Commit 52911bb6 authored by David S. Miller's avatar David S. Miller
Browse files

Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue 



Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2021-11-22

Maciej Fijalkowski says:

Here are the two fixes for issues around ethtool's set_channels()
callback for ice driver. Both are related to XDP resources. First one
corrects the size of vsi->txq_map that is used to track the usage of Tx
resources and the second one prevents the wrong refcounting of bpf_prog.
====================

Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parents 60ebd673 f65ee535

drivers/net/ethernet/intel/ice/ice_lib.c

+7 −2
Original line number Diff line number Diff line
@@ -89,8 +89,13 @@ static int ice_vsi_alloc_arrays(struct ice_vsi *vsi) 
	if (!vsi->rx_rings)

		goto err_rings;



	/* XDP will have vsi->alloc_txq Tx queues as well, so double the size */

	vsi->txq_map = devm_kcalloc(dev, (2 * vsi->alloc_txq),

	/* txq_map needs to have enough space to track both Tx (stack) rings

	 * and XDP rings; at this point vsi->num_xdp_txq might not be set,

	 * so use num_possible_cpus() as we want to always provide XDP ring

	 * per CPU, regardless of queue count settings from user that might

	 * have come from ethtool's set_channels() callback;

	 */

	vsi->txq_map = devm_kcalloc(dev, (vsi->alloc_txq + num_possible_cpus()),

				    sizeof(*vsi->txq_map), GFP_KERNEL);



	if (!vsi->txq_map)

drivers/net/ethernet/intel/ice/ice_main.c

+17 −1
Original line number Diff line number Diff line
@@ -2609,6 +2609,17 @@ int ice_prepare_xdp_rings(struct ice_vsi *vsi, struct bpf_prog *prog) 
			ice_stat_str(status));

		goto clear_xdp_rings;

	}



	/* assign the prog only when it's not already present on VSI;

	 * this flow is a subject of both ethtool -L and ndo_bpf flows;

	 * VSI rebuild that happens under ethtool -L can expose us to

	 * the bpf_prog refcount issues as we would be swapping same

	 * bpf_prog pointers from vsi->xdp_prog and calling bpf_prog_put

	 * on it as it would be treated as an 'old_prog'; for ndo_bpf

	 * this is not harmful as dev_xdp_install bumps the refcount

	 * before calling the op exposed by the driver;

	 */

	if (!ice_is_xdp_ena_vsi(vsi))

		ice_vsi_assign_bpf_prog(vsi, prog);



	return 0;
@@ -2785,6 +2796,11 @@ ice_xdp_setup_prog(struct ice_vsi *vsi, struct bpf_prog *prog, 
		if (xdp_ring_err)

			NL_SET_ERR_MSG_MOD(extack, "Freeing XDP Tx resources failed");

	} else {

		/* safe to call even when prog == vsi->xdp_prog as

		 * dev_xdp_install in net/core/dev.c incremented prog's

		 * refcount so corresponding bpf_prog_put won't cause

		 * underflow

		 */

		ice_vsi_assign_bpf_prog(vsi, prog);

	}