> On Feb 15, 2023, at 6:53 AM, Jeff Layton <jlayton@xxxxxxxxxx> wrote:
>
> On most filesystems, there is no reason to delay reaping an nfsd_file
> just because its underlying inode is still under writeback. nfsd just
> relies on client activity or the local flusher threads to do writeback.
>
> The main exception is NFS, which flushes all of its dirty data on last
> close. Add a new EXPORT_OP_FLUSH_ON_CLOSE flag to allow filesystems to
> signal that they do this, and only skip closing files under writeback on
> such filesystems.
>
> Also, remove a redundant NULL file pointer check in
> nfsd_file_check_writeback, and clean up nfs's export op flag
> definitions.
>
> Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
I assume that I'm taking this via the nfsd tree? If so,
I would like an Acked-by from the NFS client maintainers...
For the moment this is going to topic-filecache-cleanups,
not to nfsd-next.
> ---
> fs/nfs/export.c | 9 ++++++---
> fs/nfsd/filecache.c | 11 ++++++++++-
> include/linux/exportfs.h | 1 +
> 3 files changed, 17 insertions(+), 4 deletions(-)
>
> diff --git a/fs/nfs/export.c b/fs/nfs/export.c
> index 0a5ee1754d50..102a454e27c9 100644
> --- a/fs/nfs/export.c
> +++ b/fs/nfs/export.c
> @@ -156,7 +156,10 @@ const struct export_operations nfs_export_ops = {
> .fh_to_dentry = nfs_fh_to_dentry,
> .get_parent = nfs_get_parent,
> .fetch_iversion = nfs_fetch_iversion,
> - .flags = EXPORT_OP_NOWCC|EXPORT_OP_NOSUBTREECHK|
> - EXPORT_OP_CLOSE_BEFORE_UNLINK|EXPORT_OP_REMOTE_FS|
> - EXPORT_OP_NOATOMIC_ATTR,
> + .flags = EXPORT_OP_NOWCC |
> + EXPORT_OP_NOSUBTREECHK |
> + EXPORT_OP_CLOSE_BEFORE_UNLINK |
> + EXPORT_OP_REMOTE_FS |
> + EXPORT_OP_NOATOMIC_ATTR |
> + EXPORT_OP_FLUSH_ON_CLOSE,
> };
> diff --git a/fs/nfsd/filecache.c b/fs/nfsd/filecache.c
> index e6617431df7c..98e1ab013ac0 100644
> --- a/fs/nfsd/filecache.c
> +++ b/fs/nfsd/filecache.c
> @@ -302,8 +302,17 @@ nfsd_file_check_writeback(struct nfsd_file *nf)
> struct file *file = nf->nf_file;
> struct address_space *mapping;
>
> - if (!file || !(file->f_mode & FMODE_WRITE))
> + /* File not open for write? */
> + if (!(file->f_mode & FMODE_WRITE))
> return false;
> +
> + /*
> + * Some filesystems (e.g. NFS) flush all dirty data on close.
> + * On others, there is no need to wait for writeback.
> + */
> + if (!(file_inode(file)->i_sb->s_export_op->flags & EXPORT_OP_FLUSH_ON_CLOSE))
> + return false;
> +
> mapping = file->f_mapping;
> return mapping_tagged(mapping, PAGECACHE_TAG_DIRTY) ||
> mapping_tagged(mapping, PAGECACHE_TAG_WRITEBACK);
> diff --git a/include/linux/exportfs.h b/include/linux/exportfs.h
> index fe848901fcc3..218fc5c54e90 100644
> --- a/include/linux/exportfs.h
> +++ b/include/linux/exportfs.h
> @@ -221,6 +221,7 @@ struct export_operations {
> #define EXPORT_OP_NOATOMIC_ATTR (0x10) /* Filesystem cannot supply
> atomic attribute updates
> */
> +#define EXPORT_OP_FLUSH_ON_CLOSE (0x20) /* fs flushes file data on close */
> unsigned long flags;
> };
>
> --
> 2.39.1
>
--
Chuck Lever
