nfsd: use short read as well as i_size to set eof

Use the result of a local read to determine when to set the eof flag.  This
allows us to return the location of the end of the file atomically at the
time of the read.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
[bfields: add some documentation]
Signed-off-by: J. Bruce Fields <bfields@redhat.com>

diff --git a/fs/nfsd/nfs3proc.c b/fs/nfsd/nfs3proc.c
index 7b755b7f785ceb4deff9fd9a5fbafccef15efa3f..51c3b06e80365eadc9e810071e729e4669928f7f 100644 (file)
--- a/fs/nfsd/nfs3proc.c
+++ b/fs/nfsd/nfs3proc.c
@@ -147,6 +147,7 @@ nfsd3_proc_read(struct svc_rqst *rqstp, struct nfsd3_readargs *argp,
 {
        __be32  nfserr;
        u32     max_blocksize = svc_max_payload(rqstp);
+       unsigned long cnt = min(argp->count, max_blocksize);
 
        dprintk("nfsd: READ(3) %s %lu bytes at %Lu\n",
                                SVCFH_fmt(&argp->fh),
@@ -157,7 +158,7 @@ nfsd3_proc_read(struct svc_rqst *rqstp, struct nfsd3_readargs *argp,
         * 1 (status) + 22 (post_op_attr) + 1 (count) + 1 (eof)
         * + 1 (xdr opaque byte count) = 26
         */
-       resp->count = min(argp->count, max_blocksize);
+       resp->count = cnt;
        svc_reserve_auth(rqstp, ((1 + NFS3_POST_OP_ATTR_WORDS + 3)<<2) + resp->count +4);
 
        fh_copy(&resp->fh, &argp->fh);
@@ -167,8 +168,8 @@ nfsd3_proc_read(struct svc_rqst *rqstp, struct nfsd3_readargs *argp,
                                  &resp->count);
        if (nfserr == 0) {
                struct inode    *inode = d_inode(resp->fh.fh_dentry);
-
-               resp->eof = (argp->offset + resp->count) >= inode->i_size;
+               resp->eof = nfsd_eof_on_read(cnt, resp->count, argp->offset,
+                                                       inode->i_size);
        }
 
        RETURN_STATUS(nfserr);

diff --git a/fs/nfsd/nfs4xdr.c b/fs/nfsd/nfs4xdr.c
index aa87954b4af28afb4dfe5d696a42ab30338a3009..9df898ba648f73a14b1e42be47dbfddf189ec0a6 100644 (file)
--- a/fs/nfsd/nfs4xdr.c
+++ b/fs/nfsd/nfs4xdr.c
@@ -3365,6 +3365,7 @@ static __be32 nfsd4_encode_splice_read(
        struct xdr_stream *xdr = &resp->xdr;
        struct xdr_buf *buf = xdr->buf;
        u32 eof;
+       long len;
        int space_left;
        __be32 nfserr;
        __be32 *p = xdr->p - 2;
@@ -3373,6 +3374,7 @@ static __be32 nfsd4_encode_splice_read(
        if (xdr->end - xdr->p < 1)
                return nfserr_resource;
 
+       len = maxcount;
        nfserr = nfsd_splice_read(read->rd_rqstp, file,
                                  read->rd_offset, &maxcount);
        if (nfserr) {
@@ -3385,8 +3387,8 @@ static __be32 nfsd4_encode_splice_read(
                return nfserr;
        }
 
-       eof = (read->rd_offset + maxcount >=
-              d_inode(read->rd_fhp->fh_dentry)->i_size);
+       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);
@@ -3456,14 +3458,15 @@ static __be32 nfsd4_encode_readv(struct nfsd4_compoundres *resp,
        }
        read->rd_vlen = v;
 
+       len = maxcount;
        nfserr = nfsd_readv(file, read->rd_offset, resp->rqstp->rq_vec,
                        read->rd_vlen, &maxcount);
        if (nfserr)
                return nfserr;
        xdr_truncate_encode(xdr, starting_len + 8 + ((maxcount+3)&~3));
 
-       eof = (read->rd_offset + maxcount >=
-              d_inode(read->rd_fhp->fh_dentry)->i_size);
+       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);

diff --git a/fs/nfsd/vfs.h b/fs/nfsd/vfs.h
index c11ba316f23f9b36b6ff57f7d9e5be94a0771962..2d573ec057f801d6406ccf51d21549eb59009a44 100644 (file)
--- a/fs/nfsd/vfs.h
+++ b/fs/nfsd/vfs.h
@@ -139,4 +139,23 @@ 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 */