nfsd: fix nfs read eof detection

Currently, the knfsd server assumes that a short read indicates an
end of file. That assumption is incorrect. The short read means that
either we've hit the end of file, or we've hit a read error.

In the case of a read error, the client may want to retry (as per the
implementation recommendations in RFC1813 and RFC7530), but currently it
is being told that it hit an eof.

Move the code to detect eof from version specific code into the generic
nfsd read.

Report eof only in the two following cases:
1) read() returns a zero length short read with no error.
2) the offset+length of the read is >= the file size.

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>

1 files changed, 26 insertions, 11 deletions

diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 0867d5319fdb..bd0a385df3fc 100644
--- a/fs/nfsd/vfs.c
+++ b/fs/nfsd/vfs.c
@@ -834,12 +834,23 @@ static int nfsd_direct_splice_actor(struct pipe_inode_info *pipe,
 	return __splice_from_pipe(pipe, sd, nfsd_splice_actor);
 }
 
+static u32 nfsd_eof_on_read(struct file *file, loff_t offset, ssize_t len,
+		size_t expected)
+{
+	if (expected != 0 && len == 0)
+		return 1;
+	if (offset+len >= i_size_read(file_inode(file)))
+		return 1;
+	return 0;
+}
+
 static __be32 nfsd_finish_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 			       struct file *file, loff_t offset,
-			       unsigned long *count, int host_err)
+			       unsigned long *count, u32 *eof, ssize_t host_err)
 {
 	if (host_err >= 0) {
 		nfsdstats.io_read += host_err;
+		*eof = nfsd_eof_on_read(file, offset, host_err, *count);
 		*count = host_err;
 		fsnotify_access(file);
 		trace_nfsd_read_io_done(rqstp, fhp, offset, *count);
@@ -851,7 +862,8 @@ static __be32 nfsd_finish_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 }
 
 __be32 nfsd_splice_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
-			struct file *file, loff_t offset, unsigned long *count)
+			struct file *file, loff_t offset, unsigned long *count,
+			u32 *eof)
 {
 	struct splice_desc sd = {
 		.len		= 0,
@@ -859,25 +871,27 @@ __be32 nfsd_splice_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		.pos		= offset,
 		.u.data		= rqstp,
 	};
-	int host_err;
+	ssize_t host_err;
 
 	trace_nfsd_read_splice(rqstp, fhp, offset, *count);
 	rqstp->rq_next_page = rqstp->rq_respages + 1;
 	host_err = splice_direct_to_actor(file, &sd, nfsd_direct_splice_actor);
-	return nfsd_finish_read(rqstp, fhp, file, offset, count, host_err);
+	return nfsd_finish_read(rqstp, fhp, file, offset, count, eof, host_err);
 }
 
 __be32 nfsd_readv(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		  struct file *file, loff_t offset,
-		  struct kvec *vec, int vlen, unsigned long *count)
+		  struct kvec *vec, int vlen, unsigned long *count,
+		  u32 *eof)
 {
 	struct iov_iter iter;
-	int host_err;
+	loff_t ppos = offset;
+	ssize_t host_err;
 
 	trace_nfsd_read_vector(rqstp, fhp, offset, *count);
 	iov_iter_kvec(&iter, READ, vec, vlen, *count);
-	host_err = vfs_iter_read(file, &iter, &offset, 0);
-	return nfsd_finish_read(rqstp, fhp, file, offset, count, host_err);
+	host_err = vfs_iter_read(file, &iter, &ppos, 0);
+	return nfsd_finish_read(rqstp, fhp, file, offset, count, eof, host_err);
 }
 
 /*
@@ -984,7 +998,8 @@ out_nfserr:
  * 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)
+	loff_t offset, struct kvec *vec, int vlen, unsigned long *count,
+	u32 *eof)
 {
 	struct nfsd_file	*nf;
 	struct file *file;
@@ -997,9 +1012,9 @@ __be32 nfsd_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
 
 	file = nf->nf_file;
 	if (file->f_op->splice_read && test_bit(RQ_SPLICE_OK, &rqstp->rq_flags))
-		err = nfsd_splice_read(rqstp, fhp, file, offset, count);
+		err = nfsd_splice_read(rqstp, fhp, file, offset, count, eof);
 	else
-		err = nfsd_readv(rqstp, fhp, file, offset, vec, vlen, count);
+		err = nfsd_readv(rqstp, fhp, file, offset, vec, vlen, count, eof);
 
 	nfsd_file_put(nf);