cifs: fix potential buffer overrun when composing a new options string
authorJeff Layton <jlayton@redhat.com>
Fri, 24 May 2013 11:40:04 +0000 (07:40 -0400)
committerSteve French <sfrench@us.ibm.com>
Fri, 24 May 2013 18:08:19 +0000 (13:08 -0500)
Consider the case where we have a very short ip= string in the original
mount options, and when we chase a referral we end up with a very long
IPv6 address. Be sure to allow for that possibility when estimating the
size of the string to allocate.

Cc: <stable@vger.kernel.org>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
fs/cifs/cifs_dfs_ref.c

index 8e33ec6..8e5a181 100644 (file)
@@ -18,6 +18,7 @@
 #include <linux/slab.h>
 #include <linux/vfs.h>
 #include <linux/fs.h>
+#include <linux/inet.h>
 #include "cifsglob.h"
 #include "cifsproto.h"
 #include "cifsfs.h"
@@ -150,7 +151,8 @@ char *cifs_compose_mount_options(const char *sb_mountdata,
         * assuming that we have 'unc=' and 'ip=' in
         * the original sb_mountdata
         */
-       md_len = strlen(sb_mountdata) + rc + strlen(ref->node_name) + 12;
+       md_len = strlen(sb_mountdata) + rc + strlen(ref->node_name) + 12 +
+                       INET6_ADDRSTRLEN;
        mountdata = kzalloc(md_len+1, GFP_KERNEL);
        if (mountdata == NULL) {
                rc = -ENOMEM;