[Devel] [PATCH RHEL7 COMMIT] netlink/diag: report flags for netlink sockets

[Konstantin Khorenko]

The commit is pushed to "branch-rh7-3.10.0-327.18.2.vz7.14.x-ovz" and will appear at https://src.openvz.org/scm/ovz/vzkernel.git
after rh7-3.10.0-327.18.2.vz7.14.15
------>
commit cb77f92e45e65639d720387f4ae297e4e515ce9c
Author: Andrey Vagin <avagin at openvz.org>
Date:   Fri Jun 17 14:27:31 2016 +0400

    netlink/diag: report flags for netlink sockets
    
    We need to know flags for dumping and restoring netlink sockets.
    
    All flags except NDIAG_FLAG_CB_RUNNING can be received with help of
    getsockopt(), but in this case we need a socket descriptor and we need
    to call getsockopt() to get each flag.
    
    With this chages we will be able to show netlink sockets flags from the
    ss tool.
    
    In criu, we need to know if a callback is running now or not.
    
    When a socket has some data in a receive queue and doesn't have running
    callbacks, we can save all data from a receive queue on dump and queue
    them back on restore.
    
    If a socket has a running callback, a receive queue contains only a part
    of data, and as soon as we read them, the callback will generate a new
    portion. In this case, we can't be sure that all data will not exceed a
    buffer limit on restore.
    
    Now we are going to dump and restore sockets without running callbacks.
    
    https://jira.sw.ru/browse/PSBM-28386
    
    Signed-off-by: Andrey Vagin <avagin at virtuozzo.com>
    Reviewed-by: Cyrill Gorcunov <gorcunov at virtuozzo.com>
---
 include/uapi/linux/netlink_diag.h | 10 ++++++++++
 net/netlink/af_netlink.c          |  9 ---------
 net/netlink/af_netlink.h          |  9 +++++++++
 net/netlink/diag.c                | 28 +++++++++++++++++++++++++++-
 4 files changed, 46 insertions(+), 10 deletions(-)

diff --git a/include/uapi/linux/netlink_diag.h b/include/uapi/linux/netlink_diag.h
index 4e31db4..6a9108f 100644
--- a/include/uapi/linux/netlink_diag.h
+++ b/include/uapi/linux/netlink_diag.h
@@ -37,6 +37,7 @@ enum {
 	NETLINK_DIAG_GROUPS,
 	NETLINK_DIAG_RX_RING,
 	NETLINK_DIAG_TX_RING,
+	NETLINK_DIAG_FLAGS,
 
 	__NETLINK_DIAG_MAX,
 };
@@ -48,5 +49,14 @@ enum {
 #define NDIAG_SHOW_MEMINFO	0x00000001 /* show memory info of a socket */
 #define NDIAG_SHOW_GROUPS	0x00000002 /* show groups of a netlink socket */
 #define NDIAG_SHOW_RING_CFG	0x00000004 /* show ring configuration */
+#define NDIAG_SHOW_FLAGS	0x00000008 /* show flags of a netlink socket */
+
+/* flags */
+#define NDIAG_FLAG_CB_RUNNING		0x00000001
+#define NDIAG_FLAG_PKTINFO		0x00000002
+#define NDIAG_FLAG_BROADCAST_ERROR	0x00000004
+#define NDIAG_FLAG_NO_ENOBUFS		0x00000008
+#define NDIAG_FLAG_LISTEN_ALL_NSID	0x00000010
+#define NDIAG_FLAG_CAP_ACK		0x00000020
 
 #endif
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 113e2ae..ba75f32 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -77,15 +77,6 @@ struct listeners {
 /* state bits */
 #define NETLINK_S_CONGESTED		0x0
 
-/* flags */
-#define NETLINK_F_KERNEL_SOCKET		0x1
-#define NETLINK_F_RECV_PKTINFO		0x2
-#define NETLINK_F_BROADCAST_SEND_ERROR	0x4
-#define NETLINK_F_RECV_NO_ENOBUFS	0x8
-#define NETLINK_F_LISTEN_ALL_NSID	0x10
-#define NETLINK_F_CAP_ACK		0x20
-#define NETLINK_F_REPAIR		0x40
-
 static inline int netlink_is_kernel(struct sock *sk)
 {
 	return nlk_sk(sk)->flags & NETLINK_F_KERNEL_SOCKET;
diff --git a/net/netlink/af_netlink.h b/net/netlink/af_netlink.h
index 577fddf..b3ce345 100644
--- a/net/netlink/af_netlink.h
+++ b/net/netlink/af_netlink.h
@@ -4,6 +4,15 @@
 #include <linux/rhashtable.h>
 #include <net/sock.h>
 
+/* flags */
+#define NETLINK_F_KERNEL_SOCKET		0x1
+#define NETLINK_F_RECV_PKTINFO		0x2
+#define NETLINK_F_BROADCAST_SEND_ERROR	0x4
+#define NETLINK_F_RECV_NO_ENOBUFS	0x8
+#define NETLINK_F_LISTEN_ALL_NSID	0x10
+#define NETLINK_F_CAP_ACK		0x20
+#define NETLINK_F_REPAIR		0x40
+
 #define NLGRPSZ(x)	(ALIGN(x, sizeof(unsigned long) * 8) / 8)
 #define NLGRPLONGS(x)	(NLGRPSZ(x)/sizeof(unsigned long))
 
diff --git a/net/netlink/diag.c b/net/netlink/diag.c
index de8c74a..0aa8744e 100644
--- a/net/netlink/diag.c
+++ b/net/netlink/diag.c
@@ -54,6 +54,27 @@ static int sk_diag_dump_groups(struct sock *sk, struct sk_buff *nlskb)
 		       nlk->groups);
 }
 
+static int sk_diag_put_flags(struct sock *sk, struct sk_buff *skb)
+{
+	struct netlink_sock *nlk = nlk_sk(sk);
+	u32 flags = 0;
+
+	if (nlk->cb_running)
+		flags |= NDIAG_FLAG_CB_RUNNING;
+	if (nlk->flags & NETLINK_F_RECV_PKTINFO)
+		flags |= NDIAG_FLAG_PKTINFO;
+	if (nlk->flags & NETLINK_F_BROADCAST_SEND_ERROR)
+		flags |= NDIAG_FLAG_BROADCAST_ERROR;
+	if (nlk->flags & NETLINK_F_RECV_NO_ENOBUFS)
+		flags |= NDIAG_FLAG_NO_ENOBUFS;
+	if (nlk->flags & NETLINK_F_LISTEN_ALL_NSID)
+		flags |= NDIAG_FLAG_LISTEN_ALL_NSID;
+	if (nlk->flags & NETLINK_F_CAP_ACK)
+		flags |= NDIAG_FLAG_CAP_ACK;
+
+	return nla_put_u32(skb, NETLINK_DIAG_FLAGS, flags);
+}
+
 static int sk_diag_fill(struct sock *sk, struct sk_buff *skb,
 			struct netlink_diag_req *req,
 			u32 portid, u32 seq, u32 flags, int sk_ino)
@@ -91,7 +112,12 @@ static int sk_diag_fill(struct sock *sk, struct sk_buff *skb,
 	    sk_diag_put_rings_cfg(sk, skb))
 		goto out_nlmsg_trim;
 
-	return nlmsg_end(skb, nlh);
+	if ((req->ndiag_show & NDIAG_SHOW_FLAGS) &&
+	    sk_diag_put_flags(sk, skb))
+		goto out_nlmsg_trim;
+
+	nlmsg_end(skb, nlh);
+	return 0;
 
 out_nlmsg_trim:
 	nlmsg_cancel(skb, nlh);