Re: [PATCH] igc: enable HW VLAN insertion/stripping by default

From: Tony Nguyen
Date: Thu Mar 13 2025 - 11:49:23 EST




On 3/13/2025 2:23 AM, Rui Salvaterra wrote:
Hi, Tony,

On Wed, 12 Mar 2025 at 21:43, Tony Nguyen <anthony.l.nguyen@xxxxxxxxx> wrote:

Unfortunately, I'm unable check with the original author or those
involved at the time. From asking around it sounds like there may have
been some initial issues when implementing this; my theory is this was
off by default so that it would minimize the affects if additional
issues were discovered.

I thought about that possibility, but in the end it didn't seem
logical to me. If the feature was WIP and/or broken in some way, why
would the user be allowed to enable it via ethtool, ever since it was
first implemented, in commit 8d7449630e3450bc0546dc0cb692fbb57d1852c0
(almost four years ago)?

It would allow users to opt-in and use the feature; if an adverse problem was later found, it could be easily mitigated by not turning the feature on. If it were on by default, and a problem found, there would a larger impact. As you mentioned, since the feature has now been in existence for many years, it's probably safe to have on by default.

I see that you missed Intel Wired LAN (intel-wired-lan@xxxxxxxxxxxxxxxx)
on the patch. Could you resend with the list included? Also, if you
could make it 'PATCH iwl-next' to target the Intel -next tree as that
seems the appropriate tree.

Oh. That list is marked as moderated, I (wrongly, for sure) assumed
there would be restrictions when mailing to it. I'll resend correctly
soon.

Thank you.

- Tony