> On Tue, 17 Mar 2026 19:34:33 +0800 Di Zhu wrote: > > In such a setup, reporting NETIF_F_GRO_HW as available for the device > > is too optimistic and misleading to the user. > > This is not a sufficient justification to make this a fix. > As Michael was already alluding to HW GRO usually covers fewer > protocols and is less effective than SW GRO. Normal config is > to have both enabled. Please explain if there's anything special > about virtio here, because I haven't heard of people disabling > SW GRO because HW GRO is already enabled.
"Thanks, Jakub. You're right, that was a poorly chosen example to justify this as a fix, and I see your point regarding HW GRO vs SW GRO. There isn't anything special about virtio here that would necessitate disabling SW GRO in normal cases; my initial thought was more about the potential for user confusion in specific testing scenarios. > > I think this should continue to target net-next with no Fixes tag. I'll drop the Fixes tag and target net-next in V2. I'll also move the hw_features initialization before register_netdevice() as discussed, to ensure the driver follows the proper pattern."
