svcrpc: fix handling of too-short rpc's

If we detect that an rpc is too short, we abort and close the
connection.  Except, there's a bug here: we're leaving sk_datalen
nonzero without leaving any pages in the sk_pages array.  The most
likely result of the inconsistency is a subsequent crash in
svc_tcp_clear_pages.

Also demote the BUG_ON in svc_tcp_clear_pages to a WARN.

Cc: stable@kernel.org
Signed-off-by: J. Bruce Fields <bfields@redhat.com>

diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 0f679df7d072794094ece1aa48988e834395d849..df74919c81c01aa250f28fc1cc32458b4835bf1f 100644 (file)
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -917,7 +917,10 @@ static void svc_tcp_clear_pages(struct svc_sock *svsk)
        len = svsk->sk_datalen;
        npages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT;
        for (i = 0; i < npages; i++) {
-               BUG_ON(svsk->sk_pages[i] == NULL);
+               if (svsk->sk_pages[i] == NULL) {
+                       WARN_ON_ONCE(1);
+                       continue;
+               }
                put_page(svsk->sk_pages[i]);
                svsk->sk_pages[i] = NULL;
        }
@@ -1092,8 +1095,10 @@ static int svc_tcp_recvfrom(struct svc_rqst *rqstp)
                goto err_noclose;
        }
 
-       if (svc_sock_reclen(svsk) < 8)
+       if (svc_sock_reclen(svsk) < 8) {
+               svsk->sk_datalen = 0;
                goto err_delete; /* client is nuts. */
+       }
 
        rqstp->rq_arg.len = svsk->sk_datalen;
        rqstp->rq_arg.page_base = 0;