On Thu, Nov 27, 2025 at 11:15:24AM +0100, Paolo Abeni wrote:
On 11/24/25 5:33 AM, Hangbin Liu wrote:
When disabling a port’s collecting and distributing states, updating only rx_disabled is not sufficient. We also need to set AD_RX_PORT_DISABLED so that the rx_machine transitions into the AD_RX_EXPIRED state.
One example is in ad_agg_selection_logic(): when a new aggregator is selected and old active aggregator is disabled, if AD_RX_PORT_DISABLED is not set, the disabled port may remain stuck in AD_RX_CURRENT due to continuing to receive partner LACP messages.
The __disable_port() called by ad_disable_collecting_distributing() does not have this issue, since its caller also clears the collecting/distributing bits.
The __disable_port() called by bond_3ad_bind_slave() should also be fine, as the RX state machine is re-initialized to AD_RX_INITIALIZE.
Given the above, why don't you apply the change in ad_agg_selection_logic() only, to reduce the chances of unintended side effects?
/P
I think setting port->sm_rx_state = AD_RX_PORT_DISABLED and slave->rx_disabled = 1 should be an atomic operation. The later 2 functions just did similar stuff(not same the fixed one) on other places.
Thanks Hagnbin