Thread (7 messages) 7 messages, 4 authors, 2023-07-13

Re: [PATCH v1 net] bridge: Return an error when enabling STP in netns.

From: Ido Schimmel <hidden>
Date: 2023-07-12 14:48:28
Also in: bridge
Subsystem: ethernet bridge, networking [general], the rest · Maintainers: Nikolay Aleksandrov, Ido Schimmel, "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Linus Torvalds

On Tue, Jul 11, 2023 at 04:54:15PM -0700, Kuniyuki Iwashima wrote:
When we create an L2 loop on a bridge in netns, we will see packets storm
even if STP is enabled.

  # unshare -n
  # ip link add br0 type bridge
  # ip link add veth0 type veth peer name veth1
  # ip link set veth0 master br0 up
  # ip link set veth1 master br0 up
  # ip link set br0 type bridge stp_state 1
  # ip link set br0 up
  # sleep 30
  # ip -s link show br0
  2: br0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
      link/ether b6:61:98:1c:1c:b5 brd ff:ff:ff:ff:ff:ff
      RX: bytes  packets  errors  dropped missed  mcast
      956553768  12861249 0       0       0       12861249  <-. Keep
      TX: bytes  packets  errors  dropped carrier collsns     |  increasing
      1027834    11951    0       0       0       0         <-'   rapidly

This is because llc_rcv() drops all packets in non-root netns and BPDU
is dropped.

Let's show an error when enabling STP in netns.

  # unshare -n
  # ip link add br0 type bridge
  # ip link set br0 type bridge stp_state 1
  Error: bridge: STP can't be enabled in non-root netns.

Note this commit will be reverted later when we namespacify the whole LLC
infra.

Fixes: e730c15519d0 ("[NET]: Make packet reception network namespace safe")
Suggested-by: Harry Coin <redacted>
I'm not sure that's accurate. I read his response in the link below and
he says "I'd rather be warned than blocked" and "But better warned and
awaiting a fix than blocked", which I agree with. The patch has the
potential to cause a lot of regressions, but without actually fixing the
problem.

How about simply removing the error [1]? Since iproute2 commit
844c37b42373 ("libnetlink: Handle extack messages for non-error case"),
it can print extack warnings and not only errors. With the diff below:

 # unshare -n 
 # ip link add name br0 type bridge
 # ip link set dev br0 type bridge stp_state 1
 Warning: bridge: STP can't be enabled in non-root netns.
 # echo $?
 0

[1]
diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
index a807996ac56b..b5143de37938 100644
--- a/net/bridge/br_stp_if.c
+++ b/net/bridge/br_stp_if.c
@@ -201,10 +201,8 @@ int br_stp_set_enabled(struct net_bridge *br, unsigned long val,
 {
        ASSERT_RTNL();
 
-       if (!net_eq(dev_net(br->dev), &init_net)) {
+       if (!net_eq(dev_net(br->dev), &init_net))
                NL_SET_ERR_MSG_MOD(extack, "STP can't be enabled in non-root netns");
-               return -EINVAL;
-       }
 
        if (br_mrp_enabled(br)) {
                NL_SET_ERR_MSG_MOD(extack,
quoted hunk ↗ jump to hunk
Link: https://lore.kernel.org/netdev/0f531295-e289-022d-5add-5ceffa0df9bc@quietfountain.com/ (local)
Signed-off-by: Kuniyuki Iwashima <redacted>
---
 net/bridge/br_stp_if.c | 5 +++++
 1 file changed, 5 insertions(+)
diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
index 75204d36d7f9..a807996ac56b 100644
--- a/net/bridge/br_stp_if.c
+++ b/net/bridge/br_stp_if.c
@@ -201,6 +201,11 @@ int br_stp_set_enabled(struct net_bridge *br, unsigned long val,
 {
 	ASSERT_RTNL();
 
+	if (!net_eq(dev_net(br->dev), &init_net)) {
+		NL_SET_ERR_MSG_MOD(extack, "STP can't be enabled in non-root netns");
+		return -EINVAL;
+	}
+
 	if (br_mrp_enabled(br)) {
 		NL_SET_ERR_MSG_MOD(extack,
 				   "STP can't be enabled if MRP is already enabled");
-- 
2.30.2
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help