Re: [PATCH v2 net 4/4] net: enetc: let core segment GSO skb when headers exceed TSO_HEADER_SIZE

From: Vladimir Oltean

Date: Thu Nov 19 2026 - 04:43:29 EST


On Thu, Nov 19, 2026 at 10:38:14AM +0100, Eric Dumazet wrote:
> Le jeu. 19 nov. 2026 à 09:44, Vladimir Oltean
> <vladimir.oltean@xxxxxxx> a écrit :
> > Do you disagree with my analysis that only enetc risks corruption? I mean,
> > for a second layer of defense, having tso_features_check() everywhere
> > TSO_HEADER_SIZE is used probably makes sense, but I opted for a targeted
> > fix because I couldn't justify it for the other drivers.
>
> I agree with your analysis, I could not find another driver where
> headers can exceed TSO_HEADER_SIZE today. Your targeted fix is fine
> for net.
>
> The other drivers are only safe as a side effect of the
> NETIF_F_IPV6_CSUM rule in gso_features_check(), and none of them
> bounds the header length in an ndo_features_check(). I might respin
> the tso_features_check() series for net-next as a second layer
> (calling vlan_features_check() this time), and enetc can be
> converted on top.
>
> Reviewed-by: Eric Dumazet <edumazet@xxxxxxxxxx>

Sounds good to me, and I mean, I do agree that TSO_HEADER_SIZE not being
enforced as a maximum offloadable size in the core is definitely a trap
that should be addressed. But at the same time, I do have tens of other
'net' patches queued up waiting for this to go in in a minimal form, so
I also appreciate not waiting for tso_features_check() :)

Thanks for the review!