Jakub Kicinski wrote: > On Sat, 28 Feb 2026 14:46:28 -0500 Willem de Bruijn wrote: > > Jakub Kicinski wrote: > > > On Thu, 26 Feb 2026 10:46:53 -0800 Kiran Kella wrote: > > > > This is achieved by propagating the psp_dev from the lower device > > > > to the upper devices in the device stack via a netdevice notifier. > > > > The lowest device owns the psp_dev pointer while the upper devices > > > > just borrow the pointer. When the lower device is unlinked, the > > > > borrowed pointer is cleared in the upper device. > > > > Assumption being that psp_dev is set on the lowest device before > > > > any upper devices are stacked on that lowest device. > > > > > > As I mentioned in the other thread I'd like to establish some clear > > > expectation on where psd propagates automatically and where it doesn't. > > > I don't want to see a stream of patches that say "fix propagating PSP > > > onto X upper". And conversely report saying "PSP got propagated but > > > it doesn't actually work" (macvlan in bridge mode etc). > > > > > > Hence my preference was to require the user who created the device > > > to propagate PSP. Willem, WDYT? > > > > Can you give a bit more detail what you have in mind? > > > > An explicit user (netlink) API to associate a psp_dev with a > > net_device? > > Yes: > https://lore.kernel.org/all/[email protected]/
I see. Yes, fair to ask for an explicit sensible policy. It would be nice if PSP could transparently be enabled on upper devices, similar to dev->features. But this quickly runs into practical questions like when and how can a bonding device with multiple independent lower devices support PSP? dev->features also has type specific carve-outs for upper types, such as vlan_features and hw_enc_features. We probably don't want to introduce the same for PSP.
