On 09/28, Jakub Kicinski wrote: > test_xdp_bonding_attach() already covers a flat bond both ways round: > the master is refused while a slave has a program, and a slave is > refused while the master has one. The nested test only ever checked > that attaching to the outermost master succeeds. > > That left the more interesting half of the nesting untested. Only the > direct upper is consulted when deciding whether a device is already > running XDP, so with bond -> bond_nest1 -> bond_nest2 the check on > bond_nest2 looks at bond_nest1, and it only sees a program there if > bond_nest1 recorded the one it was handed. Before the preceding fix it > did not, and a second program could be attached to bond_nest2 while the > outer bond's program was already running on it. > > Signed-off-by: Jakub Kicinski <[email protected]>
Acked-by: Stanislav Fomichev <[email protected]>
