On Mon, May 27, 2024 at 02:06:26PM -0400, Tyler Stachecki wrote:
I'm attaching a patch that applies cleanly against 1.188 and was tested
against a patched bookworm initrd with success.
After a fair bit of local testing, I've applied this patch to netcfg
master, with several additional changes of my own, and uploaded them as
part of netcfg 1.204. For transparency I've kept my own changes as
separate git commits, and you can see them all here along with my
rebased version of Tyler's patch:
https://salsa.debian.org/installer-team/netcfg/-/compare/5eeeda8570c62ef8acefac5ab5b22af76a4c1416...1a3ab35c22f360fcc887863dce4fc19053142386?from_project_id=23458
Note in particular that I modernized the approach used for ifupdown, and
that I added Netplan support (since that had been added to netcfg since
the last iteration of the VLAN patch in this bug, so it seemed best to
enable it for parity, although the only way I could find to test it in
practice was to manually run "apt-install netplan.io"). The other
changes were just minor nits and glitches.
Once this reaches daily builds
(https://www.debian.org/devel/debian-installer/), I'd very much
appreciate testing by others in live environments that use VLANs to
confirm that I didn't break anything. I don't use them routinely
myself; I set up a test environment so that I could connect a test
laptop to a trunk port, but there's always the possibility that I made
mistakes in both the patch and my test environment that cancelled each
other out.
I also made a minor change to hw-detect to avoid confusion if you go
back to the "Detect network hardware" step:
https://salsa.debian.org/installer-team/hw-detect/-/commit/619ba547da555a9a87290c8ce9315c3db5376cf0
One piece that remains is documentation. I'll work on this once this
has percolated through to daily builds, and ideally once it's had a bit
of testing by others.
Thanks,
--
Colin Watson (he/him) [[email protected]]
