While poking at struct netdev_rx_queue I got annoyed by the huge rebuild times. I split it out from netdevice.h and then realized that it was the main reason we included xdp.h in there. So I removed that dependency as well. This gives us very pleasant build times for both xdp.h and struct netdev_rx_queue changes. I'm sending this for bpf-next because I think it'd be easiest if it goes in there, and then bpf-next gets flushed soon after? I can also make a branch on merge-base for net-next and bpf-next.. Jakub Kicinski (3): eth: add missing xdp.h includes in drivers net: move struct netdev_rx_queue out of netdevice.h net: invert the netdevice.h vs xdp.h dependency drivers/net/bonding/bond_main.c | 1 + drivers/net/ethernet/amazon/ena/ena_netdev.h | 1 + drivers/net/ethernet/engleder/tsnep.h | 1 + .../net/ethernet/freescale/dpaa2/dpaa2-eth.h | 1 + drivers/net/ethernet/freescale/enetc/enetc.h | 1 + drivers/net/ethernet/freescale/fec.h | 1 + .../ethernet/fungible/funeth/funeth_txrx.h | 1 + drivers/net/ethernet/google/gve/gve.h | 1 + drivers/net/ethernet/intel/igc/igc.h | 1 + .../ethernet/microchip/lan966x/lan966x_main.h | 1 + drivers/net/ethernet/microsoft/mana/mana_en.c | 1 + drivers/net/ethernet/stmicro/stmmac/stmmac.h | 1 + drivers/net/ethernet/ti/cpsw_priv.h | 1 + drivers/net/hyperv/hyperv_net.h | 1 + drivers/net/tap.c | 1 + drivers/net/virtio_net.c | 1 + include/linux/filter.h | 17 ------ include/linux/netdevice.h | 55 ++----------------- include/net/busy_poll.h | 1 + include/net/mana/mana.h | 2 + include/net/netdev_rx_queue.h | 53 ++++++++++++++++++ include/net/xdp.h | 27 ++++++++- include/trace/events/xdp.h | 1 + kernel/bpf/btf.c | 1 + kernel/bpf/offload.c | 1 + kernel/bpf/verifier.c | 1 + net/bpf/test_run.c | 1 + net/core/dev.c | 1 + net/core/net-sysfs.c | 1 + net/xdp/xsk.c | 1 + 30 files changed, 108 insertions(+), 71 deletions(-) create mode 100644 include/net/netdev_rx_queue.h -- 2.41.0
