nfsd4: fix corruption of NFSv4 read data

The calculation of page_ptr here is wrong in the case the read doesn't
start at an offset that is a multiple of a page.

The result is that nfs4svc_encode_compoundres sets rq_next_page to a
value one too small, and then the loop in svc_free_res_pages may
incorrectly fail to clear a page pointer in rq_respages[].

Pages left in rq_respages[] are available for the next rpc request to
use, so xdr data may be written to that page, which may hold data still
waiting to be transmitted to the client or data in the page cache.

The observed result was silent data corruption seen on an NFSv4 client.

We tag this as "fixing" 05638dc because that commit exposed this
bug, though the incorrect calculation predates it.

Particular thanks to Andrea Arcangeli and David Gilbert for analysis and
testing.

Fixes: 05638dc "nfsd4: simplify server xdr->next_page use"
Cc: stable@vger.kernel.org
Reported-by: Andrea Arcangeli <aarcange@redhat.com>
Tested-by: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>

fs/nfsd/nfs4xdr.c

@@ -3104,7 +3104,8 @@ static __be32 nfsd4_encode_splice_read(
 
 	buf->page_len = maxcount;
 	buf->len += maxcount;
-	xdr->page_ptr += (maxcount + PAGE_SIZE - 1) / PAGE_SIZE;
+	xdr->page_ptr += (buf->page_base + maxcount + PAGE_SIZE - 1)
+							/ PAGE_SIZE;
 
 	/* Use rest of head for padding and remaining ops: */
 	buf->tail[0].iov_base = xdr->p;