {
> +		bmval[0] |= FATTR4_WORD0_HIDDEN;
> +		len += 4;
> +	}
> +
> +	if (statx && (statx->fa_valid[0] & NFS_FA_VALID_SYSTEM) &&
> +	   (attrmask[1] & FATTR4_WORD1_SYSTEM)) {
> +		bmval[1] |= FATTR4_WORD1_SYSTEM;
> +		len += 4;
> +	}
> +
>  	if (label && (attrmask[2] & FATTR4_WORD2_SECURITY_LABEL)) {
>  		len += 4 + 4 + 4 + (XDR_QUADLEN(label->len) << 2);
>  		bmval[2] |= FATTR4_WORD2_SECURITY_LABEL;
> @@ -1163,12 +1195,21 @@ static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap,
>  
>  	if (bmval[0] & FATTR4_WORD0_SIZE)
>  		p = xdr_encode_hyper(p, iap->ia_size);
> +	if (bmval[0] & FATTR4_WORD0_ARCHIVE)
> +		*p++ = (statx->fa_flags & NFS_FA_FLAG_ARCHIVE) ?
> +			cpu_to_be32(1) : cpu_to_be32(0);
> +	if (bmval[0] & FATTR4_WORD0_HIDDEN)
> +		*p++ = (statx->fa_flags & NFS_FA_FLAG_HIDDEN) ?
> +			cpu_to_be32(1) : cpu_to_be32(0);
>  	if (bmval[1] & FATTR4_WORD1_MODE)
>  		*p++ = cpu_to_be32(iap->ia_mode & S_IALLUGO);
>  	if (bmval[1] & FATTR4_WORD1_OWNER)
>  		p = xdr_encode_opaque(p, owner_name, owner_namelen);
>  	if (bmval[1] & FATTR4_WORD1_OWNER_GROUP)
>  		p = xdr_encode_opaque(p, owner_group, owner_grouplen);
> +	if (bmval[1] & FATTR4_WORD1_SYSTEM)
> +		*p++ = (statx->fa_flags & NFS_FA_FLAG_SYSTEM) ?
> +			cpu_to_be32(1) : cpu_to_be32(0);
>  	if (bmval[1] & FATTR4_WORD1_TIME_ACCESS_SET) {
>  		if (iap->ia_valid & ATTR_ATIME_SET) {
>  			*p++ = cpu_to_be32(NFS4_SET_TO_CLIENT_TIME);
> @@ -1176,6 +1217,10 @@ static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap,
>  		} else
>  			*p++ = cpu_to_be32(NFS4_SET_TO_SERVER_TIME);
>  	}
> +	if (bmval[1] & FATTR4_WORD1_TIME_BACKUP)
> +		p = xdr_encode_nfstime4(p, &statx->fa_time_backup);
> +	if (bmval[1] & FATTR4_WORD1_TIME_CREATE)
> +		p = xdr_encode_nfstime4(p, &statx->fa_btime);
>  	if (bmval[1] & FATTR4_WORD1_TIME_MODIFY_SET) {
>  		if (iap->ia_valid & ATTR_MTIME_SET) {
>  			*p++ = cpu_to_be32(NFS4_SET_TO_CLIENT_TIME);
> @@ -1248,7 +1293,7 @@ static void encode_create(struct xdr_stream *xdr, const struct nfs4_create_arg *
>  
>  	encode_string(xdr, create->name->len, create->name->name);
>  	encode_attrs(xdr, create->attrs, create->label, &create->umask,
> -			create->server, create->server->attr_bitmask);
> +		     NULL, create->server, create->server->attr_bitmask);
>  }
>  
>  static void encode_getattr(struct xdr_stream *xdr,
> @@ -1434,12 +1479,12 @@ static inline void encode_createmode(struct xdr_stream *xdr, const struct nfs_op
>  	case NFS4_CREATE_UNCHECKED:
>  		*p = cpu_to_be32(NFS4_CREATE_UNCHECKED);
>  		encode_attrs(xdr, arg->u.attrs, arg->label, &arg->umask,
> -				arg->server, arg->server->attr_bitmask);
> +			     NULL, arg->server, arg->server->attr_bitmask);
>  		break;
>  	case NFS4_CREATE_GUARDED:
>  		*p = cpu_to_be32(NFS4_CREATE_GUARDED);
>  		encode_attrs(xdr, arg->u.attrs, arg->label, &arg->umask,
> -				arg->server, arg->server->attr_bitmask);
> +			     NULL, arg->server, arg->server->attr_bitmask);
>  		break;
>  	case NFS4_CREATE_EXCLUSIVE:
>  		*p = cpu_to_be32(NFS4_CREATE_EXCLUSIVE);
> @@ -1449,7 +1494,7 @@ static inline void encode_createmode(struct xdr_stream *xdr, const struct nfs_op
>  		*p = cpu_to_be32(NFS4_CREATE_EXCLUSIVE4_1);
>  		encode_nfs4_verifier(xdr, &arg->u.verifier);
>  		encode_attrs(xdr, arg->u.attrs, arg->label, &arg->umask,
> -				arg->server, arg->server->exclcreat_bitmask);
> +			     NULL, arg->server, arg->server->exclcreat_bitmask);
>  	}
>  }
>  
> @@ -1712,8 +1757,8 @@ static void encode_setattr(struct xdr_stream *xdr, const struct nfs_setattrargs
>  {
>  	encode_op_hdr(xdr, OP_SETATTR, decode_setattr_maxsz, hdr);
>  	encode_nfs4_stateid(xdr, &arg->stateid);
> -	encode_attrs(xdr, arg->iap, arg->label, NULL, server,
> -			server->attr_bitmask);
> +	encode_attrs(xdr, arg->iap, arg->label, NULL, arg->statx, server,
> +		     server->attr_bitmask);
>  }
>  
>  static void encode_setclientid(struct xdr_stream *xdr, const struct nfs4_setclientid *setclientid, struct compound_hdr *hdr)
> diff --git a/fs/nfs/nfstrace.c b/fs/nfs/nfstrace.c
> index 5d1bfccbb4da..0b88deb0216e 100644
> --- a/fs/nfs/nfstrace.c
> +++ b/fs/nfs/nfstrace.c
> @@ -9,6 +9,11 @@
>  #define CREATE_TRACE_POINTS
>  #include "nfstrace.h"
>  
> +EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_ioctl_file_statx_get_enter);
> +EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_ioctl_file_statx_get_exit);
> +EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_ioctl_file_statx_set_enter);
> +EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_ioctl_file_statx_set_exit);
> +
>  EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_fsync_enter);
>  EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_fsync_exit);
>  EXPORT_TRACEPOINT_SYMBOL_GPL(nfs_xdr_status);
> diff --git a/fs/nfs/nfstrace.h b/fs/nfs/nfstrace.h
> index 2ef7cff8a4ba..b67dd087fb47 100644
> --- a/fs/nfs/nfstrace.h
> +++ b/fs/nfs/nfstrace.h
> @@ -166,6 +166,11 @@ DEFINE_NFS_INODE_EVENT_DONE(nfs_fsync_exit);
>  DEFINE_NFS_INODE_EVENT(nfs_access_enter);
>  DEFINE_NFS_INODE_EVENT_DONE(nfs_set_cache_invalid);
>  
> +DEFINE_NFS_INODE_EVENT(nfs_ioctl_file_statx_get_enter);
> +DEFINE_NFS_INODE_EVENT_DONE(nfs_ioctl_file_statx_get_exit);
> +DEFINE_NFS_INODE_EVENT(nfs_ioctl_file_statx_set_enter);
> +DEFINE_NFS_INODE_EVENT_DONE(nfs_ioctl_file_statx_set_exit);
> +
>  TRACE_EVENT(nfs_access_exit,
>  		TP_PROTO(
>  			const struct inode *inode,
> diff --git a/fs/nfs/proc.c b/fs/nfs/proc.c
> index 73dcaa99fa9b..8fd96d93630a 100644
> --- a/fs/nfs/proc.c
> +++ b/fs/nfs/proc.c
> @@ -717,6 +717,7 @@ const struct nfs_rpc_ops nfs_v2_clientops = {
>  	.dir_inode_ops	= &nfs_dir_inode_operations,
>  	.file_inode_ops	= &nfs_file_inode_operations,
>  	.file_ops	= &nfs_file_operations,
> +	.dir_ops	= &nfs_dir_operations,
>  	.getroot	= nfs_proc_get_root,
>  	.submount	= nfs_submount,
>  	.try_get_tree	= nfs_try_get_tree,
> diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h
> index 058fc11338d9..0c3a5859f7f3 100644
> --- a/include/linux/nfs_fs.h
> +++ b/include/linux/nfs_fs.h
> @@ -501,6 +501,7 @@ extern __be32 root_nfs_parse_addr(char *name); /*__init*/
>  extern const struct file_operations nfs_file_operations;
>  #if IS_ENABLED(CONFIG_NFS_V4)
>  extern const struct file_operations nfs4_file_operations;
> +extern const struct file_operations nfs4_dir_operations;
>  #endif /* CONFIG_NFS_V4 */
>  extern const struct address_space_operations nfs_file_aops;
>  extern const struct address_space_operations nfs_dir_aops;
> diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
> index 0d5b11c1bfec..9ce61f680a13 100644
> --- a/include/linux/nfs_xdr.h
> +++ b/include/linux/nfs_xdr.h
> @@ -812,6 +812,7 @@ struct nfs_createargs {
>  	struct iattr *		sattr;
>  };
>  
> +struct nfs4_statx;
>  struct nfs_setattrargs {
>  	struct nfs4_sequence_args 	seq_args;
>  	struct nfs_fh *                 fh;
> @@ -820,6 +821,7 @@ struct nfs_setattrargs {
>  	const struct nfs_server *	server; /* Needed for name mapping */
>  	const u32 *			bitmask;
>  	const struct nfs4_label		*label;
> +	const struct nfs4_statx		*statx;
>  };
>  
>  struct nfs_setaclargs {
> @@ -1744,6 +1746,7 @@ struct nfs_rpc_ops {
>  	const struct inode_operations *dir_inode_ops;
>  	const struct inode_operations *file_inode_ops;
>  	const struct file_operations *file_ops;
> +	const struct file_operations *dir_ops;
>  	const struct nlmclnt_operations *nlmclnt_ops;
>  
>  	int	(*getroot) (struct nfs_server *, struct nfs_fh *,
> diff --git a/include/uapi/linux/nfs.h b/include/uapi/linux/nfs.h
> index 946cb62d64b0..df87da39bc43 100644
> --- a/include/uapi/linux/nfs.h
> +++ b/include/uapi/linux/nfs.h
> @@ -9,6 +9,8 @@
>  #define _UAPI_LINUX_NFS_H
>  
>  #include <linux/types.h>
> +#include <asm/byteorder.h>
> +#include <linux/time.h>
>  
>  #define NFS_PROGRAM	100003
>  #define NFS_PORT	2049
> @@ -35,6 +37,94 @@
>  
>  #define NFS_PIPE_DIRNAME "nfs"
>  
> +/* NFS ioctls */
> +#define NFS_IOC_FILE_STATX_GET	_IOR('N', 2, struct nfs_ioctl_nfs4_statx)
> +#define NFS_IOC_FILE_STATX_SET	_IOW('N', 3, struct nfs_ioctl_nfs4_statx)
> +
> +/* Options for struct nfs_ioctl_nfs4_statx */
> +#define NFS_FA_OPTIONS_SYNC_AS_STAT			0x0000
> +#define NFS_FA_OPTIONS_FORCE_SYNC			0x2000 /* See statx */
> +#define NFS_FA_OPTIONS_DONT_SYNC			0x4000 /* See statx */
> +
> +#define NFS_FA_VALID_TIME_CREATE			0x00001UL
> +#define NFS_FA_VALID_TIME_BACKUP			0x00002UL
> +#define NFS_FA_VALID_ARCHIVE				0x00004UL
> +#define NFS_FA_VALID_HIDDEN				0x00008UL
> +#define NFS_FA_VALID_SYSTEM				0x00010UL
> +#define NFS_FA_VALID_OWNER				0x00020UL
> +#define NFS_FA_VALID_OWNER_GROUP			0x00040UL
> +#define NFS_FA_VALID_ATIME				0x00080UL
> +#define NFS_FA_VALID_MTIME				0x00100UL
> +#define NFS_FA_VALID_CTIME				0x00200UL
> +#define NFS_FA_VALID_OFFLINE				0x00400UL
> +#define NFS_FA_VALID_MODE				0x00800UL
> +#define NFS_FA_VALID_NLINK				0x01000UL
> +#define NFS_FA_VALID_BLKSIZE				0x02000UL
> +#define NFS_FA_VALID_INO				0x04000UL
> +#define NFS_FA_VALID_DEV				0x08000UL
> +#define NFS_FA_VALID_RDEV				0x10000UL
> +#define NFS_FA_VALID_SIZE				0x20000UL
> +#define NFS_FA_VALID_BLOCKS				0x40000UL
> +
> +#define NFS_FA_VALID_ALL_ATTR_0 ( NFS_FA_VALID_TIME_CREATE | \
> +		NFS_FA_VALID_TIME_BACKUP | \
> +		NFS_FA_VALID_ARCHIVE | \
> +		NFS_FA_VALID_HIDDEN | \
> +		NFS_FA_VALID_SYSTEM | \
> +		NFS_FA_VALID_OWNER | \
> +		NFS_FA_VALID_OWNER_GROUP | \
> +		NFS_FA_VALID_ATIME | \
> +		NFS_FA_VALID_MTIME | \
> +		NFS_FA_VALID_CTIME | \
> +		NFS_FA_VALID_OFFLINE | \
> +		NFS_FA_VALID_MODE | \
> +		NFS_FA_VALID_NLINK | \
> +		NFS_FA_VALID_BLKSIZE | \
> +		NFS_FA_VALID_INO | \
> +		NFS_FA_VALID_DEV | \
> +		NFS_FA_VALID_RDEV | \
> +		NFS_FA_VALID_SIZE | \
> +		NFS_FA_VALID_BLOCKS)
> +
> +#define NFS_FA_FLAG_ARCHIVE				(1UL << 0)
> +#define NFS_FA_FLAG_HIDDEN				(1UL << 1)
> +#define NFS_FA_FLAG_SYSTEM				(1UL << 2)
> +#define NFS_FA_FLAG_OFFLINE				(1UL << 3)
> +
> +struct nfs_ioctl_timespec {
> +	__s64		tv_sec;
> +	__s64		tv_nsec;
> +};
> +
> +struct nfs_ioctl_nfs4_statx {
> +	__s32		real_fd;		/* real FD to use,
> +						   -1 means use current file */
> +	__u32		fa_options;
> +
> +	__u64		fa_request[2];		/* Attributes to retrieve */
> +	__u64		fa_valid[2];		/* Attributes set */
> +
> +	struct nfs_ioctl_timespec fa_time_backup;/* Backup time */
> +	struct nfs_ioctl_timespec fa_btime;     /* Birth time */
> +	__u64		fa_flags;		/* Flag attributes */
> +	/* Ordinary attributes follow */
> +	struct nfs_ioctl_timespec fa_atime;	/* Access time */
> +	struct nfs_ioctl_timespec fa_mtime;	/* Modify time */
> +	struct nfs_ioctl_timespec fa_ctime;	/* Change time */
> +	__u32		fa_owner_uid;		/* Owner User ID */
> +	__u32		fa_group_gid;		/* Primary Group ID */
> +	__u32		fa_mode;		/* Mode */
> +	__u32	 	fa_nlink;
> +	__u32		fa_blksize;
> +	__u32		fa_spare;		/* Alignment */
> +	__u64		fa_ino;
> +	__u32		fa_dev;
> +	__u32		fa_rdev;
> +	__s64		fa_size;
> +	__s64		fa_blocks;
> +	__u64 		fa_padding[4];
> +};
> +
>  /*
>   * NFS stats. The good thing with these values is that NFSv3 errors are
>   * a superset of NFSv2 errors (with the exception of NFSERR_WFLUSH which
> -- 
> 2.33.1