Binding a privileged port in rpc.statd.

From abdaa0c27d66c71c9500b983fc4366cb3e514fff Mon Sep 17 00:00:00 2001

Hi,

It seems that rpc.statd will fail to bind to a (free) port, if the first port it receives is listed in /etc/services. The bindresvport will keep on returning the same port if we close the socket too early.

The attached patch should fix this.

Signature · 2 lines

-- 
Hemmo

Quoted 1 line

From abdaa0c27d66c71c9500b983fc4366cb3e514fff Mon Sep 17 00:00:00 2001

From: Hemmo Nieminen <[email protected]>
Date: Thu, 3 Jan 2013 10:00:17 +0200
Subject: [PATCH] Fix socket binding loop.

Instead of closing the sockets before requesting a new one, keep them open until a suitable one is found. Otherwise bindresvport will return the same port over and over again.

---
 utils/statd/rmtcall.c | 19 +++++++++++++------
 1 file changed, 13 insertions(+), 6 deletions(-)

utils/statd/rmtcall.c +13 -6

index 4ecb03c..fd576d9 100644
@@ -68,21 +68,19 @@ statd_get_socket(void)
 {
 	struct sockaddr_in	sin;
 	struct servent *se;
-	int loopcnt = 100;
+	const int loopcnt = 100;
+	int i, tmp_sockets[loopcnt];
 
 	if (sockfd >= 0)
 		return sockfd;
 
-	while (loopcnt-- > 0) {
-
-		if (sockfd >= 0) close(sockfd);
+	for (i = 0; i < loopcnt; ++i) {
 
 		if ((sockfd = socket(AF_INET, SOCK_DGRAM, IPPROTO_UDP)) < 0) {
 			xlog(L_ERROR, "%s: Can't create socket: %m", __func__);
-			return -1;
+			break;
 		}
 
-
 		memset(&sin, 0, sizeof(sin));
 		sin.sin_family = AF_INET;
 		sin.sin_addr.s_addr = htonl(INADDR_LOOPBACK);

@@ -96,7 +94,16 @@ statd_get_socket(void)
 		if (se == NULL)
 			break;
 		/* rather not use that port, try again */
+
+		tmp_sockets[i] = sockfd;
 	}
+
+	while (--i >= 0)
+		close(tmp_sockets[i]);
+
+	if (sockfd < 0)
+		return -1;
+
 	FD_SET(sockfd, &SVC_FDSET);
 	return sockfd;
 }

Signature · 1 line

1.8.1

Hi,

It seems that rpc.statd will fail to bind to a (free) port, if the first port it receives is listed in /etc/services. The bindresvport will keep on returning the same port if we close the socket too early.

The attached patch should fix this.