From 663f146f2ecfcc47934cb9f9543f664eeb6adb46 Mon Sep 17 00:00:00 2001 From: Vu Pham Date: Wed, 31 Oct 2018 16:03:21 +0200 Subject: [PATCH 1/9] net/mlx5: E-Switch, Fix fdb cap bits swap The cap bits locations for the fdb caps of multi path to table (used for local mirroring) and multi encap (used for prio/chains) were wrongly used in swapped locations. This went unnoted so far b/c we tested the offending patch with CX5 FW that supports both of them. On different environments where not both caps are supported, we will be messed up, fix that. Fixes: b9aa0ba17af5 ('net/mlx5: Add cap bits for multi fdb encap') Signed-off-by: Vu Pham Reviewed-by: Or Gerlitz Tested-by: Or Gerlitz Signed-off-by: Saeed Mahameed --- include/linux/mlx5/mlx5_ifc.h | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 34e17e6f8942..4e77bfe0b580 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -582,11 +582,13 @@ struct mlx5_ifc_flow_table_nic_cap_bits { }; struct mlx5_ifc_flow_table_eswitch_cap_bits { - u8 reserved_at_0[0x1c]; - u8 fdb_multi_path_to_table[0x1]; - u8 reserved_at_1d[0x1]; + u8 reserved_at_0[0x1a]; u8 multi_fdb_encap[0x1]; - u8 reserved_at_1e[0x1e1]; + u8 reserved_at_1b[0x1]; + u8 fdb_multi_path_to_table[0x1]; + u8 reserved_at_1d[0x3]; + + u8 reserved_at_20[0x1e0]; struct mlx5_ifc_flow_table_prop_layout_bits flow_table_properties_nic_esw_fdb; From 8956f0014ea5ed824aae3816fd9e52cf3c85b9d4 Mon Sep 17 00:00:00 2001 From: Gavi Teitz Date: Mon, 19 Nov 2018 17:22:30 +0200 Subject: [PATCH 2/9] net/mlx5e: Fix default amount of channels for VF representors The default amount of channels a representor opens was erroneously changed from one to the maximum amount of channels, restore to its intended value. Fixes: 779d986d60de ("net/mlx5e: Do not ignore netdevice TX/RX queues number") Signed-off-by: Gavi Teitz Reviewed-by: Or Gerlitz Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c index c3c657548824..6a4d782c1de1 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c @@ -46,6 +46,7 @@ #define MLX5E_REP_PARAMS_LOG_SQ_SIZE \ max(0x6, MLX5E_PARAMS_MINIMUM_LOG_SQ_SIZE) +#define MLX5E_REP_PARAMS_DEF_NUM_CHANNELS 1 static const char mlx5e_rep_driver_name[] = "mlx5e_rep"; @@ -1083,9 +1084,7 @@ static int mlx5e_init_rep(struct mlx5_core_dev *mdev, if (err) return err; - - priv->channels.params.num_channels = - mlx5e_get_netdev_max_channels(netdev); + priv->channels.params.num_channels = MLX5E_REP_PARAMS_DEF_NUM_CHANNELS; mlx5e_build_rep_params(mdev, &priv->channels.params, netdev->mtu); mlx5e_build_rep_netdev(netdev); From e1c15b62b7015119d3e5915cd2ae3b89d59c2576 Mon Sep 17 00:00:00 2001 From: Moshe Shemesh Date: Fri, 2 Nov 2018 06:10:49 +0200 Subject: [PATCH 3/9] net/mlx5e: RX, Verify MPWQE stride size is in range Add check of MPWQE stride size is within range supported by HW. In case calculated MPWQE stride size exceed range, linear SKB can't be used and we should use non linear MPWQE instead. Fixes: 619a8f2a42f1 ("net/mlx5e: Use linear SKB in Striding RQ") Signed-off-by: Moshe Shemesh Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index 871313d6b34d..c17de6a990ff 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -128,6 +128,8 @@ static bool mlx5e_rx_is_linear_skb(struct mlx5_core_dev *mdev, return !params->lro_en && frag_sz <= PAGE_SIZE; } +#define MLX5_MAX_MPWQE_LOG_WQE_STRIDE_SZ ((BIT(__mlx5_bit_sz(wq, log_wqe_stride_size)) - 1) + \ + MLX5_MPWQE_LOG_STRIDE_SZ_BASE) static bool mlx5e_rx_mpwqe_is_linear_skb(struct mlx5_core_dev *mdev, struct mlx5e_params *params) { @@ -138,6 +140,9 @@ static bool mlx5e_rx_mpwqe_is_linear_skb(struct mlx5_core_dev *mdev, if (!mlx5e_rx_is_linear_skb(mdev, params)) return false; + if (order_base_2(frag_sz) > MLX5_MAX_MPWQE_LOG_WQE_STRIDE_SZ) + return false; + if (MLX5_CAP_GEN(mdev, ext_stride_num_range)) return true; From e88afe759a49f8d51d6cc5fc89d65b4a5768703f Mon Sep 17 00:00:00 2001 From: Or Gerlitz Date: Sun, 9 Dec 2018 17:03:36 +0200 Subject: [PATCH 4/9] net/mlx5e: Err if asked to mirror a goto chain tc eswitch rule Currently we are not supporting this and not err-ing on that either. For now, just err if asked to do that. Fixes: bf07aa730a04 ('net/mlx5e: Support offloading tc priorities and chains for eswitch flows') Signed-off-by: Or Gerlitz Reported-by: Eli Britstein Reviewed-by: Eli Britstein Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c index fca6f4132c91..1f01c905a042 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c @@ -2966,8 +2966,7 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv, struct tcf_exts *exts, NL_SET_ERR_MSG(extack, "Requested destination chain is out of supported range"); return -EOPNOTSUPP; } - action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST | - MLX5_FLOW_CONTEXT_ACTION_COUNT; + action |= MLX5_FLOW_CONTEXT_ACTION_COUNT; attr->dest_chain = dest_chain; continue; @@ -2980,6 +2979,14 @@ static int parse_tc_fdb_actions(struct mlx5e_priv *priv, struct tcf_exts *exts, if (!actions_match_supported(priv, exts, parse_attr, flow, extack)) return -EOPNOTSUPP; + if (attr->dest_chain) { + if (attr->action & MLX5_FLOW_CONTEXT_ACTION_FWD_DEST) { + NL_SET_ERR_MSG(extack, "Mirroring goto chain rules isn't supported"); + return -EOPNOTSUPP; + } + attr->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + } + if (attr->mirror_count > 0 && !mlx5_esw_has_fwd_fdb(priv->mdev)) { NL_SET_ERR_MSG_MOD(extack, "current firmware doesn't support split rule for port mirroring"); From d14f6f2a84326dd2751962167b2293fc91751ff8 Mon Sep 17 00:00:00 2001 From: Or Gerlitz Date: Sun, 9 Dec 2018 17:15:23 +0200 Subject: [PATCH 5/9] net/mlx5e: Avoid overriding the user provided priority for offloaded tc rules Just a leftover which was wrongly left there, remove it while spawning a message to suggest firmware upgrade. Fixes: bf07aa730a04 ('net/mlx5e: Support offloading tc priorities and chains for eswitch flows') Signed-off-by: Or Gerlitz Reviewed-by: Paul Blakey Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c index 1f01c905a042..e6170ddf2cc4 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c @@ -907,11 +907,10 @@ mlx5e_tc_add_fdb_flow(struct mlx5e_priv *priv, struct mlx5e_priv *out_priv; int err = 0, encap_err = 0; - /* if prios are not supported, keep the old behaviour of using same prio - * for all offloaded rules. - */ - if (!mlx5_eswitch_prios_supported(esw)) - attr->prio = 1; + if (!mlx5_eswitch_prios_supported(esw) && attr->prio != 1) { + NL_SET_ERR_MSG(extack, "E-switch priorities unsupported, upgrade FW"); + return -EOPNOTSUPP; + } if (attr->chain > max_chain) { NL_SET_ERR_MSG(extack, "Requested chain is out of supported range"); From 154e62abe9cd07c9b71735d8456a03a7c5251280 Mon Sep 17 00:00:00 2001 From: Or Gerlitz Date: Mon, 10 Dec 2018 12:31:42 +0200 Subject: [PATCH 6/9] net/mlx5e: Properly initialize flow attributes for slow path eswitch rule deletion When a neighbour is resolved, we delete the goto slow path rule from HW. The eswitch flow attributes where not properly initialized on that case, hence we mess up the eswitch refcounts for chain zero (the default one). Fix that along with making sure to use semicolons and not commas on that code; Fixes: 5dbe906ff1d5 ('net/mlx5e: Use a slow path rule instead if vxlan neighbour isn't available') Signed-off-by: Or Gerlitz Reviewed-by: Paul Blakey Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c index e6170ddf2cc4..3b732c02b5e1 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c @@ -870,9 +870,9 @@ mlx5e_tc_offload_to_slow_path(struct mlx5_eswitch *esw, struct mlx5_flow_handle *rule; memcpy(slow_attr, flow->esw_attr, sizeof(*slow_attr)); - slow_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST, - slow_attr->mirror_count = 0, - slow_attr->dest_chain = FDB_SLOW_PATH_CHAIN, + slow_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + slow_attr->mirror_count = 0; + slow_attr->dest_chain = FDB_SLOW_PATH_CHAIN; rule = mlx5e_tc_offload_fdb_rules(esw, flow, spec, slow_attr); if (!IS_ERR(rule)) @@ -887,6 +887,9 @@ mlx5e_tc_unoffload_from_slow_path(struct mlx5_eswitch *esw, struct mlx5_esw_flow_attr *slow_attr) { memcpy(slow_attr, flow->esw_attr, sizeof(*slow_attr)); + slow_attr->action = MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; + slow_attr->mirror_count = 0; + slow_attr->dest_chain = FDB_SLOW_PATH_CHAIN; mlx5e_tc_unoffload_fdb_rules(esw, flow, slow_attr); flow->flags &= ~MLX5E_TC_FLOW_SLOW; } From 61c806dafe81c58f41fec10fb5cad0d3a4390e3e Mon Sep 17 00:00:00 2001 From: Or Gerlitz Date: Mon, 10 Dec 2018 17:05:59 +0200 Subject: [PATCH 7/9] net/mlx5e: Avoid encap flows deletion attempt the 1st time a neigh is resolved Currently, we are deleting offloaded encap flows in case the relevant neigh becomes unconnected while the encap is valid (a sign that it used to be connected), or if the curr neigh mac is different from the cached mac (a sign that the remote side changed their mac). The 2nd check also applies when the neigh becomes connected on the 1st time (we start with zero mac). Before the offending commit, the deleting handler was practically no op, as no flows were offloaded. But since that commit, we offload neigh-less encap flows to slow path. Under mirroring scheme, we go into the delete handler, attempt to unoffload a mirror rule which was never set (as we were offloading to slow path) and crash. Fix that by calling the delete handler only when the encap is valid, which covers both cases mentioned above. Fixes: 5dbe906ff1d5 ('net/mlx5e: Use a slow path rule instead if vxlan neighbour isn't available') Signed-off-by: Or Gerlitz Signed-off-by: Eli Britstein Reviewed-by: Paul Blakey Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_rep.c | 4 ++-- drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 7 +++---- 2 files changed, 5 insertions(+), 6 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c index 6a4d782c1de1..820fe85100b0 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rep.c @@ -467,8 +467,8 @@ static void mlx5e_rep_update_flows(struct mlx5e_priv *priv, ASSERT_RTNL(); - if ((!neigh_connected && (e->flags & MLX5_ENCAP_ENTRY_VALID)) || - !ether_addr_equal(e->h_dest, ha)) + if ((e->flags & MLX5_ENCAP_ENTRY_VALID) && + (!neigh_connected || !ether_addr_equal(e->h_dest, ha))) mlx5e_tc_encap_flows_del(priv, e); if (neigh_connected && !(e->flags & MLX5_ENCAP_ENTRY_VALID)) { diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c index 3b732c02b5e1..9dabe9d4b279 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c @@ -1096,10 +1096,9 @@ void mlx5e_tc_encap_flows_del(struct mlx5e_priv *priv, flow->rule[0] = rule; } - if (e->flags & MLX5_ENCAP_ENTRY_VALID) { - e->flags &= ~MLX5_ENCAP_ENTRY_VALID; - mlx5_packet_reformat_dealloc(priv->mdev, e->encap_id); - } + /* we know that the encap is valid */ + e->flags &= ~MLX5_ENCAP_ENTRY_VALID; + mlx5_packet_reformat_dealloc(priv->mdev, e->encap_id); } static struct mlx5_fc *mlx5e_tc_get_counter(struct mlx5e_tc_flow *flow) From d13b224f431579fe2d712871d4265d7a22ca6c9c Mon Sep 17 00:00:00 2001 From: Mikhael Goikhman Date: Mon, 19 Nov 2018 19:11:12 +0200 Subject: [PATCH 8/9] net/mlx5e: Remove unused UDP GSO remaining counter Remove tx_udp_seg_rem counter from ethtool output, as it is no longer being updated in the driver's data flow. Fixes: 3f44899ef2ce ("net/mlx5e: Use PARTIAL_GSO for UDP segmentation") Signed-off-by: Mikhael Goikhman Reviewed-by: Eran Ben Elisha Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_stats.c | 2 -- drivers/net/ethernet/mellanox/mlx5/core/en_stats.h | 2 -- 2 files changed, 4 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c index 3e99d0728b2f..4337afd610d7 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c @@ -74,7 +74,6 @@ static const struct counter_desc sw_stats_desc[] = { { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_recover) }, { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_cqes) }, { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_queue_wake) }, - { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_udp_seg_rem) }, { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_cqe_err) }, { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_xmit) }, { MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_xdp_full) }, @@ -198,7 +197,6 @@ void mlx5e_grp_sw_update_stats(struct mlx5e_priv *priv) s->tx_nop += sq_stats->nop; s->tx_queue_stopped += sq_stats->stopped; s->tx_queue_wake += sq_stats->wake; - s->tx_udp_seg_rem += sq_stats->udp_seg_rem; s->tx_queue_dropped += sq_stats->dropped; s->tx_cqe_err += sq_stats->cqe_err; s->tx_recover += sq_stats->recover; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h index 3f8e870ef4c9..3ff69ddae2d3 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h @@ -87,7 +87,6 @@ struct mlx5e_sw_stats { u64 tx_recover; u64 tx_cqes; u64 tx_queue_wake; - u64 tx_udp_seg_rem; u64 tx_cqe_err; u64 tx_xdp_xmit; u64 tx_xdp_full; @@ -221,7 +220,6 @@ struct mlx5e_sq_stats { u64 csum_partial_inner; u64 added_vlan_packets; u64 nop; - u64 udp_seg_rem; #ifdef CONFIG_MLX5_EN_TLS u64 tls_ooo; u64 tls_resync_bytes; From fa2bf86bab4bbc61e5678a42a14e40075093a98f Mon Sep 17 00:00:00 2001 From: Tal Gilboa Date: Thu, 22 Nov 2018 14:20:45 +0200 Subject: [PATCH 9/9] net/mlx5e: Cancel DIM work on close SQ TXQ SQ closure is followed by closing the corresponding CQ. A pending DIM work would try to modify the now non-existing CQ. This would trigger an error: [85535.835926] mlx5_core 0000:af:00.0: mlx5_cmd_check:769:(pid 124399): MODIFY_CQ(0x403) op_mod(0x0) failed, status bad resource state(0x9), syndrome (0x1d7771) Fix by making sure to cancel any pending DIM work before destroying the SQ. Fixes: cbce4f444798 ("net/mlx5e: Enable adaptive-TX moderation") Signed-off-by: Tal Gilboa Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index c17de6a990ff..b70cb6fd164c 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -1401,6 +1401,7 @@ static void mlx5e_close_txqsq(struct mlx5e_txqsq *sq) struct mlx5_core_dev *mdev = c->mdev; struct mlx5_rate_limit rl = {0}; + cancel_work_sync(&sq->dim.work); mlx5e_destroy_sq(mdev, sq->sqn); if (sq->rate_limit) { rl.rate = sq->rate_limit;