All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH iptables] iptables: use flock() instead of abstract unix sockets
@ 2015-01-19 18:09 Pablo Neira Ayuso
  2015-01-19 18:36 ` Lennart Poettering
  2015-01-24 22:35 ` Phil Oester
  0 siblings, 2 replies; 3+ messages in thread
From: Pablo Neira Ayuso @ 2015-01-19 18:09 UTC (permalink / raw)
  To: netfilter-devel; +Cc: kernel, lennart

Abstract unix sockets cannot be used to synchronize several concurrent
instances of iptables since an unpriviledged process can create them and
prevent the legitimate iptables instance from running.

Use flock() and /run instead as suggested by Lennart Poettering.

Fixes: 93587a0 ("ip[6]tables: Add locking to prevent concurrent instances")
Reported-by: Lennart Poettering <lennart@poettering.net>
Cc: Phil Oester <kernel@linuxace.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 iptables/xshared.c |   22 +++++++---------------
 1 file changed, 7 insertions(+), 15 deletions(-)

diff --git a/iptables/xshared.c b/iptables/xshared.c
index b18022e..7beb86b 100644
--- a/iptables/xshared.c
+++ b/iptables/xshared.c
@@ -9,11 +9,11 @@
 #include <sys/socket.h>
 #include <sys/un.h>
 #include <unistd.h>
+#include <fcntl.h>
 #include <xtables.h>
 #include "xshared.h"
 
-#define XT_SOCKET_NAME "xtables"
-#define XT_SOCKET_LEN 8
+#define XT_LOCK_NAME	"/run/xtables.lock"
 
 /*
  * Print out any special helps. A user might like to be able to add a --help
@@ -245,22 +245,14 @@ void xs_init_match(struct xtables_match *match)
 
 bool xtables_lock(int wait)
 {
-	int i = 0, ret, xt_socket;
-	struct sockaddr_un xt_addr;
-	int waited = 0;
-
-	memset(&xt_addr, 0, sizeof(xt_addr));
-	xt_addr.sun_family = AF_UNIX;
-	strcpy(xt_addr.sun_path+1, XT_SOCKET_NAME);
-	xt_socket = socket(AF_UNIX, SOCK_STREAM, 0);
-	/* If we can't even create a socket, fall back to prior (lockless) behavior */
-	if (xt_socket < 0)
+	int fd, waited = 0, i = 0;
+
+	fd = open(XT_LOCK_NAME, O_CREAT, 0600);
+	if (fd < 0)
 		return true;
 
 	while (1) {
-		ret = bind(xt_socket, (struct sockaddr*)&xt_addr,
-			   offsetof(struct sockaddr_un, sun_path)+XT_SOCKET_LEN);
-		if (ret == 0)
+		if (flock(fd, LOCK_EX | LOCK_NB) == 0)
 			return true;
 		else if (wait >= 0 && waited >= wait)
 			return false;
-- 
1.7.10.4


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH iptables] iptables: use flock() instead of abstract unix sockets
  2015-01-19 18:09 [PATCH iptables] iptables: use flock() instead of abstract unix sockets Pablo Neira Ayuso
@ 2015-01-19 18:36 ` Lennart Poettering
  2015-01-24 22:35 ` Phil Oester
  1 sibling, 0 replies; 3+ messages in thread
From: Lennart Poettering @ 2015-01-19 18:36 UTC (permalink / raw)
  To: Pablo Neira Ayuso; +Cc: netfilter-devel, kernel

On Mon, 19.01.15 19:09, Pablo Neira Ayuso (pablo@netfilter.org) wrote:

> Abstract unix sockets cannot be used to synchronize several concurrent
> instances of iptables since an unpriviledged process can create them and
> prevent the legitimate iptables instance from running.
> 
> Use flock() and /run instead as suggested by Lennart Poettering.

Looks OK. Of course, it's a bit nasty to do the sleep() loop, but
there is no time-limited version of flock(), hence doing the sleep()
loop is kinda necessary, unless one wants to use SIGARLM, but that's
awful to do without races...

Hence, looks OK to me.

A minor optimization might be to move the lock file into its own
subdir /run/iptables/ or so, but it's OK if you don't.

Lennart

-- 
Lennart Poettering, Red Hat

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH iptables] iptables: use flock() instead of abstract unix sockets
  2015-01-19 18:09 [PATCH iptables] iptables: use flock() instead of abstract unix sockets Pablo Neira Ayuso
  2015-01-19 18:36 ` Lennart Poettering
@ 2015-01-24 22:35 ` Phil Oester
  1 sibling, 0 replies; 3+ messages in thread
From: Phil Oester @ 2015-01-24 22:35 UTC (permalink / raw)
  To: Pablo Neira Ayuso; +Cc: netfilter-devel, lennart

On Mon, Jan 19, 2015 at 07:09:17PM +0100, Pablo Neira Ayuso wrote:
> Abstract unix sockets cannot be used to synchronize several concurrent
> instances of iptables since an unpriviledged process can create them and
> prevent the legitimate iptables instance from running.
> 
> Use flock() and /run instead as suggested by Lennart Poettering.
> 
> Fixes: 93587a0 ("ip[6]tables: Add locking to prevent concurrent instances")
> Reported-by: Lennart Poettering <lennart@poettering.net>

Looks good Pablo.  Thanks for finding this Lennart.  

Phil

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2015-01-24 22:35 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-01-19 18:09 [PATCH iptables] iptables: use flock() instead of abstract unix sockets Pablo Neira Ayuso
2015-01-19 18:36 ` Lennart Poettering
2015-01-24 22:35 ` Phil Oester

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.