NFS: Fix a problem with the legacy binary mount code
authorTrond Myklebust <Trond.Myklebust@netapp.com>
Tue, 4 Sep 2012 15:05:07 +0000 (11:05 -0400)
committerTrond Myklebust <Trond.Myklebust@netapp.com>
Tue, 4 Sep 2012 18:52:43 +0000 (14:52 -0400)
Apparently, am-utils is still using the legacy binary mountdata interface,
and is having trouble parsing /proc/mounts due to the 'port=' field being
incorrectly set.

The following patch should fix up the regression.

Reported-by: Marius Tolzmann <tolzmann@molgen.mpg.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@vger.kernel.org
fs/nfs/super.c

index 239aff7338eb89ee8c0d4080694178317d84929e..b8eda700584bfbd25086d898a21b3fd03959030d 100644 (file)
@@ -1867,6 +1867,7 @@ static int nfs23_validate_mount_data(void *options,
 
                memcpy(sap, &data->addr, sizeof(data->addr));
                args->nfs_server.addrlen = sizeof(data->addr);
+               args->nfs_server.port = ntohs(data->addr.sin_port);
                if (!nfs_verify_server_address(sap))
                        goto out_no_address;
 
@@ -2564,6 +2565,7 @@ static int nfs4_validate_mount_data(void *options,
                        return -EFAULT;
                if (!nfs_verify_server_address(sap))
                        goto out_no_address;
+               args->nfs_server.port = ntohs(((struct sockaddr_in *)sap)->sin_port);
 
                if (data->auth_flavourlen) {
                        if (data->auth_flavourlen > 1)