Re: [PATCH net v2] bonding: fix missing down path on speed/duplex failure after FAIL->UP

[Timur Stukan]

On Fri, 30 Oct 2026 at 12:50, <netdev-bot+sashiko@xxxxxxxxxx> wrote:

> Does slave->link == BOND_LINK_FAIL always mean the slave is still in use?

No, but both routes described here start with the 802.3ad notifier
setting FAIL on a slave that is DOWN or BACK, because it checks
last_link_up rather than the current link state. I'll change the
notifier to set FAIL only for a slave that is currently UP, so FAIL
again means the slave was in use, and send it as patch 1 of v3.

> Can an in-use slave reach this fallback with slave->link == BOND_LINK_UP?

Yes, this is also pre-existing, but it is the same problem this patch
is fixing: a slave that may be in use being set to DOWN silently. I
will add that to v3 also, and in v3 the full down path will be taken
for a slave that is FAIL or UP if speed/duplex check fails, and only
DOWN/BACK will be reverted silently. Checking curr_active_slave alone
would miss non-active slaves in alb mode. In a normal miimon flow UP
is never proposed for a slave that is already UP, so this only affects
the racy cases.

pw-bot: cr