libceph: enable ceph in a non-default network namespace

Grab a reference on a network namespace of the 'rbd map' (in case of
rbd) or 'mount' (in case of ceph) process and use that to open sockets
instead of always using init_net and bailing if network namespace is
anything but init_net.  Be careful to not share struct ceph_client
instances between different namespaces and don't add any code in the
!CONFIG_NET_NS case.

This is based on a patch from Hong Zhiguo <zhiguohong@tencent.com>.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Sage Weil <sage@redhat.com>

2 files changed, 19 insertions, 7 deletions

diff --git a/net/ceph/ceph_common.c b/net/ceph/ceph_common.c
index cb7db320dd27..f30329f72641 100644
--- a/net/ceph/ceph_common.c
+++ b/net/ceph/ceph_common.c
@@ -9,6 +9,7 @@
 #include <keys/ceph-type.h>
 #include <linux/module.h>
 #include <linux/mount.h>
+#include <linux/nsproxy.h>
 #include <linux/parser.h>
 #include <linux/sched.h>
 #include <linux/seq_file.h>
@@ -16,8 +17,6 @@
 #include <linux/statfs.h>
 #include <linux/string.h>
 #include <linux/vmalloc.h>
-#include <linux/nsproxy.h>
-#include <net/net_namespace.h>
 
 
 #include <linux/ceph/ceph_features.h>
@@ -131,6 +130,13 @@ int ceph_compare_options(struct ceph_options *new_opt,
         int i;
         int ret;
 
+        /*
+         * Don't bother comparing options if network namespaces don't
+         * match.
+         */
+        if (!net_eq(current->nsproxy->net_ns, read_pnet(&client->msgr.net)))
+                return -1;
+
         ret = memcmp(opt1, opt2, ofs);
         if (ret)
                 return ret;
@@ -335,9 +341,6 @@ ceph_parse_options(char *options, const char *dev_name,
         int err = -ENOMEM;
         substring_t argstr[MAX_OPT_ARGS];
 
-        if (current->nsproxy->net_ns != &init_net)
-                return ERR_PTR(-EINVAL);
-
         opt = kzalloc(sizeof(*opt), GFP_KERNEL);
         if (!opt)
                 return ERR_PTR(-ENOMEM);
@@ -608,6 +611,7 @@ struct ceph_client *ceph_create_client(struct ceph_options *opt, void *private,
 fail_monc:
         ceph_monc_stop(&client->monc);
 fail:
+        ceph_messenger_fini(&client->msgr);
         kfree(client);
         return ERR_PTR(err);
 }
@@ -621,8 +625,8 @@ void ceph_destroy_client(struct ceph_client *client)
 
         /* unmount */
         ceph_osdc_stop(&client->osdc);
-
         ceph_monc_stop(&client->monc);
+        ceph_messenger_fini(&client->msgr);
 
         ceph_debugfs_client_cleanup(client);
 
diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index 1679f47280e2..5c1f98ea6741 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -6,6 +6,7 @@
 #include <linux/inet.h>
 #include <linux/kthread.h>
 #include <linux/net.h>
+#include <linux/nsproxy.h>
 #include <linux/slab.h>
 #include <linux/socket.h>
 #include <linux/string.h>
@@ -479,7 +480,7 @@ static int ceph_tcp_connect(struct ceph_connection *con)
         int ret;
 
         BUG_ON(con->sock);
-        ret = sock_create_kern(&init_net, con->peer_addr.in_addr.ss_family,
+        ret = sock_create_kern(read_pnet(&con->msgr->net), paddr->ss_family,
                                SOCK_STREAM, IPPROTO_TCP, &sock);
         if (ret)
                 return ret;
@@ -2944,11 +2945,18 @@ void ceph_messenger_init(struct ceph_messenger *msgr,
         msgr->tcp_nodelay = tcp_nodelay;
 
         atomic_set(&msgr->stopping, 0);
+        write_pnet(&msgr->net, get_net(current->nsproxy->net_ns));
 
         dout("%s %p\n", __func__, msgr);
 }
 EXPORT_SYMBOL(ceph_messenger_init);
 
+void ceph_messenger_fini(struct ceph_messenger *msgr)
+{
+        put_net(read_pnet(&msgr->net));
+}
+EXPORT_SYMBOL(ceph_messenger_fini);
+
 static void clear_standby(struct ceph_connection *con)
 {
         /* come back from STANDBY? */