On 5/20/17 8:57 AM, Hangbin Liu wrote:
On Fri, May 19, 2017 at 07:30:43PM +0200, Ivan Vecera wrote:
Current bridge code incorrectly handles starting/stopping of hello and
hold timers during STP enable/disable.
1. Timers are stopped in br_stp_start() during NO_STP->USER_STP
transition. The timers are already stopped in NO_STP state so
this is confusing no-op.
Hi Ivan,
Shouldn't we start hello timer in br_stp_start when NO_STP -> BR_KERNEL_STP ?
Please see Xin Long's recent -net patch that fixes exactly this issue.
It will answer your questions below, too.
https://patchwork.ozlabs.org/patch/764685/
2. During USER_STP->NO_STP transition the timers are started. This
does not make sense and is confusion because the timer should not be
active in NO_STP state.
Yes, but what about BR_KERNEL_STP -> NO_STP in function br_stp_stop() ?
Cc: davem@xxxxxxxxxxxxx
Cc: sashok@xxxxxxxxxxxxxxxxxxx
Cc: stephen@xxxxxxxxxxxxxxxxxx
Cc: bridge@xxxxxxxxxxxxxxxxxxxxxxxxxx
Cc: lucien.xin@xxxxxxxxx
Cc: nikolay@xxxxxxxxxxxxxxxxxxx
Signed-off-by: Ivan Vecera <cera@xxxxxxx>
---
net/bridge/br_stp_if.c | 11 -----------
1 file changed, 11 deletions(-)
diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
index 08341d2aa9c9..a05027027513 100644
--- a/net/bridge/br_stp_if.c
+++ b/net/bridge/br_stp_if.c
@@ -150,7 +150,6 @@ static int br_stp_call_user(struct net_bridge *br, char *arg)
static void br_stp_start(struct net_bridge *br)
{
- struct net_bridge_port *p;
int err = -ENOENT;
if (net_eq(dev_net(br->dev), &init_net))
@@ -169,11 +168,6 @@ static void br_stp_start(struct net_bridge *br)
if (!err) {
br->stp_enabled = BR_USER_STP;
br_debug(br, "userspace STP started\n");
-
- /* Stop hello and hold timers */
- del_timer(&br->hello_timer);
- list_for_each_entry(p, &br->port_list, list)
- del_timer(&p->hold_timer);
I'm not sure if user space daemon will send bpdu or not? In comment
76b91c32dd86 ("bridge: stp: when using userspace stp stop kernel hello and
hold timers"). Nikolay said we should not handle it with BR_USER_STP >
} else {
br->stp_enabled = BR_KERNEL_STP;
br_debug(br, "using kernel STP\n");
@@ -187,7 +181,6 @@ static void br_stp_start(struct net_bridge *br)
static void br_stp_stop(struct net_bridge *br)
{
- struct net_bridge_port *p;
int err;
if (br->stp_enabled == BR_USER_STP) {
@@ -196,10 +189,6 @@ static void br_stp_stop(struct net_bridge *br)
br_err(br, "failed to stop userspace STP (%d)\n", err);
/* To start timers on any ports left in blocking */
- mod_timer(&br->hello_timer, jiffies + br->hello_time);
- list_for_each_entry(p, &br->port_list, list)
- mod_timer(&p->hold_timer,
- round_jiffies(jiffies + BR_HOLD_TIME));
If we do not del hello_timer. after it expired in br_hello_timer_expired(),
Our state is br->dev->flags & IFF_UP and br->stp_enabled == NO_STP, it will
call mod_timer(&br->hello_timer, round_jiffies(jiffies + br->hello_time))
and we will keep sending bpdu message even after stp stoped.
Again see Xin Long's recent -net patch.
spin_lock_bh(&br->lock);
br_port_state_selection(br);
spin_unlock_bh(&br->lock);
--
So how about just like
diff --git a/net/bridge/br_stp_if.c b/net/bridge/br_stp_if.c
index d8ad73b..0198f62 100644
--- a/net/bridge/br_stp_if.c
+++ b/net/bridge/br_stp_if.c
@@ -183,6 +183,7 @@ static void br_stp_start(struct net_bridge *br)
} else {
br->stp_enabled = BR_KERNEL_STP;
br_debug(br, "using kernel STP\n");
+ mod_timer(&br->hello_timer, jiffies + br->hello_time);
/* To start timers on any ports left in blocking */
br_port_state_selection(br);
@@ -202,7 +203,6 @@ static void br_stp_stop(struct net_bridge *br)
br_err(br, "failed to stop userspace STP (%d)\n", err);
/* To start timers on any ports left in blocking */
- mod_timer(&br->hello_timer, jiffies + br->hello_time);
list_for_each_entry(p, &br->port_list, list)
mod_timer(&p->hold_timer,
round_jiffies(jiffies + BR_HOLD_TIME));
@@ -211,6 +211,7 @@ static void br_stp_stop(struct net_bridge *br)
spin_unlock_bh(&br->lock);
}
+ del_timer_sync(&br->hello_timer);
br->stp_enabled = BR_NO_STP;
}
Thanks
Hangbin
