NFSD: Fix READDIR on NFSv3 mounts of ext4 exports

I noticed that recently, simple operations like "make" started
failing on NFSv3 mounts of ext4 exports. Network capture shows that
READDIRPLUS operated correctly but READDIR failed with
NFS3ERR_INVAL. The vfs_llseek() call returned EINVAL when it is
passed a non-zero starting directory cookie.

I bisected to commit c689bdd ("nfsd: further centralize
protocol version checks.").

Turns out that nfsd3_proc_readdir() does not call fh_verify() before
it calls nfsd_readdir(), so the new fhp->fh_64bit_cookies boolean is
not set properly. This leaves the NFSD_MAY_64BIT_COOKIE unset when
the directory is opened.

For ext4, this causes the wrong "max file size" value to be used
when sanity checking the incoming directory cookie (which is a seek
offset value).

The fhp->fh_64bit_cookies boolean is /always/ properly initialized
after nfsd_open() returns. There doesn't seem to be a reason for the
generic NFSD open helper to handle the f_mode fix-up for
directories, so just move that to the one caller that tries to open
an S_IFDIR with NFSD_MAY_64BIT_COOKIE.

Suggested-by: NeilBrown <neilb@suse.de>
Fixes: c689bdd ("nfsd: further centralize protocol version checks.")
Reviewed-by: NeilBrown <neilb@suse.de>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

fs/nfsd/vfs.c

@@ -903,11 +903,6 @@ __nfsd_open(struct svc_rqst *rqstp, struct svc_fh *fhp, umode_t type,
 		goto out;
 	}
 
-	if (may_flags & NFSD_MAY_64BIT_COOKIE)
-		file->f_mode |= FMODE_64BITHASH;
-	else
-		file->f_mode |= FMODE_32BITHASH;
-
 	*filp = file;
 out:
 	return host_err;
@@ -2174,13 +2169,15 @@ nfsd_readdir(struct svc_rqst *rqstp, struct svc_fh *fhp, loff_t *offsetp,
 	loff_t		offset = *offsetp;
 	int             may_flags = NFSD_MAY_READ;
 
-	if (fhp->fh_64bit_cookies)
-		may_flags |= NFSD_MAY_64BIT_COOKIE;
-
 	err = nfsd_open(rqstp, fhp, S_IFDIR, may_flags, &file);
 	if (err)
 		goto out;
 
+	if (fhp->fh_64bit_cookies)
+		file->f_mode |= FMODE_64BITHASH;
+	else
+		file->f_mode |= FMODE_32BITHASH;
+
 	offset = vfs_llseek(file, offset, SEEK_SET);
 	if (offset < 0) {
 		err = nfserrno((int)offset);