netdev - [PATCHv2 net-next 0/3] bonding: fix 802.3ad churn machine and port state issues

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20260114064921.57686-1-liuhangbin@gmail.com>
Date: Wed, 14 Jan 2026 06:49:18 +0000
From: Hangbin Liu <liuhangbin@...il.com>
To: netdev@...r.kernel.org
Cc: Jay Vosburgh <jv@...sburgh.net>,
	Andrew Lunn <andrew+netdev@...n.ch>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>,
	Paolo Abeni <pabeni@...hat.com>,
	Simon Horman <horms@...nel.org>,
	Mahesh Bandewar <maheshb@...gle.com>,
	Shuah Khan <shuah@...nel.org>,
	linux-kselftest@...r.kernel.org,
	Hangbin Liu <liuhangbin@...il.com>
Subject: [PATCHv2 net-next 0/3] bonding: fix 802.3ad churn machine and port state issues

This series fixes two issues in the bonding 802.3ad implementation
related to port state management and churn detection:

1. When disabling a port, we need to set AD_RX_PORT_DISABLED to ensure
   proper state machine transitions, preventing ports from getting stuck
   in AD_RX_CURRENT state.

2. The ad_churn_machine implementation is restructured to follow IEEE
   802.1AX-2014 specifications correctly. The current implementation has
   several issues: it doesn't transition to "none" state immediately when
   synchronization is achieved, and can get stuck in churned state in
   multi-aggregator scenarios.

3. Selftests are enhanced to validate both mux state machine and churn
   state logic under aggregator selection and failover scenarios.

These changes ensure proper LACP state machine behavior and fix issues
where ports could remain in incorrect states during aggregator failover.

v2:
  * The changes are large and not urgent. Post to net-next as Paolo suggested
  * set AD_RX_PORT_DISABLED only in ad_agg_selection_logic to avoid side effect. (Paolo Abeni)
  * remove actor_churn as it can only be true when the state is ACTOR_CHURN (Paolo Abeni)
  * remove AD_PORT_CHURNED since we don't need it anywhere (Paolo Abeni)
  * I didn't add new helper for ad_churn_machine() as it looks not help much.

v1: https://lore.kernel.org/netdev/20251124043310.34073-1-liuhangbin@gmail.com

Hangbin Liu (3):
  bonding: set AD_RX_PORT_DISABLED when disabling a port
  bonding: restructure ad_churn_machine
  selftests: bonding: add mux and churn state testing

 drivers/net/bonding/bond_3ad.c                | 97 ++++++++++++++-----
 .../selftests/drivers/net/bonding/Makefile    |  2 +-
 ...nd_lacp_prio.sh => bond_lacp_ad_select.sh} | 73 ++++++++++++++
 3 files changed, 146 insertions(+), 26 deletions(-)
 rename tools/testing/selftests/drivers/net/bonding/{bond_lacp_prio.sh => bond_lacp_ad_select.sh} (64%)

-- 
2.50.1