Re: [PATCHv2 net-next 2/3] bonding: restructure ad_churn_machine
From: Paolo Abeni <pabeni@redhat.com>
Date: 2026-01-21 07:58:19
Also in:
linux-kselftest
On 1/21/26 12:11 AM, Jakub Kicinski wrote:
On Tue, 20 Jan 2026 09:29:14 +0100 Paolo Abeni wrote:quoted
On 1/19/26 9:22 PM, Jakub Kicinski wrote:quoted
On Wed, 14 Jan 2026 06:49:20 +0000 Hangbin Liu wrote:quoted
The current ad_churn_machine implementation only transitions the actor/partner churn state to churned or none after the churn timer expires. However, IEEE 802.1AX-2014 specifies that a port should enter the none state immediately once the actor’s port state enters synchronization.Paolo, how do you feel about his patch with 2+ weeks until final? The first patch is definitely suitable for net. If this one is not it should not have a Fixes tag. I'd lean towards getting them all into -rc7 if we can.My personal preference would be for 2/3 landing into net-next: the code looks correct to me, but refactor has IMHO still to much potential for regressions do land directly into net and the blamed commit is quite old. I suggested targeting net-next while retaining the Fixes tag as we already had complex fixes landing into net-next in the past.The appropriate way to delay propagation of the fix to add: Cc: <redacted> # after 4 weeks not to merge things into -next.
I went over the code as carefully as I could and I don't see any obvious problem, so I don't have a so strong opinion vs net, but to hopefully clarify: my thinking is that this is net-next material because it's an invasive refactor with behavior change. My preference would be let the code be tested in next/net-next for a while before landing into mainline. /P