nfsd4: separate splice and readv cases

The splice and readv cases are actually quite different--for example the
former case ignores the array of vectors we build up for the latter.

It is probably clearer to separate the two cases entirely.

There's some code duplication between the split out encoders, but this
is only temporary and will be fixed by a later patch.

Signed-off-by: J. Bruce Fields <bfields@redhat.com>

3 files changed, 214 insertions, 88 deletions

diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index f6a5cb72269..bad762b6cf1 100644
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -3069,41 +3069,84 @@ nfsd4_encode_open_downgrade(struct nfsd4_compoundres *resp, __be32 nfserr, struc
 	return nfserr;
 }
 
-static __be32
-nfsd4_encode_read(struct nfsd4_compoundres *resp, __be32 nfserr,
-		  struct nfsd4_read *read)
+static __be32 nfsd4_encode_splice_read(
+				struct nfsd4_compoundres *resp,
+				struct nfsd4_read *read,
+				struct file *file, unsigned long maxcount)
 {
-	u32 eof;
-	int v;
-	struct page *page;
-	unsigned long maxcount; 
 	struct xdr_stream *xdr = &resp->xdr;
-	int starting_len = xdr->buf->len;
+	u32 eof;
+	int starting_len = xdr->buf->len - 8;
 	int space_left;
-	long len;
+	__be32 nfserr;
+	__be32 tmp;
 	__be32 *p;
 
-	if (nfserr)
-		return nfserr;
-
-	p = xdr_reserve_space(xdr, 8); /* eof flag and byte count */
-	if (!p) {
-		WARN_ON_ONCE(resp->rqstp->rq_splice_ok);
+	/*
+	 * Don't inline pages unless we know there's room for eof,
+	 * count, and possible padding:
+	 */
+	if (xdr->end - xdr->p < 3)
 		return nfserr_resource;
+
+	nfserr = nfsd_splice_read(read->rd_rqstp, file,
+				  read->rd_offset, &maxcount);
+	if (nfserr) {
+		/*
+		 * nfsd_splice_actor may have already messed with the
+		 * page length; reset it so as not to confuse
+		 * xdr_truncate_encode:
+		 */
+		xdr->buf->page_len = 0;
+		return nfserr;
 	}
 
-	/* Make sure there will be room for padding if needed: */
-	if (xdr->end - xdr->p < 1)
-		return nfserr_resource;
+	eof = (read->rd_offset + maxcount >=
+	       read->rd_fhp->fh_dentry->d_inode->i_size);
 
-	if (resp->xdr.buf->page_len) {
-		WARN_ON_ONCE(resp->rqstp->rq_splice_ok);
-		return nfserr_resource;
+	tmp = htonl(eof);
+	write_bytes_to_xdr_buf(xdr->buf, starting_len    , &tmp, 4);
+	tmp = htonl(maxcount);
+	write_bytes_to_xdr_buf(xdr->buf, starting_len + 4, &tmp, 4);
+
+	resp->xdr.buf->page_len = maxcount;
+	xdr->buf->len += maxcount;
+	xdr->page_ptr += (maxcount + PAGE_SIZE - 1) / PAGE_SIZE;
+	xdr->iov = xdr->buf->tail;
+
+	/* Use rest of head for padding and remaining ops: */
+	resp->xdr.buf->tail[0].iov_base = xdr->p;
+	resp->xdr.buf->tail[0].iov_len = 0;
+	if (maxcount&3) {
+		p = xdr_reserve_space(xdr, 4);
+		WRITE32(0);
+		resp->xdr.buf->tail[0].iov_base += maxcount&3;
+		resp->xdr.buf->tail[0].iov_len = 4 - (maxcount&3);
+		xdr->buf->len -= (maxcount&3);
 	}
 
-	maxcount = svc_max_payload(resp->rqstp);
-	if (maxcount > read->rd_length)
-		maxcount = read->rd_length;
+	space_left = min_t(int, (void *)xdr->end - (void *)xdr->p,
+				xdr->buf->buflen - xdr->buf->len);
+	xdr->buf->buflen = xdr->buf->len + space_left;
+	xdr->end = (__be32 *)((void *)xdr->end + space_left);
+
+	return 0;
+}
+
+static __be32 nfsd4_encode_readv(struct nfsd4_compoundres *resp,
+				 struct nfsd4_read *read,
+				 struct file *file, unsigned long maxcount)
+{
+	struct xdr_stream *xdr = &resp->xdr;
+	u32 eof;
+	int v;
+	struct page *page;
+	int starting_len = xdr->buf->len - 8;
+	int space_left;
+	long len;
+	__be32 nfserr;
+	__be32 tmp;
+	__be32 *p;
 
 	len = maxcount;
 	v = 0;
@@ -3124,34 +3167,26 @@ nfsd4_encode_read(struct nfsd4_compoundres *resp, __be32 nfserr,
 	}
 	read->rd_vlen = v;
 
-	nfserr = nfsd_read_file(read->rd_rqstp, read->rd_fhp, read->rd_filp,
-			read->rd_offset, resp->rqstp->rq_vec, read->rd_vlen,
-			&maxcount);
-
-	if (nfserr) {
-		/*
-		 * nfsd_splice_actor may have already messed with the
-		 * page length; reset it so as not to confuse
-		 * xdr_truncate_encode:
-		 */
-		xdr->buf->page_len = 0;
-		xdr_truncate_encode(xdr, starting_len);
+	nfserr = nfsd_readv(file, read->rd_offset, resp->rqstp->rq_vec,
+			read->rd_vlen, &maxcount);
+	if (nfserr)
 		return nfserr;
-	}
+
 	eof = (read->rd_offset + maxcount >=
 	       read->rd_fhp->fh_dentry->d_inode->i_size);
 
-	WRITE32(eof);
-	WRITE32(maxcount);
-	WARN_ON_ONCE(resp->xdr.buf->head[0].iov_len != (char *)p
-				- (char *)resp->xdr.buf->head[0].iov_base);
+	tmp = htonl(eof);
+	write_bytes_to_xdr_buf(xdr->buf, starting_len    , &tmp, 4);
+	tmp = htonl(maxcount);
+	write_bytes_to_xdr_buf(xdr->buf, starting_len + 4, &tmp, 4);
+
 	resp->xdr.buf->page_len = maxcount;
 	xdr->buf->len += maxcount;
 	xdr->page_ptr += v;
 	xdr->iov = xdr->buf->tail;
 
 	/* Use rest of head for padding and remaining ops: */
-	resp->xdr.buf->tail[0].iov_base = p;
+	resp->xdr.buf->tail[0].iov_base = xdr->p;
 	resp->xdr.buf->tail[0].iov_len = 0;
 	if (maxcount&3) {
 		p = xdr_reserve_space(xdr, 4);
@@ -3167,6 +3202,60 @@ nfsd4_encode_read(struct nfsd4_compoundres *resp, __be32 nfserr,
 	xdr->end = (__be32 *)((void *)xdr->end + space_left);
 
 	return 0;
+
+}
+
+static __be32
+nfsd4_encode_read(struct nfsd4_compoundres *resp, __be32 nfserr,
+		  struct nfsd4_read *read)
+{
+	unsigned long maxcount;
+	struct xdr_stream *xdr = &resp->xdr;
+	struct file *file = read->rd_filp;
+	int starting_len = xdr->buf->len;
+	struct raparms *ra;
+	__be32 *p;
+	__be32 err;
+
+	if (nfserr)
+		return nfserr;
+
+	p = xdr_reserve_space(xdr, 8); /* eof flag and byte count */
+	if (!p) {
+		WARN_ON_ONCE(resp->rqstp->rq_splice_ok);
+		return nfserr_resource;
+	}
+
+	if (resp->xdr.buf->page_len) {
+		WARN_ON_ONCE(resp->rqstp->rq_splice_ok);
+		return nfserr_resource;
+	}
+
+	xdr_commit_encode(xdr);
+
+	maxcount = svc_max_payload(resp->rqstp);
+	if (maxcount > read->rd_length)
+		maxcount = read->rd_length;
+
+	if (!read->rd_filp) {
+		err = nfsd_get_tmp_read_open(resp->rqstp, read->rd_fhp,
+						&file, &ra);
+		if (err)
+			goto err_truncate;
+	}
+
+	if (file->f_op->splice_read && resp->rqstp->rq_splice_ok)
+		err = nfsd4_encode_splice_read(resp, read, file, maxcount);
+	else
+		err = nfsd4_encode_readv(resp, read, file, maxcount);
+
+	if (!read->rd_filp)
+		nfsd_put_tmp_read_open(file, ra);
+
+err_truncate:
+	if (err)
+		xdr_truncate_encode(xdr, starting_len);
+	return err;
 }
 
 static __be32
diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index f3a0f6cc429..b7d35a4afef 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -820,41 +820,54 @@ static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
 	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
 }
 
-static __be32
-nfsd_vfs_read(struct svc_rqst *rqstp, struct file *file,
-              loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
+__be32 nfsd_finish_read(struct file *file, unsigned long *count, int host_err)
 {
-	mm_segment_t	oldfs;
-	__be32		err;
-	int		host_err;
-
-	err = nfserr_perm;
-
-	if (file->f_op->splice_read && rqstp->rq_splice_ok) {
-		struct splice_desc sd = {
-			.len		= 0,
-			.total_len	= *count,
-			.pos		= offset,
-			.u.data		= rqstp,
-		};
-
-		rqstp->rq_next_page = rqstp->rq_respages + 1;
-		host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
-	} else {
-		oldfs = get_fs();
-		set_fs(KERNEL_DS);
-		host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
-		set_fs(oldfs);
-	}
-
 	if (host_err >= 0) {
 		nfsdstats.io_read += host_err;
 		*count = host_err;
-		err = 0;
 		fsnotify_access(file);
+		return 0;
 	} else 
-		err = nfserrno(host_err);
-	return err;
+		return nfserrno(host_err);
+}
+
+int nfsd_splice_read(struct svc_rqst *rqstp,
+		     struct file *file, loff_t offset, unsigned long *count)
+{
+	struct splice_desc sd = {
+		.len		= 0,
+		.total_len	= *count,
+		.pos		= offset,
+		.u.data		= rqstp,
+	};
+	int host_err;
+
+	rqstp->rq_next_page = rqstp->rq_respages + 1;
+	host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
+	return nfsd_finish_read(file, count, host_err);
+}
+
+int nfsd_readv(struct file *file, loff_t offset, struct kvec *vec, int vlen,
+		unsigned long *count)
+{
+	mm_segment_t oldfs;
+	int host_err;
+
+	oldfs = get_fs();
+	set_fs(KERNEL_DS);
+	host_err = vfs_readv(file, (struct iovec __user *)vec, vlen, &offset);
+	set_fs(oldfs);
+	return nfsd_finish_read(file, count, host_err);
+}
+
+static __be32
+nfsd_vfs_read(struct svc_rqst *rqstp, struct file *file,
+	      loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
+{
+	if (file->f_op->splice_read && rqstp->rq_splice_ok)
+		return nfsd_splice_read(rqstp, file, offset, count);
+	else
+		return nfsd_readv(file, offset, vec, vlen, count);
 }
 
 /*
@@ -956,33 +969,28 @@ out_nfserr:
 	return err;
 }
 
-/*
- * Read data from a file. count must contain the requested read count
- * on entry. On return, *count contains the number of bytes actually read.
- * N.B. After this call fhp needs an fh_put
- */
-__be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
-	loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
+__be32 nfsd_get_tmp_read_open(struct svc_rqst *rqstp, struct svc_fh *fhp,
+		struct file **file, struct raparms **ra)
 {
-	struct file *file;
 	struct inode *inode;
-	struct raparms	*ra;
 	__be32 err;
 
-	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, &file);
+	err = nfsd_open(rqstp, fhp, S_IFREG, NFSD_MAY_READ, file);
 	if (err)
 		return err;
 
-	inode = file_inode(file);
+	inode = file_inode(*file);
 
 	/* Get readahead parameters */
-	ra = nfsd_get_raparms(inode->i_sb->s_dev, inode->i_ino);
-
-	if (ra && ra->p_set)
-		file->f_ra = ra->p_ra;
+	*ra = nfsd_get_raparms(inode->i_sb->s_dev, inode->i_ino);
 
-	err = nfsd_vfs_read(rqstp, file, offset, vec, vlen, count);
+	if (*ra && (*ra)->p_set)
+		(*file)->f_ra = (*ra)->p_ra;
+	return nfs_ok;
+}
 
+void nfsd_put_tmp_read_open(struct file *file, struct raparms *ra)
+{
 	/* Write back readahead params */
 	if (ra) {
 		struct raparm_hbucket *rab = &raparm_hash[ra->p_hindex];
@@ -992,8 +1000,29 @@ __be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		ra->p_count--;
 		spin_unlock(&rab->pb_lock);
 	}
-
 	nfsd_close(file);
+}
+
+/*
+ * Read data from a file. count must contain the requested read count
+ * on entry. On return, *count contains the number of bytes actually read.
+ * N.B. After this call fhp needs an fh_put
+ */
+__be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
+	loff_t offset, struct kvec *vec, int vlen, unsigned long *count)
+{
+	struct file *file;
+	struct raparms	*ra;
+	__be32 err;
+
+	err = nfsd_get_tmp_read_open(rqstp, fhp, &file, &ra);
+	if (err)
+		return err;
+
+	err = nfsd_vfs_read(rqstp, file, offset, vec, vlen, count);
+
+	nfsd_put_tmp_read_open(file, ra);
+
 	return err;
 }
 
diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
index fbe90bdb221..7441e9655eb 100644
--- a/fs/nfsd/vfs.h
+++ b/fs/nfsd/vfs.h
@@ -70,6 +70,14 @@ __be32		nfsd_commit(struct svc_rqst *, struct svc_fh *,
 __be32		nfsd_open(struct svc_rqst *, struct svc_fh *, umode_t,
 				int, struct file **);
 void		nfsd_close(struct file *);
+struct raparms;
+__be32		nfsd_get_tmp_read_open(struct svc_rqst *, struct svc_fh *,
+				struct file **, struct raparms **);
+void		nfsd_put_tmp_read_open(struct file *, struct raparms *);
+int		nfsd_splice_read(struct svc_rqst *,
+				struct file *, loff_t, unsigned long *);
+int		nfsd_readv(struct file *, loff_t, struct kvec *, int,
+				unsigned long *);
 __be32 		nfsd_read(struct svc_rqst *, struct svc_fh *,
 				loff_t, struct kvec *, int, unsigned long *);
 __be32 		nfsd_read_file(struct svc_rqst *, struct svc_fh *, struct file *,