[PATCH] VSOCK: bind to random port for VMADDR_PORT_ANY

The old code always starts from fixed port for VMADDR_PORT_ANY. Sometimes
when VMM crashed, there is still orphaned vsock which is waiting for
close timer, then it could cause connection time out for new started VM
if they are trying to connect to same port with same guest cid since the
new packets could hit that orphaned vsock. We could also fix this by doing
more in vhost_vsock_reset_orphans, but any way, it should be better to start
from a random local port instead of a fixed one.

Signed-off-by: Lepton Wu <redacted>
---
 net/vmw_vsock/af_vsock.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index ab27a2872935..73817e846a1f 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c

@@ -107,6 +107,7 @@
 #include <linux/mutex.h>
 #include <linux/net.h>
 #include <linux/poll.h>
+#include <linux/random.h>
 #include <linux/skbuff.h>
 #include <linux/smp.h>
 #include <linux/socket.h>

@@ -504,9 +505,12 @@ static void vsock_pending_work(struct work_struct *work)
 static int __vsock_bind_stream(struct vsock_sock *vsk,
 			       struct sockaddr_vm *addr)
 {
-	static u32 port = LAST_RESERVED_PORT + 1;
+	static u32 port = 0;
 	struct sockaddr_vm new_addr;
 
+	if (!port)
+		port = prandom_u32();
+
 	vsock_addr_init(&new_addr, addr->svm_cid, addr->svm_port);
 
 	if (addr->svm_port == VMADDR_PORT_ANY) {

-- 
2.20.0.rc2.403.gdbc3b29805-goog

On Mon, Dec 10, 2018 at 11:02:35PM -0800, Lepton Wu wrote:

The old code always starts from fixed port for VMADDR_PORT_ANY. Sometimes
when VMM crashed, there is still orphaned vsock which is waiting for
close timer, then it could cause connection time out for new started VM
if they are trying to connect to same port with same guest cid since the
new packets could hit that orphaned vsock. We could also fix this by doing
more in vhost_vsock_reset_orphans, but any way, it should be better to start
from a random local port instead of a fixed one.

Signed-off-by: Lepton Wu <redacted>
---
 net/vmw_vsock/af_vsock.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Jorgen, Dexuan: Any objection to this?  It also affects the other
AF_VSOCK transports.

On Mon, Dec 10, 2018 at 11:02:35PM -0800, Lepton Wu wrote:

quoted

The old code always starts from fixed port for VMADDR_PORT_ANY. Sometimes
when VMM crashed, there is still orphaned vsock which is waiting for
close timer, then it could cause connection time out for new started VM
if they are trying to connect to same port with same guest cid since the
new packets could hit that orphaned vsock. We could also fix this by doing
more in vhost_vsock_reset_orphans, but any way, it should be better to start
from a random local port instead of a fixed one.

Signed-off-by: Lepton Wu <redacted>
---
 net/vmw_vsock/af_vsock.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Jorgen, Dexuan: Any objection to this?  It also affects the other
AF_VSOCK transports.

Makes sense to me.

quoted

diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index ab27a2872935..73817e846a1f 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c

@@ -107,6 +107,7 @@
 #include <linux/mutex.h>
 #include <linux/net.h>
 #include <linux/poll.h>
+#include <linux/random.h>
 #include <linux/skbuff.h>
 #include <linux/smp.h>
 #include <linux/socket.h>

@@ -504,9 +505,12 @@ static void vsock_pending_work(struct work_struct *work)
 static int __vsock_bind_stream(struct vsock_sock *vsk,
                             struct sockaddr_vm *addr)
 {
-     static u32 port = LAST_RESERVED_PORT + 1;
+     static u32 port = 0;
      struct sockaddr_vm new_addr;

+     if (!port)
+             port = prandom_u32();
+

How about making this:
   port = LAST_RESERVED_PORT + 1 + prandom_u32_max(U32_MAX - LAST_RESERVED_PORT);
so the initial assignment is a valid port in the unreserved range. It will be corrected in the first iteration,
but this would make the intention clearer.

quoted

      vsock_addr_init(&new_addr, addr->svm_cid, addr->svm_port);

      if (addr->svm_port == VMADDR_PORT_ANY) {
--
2.20.0.rc2.403.gdbc3b29805-goog

Thanks,
Jorgen

The old code always starts from fixed port for VMADDR_PORT_ANY. Sometimes
when VMM crashed, there is still orphaned vsock which is waiting for
close timer, then it could cause connection time out for new started VM
if they are trying to connect to same port with same guest cid since the
new packets could hit that orphaned vsock. We could also fix this by doing
more in vhost_vsock_reset_orphans, but any way, it should be better to start
from a random local port instead of a fixed one.

Signed-off-by: Lepton Wu <redacted>
---
 net/vmw_vsock/af_vsock.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index ab27a2872935..43a1dec08825 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c

@@ -107,6 +107,7 @@
 #include <linux/mutex.h>
 #include <linux/net.h>
 #include <linux/poll.h>
+#include <linux/random.h>
 #include <linux/skbuff.h>
 #include <linux/smp.h>
 #include <linux/socket.h>

@@ -504,9 +505,13 @@ static void vsock_pending_work(struct work_struct *work)
 static int __vsock_bind_stream(struct vsock_sock *vsk,
 			       struct sockaddr_vm *addr)
 {
-	static u32 port = LAST_RESERVED_PORT + 1;
+	static u32 port = 0;
 	struct sockaddr_vm new_addr;
 
+	if (!port)
+		port = LAST_RESERVED_PORT + 1 +
+			prandom_u32_max(U32_MAX - LAST_RESERVED_PORT);
+
 	vsock_addr_init(&new_addr, addr->svm_cid, addr->svm_port);
 
 	if (addr->svm_port == VMADDR_PORT_ANY) {

-- 
2.20.0.405.gbc1bbc6f85-goog

From: Stefan Hajnoczi <stefanha@redhat.com>
Sent: Tuesday, December 11, 2018 8:27 AM
To: Jorgen Hansen <redacted>; Dexuan Cui
[off-list ref]
Cc: netdev@vger.kernel.org; Lepton Wu <redacted>
Subject: Re: [PATCH] VSOCK: bind to random port for VMADDR_PORT_ANY

On Mon, Dec 10, 2018 at 11:02:35PM -0800, Lepton Wu wrote:

quoted

The old code always starts from fixed port for VMADDR_PORT_ANY.

Sometimes

quoted

when VMM crashed, there is still orphaned vsock which is waiting for
close timer, then it could cause connection time out for new started VM
if they are trying to connect to same port with same guest cid since the
new packets could hit that orphaned vsock. We could also fix this by doing
more in vhost_vsock_reset_orphans, but any way, it should be better to start
from a random local port instead of a fixed one.

Signed-off-by: Lepton Wu <redacted>
---
 net/vmw_vsock/af_vsock.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Jorgen, Dexuan: Any objection to this?  It also affects the other
AF_VSOCK transports.

Thanks for letting me notice the patch. :-)

Now I saw Lepton's v2 patch, which looks good to me. 

Thanks,
-- Dexuan

Thanks for the update - looks good to me.

Reviewed-by: Jorgen Hansen <redacted>

/jsh

From: Lepton Wu <redacted>
Date: Tue, 11 Dec 2018 11:12:55 -0800

The old code always starts from fixed port for VMADDR_PORT_ANY. Sometimes
when VMM crashed, there is still orphaned vsock which is waiting for
close timer, then it could cause connection time out for new started VM
if they are trying to connect to same port with same guest cid since the
new packets could hit that orphaned vsock. We could also fix this by doing
more in vhost_vsock_reset_orphans, but any way, it should be better to start
from a random local port instead of a fixed one.

Signed-off-by: Lepton Wu <redacted>

Applied, thanks.