From 80b2bd737d0e833e6a2b77e482e5a714a79c86a4 Mon Sep 17 00:00:00 2001 From: Leon Romanovsky Date: Tue, 24 May 2022 15:59:27 +0300 Subject: net/mlx5: Don't use already freed action pointer The call to mlx5dr_action_destroy() releases "action" memory. That pointer is set to miss_action later and generates the following smatch error: drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c:53 set_miss_action() warn: 'action' was already freed. Make sure that the pointer is always valid by setting NULL after destroy. Fixes: 6a48faeeca10 ("net/mlx5: Add direct rule fs_cmd implementation") Reported-by: Dan Carpenter Signed-off-by: Leon Romanovsky Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) (limited to 'drivers/net/ethernet') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c b/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c index 728f81882589..6a9abba92df6 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/steering/fs_dr.c @@ -44,11 +44,10 @@ static int set_miss_action(struct mlx5_flow_root_namespace *ns, err = mlx5dr_table_set_miss_action(ft->fs_dr_table.dr_table, action); if (err && action) { err = mlx5dr_action_destroy(action); - if (err) { - action = NULL; - mlx5_core_err(ns->dev, "Failed to destroy action (%d)\n", - err); - } + if (err) + mlx5_core_err(ns->dev, + "Failed to destroy action (%d)\n", err); + action = NULL; } ft->fs_dr_table.miss_action = action; if (old_miss_action) { -- cgit From 66cb64e292d21588bdb831f08a7ec0ff04d6380d Mon Sep 17 00:00:00 2001 From: Maor Dickman Date: Mon, 2 May 2022 10:51:30 +0300 Subject: net/mlx5e: TC NIC mode, fix tc chains miss table The cited commit changed promisc table to be created on demand with the highest priority in the NIC table replacing the vlan table, this caused tc NIC tables miss flow to skip the prmoisc table because it use vlan table as miss table. OVS offload in NIC mode use promisc by default so any unicast packet which will be handled by tc NIC tables miss flow will skip the promisc rule and will be dropped. Fix this by adding new empty table in new tc level with low priority and point the nic tc chain miss to it, the new table is managed so it will point to vlan table if promisc is disabled and to promisc table if enabled. Fixes: 1c46d7409f30 ("net/mlx5e: Optimize promiscuous mode") Signed-off-by: Maor Dickman Reviewed-by: Paul Blakey Reviewed-by: Ariel Levkovich Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en/fs.h | 2 ++ drivers/net/ethernet/mellanox/mlx5/core/en_tc.c | 38 +++++++++++++++++++++-- drivers/net/ethernet/mellanox/mlx5/core/fs_core.c | 2 +- 3 files changed, 39 insertions(+), 3 deletions(-) (limited to 'drivers/net/ethernet') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h index 4130a871de61..6e3a90a959e9 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/fs.h @@ -12,6 +12,7 @@ struct mlx5e_post_act; enum { MLX5E_TC_FT_LEVEL = 0, MLX5E_TC_TTC_FT_LEVEL, + MLX5E_TC_MISS_LEVEL, }; struct mlx5e_tc_table { @@ -20,6 +21,7 @@ struct mlx5e_tc_table { */ struct mutex t_lock; struct mlx5_flow_table *t; + struct mlx5_flow_table *miss_t; struct mlx5_fs_chains *chains; struct mlx5e_post_act *post_act; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c index 49dea02a12d2..34bf11cdf90f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tc.c @@ -4714,6 +4714,33 @@ static int mlx5e_tc_nic_get_ft_size(struct mlx5_core_dev *dev) return tc_tbl_size; } +static int mlx5e_tc_nic_create_miss_table(struct mlx5e_priv *priv) +{ + struct mlx5_flow_table **ft = &priv->fs.tc.miss_t; + struct mlx5_flow_table_attr ft_attr = {}; + struct mlx5_flow_namespace *ns; + int err = 0; + + ft_attr.max_fte = 1; + ft_attr.autogroup.max_num_groups = 1; + ft_attr.level = MLX5E_TC_MISS_LEVEL; + ft_attr.prio = 0; + ns = mlx5_get_flow_namespace(priv->mdev, MLX5_FLOW_NAMESPACE_KERNEL); + + *ft = mlx5_create_auto_grouped_flow_table(ns, &ft_attr); + if (IS_ERR(*ft)) { + err = PTR_ERR(*ft); + netdev_err(priv->netdev, "failed to create tc nic miss table err=%d\n", err); + } + + return err; +} + +static void mlx5e_tc_nic_destroy_miss_table(struct mlx5e_priv *priv) +{ + mlx5_destroy_flow_table(priv->fs.tc.miss_t); +} + int mlx5e_tc_nic_init(struct mlx5e_priv *priv) { struct mlx5e_tc_table *tc = &priv->fs.tc; @@ -4746,19 +4773,23 @@ int mlx5e_tc_nic_init(struct mlx5e_priv *priv) } tc->mapping = chains_mapping; + err = mlx5e_tc_nic_create_miss_table(priv); + if (err) + goto err_chains; + if (MLX5_CAP_FLOWTABLE_NIC_RX(priv->mdev, ignore_flow_level)) attr.flags = MLX5_CHAINS_AND_PRIOS_SUPPORTED | MLX5_CHAINS_IGNORE_FLOW_LEVEL_SUPPORTED; attr.ns = MLX5_FLOW_NAMESPACE_KERNEL; attr.max_ft_sz = mlx5e_tc_nic_get_ft_size(dev); attr.max_grp_num = MLX5E_TC_TABLE_NUM_GROUPS; - attr.default_ft = mlx5e_vlan_get_flowtable(priv->fs.vlan); + attr.default_ft = priv->fs.tc.miss_t; attr.mapping = chains_mapping; tc->chains = mlx5_chains_create(dev, &attr); if (IS_ERR(tc->chains)) { err = PTR_ERR(tc->chains); - goto err_chains; + goto err_miss; } tc->post_act = mlx5e_tc_post_act_init(priv, tc->chains, MLX5_FLOW_NAMESPACE_KERNEL); @@ -4781,6 +4812,8 @@ err_reg: mlx5_tc_ct_clean(tc->ct); mlx5e_tc_post_act_destroy(tc->post_act); mlx5_chains_destroy(tc->chains); +err_miss: + mlx5e_tc_nic_destroy_miss_table(priv); err_chains: mapping_destroy(chains_mapping); err_mapping: @@ -4821,6 +4854,7 @@ void mlx5e_tc_nic_cleanup(struct mlx5e_priv *priv) mlx5e_tc_post_act_destroy(tc->post_act); mapping_destroy(tc->mapping); mlx5_chains_destroy(tc->chains); + mlx5e_tc_nic_destroy_miss_table(priv); } int mlx5e_tc_ht_init(struct rhashtable *tc_ht) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c index 84caffe4c278..fdcf7f529330 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/fs_core.c @@ -114,7 +114,7 @@ #define KERNEL_MIN_LEVEL (KERNEL_NIC_PRIO_NUM_LEVELS + 1) #define KERNEL_NIC_TC_NUM_PRIOS 1 -#define KERNEL_NIC_TC_NUM_LEVELS 2 +#define KERNEL_NIC_TC_NUM_LEVELS 3 #define ANCHOR_NUM_LEVELS 1 #define ANCHOR_NUM_PRIOS 1 -- cgit From 1f2856cde64baa78475e6d3c601fb7b7f693a161 Mon Sep 17 00:00:00 2001 From: Paul Blakey Date: Mon, 23 May 2022 19:12:21 +0300 Subject: net/mlx5: CT: Fix header-rewrite re-use for tupels Tuple entries that don't have nat configured for them which are added to the ct nat table will always create a new modify header, as we don't check for possible re-use on them. The same for tuples that have nat configured for them but are added to ct table. Fix the above by only avoiding wasteful re-use lookup for actually natted entries in ct nat table. Fixes: 7fac5c2eced3 ("net/mlx5: CT: Avoid reusing modify header context for natted entries") Signed-off-by: Paul Blakey Reviewed-by: Ariel Levkovich Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c | 19 +++++++++++-------- 1 file changed, 11 insertions(+), 8 deletions(-) (limited to 'drivers/net/ethernet') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c index bceea7a1589e..25f51f80a9b4 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/tc_ct.c @@ -715,7 +715,7 @@ mlx5_tc_ct_entry_create_mod_hdr(struct mlx5_tc_ct_priv *ct_priv, struct mlx5_flow_attr *attr, struct flow_rule *flow_rule, struct mlx5e_mod_hdr_handle **mh, - u8 zone_restore_id, bool nat) + u8 zone_restore_id, bool nat_table, bool has_nat) { DECLARE_MOD_HDR_ACTS_ACTIONS(actions_arr, MLX5_CT_MIN_MOD_ACTS); DECLARE_MOD_HDR_ACTS(mod_acts, actions_arr); @@ -731,11 +731,12 @@ mlx5_tc_ct_entry_create_mod_hdr(struct mlx5_tc_ct_priv *ct_priv, &attr->ct_attr.ct_labels_id); if (err) return -EOPNOTSUPP; - if (nat) { - err = mlx5_tc_ct_entry_create_nat(ct_priv, flow_rule, - &mod_acts); - if (err) - goto err_mapping; + if (nat_table) { + if (has_nat) { + err = mlx5_tc_ct_entry_create_nat(ct_priv, flow_rule, &mod_acts); + if (err) + goto err_mapping; + } ct_state |= MLX5_CT_STATE_NAT_BIT; } @@ -750,7 +751,7 @@ mlx5_tc_ct_entry_create_mod_hdr(struct mlx5_tc_ct_priv *ct_priv, if (err) goto err_mapping; - if (nat) { + if (nat_table && has_nat) { attr->modify_hdr = mlx5_modify_header_alloc(ct_priv->dev, ct_priv->ns_type, mod_acts.num_actions, mod_acts.actions); @@ -818,7 +819,9 @@ mlx5_tc_ct_entry_add_rule(struct mlx5_tc_ct_priv *ct_priv, err = mlx5_tc_ct_entry_create_mod_hdr(ct_priv, attr, flow_rule, &zone_rule->mh, - zone_restore_id, nat); + zone_restore_id, + nat, + mlx5_tc_ct_entry_has_nat(entry)); if (err) { ct_dbg("Failed to create ct entry mod hdr"); goto err_mod_hdr; -- cgit From 2e642afb61b24401a7ec819d27ddcd69c7c29784 Mon Sep 17 00:00:00 2001 From: Maxim Mikityanskiy Date: Fri, 15 Apr 2022 16:19:15 +0300 Subject: net/mlx5e: Disable softirq in mlx5e_activate_rq to avoid race condition When the driver activates the channels, it assumes NAPI isn't running yet. mlx5e_activate_rq posts a NOP WQE to ICOSQ to trigger a hardware interrupt and start NAPI, which will run mlx5e_alloc_rx_mpwqe and post UMR WQEs to ICOSQ to be able to receive packets with striding RQ. Unfortunately, a race condition is possible if NAPI is triggered by something else (for example, TX) at a bad timing, before mlx5e_activate_rq finishes. In this case, mlx5e_alloc_rx_mpwqe may post UMR WQEs to ICOSQ, and with the bad timing, the wqe_info of the first UMR may be overwritten by the wqe_info of the NOP posted by mlx5e_activate_rq. The consequence is that icosq->db.wqe_info[0].num_wqebbs will be changed from MLX5E_UMR_WQEBBS to 1, disrupting the integrity of the array-based linked list in wqe_info[]. mlx5e_poll_ico_cq will hang in an infinite loop after processing wqe_info[0], because after the corruption, the next item to be processed will be wqe_info[1], which is filled with zeros, and `sqcc += wi->num_wqebbs` will never move further. This commit fixes this race condition by using async_icosq to post the NOP and trigger the interrupt. async_icosq is always protected with a spinlock, eliminating the race condition. Fixes: bc77b240b3c5 ("net/mlx5e: Add fragmented memory support for RX multi packet WQE") Signed-off-by: Maxim Mikityanskiy Reported-by: Karsten Nielsen Reviewed-by: Tariq Toukan Reviewed-by: Gal Pressman Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en.h | 4 ++++ drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c | 1 + .../ethernet/mellanox/mlx5/core/en/reporter_rx.c | 6 ++++++ drivers/net/ethernet/mellanox/mlx5/core/en/trap.c | 1 + .../net/ethernet/mellanox/mlx5/core/en/xsk/pool.c | 1 + .../net/ethernet/mellanox/mlx5/core/en/xsk/setup.c | 5 +---- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 24 +++++++++++++++------- 7 files changed, 31 insertions(+), 11 deletions(-) (limited to 'drivers/net/ethernet') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h index 65d3c4865abf..b6c15efe92ad 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h @@ -764,6 +764,7 @@ struct mlx5e_rq { u8 wq_type; u32 rqn; struct mlx5_core_dev *mdev; + struct mlx5e_channel *channel; u32 umr_mkey; struct mlx5e_dma_info wqe_overflow; @@ -1076,6 +1077,9 @@ void mlx5e_close_cq(struct mlx5e_cq *cq); int mlx5e_open_locked(struct net_device *netdev); int mlx5e_close_locked(struct net_device *netdev); +void mlx5e_trigger_napi_icosq(struct mlx5e_channel *c); +void mlx5e_trigger_napi_sched(struct napi_struct *napi); + int mlx5e_open_channels(struct mlx5e_priv *priv, struct mlx5e_channels *chs); void mlx5e_close_channels(struct mlx5e_channels *chs); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c index 335b20b6383b..047f88f09203 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/ptp.c @@ -736,6 +736,7 @@ void mlx5e_ptp_activate_channel(struct mlx5e_ptp *c) if (test_bit(MLX5E_PTP_STATE_RX, c->state)) { mlx5e_ptp_rx_set_fs(c->priv); mlx5e_activate_rq(&c->rq); + mlx5e_trigger_napi_sched(&c->napi); } } diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c index 2684e9da9f41..fc366e66d0b0 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/reporter_rx.c @@ -123,6 +123,8 @@ static int mlx5e_rx_reporter_err_icosq_cqe_recover(void *ctx) xskrq->stats->recover++; } + mlx5e_trigger_napi_icosq(icosq->channel); + mutex_unlock(&icosq->channel->icosq_recovery_lock); return 0; @@ -166,6 +168,10 @@ static int mlx5e_rx_reporter_err_rq_cqe_recover(void *ctx) clear_bit(MLX5E_RQ_STATE_RECOVERING, &rq->state); mlx5e_activate_rq(rq); rq->stats->recover++; + if (rq->channel) + mlx5e_trigger_napi_icosq(rq->channel); + else + mlx5e_trigger_napi_sched(rq->cq.napi); return 0; out: clear_bit(MLX5E_RQ_STATE_RECOVERING, &rq->state); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c index 857840ab1e91..11f2a7fb72a9 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/trap.c @@ -179,6 +179,7 @@ static void mlx5e_activate_trap(struct mlx5e_trap *trap) { napi_enable(&trap->napi); mlx5e_activate_rq(&trap->rq); + mlx5e_trigger_napi_sched(&trap->napi); } void mlx5e_deactivate_trap(struct mlx5e_priv *priv) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c index 279cd8f4e79f..2c520394aa1d 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/pool.c @@ -117,6 +117,7 @@ static int mlx5e_xsk_enable_locked(struct mlx5e_priv *priv, goto err_remove_pool; mlx5e_activate_xsk(c); + mlx5e_trigger_napi_icosq(c); /* Don't wait for WQEs, because the newer xdpsock sample doesn't provide * any Fill Ring entries at the setup stage. diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c index 3ad7f1301fa8..98ed9ef3a6bd 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en/xsk/setup.c @@ -64,6 +64,7 @@ static int mlx5e_init_xsk_rq(struct mlx5e_channel *c, rq->clock = &mdev->clock; rq->icosq = &c->icosq; rq->ix = c->ix; + rq->channel = c; rq->mdev = mdev; rq->hw_mtu = MLX5E_SW2HW_MTU(params, params->sw_mtu); rq->xdpsq = &c->rq_xdpsq; @@ -179,10 +180,6 @@ void mlx5e_activate_xsk(struct mlx5e_channel *c) mlx5e_reporter_icosq_resume_recovery(c); /* TX queue is created active. */ - - spin_lock_bh(&c->async_icosq_lock); - mlx5e_trigger_irq(&c->async_icosq); - spin_unlock_bh(&c->async_icosq_lock); } void mlx5e_deactivate_xsk(struct mlx5e_channel *c) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index 05c015515cce..930a5402c817 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -475,6 +475,7 @@ static int mlx5e_init_rxq_rq(struct mlx5e_channel *c, struct mlx5e_params *param rq->clock = &mdev->clock; rq->icosq = &c->icosq; rq->ix = c->ix; + rq->channel = c; rq->mdev = mdev; rq->hw_mtu = MLX5E_SW2HW_MTU(params, params->sw_mtu); rq->xdpsq = &c->rq_xdpsq; @@ -1066,13 +1067,6 @@ err_free_rq: void mlx5e_activate_rq(struct mlx5e_rq *rq) { set_bit(MLX5E_RQ_STATE_ENABLED, &rq->state); - if (rq->icosq) { - mlx5e_trigger_irq(rq->icosq); - } else { - local_bh_disable(); - napi_schedule(rq->cq.napi); - local_bh_enable(); - } } void mlx5e_deactivate_rq(struct mlx5e_rq *rq) @@ -2227,6 +2221,20 @@ static int mlx5e_channel_stats_alloc(struct mlx5e_priv *priv, int ix, int cpu) return 0; } +void mlx5e_trigger_napi_icosq(struct mlx5e_channel *c) +{ + spin_lock_bh(&c->async_icosq_lock); + mlx5e_trigger_irq(&c->async_icosq); + spin_unlock_bh(&c->async_icosq_lock); +} + +void mlx5e_trigger_napi_sched(struct napi_struct *napi) +{ + local_bh_disable(); + napi_schedule(napi); + local_bh_enable(); +} + static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix, struct mlx5e_params *params, struct mlx5e_channel_param *cparam, @@ -2308,6 +2316,8 @@ static void mlx5e_activate_channel(struct mlx5e_channel *c) if (test_bit(MLX5E_CHANNEL_STATE_XSK, c->state)) mlx5e_activate_xsk(c); + + mlx5e_trigger_napi_icosq(c); } static void mlx5e_deactivate_channel(struct mlx5e_channel *c) -- cgit From f6279f113ad593971999c877eb69dc3d36a75894 Mon Sep 17 00:00:00 2001 From: Maxim Mikityanskiy Date: Mon, 23 May 2022 15:39:13 +0300 Subject: net/mlx5e: Update netdev features after changing XDP state Some features (LRO, HW GRO) conflict with XDP. If there is an attempt to enable such features while XDP is active, they will be set to `off [requested on]`. In order to activate these features after XDP is turned off, the driver needs to call netdev_update_features(). This commit adds this missing call after XDP state changes. Fixes: cf6e34c8c22f ("net/mlx5e: Properly block LRO when XDP is enabled") Fixes: b0617e7b3500 ("net/mlx5e: Properly block HW GRO when XDP is enabled") Signed-off-by: Maxim Mikityanskiy Reviewed-by: Tariq Toukan Signed-off-by: Saeed Mahameed --- drivers/net/ethernet/mellanox/mlx5/core/en_main.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'drivers/net/ethernet') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index 930a5402c817..087952b84ccb 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -4569,6 +4569,11 @@ static int mlx5e_xdp_set(struct net_device *netdev, struct bpf_prog *prog) unlock: mutex_unlock(&priv->state_lock); + + /* Need to fix some features. */ + if (!err) + netdev_update_features(netdev); + return err; } -- cgit From 1c5de097bea31760c3f0467ac0c84ba0dc3525d5 Mon Sep 17 00:00:00 2001 From: Saeed Mahameed Date: Fri, 19 Feb 2021 23:10:47 -0800 Subject: net/mlx5: Fix mlx5_get_next_dev() peer device matching In some use-cases, mlx5 instances will need to search for their peer device (the other port on the same HCA). For that, mlx5 device matching mechanism relied on auxiliary_find_device() to search, and used a bad matching callback function. This approach has two issues: 1) next_phys_dev() the matching function, assumed all devices are of the type mlx5_adev (mlx5 auxiliary device) which is wrong and could lead to crashes, this worked for a while, since only lately other drivers started registering auxiliary devices. 2) using the auxiliary class bus (auxiliary_find_device) to search for mlx5_core_dev devices, who are actually PCIe device instances, is wrong. This works since mlx5_core always has at least one mlx5_adev instance hanging around in the aux bus. As suggested by others we can fix 1. by comparing device names prefixes if they have the string "mlx5_core" in them, which is not a best practice ! but even with that fixed, still 2. needs fixing, we are trying to match pcie device peers so we should look in the right bus (pci bus), hence this fix. The fix: 1) search the pci bus for mlx5 peer devices, instead of the aux bus 2) to validated devices are the same type "mlx5_core_dev" compare if they have the same driver, which is bulletproof. This wouldn't have worked with the aux bus since the various mlx5 aux device types don't share the same driver, even if they share the same device wrapper struct (mlx5_adev) "which helped to find the parent device" Fixes: a925b5e309c9 ("net/mlx5: Register mlx5 devices to auxiliary virtual bus") Reported-by: Alexander Lobakin Reported-by: Maher Sanalla Signed-off-by: Saeed Mahameed Reviewed-by: Leon Romanovsky Reviewed-by: Mark Bloch Reviewed-by: Maher Sanalla --- drivers/net/ethernet/mellanox/mlx5/core/dev.c | 34 ++++++++++++++++++--------- 1 file changed, 23 insertions(+), 11 deletions(-) (limited to 'drivers/net/ethernet') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/dev.c index 11f7c03ae81b..0eb9d74547f8 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/dev.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c @@ -571,18 +571,32 @@ static int _next_phys_dev(struct mlx5_core_dev *mdev, return 1; } +static void *pci_get_other_drvdata(struct device *this, struct device *other) +{ + if (this->driver != other->driver) + return NULL; + + return pci_get_drvdata(to_pci_dev(other)); +} + static int next_phys_dev(struct device *dev, const void *data) { - struct mlx5_adev *madev = container_of(dev, struct mlx5_adev, adev.dev); - struct mlx5_core_dev *mdev = madev->mdev; + struct mlx5_core_dev *mdev, *this = (struct mlx5_core_dev *)data; + + mdev = pci_get_other_drvdata(this->device, dev); + if (!mdev) + return 0; return _next_phys_dev(mdev, data); } static int next_phys_dev_lag(struct device *dev, const void *data) { - struct mlx5_adev *madev = container_of(dev, struct mlx5_adev, adev.dev); - struct mlx5_core_dev *mdev = madev->mdev; + struct mlx5_core_dev *mdev, *this = (struct mlx5_core_dev *)data; + + mdev = pci_get_other_drvdata(this->device, dev); + if (!mdev) + return 0; if (!MLX5_CAP_GEN(mdev, vport_group_manager) || !MLX5_CAP_GEN(mdev, lag_master) || @@ -596,19 +610,17 @@ static int next_phys_dev_lag(struct device *dev, const void *data) static struct mlx5_core_dev *mlx5_get_next_dev(struct mlx5_core_dev *dev, int (*match)(struct device *dev, const void *data)) { - struct auxiliary_device *adev; - struct mlx5_adev *madev; + struct device *next; if (!mlx5_core_is_pf(dev)) return NULL; - adev = auxiliary_find_device(NULL, dev, match); - if (!adev) + next = bus_find_device(&pci_bus_type, NULL, dev, match); + if (!next) return NULL; - madev = container_of(adev, struct mlx5_adev, adev); - put_device(&adev->dev); - return madev->mdev; + put_device(next); + return pci_get_drvdata(to_pci_dev(next)); } /* Must be called with intf_mutex held */ -- cgit