ixgbe, xsk: Terminate Rx side of NAPI when XSK Rx queue gets full
authorMaciej Fijalkowski <maciej.fijalkowski@intel.com>
Wed, 13 Apr 2022 15:30:08 +0000 (17:30 +0200)
committerDaniel Borkmann <daniel@iogearbox.net>
Fri, 15 Apr 2022 19:10:45 +0000 (21:10 +0200)
commitc7dd09fd46283029a41615b2b7034aea26b22ee0
tree30e103787898e807de9af1efb928c686bf2d2787
parentb8aef650e54982728660919a0cf9cdacb079ef86
ixgbe, xsk: Terminate Rx side of NAPI when XSK Rx queue gets full

When XSK pool uses need_wakeup feature, correlate -ENOBUFS that was
returned from xdp_do_redirect() with a XSK Rx queue being full. In such
case, terminate the Rx processing that is being done on the current HW
Rx ring and let the user space consume descriptors from XSK Rx queue so
that there is room that driver can use later on.

Introduce new internal return code IXGBE_XDP_EXIT that will indicate case
described above.

Note that it does not affect Tx processing that is bound to the same
NAPI context, nor the other Rx rings.

Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20220413153015.453864-8-maciej.fijalkowski@intel.com
drivers/net/ethernet/intel/ixgbe/ixgbe_txrx_common.h
drivers/net/ethernet/intel/ixgbe/ixgbe_xsk.c