Re: git: 349fcf079ca3 - main - net: add ifnet_rename_event EVENTHANDLER(9) for interface renaming

From: Gleb Smirnoff <glebius_at_freebsd.org>
Date: Wed, 04 Feb 2026 22:40:07 UTC
On Wed, Feb 04, 2026 at 09:21:37AM +0800, Zhenlei Huang wrote:
Z> >> commit 349fcf079ca32d5c93e45366d2b27638747affeb
Z> >> Author:     Gleb Smirnoff <glebius@FreeBSD.org>
Z> >> AuthorDate: 2025-12-21 21:31:43 +0000
Z> >> Commit:     Gleb Smirnoff <glebius@FreeBSD.org>
Z> >> CommitDate: 2025-12-22 02:23:14 +0000
Z> >> 
Z> >>    net: add ifnet_rename_event EVENTHANDLER(9) for interface renaming
Z> >> 
Z> >>    and don't trigger ifnet_arrival_event and ifnet_departure_event for a
Z> >>    rename, as the interface isn't being detached from any protocol.  The
Z> >>    consumers of the arrival/departure events are divided into a few
Z> >>    categories:
Z> >>    - which indeed need to do the same actions as if interface was fully
Z> >>      detached and attached: routing socket and netlink notifications to
Z> >>      userland and the Linux sysfs.  All addressed by this commit.
Z> >>    - which build their logic based on an interface name, but should actually
Z> >>      update their database on rename: packet filters.  This commit leaves
Z> >>      them with the old behavior - emulate full detach & attach, but this
Z> >>      should be improved.
Z> >>    - which shouldn't do anything on rename, not touched by the commit.
Z> >>    - ng_ether and if_tuntap, that are special and will be addressed by
Z> >>      separate commits.
Z> > 
Z> > Can we get rid of the IFF_RENAMING flag now?  IIUC all of its uses are
Z> > in ifnet_departure_event handlers to short-circuit the handler because
Z> > the interface is not actually going away.
Z> 
Z> Now that we have ifnet_rename_event, I think the flag IFF_RENAMING is
Z> not longer required.
Z> 
Z> Well there're still some consumers, they should be cleaned up prior to retire
Z> IFF_RENAMING.

100% correct.  Fell free to do that, or I can do that.

-- 
Gleb Smirnoff