Maciej Fijalkowski [off-list ref] writes:
xskxceiver -i ethX -i ethX, used for HW tests, bound both sockets to
queue 0 of one interface, which the kernel only allows with a shared
UMEM. With the endpoints about to become separate processes, they would
have to share the UMEM mapping and the socket fd.
The mode also relied on the interface being set up by hand with a
single queue and promiscuous mode, which does not suit CI. The hardware
test added later in this series steers its traffic with ntuple and RSS
configuration instead.
Remove the shared-UMEM TX setup, the split UMEM addressing and the
POLL_TXQ_FULL workaround it needed, and drop the -i IFACE option of
test_xsk.sh that used it. XDP_SHARED_UMEM (two sockets of one endpoint)
is unaffected.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Reviewed-by: Björn Töpel <bjorn@kernel.org>