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>

diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
index 9bc32af4e2daff14817c4306833009c1d9ab92aa..cea68d8411ac520a02f671ef418027ce5064b831 100644 (file)
--- a/fs/nfsd/nfs3proc.c
+++ b/fs/nfsd/nfs3proc.c
@@ -172,13 +172,8 @@ nfsd3_proc_read(struct svc_rqst *rqstp)
        nfserr = nfsd_read(rqstp, &resp->fh,
                                  argp->offset,
                                  rqstp->rq_vec, argp->vlen,
-                                 &resp->count);
-       if (nfserr == 0) {
-               struct inode    *inode = d_inode(resp->fh.fh_dentry);
-               resp->eof = nfsd_eof_on_read(cnt, resp->count, argp->offset,
-                                                       inode->i_size);
-       }
-
+                                 &resp->count,
+                                 &resp->eof);
        RETURN_STATUS(nfserr);
 }
 

diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index c1fc2641e3e72e62b53ddeded060704f9acfbe3a..533d0fc3c96b8fa43c2465489d00ab8f326085e0 100644 (file)
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -3472,7 +3472,7 @@ static __be32 nfsd4_encode_splice_read(
 
        len = maxcount;
        nfserr = nfsd_splice_read(read->rd_rqstp, read->rd_fhp,
-                                 file, read->rd_offset, &maxcount);
+                                 file, read->rd_offset, &maxcount, &eof);
        read->rd_length = maxcount;
        if (nfserr) {
                /*
@@ -3484,9 +3484,6 @@ static __be32 nfsd4_encode_splice_read(
                return nfserr;
        }
 
-       eof = nfsd_eof_on_read(len, maxcount, read->rd_offset,
-                               d_inode(read->rd_fhp->fh_dentry)->i_size);
-
        *(p++) = htonl(eof);
        *(p++) = htonl(maxcount);
 
@@ -3557,15 +3554,13 @@ static __be32 nfsd4_encode_readv(struct nfsd4_compoundres *resp,
 
        len = maxcount;
        nfserr = nfsd_readv(resp->rqstp, read->rd_fhp, file, read->rd_offset,
-                           resp->rqstp->rq_vec, read->rd_vlen, &maxcount);
+                           resp->rqstp->rq_vec, read->rd_vlen, &maxcount,
+                           &eof);
        read->rd_length = maxcount;
        if (nfserr)
                return nfserr;
        xdr_truncate_encode(xdr, starting_len + 8 + ((maxcount+3)&~3));
 
-       eof = nfsd_eof_on_read(len, maxcount, read->rd_offset,
-                               d_inode(read->rd_fhp->fh_dentry)->i_size);
-
        tmp = htonl(eof);
        write_bytes_to_xdr_buf(xdr->buf, starting_len    , &tmp, 4);
        tmp = htonl(maxcount);

diff --git a/fs/nfsd/nfsproc.c b/fs/nfsd/nfsproc.c
index 0d20fd161225a3d97b7a7a046bb4cc8d53f404f4..c83ddac22f38fe024029c308b078f6ca9f73e5b2 100644 (file)
--- a/fs/nfsd/nfsproc.c
+++ b/fs/nfsd/nfsproc.c
@@ -172,6 +172,7 @@ nfsd_proc_read(struct svc_rqst *rqstp)
        struct nfsd_readargs *argp = rqstp->rq_argp;
        struct nfsd_readres *resp = rqstp->rq_resp;
        __be32  nfserr;
+       u32 eof;
 
        dprintk("nfsd: READ    %s %d bytes at %d\n",
                SVCFH_fmt(&argp->fh),
@@ -195,7 +196,8 @@ nfsd_proc_read(struct svc_rqst *rqstp)
        nfserr = nfsd_read(rqstp, fh_copy(&resp->fh, &argp->fh),
                                  argp->offset,
                                  rqstp->rq_vec, argp->vlen,
-                                 &resp->count);
+                                 &resp->count,
+                                 &eof);
 
        if (nfserr) return nfserr;
        return fh_getattr(&resp->fh, &resp->stat);

diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
index 0867d5319fdb49f13f87bff73f86b1f912cf28ca..bd0a385df3fc6d632be8bed9f0e298a0da24d66f 100644 (file)
--- 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);
 

diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
index e0f7792165a6855aefd15c10954a045d1dbe7379..a13fd9d7e1f5eebe6a20613d9adf75a715dac0cf 100644 (file)
--- a/fs/nfsd/vfs.h
+++ b/fs/nfsd/vfs.h
@@ -80,13 +80,16 @@ __be32              nfsd_open_verified(struct svc_rqst *, struct svc_fh *, umode_t,
                                int, struct file **);
 __be32         nfsd_splice_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
                                struct file *file, loff_t offset,
-                               unsigned long *count);
+                               unsigned long *count,
+                               u32 *eof);
 __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);
+                               unsigned long *count,
+                               u32 *eof);
 __be32                 nfsd_read(struct svc_rqst *, struct svc_fh *,
-                               loff_t, struct kvec *, int, unsigned long *);
+                               loff_t, struct kvec *, int, unsigned long *,
+                               u32 *eof);
 __be32                 nfsd_write(struct svc_rqst *, struct svc_fh *, loff_t,
                                struct kvec *, int, unsigned long *, int);
 __be32         nfsd_vfs_write(struct svc_rqst *rqstp, struct svc_fh *fhp,
@@ -149,23 +152,4 @@ static inline int nfsd_create_is_exclusive(int createmode)
               || createmode == NFS4_CREATE_EXCLUSIVE4_1;
 }
 
-static inline bool nfsd_eof_on_read(long requested, long read,
-                               loff_t offset, loff_t size)
-{
-       /* We assume a short read means eof: */
-       if (requested > read)
-               return true;
-       /*
-        * A non-short read might also reach end of file.  The spec
-        * still requires us to set eof in that case.
-        *
-        * Further operations may have modified the file size since
-        * the read, so the following check is not atomic with the read.
-        * We've only seen that cause a problem for a client in the case
-        * where the read returned a count of 0 without setting eof.
-        * That case was fixed by the addition of the above check.
-        */
-       return (offset + read >= size);
-}
-
 #endif /* LINUX_NFSD_VFS_H */

diff --git a/fs/nfsd/xdr3.h b/fs/nfsd/xdr3.h
index 2cb29e961a760f6797abfe9317c4903fba868c93..99ff9f403ff18365b501f8c18fb43c2b538eae59 100644 (file)
--- a/fs/nfsd/xdr3.h
+++ b/fs/nfsd/xdr3.h
@@ -151,7 +151,7 @@ struct nfsd3_readres {
        __be32                  status;
        struct svc_fh           fh;
        unsigned long           count;
-       int                     eof;
+       __u32                   eof;
 };
 
 struct nfsd3_writeres {